Study of erosion,

river-bed deformation
and sediment transport
in river basins
as related to natural
and man-made changes
INTERNATIONAL HYDROLOGICAL PROGRAMME

Study of erosion, river bed

deformation and sediment transport
in river bakins as related to natural
and man-made changes

IHP-IV Project H-l -2

IHP-V 1Technical Documents in Hydrology 1 No. 10

UNESCO, Paris, 1997

SC-97lWSl42
The designations employed and the presentation of material throughout
the publication do not imply the expression of any opinion
whatsoever on the part of UNESCO concerning the legal status
of any country, territory, city or of its authorities, or
concerning the delimitation of its frontiers or boundaries.
CONTENTS

INTRODUCTION V

1. EROSION IN RIVER BASINS

Erosion and sedimentyield in a changingenvironment

D. E. Walling 1

WatershedManagementin China: Conceptsand Techniques

Ding Lianzhen 54

The application of geographicalinformation systemsto soil conservationstrategies

W. Summer and E. Klaghofer 77

Erosion and sedimentyield on plains in the temperatezone

A. P. Dedcov and V.I. Mozzherin 84

Improved methodologyfor the computationof normal annual

yield of suspendedsedimentsfrom rivers
N.N. Bobrovitskaya and KM. zubkova 92

Erosion of cohesivematerials
T. E. Mirtskhoulava 104
2. SEDIMENT TRANSPORT

Sedimentnon-uniformity effects on entrainment and transport

K. G. Ranga Raju and M. K. Mittal 116

Sedimenttransport patternsobservedin SouthernAfrican rivers

A. Rooseboom 135

Theoretical premisesfor determining of bed scour and accretionareas

A. N. Butakov 143

Sedimentassociatedtransport of Chernobyl radionuclidesin the Pripyat river, Ukraine

0. Voitsekhovitch, V. Kanivets and V. Vishnevsky 155

Fluvial sedimenttransport in the arid regions of Central Asia

H.A. Ismagilov 165

3. RIVER CHANNEL DYNAMICS

River responseto natural and man-madechange

S. Bruk 174

The relation betweenriver channeldimensionsand dischargeof water

K. V. Grishanin 195

On the fractal sinuous&yof rivers

V. N. Nikora, D. M. Hicks and G.M. Smart 205

Channelprocessesand their role in river ecosystems

R.S. Chalov and A.M. Alabayan 216

Impact of gravel excavationon channelprocessesin the Laba river, North Caucasus

A.B. Shvidchenko and Z. D. Kopaliani 227

Changesof channelmorphology of the Rioni river Central Caucasus

0. G. Khmaladze and O.D. Shautidze 239

Airborne data for river bed deformationsstudy

D. V. Snishchenko 245
4. RIVER CHANNEL DESIGN CONCEPTS AND APPLICATIONS

Theory and practice of river channelprocesses

B.F. Snishchenko and Z.D. Kopaliani 251

Methodology for the inventory of channelprocessesfor water projects

B. F. Snishchenko 268

Changesin sedimenttransport and river engineering concepts,casestudy of the

river Drau in Austria
H.M. Habersack and H. P. Nachtnebel 277

Improvement of the navigablewidth in river bendsby periodic dredging,

casestudy of the river Waal, The Netherlands
M. Tml, H. Barneveld and T. Swanenberg 287

Program for computationof channeldeformations downstreamof dams:

casestudy of the Votkinsk Hydropower Plant on the Kama river, Russia
A. B. Veksler, V M. Donenberg, Y.L. Manuilov and R. S. Fried 298

Design of bank revetmentbasedon reliability concept

M. Bozinovic 313

5. MODELING STRATEGIES AND APPROACHES

River modeling
G. Di Silvio 322

On methodsfor the prediction of river planform changes

E. Mosselman 345

Prediction of long-term evolution of lowland river channels

V.A. Bazilevich, V. V. Kozitsky and J. A. Gaiduchenko 354

Hydromorphological aspectof channelprocessmodeling

A. B. Klaven 361

Flume investigationsinto the influence of river training structureson

sedimenttransport
S. Wieprecht, W. Bechteler 368

A perspectiveof mobile bed river models

G. Glazik 380

...
111
INTRODUCTION

Erosion and sedimentation were represented in the Fourth Phase (1990-1995) of the
International Hydrological Programmeby IHP-IV Project H-l-2, entitled “Study of erosion,
river bed deformation and sediment transport in river basins as related to natural and man
made changes “.

For the executionof the Project, the IHP IntergovernmentalCouncil set up a Working
Group. Based on the proposals of IHP National Committeesand NGOs, the following seven
Working Group members were nominated (in alphabetical order): S. Bruk, G. Di Silvio,
Ding Lianzhen, KG. Ranga Raju, A. Rooseboom, B.F. Snishchenkoand D.E. Walling.
Coordination was provided by S. Bruk.

The Working Group met on 7 April 1992 during the International Symposium on
River Sedimentationin Karlsruhe, in cooperation with the Fluvial Hydraulics Committee of
the IAHR. The main principles of the project were then discussedand it was concluded that
the main objective of the Project was to synthesizeexisting information on various aspectsof
erosion and sedimentation in different regions of the world, emphasizing those points in
which differences in opinions and methods are particularly evident. The synthesis was
expectedto be achievedthrough exchangeof views and discussionsat international symposia.
In order to incite discussionsand responsesof the scientific community, it was requestedthat
the Working Group membersprovide thought-provokingpersonalperspectiveson the subject,
rather than text-book type contributions or state-of-the-artreports.

The first opportunity for an exchange of views was offered by the international
symposium, convened in the framework of IHP-IV Project H-l-2, entitled “East-West,
North-South Encounter on the State-of-the-art in River Engineering Methods and Design
Philosophies”. The Symposium took place on 16-20 May 1994 in the State Hydrological
Institute, St.Petersburg, Russia, organized by the Russian IHP National Committee and
UNESCO, and sponsored by IAHR, IAHS and UNEP. The Working Group members
presentedtheir contributions as speciallecturesat the Symposium.

The Proceedings of the St.Petersburg Symposium were published by the State

Hydrological Institute, and comprise75 papers, in three chapters:

A. Physical processesand their elementsin watershedsand rivers - 24 papers

B. Modelling of alluvial processesand sedimenttransportation- 26 papers
C. River responseto hydraulic structuresand casestudiesrelated to projects
affecting sedimenttransport - 25 papers.

V
 The Working Group met the secondtime during the Symposium. It decided then to
include into the Technical Report a representativeselection of the papers presentedat the
Symposium, along with the contributions of the Working Group members, bearing in mind
that the Symposium Proceedingsare practically accessibleonly to the participants, whereas
the Technical Report will easily reach all interestedresearchers,through the regular channels
of UNESCO.

The present publication is result of the above endeavours. The authors alone are
responsiblefor any statementsand opinions, containedin their contributions. The writers trust
that their report will invite responsesfrom researchersfrom various fields of activity in
different regions of the World, in accordancewith the objectivesof the project.

Acknowledgement

Most authors took personal care to get their contributions into camera-readyform. Some of
the papers, nevertheless, needed language editing and had to be re-typed. This task was
accomplishedthrough the efforts and competenceof the Department of Fluvial Processesof
the State Hydrological Institute, St.Petersburg, under the direction of Professor B.F.
Snishchenko.

Dr. Habib Zebidi, Programme Specialist, Division of Water Sciences, UNESCO,

organized the activities of the Working Group, and took part in all its meetings. His
contribution to the project was essentialand highly valuable.

Dr. StevanBruk

vi
1. EROSION IN RIVER BASINS
 EROSION AND SEDlHENTYIELD IN A CHANGING E3TVlRONMENT

D.E. Walling

Department of Geography, University of Exeter, Exeter, UK.

In a book published in 1976, Eckholm contended that " excess sediment is the

major form of human-induced water pollution in the world today and exacts a

heavier cost . . . . . . possibly more than all other pollutants combined."

Similar sentiments have been used to emphasise the importance of problems of

loss of reservoir storage due to sedimentation (cf. Stevens, 1936; Dendy,

1968; Mahmood, 1987), the off-farm impact of eroded sediment (cf. Clark -et

al., 1985>, the role of sediment in the transport of contaminants (cf. Allan,

1986; Novotny and Chesters, 1982) and various other environmental and

operational problems associated with enhanced suspended sediment transport by

water courses (cf. Table 1). These problems have a very significant economic

dimension, since, for example, Clark et al. (1985) estimated that the annual

economic cost of off-farm sediment problems in the United States was of the

order of US $6.1 billion at 1980 prices and similar calculations undertaken

for South Africa by Braune and Looser (1989) have estimated the cost of the

off-site damage caused by soil erosion to be of the order of US $36 million.

Both estimates exclude the less tangible environmental damage, which is

extremely difficult to quantify, and are therefore likely to significantly

underestimate the true cost. Mahmood (1987) has also estimated that, as a

result of sedimentation, the major reservoirs of the world are currently

losing storage at the rate of 1 % of gross capacity, or 50 km3, per year.

Viewed in terms of replacement costs, this loss is equivalent to an annual

cost of $6 billion.

1
 The problems associated with increased sediment loads in rivers consequent

upon increased erosion within their drainage basins are now widely recognised

and this recognition has been paralleled by growing evidence of greatly

increased rates of soil loss and sediment yield in many areas of the world as

a result of human activity and particularly land use change. Table 2, for

example, contrasts rates of soil erosion documented under natural undisturbed

conditions with those occurring in cultivated areas. In all cases there is an

order of magnitude increase, and in several instances the increases are even

greater. The global implications of the data presented in Table 2 may be

highlighted by recognising that over the past 200 years the area of the

earth's surface given over to crop production and livestock grazing has

increased by more than five-fold (Buringh and Dudal, 1987) and that the

recent ISRIC / UNEP global survey of human-induced soil degradation (Oldeman

et al., 1991) has shown that nearly 10% of the total land surface of the

globe is currently adversely affected by water erosion. Table 3 provides

further information on the degree of soil degradation associated with these

areas and Table 4 indicates the relative importance of the major causative

factors of soil degradation (including also wind erosion and chemical and

physical damage) in different areas of the world. According to Table 4,

deforestation, overgrazing and agricultural mismanagement represent the

primary cause of >90% of the current global soil degradation.

A substantial proportion of the eroded material generated by this increase

in both the incidence and intensity of water erosion finds its way into

rivers and there are reports of greatly increased sediment yields in many

parts of the world and particularly in developing countries. Figure 1, based

on the work of Abernethy (1990), provides an example of the magnitude of the

increase documented for several reservoir catchments in Southeast Asia. Data

obtained from reservoir surveys undertaken at different times in the past

have been used to indicate that sediment yields in these catchments which

2
Table 1 Potential physical impacts of increased suspended sediment loads in
rivers.

A IN-STREAM EFFECTS

Biological impacts e.g. turbidity, sedimentaton, reduced productivity and

species diversity.
Recreational impacts e.g. restriction of swimming, boating and fishing and
reduction of overall aesthetics.
Sedimentation of chauels and water storage bodies e.g. reservoir
sedimentation, impairment of navigation, siltation of training structures.
Increased abrasion of hydraulic equipment e.g. REP turbines.

B OFF-STREAM EFFECTS

Flood damage e.g. aggradation, increased damage from muddy water.

Sedimentation of conveyance systems e.g. irrigation and drainage channels.
Increased cost of water treatment e.g. increased sedimentation times,
clogging of filters.
Impairment of industrial water use e.g. reduced cooling efficiency, abrasion
of pumps and turbines.
Sealing of irrigated soils

Based on Walling (1989a)

Table 2 A comparison of soil erosion rates under natural undisturbed

conditions and under cultivation in selected areas of the world.

Country Na ural Cul$vated-1

-5
(kg m year-l) (kg m year )

China < 0.20 15.00 - 20.00

USA 0.003 - 0.30 0.50 - 17.00
Ivory Coast 0.003 - 0.02 0.01 - 9.00
Nigeria 0.05 - 0.10 0.01 - 3.50
India 0.05 - 0.10 0.03 - 2.00
Belgium 0.01 - 0.05 0.30 - 3.00
UK 0.01 - 0.05 0.01 - 0.30

Based on Morgan (1979)

3
Table 3 Global human-induced soil degradation by water erosion
(10 ha of terrain affected)

Tn= Degree Total

Light Moderate Strong Extreme

Loss of topsoil 301 454 161 3.8 920

Terrain deformation 42 72 56 2.8 173

TOTAL 343 526 217 6.6 1093

Based on Oldeman et al. (1991)

Table 4 Factors cgntrolling global soil degradation.

(10 ha of terrain affected)

Continent Deforestation Overgrazing Agricultural Over- Bio-industrial

Mismanagement exploitation

Africa 67 243 121 63 +

Asia 298 197 204 46 1
S America 100 68 64 12
N & C America 18 38 91 11 +
Europe 84 50 64 1 21
Australasia 12 83 8 +

WORLD 579 679 552 133 23

Based on Oldeman et al. (1991)

Table 5 Lake sediment-based evidence of increases in sediment flux due to

catchment disturbance by human activity from tropical environments.

Lake Location Documented increase Source

in sediment flux

Lake Patzcuaro Mexico x7 O'Hara et al. (1980)

Lake Sacnao Mexico x35 Deevey et al. (1979)
Lake Ipea Papua New Guinea x10 Oldfield et al. (1980)
Lac Azigza Morocco x5 Flower et al. (1989)

4
have experienced substantial land clearance and intensification of land use

have evidenced annual rates of increase of between 2.5 and 6.0 percent.

Abernethy (1990) suggested that these increases closely paralleled the rates

of population growth in the catchments concerned (cf. Figure 1, inset),

although the ratio of the rate of increase in sediment yield to that for the

population was greater than unity. Based on this evidence he suggested that

in many developing countries annual sediment yields are currently increasing

at a rate equivalent to 1.6 times the rate of population increase and that

sediment yields could be expected to double in about 20 years. Contrasting

situations will, however, clearly exist in some other areas of the world

where reservoirs now trap a major proportion of the sediment formerly

transported by the rivers. In the classic case of the lower Nile,

construction of the Aswan High Dam has caused the annual sediment load at the

mouth of this major river to decrease from c. 100 million tonnes to near zero.

With the current concern for global change and the impact of both climate

change and human activity on the global system promoted by international

scientific programmes such as the IGBP, there is clearly a need to consider

changes in erosion and sediment transport as a key component of such change.

These changes are important from a scientific and environmental viewpoint

related, for example, to land degradation, terrestrial inputs to the oceans

and global element budgets, but, as noted above, they also have important

economic and management implications relating to river management, loss of

valuable reservoir storage and water quality degradation. It has, for

example, frequently been suggested that the major flood disasters which have

ravaged Bangladesh in recent years are, at least in part, the result of

excessive sedimentation within the lower reaches of its rivers consequent

upon land use change and increased erosion in the foothill and mountain

regions upstream. This contribution attempts to review available evidence

concerning the sensitivity of sediment yields to global change and the

magnitude of the associated changes.

5
200.

100'
1910 1920 1930 1940 1950 1960 1970 1980 1990
Year

Figure 1 Trends of increasing sediment yields in selected reservoir

catchments in Southeast Asia (based on Abernethy, 1990).

6
CHANGING SKDIMENT YIELDS

In comparison with other hydrological and hydrometeorological parameters,

such as river floods and annual precipitation, information on the sensitivity

of river basin sediment yields to environmental change is difficult to

assemble, due to the general absence of reliable long-term records of

sediment yield in most areas of the world. Few, if any, records extend back

beyond the present century and, in view of the many problems associated with

obtaining accurate estimates of annual loads (cf. Walling and Webb, 1981),

the reliability of early records is frequently open to question. Other

sources of information must therefore also be exploited. These sources

reflect a variety of timescales ranging from a long-term geological

perspective, through the evidence afforded by lake sediments which extends

back over lo2 to lo3 years, to recent catchment experiments and attempts to

use current data to provide a longer-term perspective by means of space-time

substitution. The information furnished by these various sources of

information, including available long-term records, will be reviewed in turn.

RECONSTRUCTING PAST SEDIMENT YIEDS

The long-term geological perspective

Information on the present-day distribution of sedimentary rocks of different

ages on the earth's surface and their associated mass can provide a basis for

estimating the.sedimentation rate at different periods in the past (cf.

Gregor, 1985). Since in broad terms the sedimentation rate can be equated

with the global erosion rate, it is possible to derive estimates of global

denudation rates in the past. These values will include both mechamical and

chemical denudation, but because the two are closely related (cf. Walling and

Webb, 1983), and the former dominates the total denudation rate, they can

provide a useful indication of gross variations in global erosion rates over

7
the past 500 million years. Coupled with estimates of the area1 extent of the

land area at different times in the geological past, these values can in turn

provide estimates of changing rates of specific sediment yield over this

period. This approach has been used by Tardy et al., (1989) to reconstruct

the temporal pattern of global specific sediment yield over the past 500

million years depicted in Figure 2A. Figure 2A indicates that global specific
-2 -1 -2
sediment yields have ranged between about 30 t km year and 70 t km
-1
year during the geological past. These variations reflect fluctuations in

the global climate, and more particularly in runoff amounts, as well as in

vegetation cover, in the relief of the land masses and in tectonic activity.

The Cambrian, Devonian, Cretaceous and Tertiary periods were thus

characterised by relatively high sediment yields, whereas values in the

Carboniferous, Permian and Triassic were substantially lower. The Devonian

period ( c. 350-400 x lo6 years BP.) which is marked by relatively high

sediment yields is known to have been a particularly wet period with high

runoff rates.

Also of significance in Figure 2A is the trend of increasing sediment

yields towards the present, which has been represented by an estimate of

contemporary transport of sediment from the land surface of the globe to the
-2 -1
oceans and which is characterised by values in excess of 100 t km year .

Whilst these higher values are, at least in part, a reflection of the

increasing rates of tectonic activity that have prevailed since the end of

the Jurassic period (c. 130 x lo6 years BP.) they must also reflect the

influence of human activity and more particularly forest clearance and land

use change in increasing sediment yields. The importance of these latter

effects have been emphasised by Ye et al. (1983) in their study of rates

of sedimentation on the Huang-Hai-Huai Plain in China by the Lower Yellow

River. In this case annual sediment yields from the Loess Plateau were

estimated to have more than doubled between the Middle Holocene and the

8
 OROGENIC CYCLES
ALPINE ) HERCYNIAN 1CALEDONIAN

Global sediment yields

600
Time ( lo6 Years B.P.)

5 lb 115 r‘ 0

Time (lo3 Years B.P.)

The Umberumberka Catchment,

NS W, Australia

3000 6000
Time (Years B.P.)

Figure 2 Variations of sediment yield over geological time. Based (A) on

Tardy et al. (1989), (B) on Degens et al. (1991), and (C) on Wasson
and Galloway (1986).

9
present as a result of human activity within the region. Similarly, in their

study of rates of Holocene sedimentation in the Yellow and East China Seas,

which receive sediment from the Yellow River, Milliman et al. (1987)

estimated that the recent river input was approaching an order of magnitude

greater than that existing in the early and Middle Holocene.

The estimates of past global sediment yields presented in Figure 2A serve

to emphasise the 'natural' variability of sediment yields in response to

long-term environmental change and more particularly changes in climate and

tectonic activity. However, they also highlight the important role of human

activity in causing the the total sediment yield from the land surface of

the earth to increase approximately threefold relative to the geological

background. More detailed information regarding fluctuations in sediment

yields over the recent geological past may be usefully introduced by

considering two further case studies which have analysed long-term changes in

deposition rates in sedimentary basins. The first relates to the Black Sea

where detailed analysis of sediment cores (cf. Ross and Degens, 1974) have

been used by Degens et al. (1976,1991) to reconstruct the record of sediment

input from its ca. 2.3 x lo6 catchment area over the past 20000 years (cf.

Figure 2B). This record demonstrates that the sediment input to the Black

Sea, and therefore sediment yields from its catchment area, were relatively

low during the Weichselian glaciation. The sediment input increased

dramatically during the subsequent period of deglaciation in response to the

increased runoff, the abundant sediment supplies exposed by the retreating

ice and the lack of vegetation cover, but slowly declined towards the

Atlantic climatic optimum when a relatively dense vegetation cover would have

existed. The significant increase in sediment inputs during the past 2000

years has been directly related to the impact of human activity, and more

particularly deforestation and development of agriculture within the area,

which has caused sediment yields to increase by a factor of about 3. The data

10
presented in Figure 2B thus afford a useful means of demonstrating the impact

of human activity on sediment yields in this region, but they also emphasise

that natural variations associated with periods of glaciation and

deglaciation may cause even greater changes in sediment yield.

The second case study relates to an investigation of sediment yields over

the past 6000 years from the 420 km2 Umberumberka catchment near Broken Hill

in western New South Wales, Australia, reported by Wasson and Galloway

(1986). This was based on an analysis of the development of the alluvial fan

deposited where the stream flowed onto the Mundi Mundi plain. Lack of organic

remains precluded high resolution dating, but the results presented in Figure

2C emphasise the importance of human activity in increasing sediment yields

during the period following the first European settlement in about 1850. In

addition, the data also evidence substantial shifts in sediment yields during

the period prior to European settlement. These shifts were probably

climatically modulated, but they again emphasise the need to recognise that

the natural system may be characterized by significant variations in sediment

yield in response to climatic variability.

Although it is inevitably fraught with uncertainties, particularly as the

timescale involved increases, reconstruction of temporal patterns of sediment

yield during the geological past provides a useful mean of demonstrating both

the long-term 'natural' variability of sediment yields and the importance of

human activity in perturbing the system and causing increased sediment

yields. In the latter case, increases in sediment yield of the order of 2-3

times have been documented by several studies. In many situations such

increases will exceed the long-term 'natural' variability of the system, but

in the example of the Black Sea cited above, the natural variability can be

seen to be as high as an order of magnitude and thus to be substantially

greater than the shift caused by man-induced perturbations.

11
Evidence from lake sediments

Where lakes occur at the outlet of a drainage basin and trap a large

proportion of the sediment output, analysis of the sedimentary record can

provide valuable evidence concerning past fluctuations in sediment loads. The

Black Sea example cited above demonstrates the value of this approach for

reconstructing long-term trends, but it is also capable of providing more

detailed information on the magnitude and timing of changes in sediment

inputs from the lake catchment over periods of lo2 and lo3 years.

Furthermore, where the catchment draining to the lake is relatively small in

size it may be possible to relate the reconstructed record of sediment yield

to documentary evidence regarding land use changes and other human impacts

within the catchment. The reliability of the reconstructed record will depend

on the accuracy of the core dating techniques employed, the number of cores

and the accuracy of the core correlation procedures used to estimate the

total volumes of sediment deposited during particular periods. Evidence

of changing rates of deposition obtained from a single core can provide a

basis for evaluating changes in the relative magnitude of sediment inputs

through time, but multiple cores and core correlation techniques are an

essential prerequisite for estimating the volumes of sediment involved and

therefore the absolute values of sediment yield (cf. Dearing and Foster,

1986). Because of the increased temporal resolution associated with dating

techniques applicable to the recent past (eg. Lead-210 dating) this approach

to reconstructing the sediment yield record has been most widely applied to

the historical period and particularly to periods of extending back several

hundreds of years. Several examples of the evidence provided by this approach

are presented in Figure 3 and these will be considered in turn.

The classic example of the Frains Lake catchment in Michigan, USA,

illustrated in Figure 3A is based on the work of Davis (1976). This lake is

located at the outlet of a small 0.18 km2 drainage basin and a detailed

12
 n

‘2 500 -
K Frains Lake Catchment,
? 400 - M ichigan, USA
E
24 LAND USE
z 300-
a Forest Crops
3 200- I

v3 1800 1850 1900 1950 2000

n
Year

z
25 10

.3E
a 200 I
G4 1750 1800 1850 1900 1950
n
Year
‘2 300
6, 1 Lake Havgtidssjb, S. Skane,Sweden
Estimatesbasedon singlecore

Estimatesbasedon core grid

2000 1000 0 1000 2000

BC AD
Year
F igure 3 Lake sediment-based evidence of historical trends in sediment
yields. Based (A) on Davis (1975), (B) on Dearing et al. (1981) and
(C) on Dearing et a1.(1987).

13
programme of sediment coring, core analysis and dating enabled the record

of sediment yield for the past 200 years to be reconstructed. The

reconstructed sediment yield record shows low rates of sediment yield in

pre-settlement times, rising by a factor of up to 70 with the onset of

settlement and agricultural clearance after 1830, and stabilizing after 1900

at a rate about 10 times the pre-settlement rate.

The reconstructed record of sediment yields from the 38 km2 catchment of

Lyn Peris in North Wales, UK, illustrated in Figure 3B was generated by

Dearing et al. (1981). Again substantial changes in sediment inputs to the

lake are apparent, with sediment yields increasing about 8-fold, from about 5

t km
-2 -1 -2 -1
year in the earliest period to about 40 t km year in recent

years. These increases have been ascribed to the erosional impact of mining,

quarrying, overgrazing and recent constructional activity in this upland

catchment. Maximum and minimum estimates of sediment yield have been provided

for the individual periods in order to take account of some of the

uncertainties involved in the calculations.

An example of a reconstructed record of sediment yield extending back over

a longer period of several thousand years is provided by Figure 3C which is

based on the work of Dearing et al. (1987) on Lake Havgardssjon, a small lake

set in the hummocky moraine landscape of southern Skane, Sweden. In this

study a grid of 47 cores was used to reconstruct the record of sediment

yield for the period post-1550, and evidence from a single 4m core was used

to extend the record on a more tentative basis back to 3050 BC. Estimates

based on the single core have been ascribed precision limits in order to take

account of uncertainties in sediment dating. Based on this record, it can be

seen that sediment yields during the period 3OOO:SO BC were of the order of

25 t kmm2 year-', a level which is consistent with an area of essentially

undisturbed woodland. From about 50 BC sediment yields increased, rising to a

-2 year -1 during the period AD 950-1300. This is again
peak of 86-250 t km

14
consistent with the known history of the area, since this was a period of

forest clearance, village establishment, and agricultural expansion,

following the introduction of the heavy-wheeled plough. The decrease in

sediment yields in the subsequent period from AD 1300-1550 also coincides

with the agrarian depression documented for many areas of N.W. Europe and

with the climatic deterioration of the early part of the 'Little Ice Age'.

More recent increases in sediment yield in the period post-1550 again

correspond closely with the historical records which point to an expansion in

the area under cultivation and therefore susceptible to higher rates of

surface runoff and erosion.

The examples of lake sediment-based reconstructions of sediment yield

cited above relate to temperate environments in Europe and North America.

Similar investigations have been undertaken in other areas of the world where

it has again proved possible to document changes in sediment yields

reflecting anthropogenic disturbance. For example, O'Hara et al. (1993)

report results obtained from Lake Patzcuaro in the severely degraded

landscape of the volcanic highlands of central Mexico which permitted

reconstruction of sediment inputs to the lake over the past 4000 years. The

findings indicate that sediment yields increased more than five-fold as a

result of extensive land clearance and were at least as high under the land

management of the indigenous population as after the Spanish conquest of the

region. These findings and those from several other studies in tropical

environments are summarised in Table 5. The trends evidenced in Figure 3 and

Table 5 again demonstrate the sensitivity of sediment yield to land use

change. This sensitivity is reflected in both the substantial gross increases

evident from a comparison of sediment yields before and after human-induced

land disturbance, and the variations reflecting particular phases of human

activity.

15

--
THE EVIDENCE FRON UING-TRRM RECORDS

AS previously indicated, there is a general lack of reliable long-term

records of suspended sediment load which can be used to analyse long-term

trends in sediment yields and their controlling factors. Changes in sampling

equipment, in sampling frequency and sampling protocols and in load

calculation procedures can all introduce uncertainties into the consistency

and continuity of log-term records. In addition, major changes in river

behaviour, such as produced by reservoirs or river regulation, may introduce

further discontinuities into the records such that it becomes increasingly

difficult to interpret measured loads in terms of the sediment yields from

the upstream catchment. More work is required to collate available long-term

records and to evaluate their potential for documenting long-term trends.

Several examples based on available analyses may, however, be usefully

introduced.

Long-term records of suspended sediment load stretching back to the 1940s

and 1950s are available for a number of rivers in the former Soviet Union and

a general assessment of the trends exhibited by these data undertaken by

Bobrovitskaya (1994) indicated that of the order of 70% of the rivers were

characterised by non-stationary sediment load series. About 40% of the rivers

evidenced decreases in sediment load which were in most cases the result of

dam and reservoir construction and diversion of water for irrigation schemes.

The construction of the Krasnoyarsk power generation plant on the Yenesei

river in 1967, for example, caused a decrease in sediment loads for a

distance of more than 900 km downstream. Similar decreases were recorded on

the Ob, Don and Dneiper Rivers. Abstraction of water from the Kuban River for

irrigation schemes by the Nevinnomyssk and Great Stavropol Canals has

likewise caused the sediment loads of this river to decrease by 2-4 times. In

contrast, ca. 30% of the rivers evidenced significant increases in sediment

16
loads which were linked to land use change and catchment disturbance,

including forest clearance, expansion of cultivation, gold mining activity,

and disturbance of permafrost by pipelines.

Figure 4, based on data supplied by Bobrovitskaya (personal

communication), presents examples of four rivers in the former Soviet Union

where sediment loads have increased over the period of record as a result of

land use change. The Dema River (12500 km2), which is a tributary of the

Volga, provides an example of a river impacted by expansion of cultivation

within its drainage basin, whereas the Dnestr River (850 km2 > in the Ukraine

has been influenced by forest clearance in its headwaters. In the case of the

Kolyma River (99400 km21 in western Siberia and the Yazgulem River (1940 km2)

in Kazahkstan, the increased sediment loads can be related to gold mining

activity and expansion of irrigated agriculture respectively. The linear

trend lines fitted to the records from the four rivers and plotted on Figure

4, each evidence an upward trend which is significant at the 95% level of

confidence. In the case of the Denestr, Kolyma and Yazgulem Rivers, the

upward trends are significant at the 99% level of confidence. The trend of

increasing annual suspended sediment loads evidenced by the Dnestr River is

the most marked out of the four rivers and suggests that sediment loads in

this river have increased by about S-fold since the early 1950s. Whilst this

increase can be related to forest clearance in its upstream catchment, it

also reflects climatic change and the general increase in runoff amounts that

has occurred in this river over the period of record and more particularly

since the late 1960s (cf. Figure 5). In the case of the other three rivers

included in Figure 5, there is no evidence of a significant trend in runoff

over the period of record and the changes in sediment load may be ascribed to

land use change and catchment disturbance alone.

Double mass plots of cumulative annual sediment yield versus cumulative

annual runff provide a useful means of establishing the timing of the changes

17
 Annual suspendedsedimentyield Annual suspendedsedimentyield 13
(t km-*) G (t km-‘)
0 0d VIA 8
1950

Y 1955

1960

1965

s
1970
1970
P

5
1975
1975 e

1980
1980

1985

Annual suspendedsedimentyield Annual suspendedsedimentyield

(t km-‘) (t km-‘)
3 s g 8 “0

1945

1950
K----
1960
1965 g
R
1965
2
1970 s
" 1970

1975 F
z 1980
1980 cx,
m
1985
Dema River at Bochkarevo, 1949-l 985 Kolyma River at Srednekansk, 194 l-l 988

Dnestr River at Sambur, 1950- I983 Yazgulem River at Motravn, 1950-l 986

/
1Ill
Figure 5 Longer-term trends in the annual runoff of the four rivers
represented in Figure 4 (based on data provided by Dr Nelly
Bobrovitskaya, State Hydrological Institute, St Petersburg).
in sediment yield that have occurred in the fol~r rivers represented in Figure

4 and of tentatively distinguishing the effects of land use and climate

change (cf. Figure 6). In each case a major shift in the sediment response

appears to occur in the 1960s. If it is assumed that there is an

approximately linear relationship between runoff and sediment load, it can be

argued that the.effects of increased runoff will not be reflected by a change

in the slope of the double mass plot. Therefore, in the case of the Dnestr

River, land use change and the associated catchment disturbance can be

estimated to have caused a 1.8 fold increase in annual sediment loads. This

is of the same order as the increases of 1.4, 1.5 and 1.7 fold associated

with the impact of increased cultivation, gold mining and expansion of

irrigated agriculture, respectively, observed in the Dema, Kolyma and

Yazgulem Rivers.

The suspended sediment loads of many rivers in the USA have now been

monitored for up to 50 years or more and such records also afford valuable

evidence of recent trends in sediment yield and their controls. Figure 7,

based on the work of Uri (1991), presents a plot of the annual mean suspended

sediment concentrations recorded in the Iowa River above Coralville Reservoir

near Iowa City, for the period 1948-1985. Statistical analysis undertaken by

Uri (1991) indicates that the suspended sediment loads of this river have

demonstrated a progressive increase over time during this period. This

increase has in turn been related to the expansion of the cultivated area

within the catchment and to the increasing dominance of corn and soybean in

the cropping programme. The area planted in soybeans almost doubled between

1960 and 1985 and Uri (1991) has calculated that for each 1% increase in the

area planted to corn and soybean, the mean suspended sediment concentration

has increased by approximately 0.42%.

For many other rivers in the USA, the long-term records demonstrate a

marked decrease in sediment load consequent upon their damming for reservoir

20
Cumulative annual suspended Cumulative annual suspended
sediment yield (t km-z) sediment yield (t km-a)
c

Cumulative annual suspended Cumulative annual suspended

sediment sediment yield (t km-z)
5; G
0zl 8 8
I I I
Mean suspendedsediment concentration
(mg l-9
development. Meade and Parker (1985) have, for example, documented how the

construction of five major dams on the Missouri River between 1953 and 1963

caused a marked reduction in the sediment loads transported by this river,

such that the load entering the Mississippi was reduced to only about 25% of

its former value (cf. Figure 8).. Since the Missouri River formerly

represented the major supply of sediment to the Mississippi, the sediment

load of that river has also declined, and the load at its mouth in 1984 was

less than one-half of the value before 1953. Similarly, the same authors have

used available data for the five major rivers of Georgia and the Carolinas

from the early part of the century and from the early 1980s to demonstrate

that the loads of these rivers are now only about one-third of those in 1910.

Even greater reductions in suspended sediment transport have been documented

by Meade and Parker (1965) for rivers draining the arid and semi-arid regions

of Southwest USA, where reservoir development has again caused major changes

in river behavior. In the case of the Rio Grande, its annual sediment

discharge to the Gulf of Mexico has declined from about 20 million tonnes in

1940 to something less than 1 million tonnes at present. Similarly, the

Colorado River now discharges about 100000 tonnes of sediment to the Gulf of

California each year, whereas before about 1930 the load was more than three

orders of magnitude greater and averaged 125-150 million tonnes.

It is clear from the above examples that reservoir construction on many of

the worlds major rivers will have caused marked reductions in sediment loads.

The River Nile probably affords an even more extreme example, because the

annual sediment load transported into the delta has decreased from c. 100

million tonnes to an almost negligible value as a result of the closure of

the Aswan Dam. Data compiled by UNESCO (1978) relating to the major

reservoirs of the world indicate that these reservoirs now control c. 10% of

the total runoff from the land to the oceans. Although it is reasonable to

assume that the proportion of the total sediment flux from the land to the

23
 Sharpe Dam (1963)

FOII RandaN Dam (1953)

awns Pomt Dam (1

.-
-0

600

500

400

300

203

100

W a ter year

F igure 8 Changes in the annual suspended sediment loads transported by the

M issouri and M ississippi Rivers resulting from reservoir
construction (based on Meade and Parker, 1985)

24
oceans which is trapped in reservoirs will be similar to the proportion of

the total runoff which is controlled, it is important to note that many of

the major reservoirs of the world are located in arid and semi-arid regions

where where sediment yields are relatively high. Furthermore, the reductions

in sediment transport caused by reservoir development in many areas of the

world must be balanced against increases caused by land clearance and land

use change, to the extent that current sediment yields may be fairly close to

those existing before the onset of widespread land disturbance by human

activity. Further work is also clearly required to exploit the potential

afforded by long-term records in terms of isolating the impacts of climatic

variability from those of catchment disturbance and land use and in relation

to interactions between these two sets of forcing factors.

IzvImNCE FROH OTHER CONTEMPO

RARY HEAS-s

Catchment experiments

Most catchment experiments are, by design, concerned with relatively small

areas and specific treatments or land use changes applied to the entire

catchment area in order to monitor their impact (cf. Walling, 1979; Ward,

1971). The results obtained cannot, therefore, generally be directly

extrapolated to larger more heterogeneous drainage basins, in order to

examine longer-term trends in sediment yields. Available results do, however,

provide valuable information on the likely magnitude of the changes in

sediment response associated with particular types of catchment disturbance

or land use practices. The examples of increases in sediment yield documented

by experimental catchment studies listed listed in Table 6 provide an example

of information of this type and again emphasise the sensitivity of sediment

yields to catchment disturbance. In these cases increases of up to more than

two orders of magnitude have been documented. As a further example, attention

25
Table 6 Some results from experimental basin studies of the impact of land
use change on sediment yield.

Region Land use change Increase in sediment Reference

yield

Westland, New Clearfelling x8 O'Loughlin et al. (1980)

Zealand

Oregon, USA Clearfelling x39 Fredriksen (1970)

Northern Afforestation Painter et al. (1974)

England (ditching and ploughing) xl00

Texas, USA Forest clearance and x310 Chang et al. (1982)

cultivation

Maryland, USA Building construction x126-375 Wolman and Schick (1967)

Table 7 The impact of soil and water conservation measures on catchment

sediment yields in the Loess Region of the Middle Yellow River,
China.

Catchment Ar?a Area controlled Sediment Yield Reference

(km ) (2) Reduction (X)

Wangmao Gully 5.97 68 89 Mou (1991)

Wangjia Gully 9.1 71 91 Mou (1991)
Nanxiaohe Gully 36.3 58 97 Mou (1991)
Yangjiagou Gully 0.87 40 93 Li (1992)

Table 8 Increases in the sediment yields of world rivers resulting from

catchment disturbance by land use activities, based on data
assembled by Dedkov and Mozzherin (1984)

Group Small Basins Large Basins All Basins

Lowland Rivers x 13.0 x 8.1 x 10.0

(N = 1854)
Mountain Rivers x 2.2 x 3.8 x 2.8
(N = 1811)

26
may be directed to catchment experiments undertaken in forest areas, aimed at

evaluating the impact of forest management practices on sediment yields and

other water quality parameters. Binkley and Brown (1993) have reviewed the

large body of data available from watershed experiments undertaken in North

America and a selection of results expressed in terms of increases in mean

suspended sediment concentrations is presented in Figure 9. Increases in mean

suspended sediment concentration of up to an order of magnitude and even more

are demonstrated.

Whilst most experimental catchment studies have focused on assessing the

magnitude of increases in sediment yield caused by land disturbance and land

use practices, it also important to consider studies that have documented the

impact of soil conservation and other catchment management practices aimed at

reducing suspended sediment yields. Such results provide a contrasting

perspective on the sensitivity of sediment yields to change and indicate the

potential for reversing the detrimental effects of land use change and land

use practices. Table 7 lists some results obtained from four small watersheds

in the severely eroded loess region of the Middle Yellow River where a range

of soil conservation measures, including both tree planting and construction

of teraces and gully check dams, have been employed to achieve a marked

reduction in sediment yields. In this case, sediment yields have been reduced

by c. 90% or more.

Space-time substitution

In the absence of reliable long-term records of sediment yield capable of

demonstrating trends associated with specific environmental changes,

space-time substitution or the ergodic hypothesis can afford an effective

means of demonstrating the likely trends involved. In essence the approach

involves assembling information on spatial variations in sediment yield in

response to, for example, variations in land use, and using these variations

27
 Treatment - Mean suspendedsedimentconcentration
(mg 1-l)
A
c3 3s E a
0
s0 s 0 5 g0

:
:
:
:
:
:.
:
:
:
:
:
‘I :
:
:
:
: ‘.:
ml,
:
:
: ::
: ‘.
:
:
*.
xi &*..**
:
t-4 : ‘cfl ‘*.
-=-;J
to examine the likely magnitude of changes in sediment yield associated with

known changes in catchment condition. Figure 10 provides several examples of

the potential of this approach. In each case, by assuming that the low

sediment yields associated with forested or natural areas afford a benchmark

against which changes associated with alternative land uses can be assessed,

increases in sediment yield of up to several orders of magnitude can be

inferred. It is necessary to assume that other factors influencing the

sediment yields remain constant and the ergodic hypothesis can clearly only

be applied with confidence within a relatively small area. Thus, for example,

in the case of the data presented for Kenya in Figure 10, which are based on

the work of Dunne (19791, it may not be strictly correct to infer that

sediment yields from the grazing areas were once as low as those in the

forest areas, since the former occur primarily in semi-arid areas, whilst the

latter are found mainly in the more humid regions.

A classic example of the potential of this approach for reconstructing the

long-term record of sediment yield for an area is provided by a study

undertaken by Wolman (1967) in the Piedmont region of Maryland, USA. In this

case data such as those shown in Figure 10 were combined with a generalised

record of land use change in the region to synthesise the record of temporal

changes in sediment yield depicted in Figure 11. In this case, sediment

yields can be seen to have increased by an order of magnitude as a result of

forest clearance by European settlers and the expansion of cropping, reaching

a maximum around 1900. After this date sediment yields declined somewhat, in

response to the introduction of conservation measures and the decline of the

area under cultivation. Construction activities associated with urban and

suburban expansion caused a further increase in sediment yields in the 1960s

but values subsequently declined to values approaching those associated with

the natural forest land.use as a result of the widespread existence of urban

surfaces which tend to protect the soil from erosion.

29
 h Northern M ississippi, USA
-Lid
10000 Lower limit of sediment
z ’ / yk~~r~tscultivated
x
e :
&lo@J /
24 0

- 1 0 Upper limit of sediment

/-----T
z 0
yield from abandoned
lOO- fields and forest
Upper limit of sediment
yield from pine plantations
& mature pine-hardwoods
10 -
fl0 *
0 Mature pine-hardwoods
Depleted hardwoods
l-
* *
=
Pine plantations
Abandoned fields
0 Cultivated
+& * 0 Pasture J

1000 1250 1500 1750

Annual precipitation (mm)

;
Kenya
K
y.l 10000
s\
I/ m /f-- 0 I

Dominant land use

l Forest
o Forest>Agriculture
o Agriculture>Forest
n Grazing
1 I I I I I I

0 200 400 600 800 1000 1200

M ean annual runoff (mm)

Piedmont Region, E. USA

o Watershedsundergoing construction
l Natural watersheds

Drainage area(km')
Figure 10 Examples of studies which have investigated the spatial variation
of annual suspended sediment yields in response to land use
activities. Based (A) on Ursic and Dendy (1965) (B) on Dunne (1979)
and (C) on Wolman and Schick (1967).

30
 Woods 2
Land use Forest Cropping &. z Urban
grazing 3
r\
800

600

400

200

0
1820

Figure 11 The reconstruction of the historical record of suspended sediment

yields in the Piedmont region of Maryland, USA, proposed by Wolman
(1967) on the basis of space-time substitution.
 A further example of the information that can be obtained using this

approach, but this time at the global scale, is provided by the work of the

Soviet scientists Dedkov and Mozzherin (1984). These workers gathered

suspended sediment yield data from >3600 river basins in the mountainous and

non-mountainous (plains) regions of the world. These basins were grouped into

large (>5000 km') and small (<5000 km2) basins and further classified into

three groups, according to the degree of disturbance by agricultural

activity. Category I basins were largely undisturbed and were characterised

by either a forest cover >70% or a ploughed area <30%. Category II basins

exhibited an average degree of landscape change and were characterised by an

area under forest or cultivation of between 30% and 70%. In category III

basins, the degree of disturbance by land use activities increased still

further, with an area under forest <30% or a ploughed area >70%. By comparing

the average sediment yield of category I basins with those of category III

basins, it was possible to derive an approximate measure of the magnitude of

the increase in sediment yield associated with land disturbance by human

activity (cf. Table 8). In the plains regions of the world the average

increases were 8.1 and 13.0 for large and small river basins respectively,

whereas in the mountain areas the equivalent values were 3.8 and 2.2. Whilst

these values are necessarily limited by the fact that they represent simple

comparisons of mean values of sediment yield for drainage basins in different

categories, and are therefore heavily dependent on the representativeness of

the sample of river basins, they nevertheless provide a useful indication of

the general magnitude of the increase in the sediment yields of world rivers

resulting from catchment disturbance by land use activities. The values

involved are in line with those presented in Figure 3 where lake sediments

have provided the opportunity to reconstruct the detailed record of change

for particular drainage basins.

32
SOME PROBIJMS OF INTKRIWZATION AND PREDICTION

The preceding review of sources of information regarding the response of

suspended sediment yields to environmental change, and the magnitude of the

changes involved, has highlighted the sensitivity of this parameter to global

change. In many instances, increases of more than an order of magnitude have

been documented and it has been suggested that at the global scale, the

sediment yields of river basins in lowland areas that have been heavily

impacted by agricultural activity could now be an order of magnitude greater

than they were under undisturbed conditions. Further consideration of the

response of sediment yields to global change can usefully focus on the

problems of relating changes in downstream sediment yield to changing erosion

rates in the upstream drainage basin, the potential complexity of sediment

yield response to climatic change and interactions between climate and land

use change.

Linking erosion and sediment yield

The incidence of increased suspended sediment yields in river basins will, in

most cases, be closely linked to increasing rates of erosion within the

upstream drainage basin. The precise relationship between upstream erosion

and downstream sediment yield will, however, depend on the sediment delivery

or conveyance processes interposed between upstream erosion and downstream

sediment yields. The complexity of the drainage basin sediment delivery

system may attenuate the linkage between changing erosion rates and sediment

output, such that a close and synchronous link between changing sediment

loads and the forcing variables may not exist (cf. Robinson, 1977; Walling,

1983, 1989b). A useful example of the potential lack of correspondence

between the sediment yield of a drainage basin and the local on-site erosion

rates is afforded by Table 9 which compares the measured sediment yields from

a number of African river basins impacted by human disturbance with estimates

33
Table 9 Comparison of the measured suspended sediment yields of selected
African rivers with estimated rates of contemporary soil loss by
water erosion depicted on the FA0 map of Soil Degradation (FA0,1979)

River Country Basin2Area Suspended Sc$iment-1 FA0 Estimats of Sofl

(km > Yield (t km year Loss (t km year >

Watari Nigeria 1450 483 1000 - 5000

Bunsuru Nigeria 5900 438 1000 - 5000
Senegal Mali 157400 14.6 1000 - 5000
Faleme Mali 15000 40 1000 - 5000
Hamman Algeria 485 198 1000 - 5000
Kebir Ouest Algeria 1130 92 1000 - 5000
Mesanu Ethiopia 150 1680 5000 - 20000

Based on Walling (19898)

Table 10 The relative importance of climatic variation and watershed

management in accounting for the recent reduction in the sediment
yield of the Sanchuanhe River, China.

Annual sediment yield 1957-1969 36.81 x lo4 t.

Annual sediment yield 1980-1989 9.63 x lo4 t.
Total reduction 74 %
Reduction due to reduced rainfall 33-38 %
Reduction due to watershed management 36-41 %

Based on Zhao et al. (1992)

34
of the contemporary soil erosion rates within these basins derived from the

FA0 map of current rates of soil degradation by water erosion in Africa north

of the Equator (FAO, 1979). In all cases, the estimates of soil erosion rates

are approximately an order of magnitude or more greater than the reported

sediment yields. Thus, while a considerable body of information now exists

concerning the impact of specific land use practices on rates of soil loss

(cf. Figure 12) and soil loss equations such as the USLE and SLEMSA (cf.

Wischrneier and Smith, 1978; Elwell, 1981) can provide a means of estimating

the impact of particular cropping and land management practices on soil

erosion rates, it is much more difficult to convert such values of soil loss

into estimates of downstream sediment yield and of the changes in sediment

yield associated with a given land use change scenario, particularly in

larger drainage basins.

A classic example of the complexities involved in attempting to relate

longer-term changes in downstream sediment yield to historical records of

land use change and therefore erosional activity in the upstream basin is

provided by the work of Trimble (1976,1981) in the 360 km" basin of Coon

Creek in the Driftless Area of Wisconsin, USA. Using information from a range

of field and documentary sources he was able to reconstruct the sediment

budget of this catchment for two periods in the past, namely, 1853-1938 and

1938-1975 (cf. Figure 13). The first period represented one of poor land

management which resulted in severe erosion, whereas the second was

characterised by the introduction of conservation measures. During the first

period, large volumes of soil were eroded from the slopes of the basin, but

only a small proportion (ca. 5%) of this was transported out of the basin.

Most of the eroded material was stored on the lower slopes as colluvium, and

in the valley floors. During the latter period, when widespread conservation

measures were introduced, rates of soil loss from upland sheet and rill

erosion were reduced by about 25%, but sediment yields at the basin outlet

35
Pacific Northwest, USA
Permanent pasture
Range or seededpasture (good)
Legumes - grass hayland
Range or seededpasture (poor)
Alfalfa
Small grain
Wheat - peas
Small grain (poor)
Wheat fallow
Burnt wheat fallow
Orchards - vinyards (clean tilled)
Row crops and fallow
I I I I I I
0 20 40 GO 80 100
Relative erosion rates

Figure 12 Variation of relative rates of soil loss in the Pacific Northwest

region of the USA according to land use and crop type (based on
Brown, 1950).
 Sources (t x 103) A 1853-1938
Unland sheet

Sediment
42 discharge
at mouth

Sinks and stores (t x 103)

Sources (t x 103) B 1938-1975

Upland sheet
and rill Upland
erosion gullies Middle

Sediment
40 discharge
at mouth

Hillslopes
Sinks and stores (t x 103)

Figure 13 The sediment budgets of Coon Creek, Wisconsin, USA, reconstructed

by Trimble (1983) for the periods 1853-1938 and 1938-1975.

37
remained essentially the same, because sediment stored in the tributary

valleys and upper main valley was remobilised. In this case, therefore,

substantial changes in land use and land management within the upstream basin

were not reflected in equivalent changes in downstream sediment yield. A

similar situation has been described by Meade (1982) for other rivers in the

eastern USA.

Recent work undertaken within the agricultural area of the Russian Plain

reported by Golosov et al. (1992) p rovides further evidence of the complexity

of the sediment delivery system and of the problems associated with

attempting to link changing erosion rates with changing sediment loads in

local rivers. In their study, Golosov et al. (1992) documented the

contemporary sediment budgets of six representative small and medium-sized

drainage basins within the Russian Plain. The results which are depicted in

Figure 14 emphasise the considerable contrasts in sediment budget that may

exist within a relatively small area. Thus, whereas a major proportion of the

sediment mobilised within the Balka Sukhoj Jaz and Kijuchi Creek basins is

transported to the basin outlet, in the case of the Elkhovka Creek, Balka

Rolzavets and Veduga Creek basins most, and in one case all, of the sediment

mobilised within the basin is deposited at the foot of the slopes, on the

floodplains or in linear depressions known locally as balkas, and only a

small proportion reaches the basin outlet. Any change in rates of erosion and

sediment mobilisation within the latter three basins would be likely to be

reflected in only limited changes in sediment output, whereas in the former

basins changes in erosion would be likely to be closely reflected by changes

in sediment output.

A more specific example of the potential significance of temporal

discontinuities in sediment delivery associated with the storage and

remobilisation of large volumes of sediment in the channel system is provided

by the difficulties involved in attempting to interpret the long-term (>40

38
 Elkhovka Creek (12 km2) Veduga Creek (86.9 km2)

Balka Rolzavets (181.5 km2) KiJuchi Creek (8 km2)

Slope
42%
Rill
4.5% kas
Gully %
Balkas 53.5%
91%

9% 89%

Little Kolyshley River (181.5 km”) Balka SukhoJ Jaz (11 km2)
Slope
21%

Rill
79% Slope
23%
Balkas
1%

28% 76%

Figure 14 The sediment budgets of six small and medium-sized drainage basins
on the Russian Plain, as documented by Golosov et al. (1992).

39
year) sediment yield records available for the Orange River in South Africa

and the San Juan River, a tributary of the Colorado River in southwest USA.

Both drain large areas of semi-arid rangeland which were subjected to

considerable grazing pressure during the latter part of the nineteenth

century and the early part of the twentieth century. This grazing pressure

was subsequently reduced in an attempt to reduce rates of soil loss and

downstream sediment transport. Inspection of the long-term sediment yield

records for these two rivers, which date from 1930 (Figure 15), provide some

evidence of declining sediment yields which is confirmed by double mass

plotting of accumulated sediment yield versus accumulated runoff.Hadley

(1974) has suggested that the decline in the annual sediment yields of the

San Juan River which was initiated in 1942 can be linked to the reduced

grazing pressure and consequent improved vegetation cover and reduced soil

loss which resulted from the Taylor Grazing Act of 1934. More detailed work

by Graf (1985) has, however, cast doubt on this simple interpretation. Graf

argues that the reduction in sediment loads in the Colorado and its

tributaries which has occurred since the 1940s is due primarily to a dramatic

reversal in the behaviour of the alluvial valleys and channel upstream,

whereby erosional or throughput channel conditions were replaced by

aggradation. Much of the reduction in sediment yield which occurred after

1940 can therefore be attributed to increased losses to channel storage and

Graf (1985) indicated that the amount of sediment currently stored in the

canyon floors of the Colorado and its tributaries is equal to about 200 times

the present annual sediment yield.

The long-term sediment yield record for the Orange River presented in

Figure 15 demonstrates a similar reduction in sediment yield since 1936 to

that evidenced by the San Juan River. This reduction has been noted and

discussed by Rooseboom and Von Harmse (1979) who ascribed it to a progressive

exhaustion of erodible sediments following a period of accelerated erosion

40
 Z:, Orange River at UpingtodPrieska, South Africa
gz-
.=
=x 200 -
% = loo-
53
Q)
0 P- ~‘Ll~.dl; ..I, ,.I n,
CX.-
8
1930 1935 1940 1945 1950 1955 1 YbU IYb3 1YlU
2000
/’
18oQ- ,I’
,’
/’ 50%
1600-
,' /' ,4
1400 -

1200 -

1000-
800 -

600-

400-

200 -
0 ' &@
I I I I I I I I
0 50 100 150 200 250 300 350 400 450
Accumulated runoff (m3 x 109)

San Juan River at Bluff, Utah, USA

llqlLLLJ’l1lJlI +
1960 1965 1970 1975

18

6 10 20 30 40 50 60 70 80 90

Accumulated runoff (m3 x 109)

Figure 15 Temporal discontinuities in the long term records of suspended

sediment yield from the Orange River, South Africa and the San Juan
River, USA, demonstrated by double mass plotting. Based (A) on data
contained in Rooseboom and Maas (1974) and (B) on data from US
Geological Survey records.

41
initiated by human activity. This is consistent with the analysis of the

alluvial valley fill deposits reported by Butzer (1971), who suggested that

the intensive anthropogenic disturbance by burning and overgrazing which

occurred during the period 1880-1930 and which corresponded with a run of dry

years, caused increased surface runoff and a rapid dissection of the valley

fill deposits. The decline in sediment yields post 1936 can therefore be

tentatively ascribed to the completion of a 'cut' cycle and the further

reduction apparent since 1954 may represent the initial stages of a 'fill'

cycle, since Butzer (1971) refers to recent alluviation of the major valleys.

Both the examples cited above emphasise the potential importance of the

storage of fine sediment within alluvial valley floors in controlling the

downstream record of sediment yield and that this record may not be a direct

reflection of the temporal variation of erosion rates within the upstream

basin. In this context, the importance of scale to the interpretation of the

impact of catchment changes on downstream sediment yield must also be

recognised. Whereas such changes may be rapidly evident in small basins, the

timescale involved may be considerably greater in larger basins where large

stores of sediment exist in alluvial valleys and.channel to buffer the

upstream change.

Response to climatic change

Although most of the information and discussion presented in this

contribution has been concerned primarily with the impact of land use change

and catchment disturbance on sediment yields, in view of current interest and

concern for the potential impact of climatic change it is important also to

direct explicit attention to this component of global change. The general

absence of reliable long-term records of river sediment loads highlighted

previously has effectively precluded detailed analysis of trends related to

recent climatic fluctuations, but a variety of potential scenarios may be

42
considered in relation to future climatic changes. In this context it is

important to recognise the multiple controls on catchment sediment yields and

their interrelationships. For example, to consider the case of reduced annual

precipitation and therefore a shift towards a drier climate, it is possible

to suggest that sediment yields would decline in response to reduced runoff.

Conversely, if the interaction of precipitation amounts and vegetation cover

density is considered, it is possible to argue that sediment yields might

increase in response to reduced cover density, in line with the classic

Langbein-Schumm rule (cf. Langbein and Schumm, 1958). Changes in the

frequency and intensity and rainfall, as well as changes in landuse

consequent upon reduced moisture availability could add further complexity to

the ultimate scenario.

The rapidity of any climatic fluctuation, as well as the direction of

change, could also prove significant, since this will influence the degree of

adjustment between preciptation inputs and land cover density. This may be

illustrated further by considering Figure 16A which provides a schematic

representation of possible changes in sediment yield associated with

increases and decreases in effective preciptation. Where precipitation

increases, and the shift occurs rapidly, a significant short-term increase in

sediment yield could result from the increased erosive energy impacting on

catchment surfaces where the vegetation cover density reflects the previous

drier conditions. As the vegetation cover density increases, in response to

the increased moisture availability, sediment yields could decline to a level

below those existing originally. At the end of this 'wet' period, the shift

in effective precipitation back to the original conditions could have little

or no impact on sediment yield, although yields could increase slightly once

the vegetation cover density had reduced in response to the reduced moisture

availability. In the case of a decrease in effective precipitation, sediment

yields are shown in Figure 16A to increase slightly in response to the

43
 Effective precipitation

1 Decrease 1

Vegetation cover

Sediment yield

Time7 Time
4
Annual precipitation Annual suspended
3

g2
x River Tana
at Kamburu, Kenya
‘51
3 year running mean
I I 1 1 I I 0 1 I I I 1 I 1 I I
1950 52 54 S6 521 60 62 64 66 6X 70 1950 52 54 56 58 60 62 64 66 68 70
separating the effects of catchment management and climate change on sediment

yields. The Sanchuanhe River drains a 4161 km2 basin in the gullied-hilly

loess region of the Middle Yellow River basin and has been the focus of

considerable soil conservation and water conservancy activities during the

1980s. Thus by the end of 1989, 267 km* of bench terrace had been constructed

in the catchment, 703 km* had been planted with forest for soil conservation

purposes and nine reservoirs had been commissioned. A comparison of the mean

annual sediment loads for the periods 1957-1969 and 1980-1989 indicated that

sediment yields in the latter period had been reduced to only about 25% of

those in the period 1957-1969. This reduction was partly a reflection of the

soil and water conservation works undertaken in the river basin, but it also

reflected a shift to drier conditions, with the mean annual precipitation in

1980-1989 being only 509 mm as compared to 542 mm in 1957-1969. By using a

range of data analysis techniques, Zhao et al. (1992) were able to

demonstrate that in the case of the Sanchuanhe River basin, approximately 50%

of the reduction in sediment yield noted for the period 1980-1989 could be

attributed to the drier conditions and approximately 50% to the soil and

water conservation measures (cf. Table 10).

In some situations it is likely that changes in catchment condition and

changes in climate may operate synergistically to produce changing sediment

yields. Thus it is possible to conceive of a situation where land use change

causes general degradation within catchment leading to instability, but a

shift in climate towards wetter conditions is needed to trigger a major

change in erosion and sediment yields. Climatic change could therefore

provide an important forcing function for more general catchment disturbance.

PERSPECTIVE

The preceding review of sources of information regarding the response

of erosion and suspended sediment yields to environmental change, and

45
reduced vegetation cover density. The general reduction in erosive energy

limits the magnitude of this increase. However, at the end of this 'dry'

period, the restoration of the precipitation back to its original level is

marked by a rapid, but again short-lived, increase in sediment yield

reflecting the delay associated with the increase in vegetation cover. Thus,

although in the longer-term, increases and decreases in effective

precipitation are shown as giving rise to, respectively, a slight decrease

and increase in sediment yield, the periods of transition to increased

precipitation are marked by substantial short-lived increases in sediment

yield.

The schematic trends portrayed in Figure 16A are aimed primarily at

demonstrating the potential complexity of climatic change / sediment yield

interactions, but some evidence for the occurrence of the responses suggested

is available. Roberts and Barker (1993) cite the case of the response of the

sediment yields of two Kenyan rivers, documented by Walling (1984), to

changing climatic conditions in this region. The period at the end of the

1950s and into the early 1960s was characterised by a shift to wetter

conditions (cf. Figure 16B) consequent upon changes in atmospheric

circulation (cf. Lamb, 1966) and this was reflected by a marked, but

essentially temporary, increase in the sediment loads of the Tana and Thiba

rivers (cf. Figure 16C) which correponds closely to the schematic trends

postulated in Figure 16A.

Interactions between climate and land use change

It is also important to recognise the potential interactions between land

use change and changes in climate, both in terms of isolating their

respective impacts and and in relation to synergistic effects. The detailed

studies of the changes in the sediment load of the Sanchuanhe River in China

reported by Zhao et al. (1992) afford a useful example of the problems of

46
the magnitude of the changes involved, has highlighted the sensitivity of

sediment yields to global change. Evidence from the geological past

emphasises that sediment yields may exhibit substantial temporal variations,

even under natural conditions, in response to changes in climate and tectonic

activity.In more recent times, increases in sediment loads of several-fold

have been widely documented in river basins impacted by land use change and

other forms of human-induced disturbance and it has been suggested that at

the global scale, the sediment yields of river basins in lowland areas that

have been heavily influenced by agricultural activity could now be

approaching an order of magnitude greater than they were under undisturbed

conditions. Such increases have been particularly marked in many areas of

Asia and Oceania which account for a major proportion of the total transport

of sediment from the land to the oceans and therefore exert an important

control on the magnitude of this flux (cf. Milliman and Meade, 1983).

Furthermore there is a growing body of evidence that suggests that recent

climatic change has also caused significant increases in sediment yield in

several areas of the world, including eastern Africa and the former Soviet

Union. However, it is important to recognise that there are also many

instances where the sediment loads of rivers have declined, or at least

remained stable, in the recent past. In the case of the Sanchuanhe River in

China, cited in Table 7, climate change, as reflected in reduced annual

precipitation, has been shown to have caused reduced sediment yield and there

are numerous examples of rivers where reservoir construction has resulted in

reduced sediment loads.

In view of the economic and environmental significance of increased

sediment loads in rivers, it is important that this sensitivity is viewed as

a key component of the phenomenon of global change which is currently

attracting widespread attention. Lack of reliable long-term records has

generally precluded detailed assessments of the problem, but there is clearly

47
scope to analyse available data in greater detail. Further work is

undoubtedly required to elucidate the complexities of relating changes in

upstream erosion rates to changes in downstream sediment yields, to establish

the likely impact of existing climatic change scenarios, to evaluate the

potential effectiveness of catchment management techniques for reducing

sediment yields and to develop effective prediction procedures. There is

clearly also considerable scope to link such analysis to other aspects of

global change which are currently attracting attention. The increasing rate

of degradation of the global soil resource and changing material fluxes

between the land and the oceans are related issues that would in turn clearly

benefit from closer integration.

Acknowledgement The assistance of Dr Nelly Bobrovitskaya, State Hydrological

Institute, St Petersburg, Russia and many other individuals in providing the

author with information on river sediment loads, and of Dr Phil Owens in data

collation and processing are gratefully acknowledged.

48
Abernethy, C. (1990) The use of river and reservoir sediment data for the
study of regional soil erosion rates and trends. Paper presented at the
International Symposium on Water Erosion, Sedimentation and
Resource Conservation, Dehradun, October, 1990.

Allan, R.J. (1986) The role of particulate matter in the fate of contaminants
in aquatic ecosystems. Inland Waters Directorate, Environment Canada,
Scientific Series no. 142.

Binkley, D. and Brown, T.C. (1993) Forest practices as nonpoint sources of

pollution in North America. Water Resources Bulletin, 29, 729-740.

Bobrovitskaya, N.N. (1994) A ssessment of trends to sediment discharge

variations in rivers of the former Soviet Union (FSU). In Proceedings of the
International Symposium on East-West, North-South encounter on the
State-of-the-art in River Engineering Methods and Design Philosophies, St
Petersburg, Russia, May 1989) IAHS Publication no. 184, 131-143

Braune, E. and Looser, U. (1989) Cost impacts of sediments in South African

rivers. In Sediment and the Environment (Proceedings of the Baltimore
Symposium, May 1989) IAHS Publication no. 184, 131-143.

Brown, C.B. (1950) Effects of soil conservation. In P.D. Trask (ed.), Applied
Sedimentation, New York, 3870-406.

Buringh, P. and Dudal, R. (1987) Agricultural land use in space and time. In
M.G. Wolman and F.G.A. Fournier (eds.) Land Transformation in Agriculture,
SCOPE Report no. 32, John Wiley, Chichester, 9-43.

Butzer, K.W. (1971) Fine alluvial fills in the Orange and Vaal basins of
South Africa. Proc. Assoc. Amer. Geographers, 3. 41-48.

Chang, M., Roth, F.A. and Hunt, E.V. (1982) Sediment production under various
forest-site conditions. In Recent Developments in the Explanation and
Prediction of Erosion and Sediment Yield (Proceedings of the Exeter
Symposium, July, 1982), IAHS Publication no.137, pp. 13-22.

Clark, E.H., Haverkamp, J.A. and Chapman, W. (1985) Eroding Soils: The
Off-Farm Impacts, The Conservation Foundation, Washington D.C.

Davis, M.B. (1976) Erosion rates and landuse history in southern Michigan.
Environmental Conservation 3, 139-148.

Dearing, J.A., Elner, J.K. and Happey-Wood, C.M. (1981) Recentsedimentflux

and erosional processes in a Welsh upland lake catchment based on magnetic
susceptibility measurements. Quat. Res., 16, 356-372.

Dearing, J.A. and Foster, I.D.L. (1986) Lake sediments and palaeohydrological
studies. In B.E. Berglund (ed.) Handbook of Holocene Palaeoecology and
Palaeohydrology, John Wiley, Chichester, 67-90.

Dearing, J.A., Hakansson, H., Liedberg-Jonsson, B., Persson, A., Skansjo, S.,
Widholm, D. and El Daoushy, F. (1987) Lake sediments used to quantify the
erosional response to land use change in southern Sweden. Oikos, 50, 60-78.

49

-__
Dedkov, A.P. and Mozzherin, V.T. (1984) Eroziya i Stok Nanosov na Zemle,
Izdatelstvo Kazanskogo Universiteta.

Deevey, E.S., Rice, D.S., Rice, P.M., Vaughan, H.H., Brenner, M. and
Flannery, M.S. (1979) Mayan urbanism: Impact on a tropical karst environment.
Science 206, 298-306.

Degens, E.T., Paluska, A. and Eriksson, E. (1976) Rates of soil erosion. In

Svensson, B.H. and Soderlund, R. (eds.) Nitrogen, Phosphorus and Sulphur -
Global Cycles, SCOPE Rept. no. 7, Ecol. Bull. (Stockholm) 22, 185-191.

Degens, E.T., Kempe, S. and Richey, J.E. (1991) Summary: Biogeochemistry of

major world rivers. In E.T. Degens, S. Kempe and J.E. Richey (eds.)
Biogeochemistry of Ma.jar World Rivers, John Wiley, Chichester, 323-347.

Dendy, F.E. (1968) Sedimentation in the nation's reservoirs. J. Soil and

Water Conservation, 23, 135-137.

Dunne, T. (1979) Sediment yield and land use in tropical catchments. -J.
Hydrology, 42. 281-300.

Eckholm, E.P. (1976) Losing Ground, Norton, New York.

Elwell, H. A. (1981) A soil loss estimation technique for southern Africa. In

R.P.C. Morgan (ed.) Soil Conservation: Problems and Prospects, John Wiley,
Chichester, 281-292.

FA0 (1979) A Provisional Methodology for Soil Degradation Assessment, FAO,

Rome.

Flower, R.J., Stevenson, A.C., Dearing, J.A., Foster, I.D.L., Airey, A.,
Rippey, B., Wilson, J.P.F. and Appleby, P.G. (1989) Catchment disturbance
inferred from palaeolimnological studies of three contrasted sub-humid
environments in Morocco. J. Palaeolimnology 1, 293-322.

Fredriksen, R.L. (1970) Erosion and sedimentation following road construction

and timber harvest on unstable soils in three small western Oregon
watersheds. US ForestService Research Paper, PNW 104.

Golosov, V.N., Ivanova, N.N., Litvin, L.F. and Sidorchuk, A. Yu. (1992)
Sediment budgets of river catchments and river channel aggradation on the
Russian Plain. Geomorphology 4, 62-71.

Graf, W. (1985) Channel processes and sediment yield in arid region drainage
basins. In Abstracts of Papers for the First International Geomorphology
Conference (Manchester, UK, Sept. 1985), T. Spencer (ed.) p. 223.

Gregor, C.B. (1985) The mass-age distribution of Phanerozoic sediments. In

N.J. Snelling et al. (eds.) The Geochronology and the Geological Record, The
Geological Society of London, 284-289.

Hadley, R.F. (1974) Sediment yield and land use in southwest United States.
In Effects of Man on the Interface of the Hydrological Cycle with the
Physical Environment (Proceedings of the Paris Symposium, September, 1974)
IAHS Publication no. 113, 96-98.

Kite, G.W. (1981) Recent changes of level of Lake Victoria. Hydrological

Sciences Bulletin, 26, 233-243.

50
Knox, J.C. (1972) Valley alluviation in southwestern Wisconsin. Annals of
the Association of Americam Geographers, 62, 401-410.

Lamb, H.H. (1966) Climate in the 1960s: Changes in the world's wind
circulation reflected in prevailing temperatures, rainfall patterns and the
levels of African lakes. Geogr. J., 132, 183-212.

Langbein, W.B. and Schumm, S.A. (1958) Yield of sediment in relation to mean
annual precipitation. Trans. Amer. Geophys. Union, 39, 1076-1084.

Li, Z. (1992) The effects of forest in controlling gully erosion. In Erosion,

Debris Flows and Environment in Mountain Regions (Proceedings of the Chengdu
Symposium, July, 1992), IAHS Publication no. 209, 429-437,

Mahmood, K. (1987) Reservoir Sedimentation: Impact, Extent and Mitigation,

World Bank Technical Paper no., 71.

Meade, R.H. (1982) Sources, sinks and storage of river sediment in the
Atlantic drainage of the United States. J. Geology, 90, 235-252.

Meade, R.H. and Parker, R.S. (1985) Sediment in rivers of the United States.
In National Water Summary 1984, U.S. Geological Survey Water Supply Paper
2275, 49-60.

Milliman, J.D. and Meade, R.H. (1983) World-wide delivery of river sediment
to the oceans. Journal of Geology 91, 1-21.

Milliman, J.D., Qin, Y-S., lien, M-E. and Saito, Y. (1987) Man's influence on
the erosion and transport of sediment by Asian rivers: the Yellow River
(Huanghe) example. Journal of Geology 95, 751--762.

Morgan, R.P.C. (1986) Soil Erosion and Conservation, Longman, Harlow, 298pp.

Mou, J. (1991) THe impact of environmental change and conervation measures on

erosion and sediment loads in the Yellow River basin. In Sediment and Stream
Water Quality in a Changing Environment (Proceedings of the Vienna Symposium,
August, 1991) IAHS Publication no.203, pp. 47-52.

Novotny, V and Chesters, G. (1981) Handbook of Nonpoint Pollution: Sourcesand

Management, Van Nostrand Reinhold, New York.

O'Hara, S.L., Street-Perrott, F.A. and Burt, T.P. (1993) Accelerated soil
erosion around.a Mexican highland lake caused by prehispanic agriculture.
Nature 362, 48-51.

O'Loughlin, C.: L ., Rowe, L.K. and Pearce, A.J. (1980) Sediment yield and water
quality responses to clearfelling of evergreen mixed forests in western New
Zealand. In The Influence of Man on the Hydrological Regime, with Special i
Reference to Representative and Experimental Basins, IAHS Publicaton no. 130,
pp. 285-292.

Oldeman, L.R., Hakkeling, R.T.A. and Sombroek, W.G. (1991) World Map of the
Status of Human-Induced Soil Degradation: An Explanatory Note, ISRIC,
Wageningen.

Oldfield, F., Appleby, P.G. and Thompson R. (1980) Palaeoecological studies

of lakes in the Highlands of Papua New Guinea. I. The chronology of
sedimentation. J. Ecology 68, 457-477.

51
Painter, R.B., Blyth, K., Mosedale, J.C. and Kelly, M. (1974) The effect of
afforestation on erosion processes and sediment yield. In Effects of Man on
the Interface of the Hydrological Cycle with the Physical Environment, IAHS
Publication no. 113, 150-157.

Roberts, N. and Barker, P. (1993) Landscape stability and biogeomorphic

response to past and future climatic shifts in intertropical Africa. In
D.S.G. Thomas and R.J. Allison (eds.) Landscape Sensitivity, John Wiley,
Chichester, 65-82.

Robinson, A.R. (1977) Relationship between soil erosion and sediment

delivery. In Erosion and Solid Matter Transport in Inland Waters (Proceedings
of the Paris Symposium, July, 1977) IAHS Publication no. 122, 159-167.

Rooseboom, A. and Maas, N.F. (1974) Sedimentafvoergegewens vir die

Orange-Tugelaen Pongolariviere, Tech. Report no. 59, Dept. of Water Affairs,
Pretoria, South Africa.

Rooseboom, A. and Harmse, H.J. von K. (1979) Changes in the sediment load of
the Orange River during the period 1939-1969. In: The Hydrology of Areas of
Low Precipitation (Proceedings of the Canberra Symposium, Dec. 1979) IAHS
Publ. no. 128, 459-470.

Ross, D.A. and Degens, E.T. (1974) Recent sediments of the Black Sea. In The
Black Sea - Geology, Chemistry and Biology, Amer. Assoc. Petrol. Geol. Memz
no. 20, 249-278.

Stevens, J.C. (1936) The silt problem. ASCE Trans. 101, 207-250.

Tardy, Y., N'Kounkou, R., and Probst, J.L. (1989) The global water cycle and
continental erosion during Phanerozoic time. American Journal of Science 283,
454-474.

Trimble, S.W. (1976) Sedimentation in Coon Valley Creek, Wisconsin. In:

Proceedings of the Third Federal Interagency Sedimentation Conference, US
Water Resources Council, Washington, D.C., 5-100 - 5-112.

Trimble, S.W. (1981) Changes in sediment storage in Coon Creek Basin,

Driftless Area, Wisconsin, 1853-1975. Science, 214, 181-183.

Trimble, S.W. (1983) A sediment budget for Coon Creek basin in the Driftless
Area, Wisconsin, 1853-1977. American Journal of Science, 283, 454-474.

UNESCO (1978) World Water Balance and Water Resources of the Earth. UNESCO,
Paris.

Ursic, S.J. and Dendy, F.E. (1965) Sediment yields from small watersheds
under various land uses and forest covers. Proc. Federal Interapency
Sedimentation Conference, USDA Misc. Pub. 970, 47-52.

Walling, D.E. (1979) Hydrological processes, In K.J. Gregory and D.E. Walling
(eds.) Man and Environmental Processes, Dawson, Folkestone, 57-81.

Walling, D.E. and Webb, B.W. (1981) The reliability of suspended sediment
load data. In Erosion and Sediment Transport Measurement (Proceedings of the
Florence Symposium, June, 1981) IAHS Publication no. 133, pp. 79-88.

52
Walling, D.E. (1984) The sedimeny yields of African rivers. In Challenges inb
African Hydrology (Proceedingsof the Harare Symposium, July, 1984) IAHS
Publication no. 144, pp. 265-283.

Walling, D.E. (1983) The sediment delivery problem. J. Hydrology, 65,

209-237.

Walling, D.E. (1989a) The erosion problem. International Journal of Sediment

Research, 4, l-11.

Walling, D.E. (1989b) Linking erosion and sediment yield: some problems of
interpretation. International Journal of Sediment Research, 4, 13-26.

Walling, D.E. and Webb, B.W. (1983) The dissolved loads of rivers: a global
overview. In Dissolved Loads of Rivers and Surface Water Quantity / Quality
Relationships (Proceedings of the Hamburg Symposium, August, 1983), IAHS
Publication no. 141, pp. 3-20.

Ward, R.C. (1971) Small Watershed Experiments: An Appraisal of concepts and

Research Developments, University of Hull Occasional Papers in Geography, no.
18.

Wasson, R.J. and Galloway, R.W. (1986) Sediment yield in the Barrier Range
before and after European settlement. Australian Rangelands Journal, 8,
79-90.

Wischmeier, W.H. and Smith, D.D. (1978) Predicting Rainfall Erosion Losses.
USDA ARS Handbook no. 537.

Wolman, M.G. (1967) A cycle of erosion and sedimentation in urban river

channels. Geografiska Annaler 49A, 385-395.

Wolman, M.G. and Schick, A.P. (1967) Effects of construction on fluvial

sediment: urban and suburban areas of Maryland. Water Resources Research 6,
1312-1326.

Ye, Q., Jing, K., Yung, Y. Chen, Y. and Zhang, Y. (1983) Changes of the river
course of the Lower Yellow River with respect to erosion of Loess Plateau. In
Proceedings of the Second International Symposium on River Sedimentation,
Nanjing, China, October, 1983, 597-607.

Zhao, W., Jiao, E., Wang, G. and Meng, X. (1992) Analysis on the variation of
sediment yield in Sanchuanhe River basin in 1980s. International Journal
of Sediment Research, 7, 1-19.
Watershed management in China:
concepts and techniques
Dr. Ding Lianzhen
International Researchand Training Center for Erosion and Sedimentation
Beijing, China

1. INTRODUCTION

Soil erosion aggravatesthe environment and ecology and brings about serious damagesto
economic development. In most countries the seriously eroded regions are also the poorest and
the most underdevelopedareas. Soil erosion results in sterility of land and backwardnessof
productivity, and in turn, poverty and backwardnessoften lead to further deforestation and
excessivereclamation; a vicious circle is thus formed. What is more, the rate of population
growth in poor mountainous regions is usually hardly under control, making the situation more
miserable. A rational approach to solving the problem is to consider resources, environment,
society and economics as a whole, that is, the concept of the watershedmanagementshould be
based on the facts just mentioned. In “Guidelines for watershed management”by FAO, the
watershed managementis defined as “the comprehensivedevelopmentof a basin so as to make
productive use of all its natural resourcesand also protect them” In “Watershed Managementin
India” by Indian National Committee for IHP, the watershed managementis defined as “the
rational utilization of land and water resources of a watershed for optimum production with
minimum hazard to nature resources”. Practical experiencein China is fully in conformity with
these definitions and shows that more detailed principles and workable procedure of so-called
small watershedmanagementhave been well developedand have proved to be quite effective. In
this paper, all the materials are selectedfrom practice in China. Correspondingly, the term
small watershedmanagementis used hereinafter.
In the past, though large amounts of money and man-power were spent on soil
conservationin China, the situation was not improved much and in many places the erosion even
becamemore serious. One of the reasonsis probably that soil conservationworks were carried
out isolatedly. Since the late 1970’s, through successful implementation of small watershed
management,not only excessivesoil losseshave been checked, but also land, water and other
resourceshave been developed in a planned and coordinated way. According to statistics, up to
1989, 846 small watershedstotaling 34,040 km’ were enlisted as priority key-watershedsat state
level, another 5504 small watershedscovering 144,265 km* were enlisted as key watershedsat
provincial, city and county levels. The spread of small watershedmanagementon a nation-wide
scalehas proven to be successful,and vigorously advancedsoil conservationmany stepsforward.

54
 The small watershedrefers to a naturally closed hilly plot covering an area of 5 - 30 km*
or sometimeseven 60 - 80 km*. It functions as a unitary area in soil erosion processand at the
same time it may be consideredas unitary area for resourcedevelopmentin the local society.
Small watershedmanagementin China has succeededin combining soil conservationworks and
small-sized water resources projects and others and can serve to promote local economic
development. A brief introduction to the concepts and techniques used in planning and
managementis presentedin the following.

2. BASIC PRINCIPLES OF. WATERSHED PLANNING

With the aim of a radical change in natural conditions and associatedsocio-economic

structure, from the very beginning the watershedmanagementshould start with a thoughtful and
overall planning. The following aspectsare of great importanceand should be taken as a guiding
principle in the planning.

2.1 Simultaneous implementation of soil conservation works and development of natural

resources
Being the origin of poverty and backwardnessin mountainous regions, soil erosion is
induced and acceleratedjointly by natural and man-madefactors. Soil erosion control measures
play an important role in watershedplanning. Various erosion control measuresprovide the
possibility for resource development, and in turn, economic development will furthermore
promote soil conservationpractice. In fact, soil conservationmeasuresthemselvesfunction as an
indispensablelink for the construction of mountainousregions. So simultaneousimplementation
of soil conservationworks and developmentof various resourcesshould be consideredas one of
the fundamentalprinciples in the planning. The general objective of watershedmanagementis
radically to put an end once and for all to the poverty in theseregions.
Soil conservationworks in China were once solely for the reduction in sedimentyield.
Such a viewpoint can not mobilize local people to actively participate in the work. On the
contrary, if evident economic benefits can be shown directly through development of
mountainousregions, then more people will be attractedto the causeof soil conservationand in
the meantimethe constructionof mountainousregions will be spedup.

2.2 Optimal combinationof ecological and economicsystemsin a watershed

Watershed ‘managementis a complicated engineering system which embodies various

contentsof natural ecology and social economy. The system includes input of natural resources
and investment and man-power, and is strongly influenced by human activities especially the
short-sightedexploitation of local resources. Different combinationsof objectives in ecology
and economy lead up to different modes in the balanceof ecology and economic development.
To searchfor an optimal mode is the kernel of a successfulwatershedmanagement,enabling one
to realize low consumption, high efficiency, stable and protracted economic benefits, i.e., the
interplay among balance and coordination between population and natural resources under
various conditions must be thoroughly dealt with.

55
2.3 Optimal composition of agriculture, forestry, animal husbandryand sideline production in
a watershed

During watershedplanning, focus should be put on the comprehensiveerosion control

and the improvement of ecological environment. The break-through is to start with rational
readjustmentof land uses. Rational utilization of land and water resources will promote the
developmentof agriculture, forestry, animal husbandryand sideline productions in a coordinated
way. In fact, after completion of a well-planned schemefor a watershed,a large portion of the
once abandonedwaste land can be convertedinto productive land resource,a considerablepart of
land area can be better utilized with higher productivity, and, above all, a diversified economy
will be gradually developed and much higher benefits can be achieved. A unification of
economic, ecologic and social benefits can be in the long run realized through watershed
management. Becauseof the great difference in natural and social conditions, the successful
watershedmanagementmay be of quite different types. So the preferential production which will
play a leading role in a designatedwatershedmust be carefully studied,justified and identified in
the planning. Practical experience in China shows that there exist many different types: the
mountainoushorticulture type, the featuredcommodity production base type, the combination of
agriculture, forestry, animal husbandry type, the combination of plantation, breeding and
processingtype, etc.

2.4 Significanceof self-supportingin grain production

In denselypopulated areas food supply used to be one of the most challenging problems
and must be carefully managedin decision-making. Deforestation, excessivereclamation and
overgrazing usually come from food shortages. Concerning land use for grain production,
perhaps,there is no generalrule applicableto all countries. However, as far as soil conservation
is concerned,due attention must be paid to bearablesolution of grain production in the local area.
In the case of China, depending upon the situation of population growth, certain amount of
farmland must be preservedas a prerequisite of implementationof soil conservationworks. In
seriously eroded regions, effects should first be made to the constructionof high-quality farmland
such as bench terrace, dammedfield (reclaimed field of depositedarea behind check-dams)etc.
Unitary grain yield of bench terrace and the like is several times higher then that from
extensively cultivated slope land. Constructionof one hectareof high-quality farmland provides
the possibility to restore several hectares of slope land for afforestation or grass plantation.
Meanwhile small-sizedreservoirs, ponds, and various devices for irrigation purpose should also
be included and coordinatedwith soil conservationworks in watershedplanning to upgrade and
improve conditions for grain production.

Comprehensive planning of a watershed may cover several stages during its

implementation. According to the real situation, certain items may be modified or altered, but
the generalprinciples as describedabovemust be observed.

56
3. WATERSHED PLANNING

Planning of a designated watershed is usually worked out on the basis of previous

researchresults and/or the zonal planning of soil conservation. Nearly in all provinces in China
and for all its major rivers, zonal planning of soil conservation has long been prepared and
supplementedwith time, particularly in the seriously eroded regions. Such kind of zonal
planning, covering an area of severalthousandsto tens of thousandskm’, outlines basic features
of the region, including climate, geology, pedology, geomorphology,topography and also socio-
economic conditions of each zone or subzone. It provides a qualitative but overall picture of soil
erosion in these areas, the extent and peculiarities of soil erosion, its spatial distribution,
suggestionsto erosion control measures,and the macroscopicdecision-makingconcerning the
future developmentof the region. On the basis of systematicstudieson source areasof sediment
yield of all heavily sediment-ladenrivers in China, thirteen rivers and districts have been selected
for the first stageprojects at the state level. With such basic materials in hand, it becomeseasier
to work out a watershedplanning, and then a more detailed outline of study on a much smaller
area will be conducted for the small watershedplanning. Experience shows that the small
watershed managementfunctions as a link between large river basins and land users of small
areas. Small watershed managementis particularly in favor of the implementation of soil
conservationworks in a concentratedarea and in a continuousway. It also has the advantageto
unify administrativemanagementand constructionof hilly and mountainousregions with a higher
efficiency and better effects. In most cases within 5-10 years remarkable progress can be
achieved. Relying upon practice in severalthousandsof small watershedmanagementin China,
the generalmethod and procedureare describedas follows.

3.1 Long-term and short-term watershedplanning

Soil erosion is an old burden left over by history, so soil conservation is a great
undertaking and a strenuous task, which must be well understood by all. Even for a small
watershedof definite area, it may take a considerabletime to performthe task completely. Thus,
watershedplanning is usually carried out by several phasesconsidering the limited amount of
fund available to be allocated to the project and the possible labor forces to be mobilized. The
long-term planning outlines general picture of the watershedand the overall objectives at the
completion of related works, the key engineeringprojects and total amount of funds and man-
power needed. The short-term planning provides more details of the planned items with a
punctual timetable of the implementationof the enlisted works within a certain time period, say,
five years. In a short-term planning the objectives should also be well stipulated. The most
important thing in short-term watershedplan is that all items should be practicable with concrete
aim. A short-term watershed planning can be modified according to the real situation, if
necessary. For example, when certain technology is popularized, the execution can be sped up;
or on the contrary, when some unforseen problems occur, the whole procedure may also be
affected. Experience in soil conservationpractice in the past shows that one should never be
overanxiousfor the results of implementation,otherwise it may discouragethe enthusiasmof the
masses. On the contrary, when drawing a plan one should better leave some margin, so that
when people seethat they have surpassedthe target they can do a better job.

57
3.2 Priority objectivesand constraintsin watershedplanning

During watershedplanning proper selectionof priority objectivesand constraintsis of the

fast importance. Different solutionsmay be obtainedif the decisive conditions are given slightly
different weighted emphases. Nearly in all caseserosion control functions as one of the basic
requirementsin watershedplanning, but it is not the unique one. Other objectives concerning
resourcedevelopmentdepend heavily upon the thorough study of environment. Oftentimes there
exist contradictionsbetweenresourcedevelopmentand environmentprotection, or more exactly,
between expectedoutput and sustainability of environmentalcapacity which is closely related to
population density and other social factors. In watershed planning determination of priority
objectives must be coordinatedwith natural and socio-economicconstraints derived from local
conditions. Thus, selection of priority objectives and constraints in watershed planning is
actually related to the adopted method of planning. If a computerizedapproach is used, say,
linear programming or multi-objective programming, then both the objective and constraint
equations are operated through the given model itself. It means the selection comes from the
outside world and should reflect to the best of the reality. If an experience-basedapproach is
used, then both the priority objective and constraint have to be determined by the planner
himself. In either cases,subjectivity should be excludedas mush as possible.

3.3 Methods of watershedplanning

Watershedplanning can be carried out either by primitive method, the experience-based

approach or advancedmethod, the computerizedapproach. The latter can further be classified
into a) man-computercommunicationapproach, b) linear programming approach and c) multi-
objective programming approach.

3.3.1 Experience-basedapproach

Experience-basedapproach is of qualitative nature and can not provide specific

standard of quantification. This method is easy to handle for a smaller area, especially when
some typical examples of similar conditions have been successfully implemented to serve as
reference. The main procedureof planning by this method includes:

a) Field investigationto collect data of land resources,presentsituation of land uses,

data of soil erosion, existing soil conservationmeasures,etc.

b) Identification of constraintconditions, determinationof future development,

c) Assessmentof productivity of land resourcesfor each of the subdivided small

plots.

4 Stipulation of soil conservationplanning, small-sized reservoirs and other water

conservancyproject planning as well as other resource development planning;
then theseplanning should be coordinated.

58
 e> Estimation of benefits and cost-effectivenessanalysis.

Experience-basedapproach in watershedplanning is simple, directly perceivable

and convenient for popularization, and no complicatedcomputation is needed. However, being a
primitive method, it has the following imperfections: 1) Insufficient precision in the quantitative
analysis. Whether the planned schemeis superior or inferior to other schemecan not be shown
by itself. 2) If the watershed covers a large area or if the natural conditions are very
complicatedand diversified, then the said method is difficult to apply.

3.3.2 Computerizedapproach

Computerized approach has the advantage to apply computer technology and

databank techniques, so that more complicated casescan be managed and quantitative solution
can be obtained instead of qualitative one. As we know, computer technology develops
extremely fast and endlessly. What is describedhere is the most simple application and also the
least expensive one. The difference between man-computercommunication and linear or multi-
objective programming lies in the completenessof optimization and the extent of complexity in
computation. Man-computer communication approach can not reply whether there exists other
alternative better than the obtained one, while linear programming and multi-objective
programming approachesrequire firstly to set up a model to be used in planning, including
selectionof objective functions and constraint equations.

General procedure of man-computercommunicationapproachincludes:

a> Field investigation. Collection of relevant data of both natural and socio-
economic conditions. Specialregulations are issuedfor this purpose.

b) Set-up of a data bank.

c> Computation of economic benefits of the proposed scheme. This will be repeated
for each of the small plots. If the presumed objective can not be achieved, then
the whole plan has to be adjusted.

d) Cost-effectivenessanalysisof the proposed scheme.

e) Final selectionof schemeafter comparisonamong various alternatives(usually 3-5

alternativesare prepared for selection).

When various alternatives are submitted for selection, evidently both economic and
ecologic benefits of the alternatives differ each other. To which should priority be given, soil
conservationbenefits or economic benefits? It remains unsolved and decision has to be made by
the planner’s own experienceand judgement, i.e., subjectivity is inevitable.
In comparison with man-computer communication approach, linear programming has a
strict mathematicalbasis for computation. Various softwares have been developedfor watershed
planning and are widely used in China. This method is capableof performing analysis on usable

59

--
resources within a period of time, including land resources, labour force resource, fund
available, quota of labor force, etc. also the interplay betweenthe designatedwatershedand the
outside world, namely the market and supply of productionsetc. When this method was used in
planning of a watershed “Jixinhe” of Binxian County, Heilongjiang Province, 25 variables were
selected;the overall income of the watershedwas taken as objective function while soil loss was
consideredas a constraint condition, other constraint conditions were: land resource, labor force
resource, food supply and fertilizer supply. When linear programming approachis used, it may
yield much more alternativesfor selection, so some of them have to be screenedout before the
final selection. As there is the objective function to assessthe optimization, theoretically
speaking, the final solution should be the optimal one in the model. But sometimesthe optimal
solution might not be practicable,henceit is advisableto be recommendedafter consultationwith
the user.

Linear programming approach has many advantagessuch as more variables and more
constraints equation can be taken into consideration;but it also has its own limitations: 1) This
method is a deterministic mathematicalmodel, occasionalfactors can not be considered. 2) This
method is fit for solving linear problems, but ecological problems are usually nonlinear. 3)
Optimization is made for a single objective, but in watershedplanning real objectives are more
than one.

Multi-objective programming is a newly developedbranch of the optimization theory.

Application of this method in watershedplanning is just in the beginning stage. Prior to the set-
up of a mathematicalmodel using this method, the following analysisshould first be performed.
a) interpretation of environmental systems: natural environment in the watershed and socio-
economic environment; b) diagnosis of the systems: analysis on soil and water resourcesand
structure of various productions in the watershed,to find out causesof deterioration of ecologic
environment in the systemsof the watershed; c) prediction of systemsin the watershed,such as
prediction of population growth, prediction of grain yield and prediction of market changes,etc.
With thesebasic materialsin hand, the perspectiveof watersheddevelopmentcan be estimatedon
a sound basis. Multi-objective programming involves large amount of computation and can be
carried out only on a medium or large-sizedcomputer. When this method was used in watershed
planning of Baicaogou of Zhongyang County, Shanxi Province, there were 66 variables, 62
constraint equations and 4 objective functions. These objective functions are: Maximum
economic income, minimum soil loss, minimum total investmentand maximum bio-production.
Multi-objective programming approach is capable of generating the needed data with higher
accuracy in quantification to satisfy multiple purpose requirement in watershedplanning. Yet
becauseof the mentioneddifficulty in techniqueswe still have a long way to go before it can be
popularizedfor common application.

3.4 Land resourceassessment

Land resource assessmentfunctions as the basis in watershedplanning. Based on the

assessmentthe adaptability of land uses and the constraint conditions of land resource
developmentare studied, and then type of land use is to be identified for each of the subdivided

60
plot area. The working plan for land resource assessmentgenerally consists of three parts:
description of current land use, collection of socio-economicdata and the relevant analysis and
detailed analysis through field investigation. Quality of land resource together with specific
featureswill be studiedby specializedgroup. Then comparisonbetweenthe present land use and
the potential of land resource is made according to the adaptability, the environmental impacts
and social economic analysis. The final results of land resource assessmentsummarizesthe
following: a) recommendedmode of land use for each of the plots; b) degreeof adaptability of
land use; and c) technicalrequirementto eachplot of the watershed.

3.4.1 Field investigation

Field investigation is carried out according to specific regulations and by-laws

issued for watershed planning by river catchment administrations or local authorities.
Completenessof data collection depends strongly on method of watershed planning. When
computerizedapproach is adopted, more detailed data are preferable. The following items are
commonly included: geomorphologicalposition, present land use, elevation, soil type, slope
direction, slope gradient, depth of top soil, speciesof vegetative cover, bed rock, type of soil
erosion, severenessof soil erosion, etc. Sometimes field investigation is carried out in
combination with remote sensingtechniques,including air photos.

3.4.2 Classificationof land resourcequality

Items of data collection should relate to land resource quality. For example, in
most cases,geomorphologicalposition (hillside or flat gully bed) together with slope gradient and
the present land use can represent the overall land resource quality. In order to facilitate
assessmentof various conditions in a watershed,basedon Quantification Theory, an eight-grade
classification is adoptedin the watershedplanning for Zhuanghugouof Beijing. Grade I refers to
irrigated paddy field on floodplain, grade II - reclaimed farmland on depositsbehind check-dams
in gulled areas, grade III - bench terrace at hill foots, usually for nursery, grade IV - slope land
for orchards or grass plantation, grade V - slope land for afforestedeconomic forest, grade VI -
slope land for shrubs or natural forest, grade VII - slope land, barren and waste, grade VIII -
gully bed with outcrops, or covered by gravels and boulders. Then different weighted
coefficients are given to indice describing quality of land resources,high values meansthe land is
of higher quality and vice versa. Finally, a comprehensiveweighted quantification is worked out
for the assessmentof quality of land resourcein the watershed. Making use of this method, final
results of land resource assessmentin Zhuanghugouwatershed show that there is limited land
resource adaptableto agriculture, but the potential of developmentof orchards and economic
forest is encouraging. Hence it should be consideredas the main production in future.

3.4.3 Analysis on economic structurein watershed

In agricultural economy the current land uses can directly reflect the economic
structure of a watershed. Agriculture, forestry, animal husbandry and sideline production are
commonly the main productions in hilly and mountainous areas of a watershed. Taking the
mentionedfour productions as independentvariables, one can detect the correlation betweeneach

61
of’the four variables to the total output of the watershed. A remarkablehigh value of correlation
in agriculture but much lower values in the rest indicate the feature of monotonousagricultural
economy which means the preferential potential of land resources have not yet been duly
developed. As a matter of fact, ecological systemand economic systemin a watershedis closely
interrelated and interacted to form a compound system. Irrational developmentof agriculture
will undoubtedly lead to unbalancedecology. On the basis of land resourceassessmentand the
analysisof economic structure in a watershed,orientation of land use adjustmentcan be made.

4. Soil conservationmeasuresin watershedmanagement

Soil conservation measuresfunction as the essentialpart in watershed management.In

order to maximize comprehensiveeffects, conservation measures are usually applied in an
integratedway, i .e . , a combination of engineering(structural) measures,biological measuresand
tillage practice.
Engineering measuresinclude bench terrace on slope land, check-damsin gullies, gully
head protection works, small-sizedreservoirs and other hydraulic structures. Detail construction
of such works can be found in soil conservation manuals and in handbooks for hydraulic
engineers. Engineering measuresare more expensive and labour consuming than biological
measures,but they can fast respondto erosion control. In semi-arid regions in China where most
of the annual precipitation is concentratedwithin a few big storms and the rest part is not enough
to grow plants, bench terrace, alternate terraced field with slope . . . etc. play important role
against erosion on slopes in these areas. Construction of check dams and sediment barriers
becomes indispensablemeans in the compound project of soil conservationpractice. As to a
specific type of soil erosion, the slope disintegration, which occurs in heavily weatheredgranite
regions with heavy rainfall, specially designedengineering structure plays irreplaceable role in
the protection againsterosion.

Gully beds usually have higher moisture content, it provides better conditions for growing
plants and sometimes even for reclamation purpose. Using engineering measuresto convert
farmland on slopesinto bench terrace and to build up check-damsand sedimentbarriers in gullies
is usually an important constituentin watershed planning under certain conditions, despite of its
shortcomingof labor-consumingin practice.

Biological measuresrefer to arbors, shrubs and grassplantation, including closing hillside

and slopes, afforestation, etc. If climate and soil conditions are favourable, closing hillsides and
slopes is less expensivethan other measures. Selectionof plant speciesis of great importance,
particularly in seriously eroded areas. For extremely sterile soils such as the coarse-grained
laterite formed by weatheredgranite, the first step is to introduce drought-and sterility-resistant
and fast-growing speciesas pioneer plant, then gradually theseplant specieswill be replaced by
other speciesof higher economic value. Besides, attention should also be paid to community
designationso that a multi-layer protection of vegetativecover can be formed to achieve a better
effect. Another important aspectconcerning selectionof plant speciesis to unify ecologic effect
with economic benefits. For example, a number of herbal medicinesand fruit specieshave been
planted as biological measuresin the wet regions in China. As’a result, economic benefits
increaseremarkably. Also the Hippophae rhamnoidesL. has been found adaptableto dry and

62
 k-cc;? terrace

F ig 1. Profile of alternateterracedfield with barrensIope

_Check dam
I

-se---

----

F ig 2. Profile of cascadecheck-dams(sedimentbarriers)

F ig 3. Profile of comprehensive
measuresagainstslopedisintegration

63
I
I
I
I
I
I
I
I
I
I
I
I I
I I
I
I I
I I I I
I
I
I
I, CtX.+d I
I Ikonomic forest 1
, ZUKj Odtird 1
I_-
sterile soils such as loess, and its seedsand fruit are of high economic value. Popularization of
the wild plant in loessregions has achievedgreat success.

Cultivation on slope land is the source of sediment yield in vast eroded areas.
Conservation tillage practice is mainly adopted as remedial measure in such cases. Tillage
practice with conservation purpose includes: contour farming - furrows are ploughing along
contours; contour belt intercropping - belts are aligned along contours intercroppedwith different
combination of crops; ridge and furrow farmland - soil ridges and furrows are aligned alternately
along contours with traverse lower ridges at 1 - 3 m spacing to prevent transverse erosion;
contour ridge-furrow - ridge furrows of 80 cm wide and 40 cm high are constructed along
contours, furrows are designedwith proper gradient to drain excesswater and to prevent erosion
as well; pit planting - pit 70 cm in diameter and 50 cm deep dug in a staggeredpattern on
hillside, filled with soil containing fertilizer before planting; rotation cropping of grain and
fodder, etc. In order to provide favorable conditions for tree plantation, certain forms of land
preparation are needed, such as narrow terrace for afforestation, horizontal ditches for tree
plantation, etc. All thesemethodshave been widely practisedand proved to be effective.
Various combination of the above-mentionedmeasurescan be adopted according to the
local conditions. Examplesare given in the following casestudiesin section6.

5. SOME ASPECTS CONCERNING IMPLEMENTATION AND MANAGEMENT OF

SOIL CONSERVATION WORKS

5.1 Setting up demonstrationprojects or plots

Prior to large-scale implementation of watershed managementover large areas, it is

advisable first to concentrate funds and man-power to build up a few typical demonstration
projects or plots. In addition to its own specified tasks, each of the projects or plots has the
responsibility of demonstrationand extension. Then, experiencegained in theseprojects or plots
can be well summedup for popularizationin large areaswith efficient results. Even for a single
project of watershedmanagementit is meaningful in doing so. For example, the Zhuanghugou
watershedof Beijing covers an area of 85.21 km*. The comprehensivewatershedmanagementof
Zhuanghugouwas started in 1980. Prior to start the construction in full scale, two typical small
plots covering 0.63 and 5.36 km* respectively were selectedas demonstrationplots. Significant
economic benefit in three years and remarkably improved ecologic environment convinced
people that historical burden of poverty and backwardnesscould be overcome and the concrete
experiencegained in technical aspectsgreatly promoted the execution of watershedplanning.
The project of Zuanghugouwatershedmanagementachievedgood successwithin a short period
of ten years. (see6-l casestudy of Zhuanghugouwatershedmanagement)

5.2 Publicity and education

Being a basic constituent in the construction of mountainous regions, small watershed

managementalso leads to profound changes in conceptionsand habits against the traditional
backward ideas such as excessive reclamation on slope land, overgrazing and reckless

65
deforestation. In order to acquire supports from the masses,education and publicity play an
important role. Soil conservationhas to get support not only from relevant agenciesof water
conservancy,forestry, animal husbandryetc, but also from educationand scientific and technical
departments. To spread primary and middle school education is a long-term task, yet it is a
fundamentalstep to get rid of ignorance. In addition, to make use of various meansof publicity
such as pamphlet, exhibition, blackboardnews etc. especiallythe afore-mentioneddemonstration
plots in neighborhoodcan strongly dispel various misgivings and establishone’s confidence. On
the other hand, systematic training coursesfor transferring advancedtechniquesand know-how
to professional staff of soil conservationinstitutions at different levels are of first importance,
becausethe implementationof watershedplanning relies, in the final analysis, directly on their
efforts.

5.3 Perfection of management

As mentioned before, soil conservation is a strenuous and protracted task. It is not

occasionalthat along with the implementationof soil conservationworks, some completed works
suffer destruction, especially the reckless deforestation. So equal emphasis must be paid to
construction and protection, that is to the perfection of management,including legislation of laws
and relevant regulations. All kind of deforestation,irrational reclamationon slope land should be
strictly prohibited. So far as resource development is concerned, timber logging must be
supportedby renewal plan of afforestation sites and detailed concretemeasures. Industrial and
transport construction should minimize the destruction of vegetativecover. Waste disposal site
should be well treated and rationally located. Denudedground surfaceof construction site should
be well treated after the completion of construction. In China the policy “whoever causeserosion
is responsiblefor its rehabilitation” “whoever makesdamagesis obliged to compensate”has been
duly executednation-wide. The compensationfee is used in the prevention and control measures
in situ. Good results have been obtained.

5.4 Form of watershedmanagement

The main form of small watershed managementin China is the household contract
responsibility system. Since the beginning of 1980s’) the rural reform in China characterizedby
householdcontract responsibility systemhas been formed and spreadover the whole country. It
resulted in unforseenprogressin grain production and eventually led to a radical change in rural
economy. Similar concept was applied to the implementationof small watershedmanagement.
In 1982 the first household contract responsibility system appearedin Hequ County of Shanxi
Province. Not long after, this form was spreadto the whole loess plateau and in a nation-wide
scale. The signed contract is usually defined to be valid in 50 years. Contractors are obliged to
fulfil1 certain amount of soil conservationworks and to be in charge of the contractedplot and at
the sametime have the right to take out the merit from the plot. Quality of the completed works
are under supervisionof professionalgroup or personnel. Only after the check and acceptanceof
the completed works, the contractor can get the monetary subsidies from the local authority
according to the contract. Successorof a contractor has the right to make use of the plot as
heritage. Through the contract farmers and the small watershedare closely connectedtogether.

66
Being responsible in management,the farmer householdhas all the right to get the economic
benefits. Thus, it effectively motivate the enthusiasmof the massesand shows strong vitality.

On the basis of householdcontract, diversified forms of the implementationof watershed

management have developed to adapt different conditions. These forms are: combined-
household contract, professional team contract, collective contract, and land resource
development joint-stock company etc. This provides favorable conditions to promote soil
conservationas well as land resourcewith higher efficiency and better results.

6. CASE STUDIES

The following case studies show that under different natural and socio-economic
conditions a well-planned schemeof protection system can stop the excessiveerosion and bring
about considerableeconomic benefits. Even for such extremely difficult conditions as the loess
region and the “red desert”still one can succeedin changingthe unfavorableenvironment.

6.1 Zhuanghugouwatershedmanagement

The Zhuanghugouwatershedcovering an area of 85.21 km* is located at 40” 30’ - 40”

54’ N, 116” 38’ - 116”42’ E in Beijing City. The watershedlies in the earth-rocky region of
Yanshan Mountain Chain, at an elevation ranging from 278 to 1501 m. The watershed is
characterizedby undulating landscapewith densely distributed steep gullies. Gully density is
4.38 km/km*. Main streamin the watershedis 3 1.25 km long with a slope of 1.5% . Slope land
steeper than 26” makes up 67.85% of the total surface area. Outcropping strata belong to
volcanic clasolite of Jurraissic System,gneissand granite of ArchaeozoicEra. Parent material of
surface soil is mainly weatheredcoarse-grainedgranite and gneiss. The wide spread surface soil
belongs to mountainousbrown soil, leaching brown soil and carbonite brown soil. More than
20% of ground surface is occupied by outcrops, 60% is coarse-grainedsoil and mixture of soil
and fragment. As to natural plant species,there grow Quercus dentada, Platycladusorientalis,
Vitex negundo, Ziziphus jujuba var spinosa, Rhamnusparvifolia, Lespedezabicolor, etc. on
sunny slopes, covering 40 % . On the backside of sunny slope there grows afforested Pinus
tabulaeformis. Zhuanghugouwatershedbelongsto continentalmonsoon-affectedclimate, annual
meanprecipitation amountsto 5 11 mm, which is concentratedin flood season. Storms and heavy
rains frequently occur. Annual mean temperatureis 9” C with accumulatedtemperatureof 3900”
- 4050”C. Frost-free period lasts 160 days. Annual evaporationis 1200 mm. Eroded area in the
watershedtotals 73.26 km2, accountingfor 86%. About 30% of farmland is located on slopes
and results in averagerate of erosion of 1456 t/km*. Main form of erosion is sheeterosion, but
gully erosion and debris flow also occur. Population density is 43 person/km2. Local people
have long suffered from frequent drought and flood disastersaccompaniedwith increasingthreat
of soil erosion. There is also a shortageof food and fuel supply, 30% of the local inhabitant has
difficulties in drinking water supply.

In Zhuanghugouwatershedmanagement,soil conservationmeasuresconstitute the main

part, and at the same time, according to land resource assessmentstress is put on the
developmentof forest resources. Basedon the natural conditions, three subregionsare classified
in accordance with elevations: the upper region of high elevation with steep slopes, the
intermediate hilly region of undulating relief and the lower region to be developed into
agricultural zone. In the upper region, closing hillsides in combination with aerial sowing of
trees and nursing shrubs is carried out to form the first protective line. In the intermediatehilly
region, check-dams, sediment barriers, small-sized reservoirs and ponds are constructed in
addition to large-scaleafforestation of economic trees and orchards to form a secondprotective
line for flood control and erosion control. These measuresalso serve to develop forestry as the
main production in the watershed. In the lower region main efforts are given to convert slope
land into bench terrace to develop paddy field and also to construct protective embankmentfor
the safety of villages and farmland.

During the watershed planning multi-objective programming method is used. The

objectives are: maximum profit, minimum investment,minimum soil loss and maximum biomass
production in the watershedsystem. The drainage area of Zhuanghugouis subdivided into 949
plots and a data bank is set up for planning and management. Assessmentof land resource
quality has been thoughtfully studied using the Quantification Theory III. Eleven indices are
used to describe land resource quality referring to each of the eight grades. Then scores are
given to each of the 949 plots using different weights. To maximize the correlation betweenthe
score and the given weight yields the final groups of samplesfor a quantified analysis. Basedon
the results of land resourceassessment,the final watershedplanning is worked out.

Large-scaleimplementationof the watershedplanning was started in 1980. By the end of

1990 most of the planned items had been carried out. A total of 67 64 km* has been put under
control, making up 92% of the original erodedarea.

closing hillsides 5.00 km*

afforestation 44.88 km*
economicforest 6.16 km*
grassplantation 3.58 km*
upgrading shrub forest 5.00 km*
reclamationon dammedfield 40.00 ha
constructionof check-dams 8317 sets
constructionof pumping stations 20 sets

Comparisonof land usesin 1980 and 1990

Farmland Forest land Grazing land

307.27 ha 1885.13ha 2182.60 ha

311.80 ha 4426.40 ha 2026.70 ha

68
 Proportion among various land uses was 1:6:7 in 1980, it changedinto 1:15:6 in 1990,
which is close to the plannedoptimal proportion of 1: 19:3.

Comparisonof annual income from various production, in Yuan RMB

-
Agriculture Forestry Animal Sideline Total
husbandry production

392,000 41,600 320,000 813,000

1990 758,200 362,400 639,400 1,750,OOO 3511,000

Proportions of income among various productions was l:O.l:O. 15:0.82 in 1980, it

changedinto 1:0.48:0.84:2.30 in 1989, i.e., the agriculture increasesto 1.93 times, the forestry -
8.71 times, animal husbandry - 10.87 times, sideline production - 5.47 times, while the total
income increasesto 4.3 1 times. A diversified economyis gradually developing.

In short, watershedmanagementin Zhuanghugouachieved a preliminary success. The

overall vegetativecover increasesfrom 45 % to 80%) the rate of soil erosion decreasesfrom 1456
t/km* to 348 t/km*. Living standard of local resident also improved. The per capita annual
income increasesfrom 104.9 yuan to 591 yuan. They no longer worry about food and fuel
supply, the drinking water supply problem has been solved too.

6.2 Sunjiachawatershedmanagement

Sunjiachawatershedcovering an area of 42.08 km2 is located at 35” 50’30” - 35” 55’ 10”
N, 104”20’ 38” - 104”26’20” E in GansuProvince. The watershedlies in the gulled-hilly loess
region with elevations ranging from 1885 to 2310 m. The geomorphological feature of the
watershedis characterizedby high plateauwith deeply cutting gullies, 97% of the watershedhas
an elevation higher than 2000 m. Gully density is 1.88 km/km*. The parent material is loessial
sand and the surface soil is grey calcareoussoil with pH-value of 7.8-8.5. Depth of loess strata
ranges from tens to 300 m. Becauseof irrational reclamationon loess regions for thousandsof
years soil fertility has greatly degenerated. There is scarcely scatteredvegetativecover of Stipa
bungeanaTrin, S..Artemisia frigida etc. Main tree speciesare Uhnus punila L. Prunus sibirica,
Caraganakorshinskii Kom. Hippophaerhamnoidessubsp, Sivensis, AstragalnusadsurgensPall,
Medick sativa L., Melilotus albus Desr. Etc.

The watershedis of continental climate of semi-arid region, annual precipitation is 364

mm which usually concentratedin a few heavy rain and storm events, resulting in heavy soil
loss. Rate of erosion is 8300 t/km*. Population density of the watershedis 49 persons/km2.
Relatively speaking, the per capita farmland of 0.85 hectare is deemedabundant, yet Sunjiacha
used to be classified as underdevelopedregion, local people suffer from drought and poverty.

69
 Sunjiachawatershedcovers two parts: the upper and the lower parts. The upper part of
Sunjiacha covers an area of 29 km*, mainly consists of ridges and mounds with gentle slopes
while the lower part covering an area of 13.08 km2 consists of deep gullies with steep slopes,
being mostly waste land and waste gullies. In the watershedplanning, the guiding principle is to
construct protective system as a whole, meanwhile to convert farmland on slopes into bench
terrace and alternateterrace with slopes,as well as to carry out large-scaleafforestationand grass
plantation. The planning was performed by man-computercommunicationmethod. In order to
evaluateland resourcequality of the watershed,all the existing agricultural, forest and grass land
have been classified into 4 grades, lst, 2nd, 3rd, and not adaptable. Then the readjustmentof
land uses was worked out. Implementation of watershed planning was started in 1983. A
continuousand concentratedbench terrace covering 541.7 ha was built up not far from residence
of inhabitant. Closing hillsides and slopeswas carried out on remote areas,grassplantation was
implemented on waste hills and slopes. By the end of 1988, accumulated controlled area
amountedto 24.89 km2. The completedworks include

Bench terrace 6.76 km2

Afforestation 5.02 km*

Closing hillsides and slopesand grassplantation 12.39 km*

Construction ofchech-dams 290 sets

Construction of large-sizedcheck-dams 4 setswith overall storage

capacity of 564,800 m3

Gully head protection works 8 sets

Upon the completion of theseworks, proportion of various land uses of the watershedchangeda
lot.

Comparisonof land uses in 1983 and 1988

Farmland Forest land Grassland Waste land

1983 1769.2 ha 64 ha 37.5 ha 1595.5 ha

1988 1631.7 ha 502 ha 1233.9 ha 423.7 ha

70
 Comparisonof annual income from various production, in Yuan RMB

-
Agriculture Forestry Animal Sideline Total
husbandry production

1983 466,873 182,923 771,178

1,123,852

Remarkable progress has been achieved in the reduction of sediment yield, the rate of
erosion dropped to 380t/km2; the fuel supply problem has also been solved and the living
standardof local resident was mush improved. Total income increasedby 1.46 times within a
short time period of 6 years.

6.3 Tangbeihewatershedmanagement

Tangbeihewatershedcovering an area of 16.38 km2 is located at 26” 13’00” - 26” 13’

30” N, 115” 14’00” - 115” 18’20” E in Jiangxi Province. It is known as one of the most
seriously eroded regions in laterite zone originated from weatheredgranite. Geomorphologyof
the watershed is characterizedby undulating hilly area of low elevation of 130 - 300 m. The
outcrop strata belong to migmatite of granite of Simian Period and coarse-grainedporphyritic
biotite of YanshanOrogeny of CretaceousPeriod. Strongly weatheredcrust ranges from several
meters to tens of meters deep. After the destruction of vegetative cover for a long time,
weathered material suffers intensive erosion with an annual rate of up to 13500 t/km* in
localities, i.e., soil loss of surface layer amountsto 10 - 13 mm per year. Besides, large-scale
slope disintegrationalso occurs. As both the A and B horizons are completely lost, there remains
only parent material of coarse sand particles and fragments of broken-rocks; the surface layer is
in lack of organic matter and has poor water holding capacity. Trees and even grass grow with
difficulty. According to field investigationheld in 1980, eroded area accountsfor 99.8 % of the
total hilly slopes, areas subject to intensive and strong erosion make up 66.7 % and 15%
respectively. Before the watershedmanagementwas started, productive land resource made up
24.7% while waste barren land occupied 58% of the total. Population density of the watershed
was 195 persons/km*with very low grain yield of 196.3 kg/person. The watershedbelongs to
the wet monsoon-affectedsubtropical zone with annual precipitation of 1411 mm, annual
evaporation- 992.4 mm. Annual mean temperatureis 19.5” C, frost-free period lasts 284 days,
accumulated temperature is as high as 6649” C. Becauseof the soil condition there grow
scarcely Pinus massoniana,Schima superba,Symploeospaniculataand Arundinella Setosa,etc.
Local resident suffers acute shortageof fuel, fodder, waste manure and timber. The extremely
serious soil erosion in Tangbeihewatershedwas causedby both natural conditions and man-made
factors, and the latter was even more important, for example, the inadequateadvocacy of
converting forested land into tea garden, the recklessdeforestationand also the sharp population

71
growth. The situation remained unchangeduntil 1980 when the watershed managementwas
started.

The watershed planning was composed of six parts, namely, soil conservation works,
water resource development projects, scheme of agricultural development, road construction,
planning for rural energy supply (fuel supply), and setting up experimental plots for runoff
observation and meteorological observation. Considering the extremely stern conditions in the
watershed,the emphasiswas put on the adaptabilityof combination of engineeringmeasureswith
biological measures. In intensively and strongly eroded regions engineering measures are
indispensablefor erosion control on slopes so as to create prerequisite conditions for growing
plants. Experience shows the selection of fast-growing speciesor pioneer species is of vital
importance. Lespedezaformosa, Gardeniajasminoides etc are recommended. For moderately
eroded regions refilling of broad-leaf specieswas adoptedand for slightly eroded regions, closing
hillside and slopes were effective, meanwhile, construction of check-dams,small reservoirs and
ponds were carried out for gully control. Special treatment was performed to deal with slope
disintegration. By the end of 1989 a great part of soil conservationworks had been completed.

Afforestation and grassplantation 975.50 ha

(including mingled forest of conifer 834.90 ha
and broad-leaf species)
Closing hillside 156.70 ha
Economic forest 119.00 ha
Timber stand 21.53 ha
Nursery 18.80 ha
Check-dams 18 set
Pondsof various sizes 14
Treatmentof slope disintegration 3

In addition, irrigated farmland increasesby 40.00 ha and improvement of irrigated farmland

increasesby 41.30 ha. The above-mentionedmeasureshave been proved to be quite effective, a
more rational land use schemehas been realized and some of once abandonedwaste land has
been convertedinto productive resources

Comparisonof land usesin Tangbeihewatershed

Farmland Forest land Grassland Waste land

1979 326.40 ha 85.30 ha 82.10 ha 950.00 ha

1984 392.40 ha 670.00 ha 157.13 ha 327.40 ha

72
Proportions among various land uses changedfrom 1:0.25:0.25:2.9 in 1979 into 1:1.7:0.4:0.83
in 1984. Besides,infrastructure of agriculture has also beenchanged.

Farmland for grain crop Farmland for cash crop

1979 308.70 ha 17.70 ha

1988 305.50 ha 68.90 ha

Despite that farmland for grain crop was reducedby 3 ha, grain yield increasedby 46% due to
improvementof land quality and alleviation of drought and waterlogging.

Comparisonof annual income from various productions

Agriculture Forestry Animal Sideline Fishery Total

husbandry production

1979 185,935 10,366 1,869 26,752 1,773

1984 348,482 9,398 105,897 78,997 7,026 549,891

1988 708,304 127,337 538,608 250,780 8,012 1,633,041

During the period of 1979 - 1988, the total income increasedby 7.2 times, the per capita
income increasedfrom 43 yuan to 320.9 yuan. As to environmentalchanges,according to field
survey, the averagesedimentdischargeduring 1966 - 1980 was 53,098 m3, it dropped to 10,822
m3 during 1988 - 1989 with a reduction of 79.6%. Form 1966 to 1980 the river bed of
Tangbeiheraised by 0.96 m, the averageaggradationwas 6.4 cm per year. As a responseto the
completedworks on the watershed,the river bed lowered down by 0.235 m in July 1989, with an
average annual rate of degradationof 2.6 cm. Rate of soil erosion of the watershed decreased
from 4367 t/km* to 1097 t/km*, the increasingtendency of serious soil erosion began to stop.
Besides, the local micro climate showed some changesin forested area, the relative humidity
increasedand ground surface temperaturedecreased. One of the most encouragingeffects was
the appearenceof secondarysoil layer of 0.5 - 1.5 cm deep due to the existanceof fallen leaves.
Organic matter increasedby 8 times. Nitrogen content increasedby 11 times. Wildlife species
increasedtoo.

73
6.4 Mengpu River watershedmanagement

Mengpu River watershedcovering an area of 69.43 km* is located at 26” 22’39”- 26”
39’52”N, 105”44’52” - 105”51’55” E in Guizhou Province of southwesternChina, inhabited by
minority nationality. The watershedlies at an elevation of 1091.5 - 1793.8 m. Mengpu River is
12.1 km long, a third order tributary of the upper Yangtze River. It is characterizedby high
mountains and deep gorges with Karst formation, 85% of the drainage area is of limestone.
Surface soil is composed of lime soil, yellow soil, paddy soil and yellow brown soil, all
originated from limestone and shale. On steep slopes the cultivated layer is rather thin and
petrification is going on. Owing to the favorable climate conditions the watershedused to be
luxuriantly forested with thick plant communities. However, becauseof reckless deforestation
the primary forest has been destroyedcompletely, there exists only secondaryforest of Ailanthus
altissima, Catalpabungei, Cunninghamialanceolata,Betula platyphylla suk and Populus etc.

The watershed is located in the wet region of monsoon climate in subtropical zone.
Annual precipitation is 1450 mm with frequent storms. Mean annual temperatureis 14.5” C with
the annual accumulatedtemperatureof 5311”C. Frost-free period lasts 280 days.

According to field investigationheld in 1983 the eroded area totaled 52.4 km*, making up
75.5 % of the drainagearea. Intensively eroded area occupied 15% , strongly eroded area - 33 % ,
moderately eroded area - 21.7 % , and slightly eroded area - 28 % . The overall rate of erosion in
the watershedwas estimatedto be 3 117 t/km2. The watershedplanning is featuredby large-scale
afforestation and closing hillsides to the recovery of vegetationover the whole watershed,on the
basis of self - supporting in food supply. During 1982 - 1987 considerable progress was
achieved. the completedworks included:

Closing hillside and slopes 27.97 km*

Afforestation 26.20 km*
Farmland convertedinto afforested 5;07 km*
and grass land

Up to 1988, vegetative cover increasedfrom 22.17 % to 55.4 % in the watershed, and

vegetationarea reached98% of the total plantable area. The growing stock was estimatedto be
24,812 cu.m. During this period, 4 ponds and 203 check-damswere constructed,51 projects for
drinking water supply were build too. The existing farmland has been improved. Comparisonof
land uses in Mengpu River watershedis shown as follows:

Farmland Forest land Grassland Waste land

1982 3542 ha 732.7 ha 1088.2 ha 1580.0 ha

1987 2893 ha 920.0 ha 1123.0 ha 1411.6 ha

74
 Annual income from various productions:

Agriculture Forestry Animal Sideline Fishery Total

husbandry production

1982 623,000 56,200 289,300 71,700 - 1,040,000

1987 2,090,OOO 142,000 930,800 250,100 18,000 3,546,800

Total income increasedby 3.41 times, the per capita income increasedby 4.25 times.
The rate of erosion reduced from 3 117 t/km* in 1982 to 851 t/km2 in 1988. As a result, river
bed of the Mengpu River degradedby 84 cm on average.

REFERENCES

DING Lianzhen, 1990, Erosion and sedimentationproblems in China, Proc. of the First
Colloquium, Regional Training Programmeon erosion and sedimentationfor Asia (RAS/88/026)
p. l-24

FA0 Conservation guide, Guidelines for watershedmanagement,1977, Food and Agriculture

Organizationof the United Nations, Rome

GUO Tingfu, 1988, Soil and water conservationin China, Lecture note for Regional Training
Course on Soil Erosion and Its Control, IRTCES

GUO Tingfu, FANG Huarong and XIANG Yuzhang, 1988, Managementof Soil and Water
Conservationon Small Watershedin China

Hydcom Report, 1981, Watershedmanagementin India, India national Committee for IHP,
New Delhi

LI Huaifu, 1989, On the theory and method of small watershedmanagement,1989. Acta

Conservationis Soli et Aquae Sinica vol. 3 No.3 (in Chinese)

LI Yaquan, WANG JIULI and BI Xiaogang, 1990, Land resource assessmentand optimal
structural of land uses in Zhuanghugouwatershed,Compilation of comprehensive soil
conservationmeasuresin ZhuanghugouWatershed Management,Haihe River Conservancyand
Beijing Municipal Bureau of Water Conservancy(in Chinese)

LIU Shanjian, 1993, Soil conservationin China, In: QIAN Zhenyin (ed) ” Water Conservancy
in China” (under print)

75
JIN Zhengping, Shi Peijun, HOU Fuchang and ZHAO Huanxun, 1992, Model for soil erosion
systemin HuangfuchuanRiver Basin and its comprehensive management, Ocean Press.
(in Chinese)

Lanzhou Experimental Station for soil conservation and Headquarters of construction of

Sunjiacha watershed, 1990 , Compilation of documents for the acceptance of Sunjiacha
WatershedManagement. (in Chinese)

Bureau of soil conservation, Yangtze River Conservancy Commission, 1991, Report of the
Comprehensivemanagementof Tangbeihe Watershed, Compilation of documentation of Trial
Small WatershedManagementin YangtzeRiver Catchment1991

Bureau of soil conservation, Yangtze River Conservancy Commission, 1991, Report of the
Comprehensivemanagementof Mengpu River Watershed, Compilation of documentation of
Trial Small WatershedManagementin Yangtze River Catchment1991

76
The application of geographic information
systems to soil conservation strategies

W. SUMMERI) & E. KLAGHOFERZ)

Austrian IAHSICASVRI) & ICCE*)Working Group,
c/o 1)Technical University Vienna, Institute for Hydraulics, Hydrology
& Water Resources Management, Karlsplatz 13/223, A-1040 Vienna,
Austria
c/o 2)Federal Institute for Land and Water Management Research
A-3252 Petzenkirchen, Austria

ABSTRACT A very important factor in soil conservation became the

application of Geographic Information Systems (GIS). Some aspects on
the advantages as well as limitations of vector based GIS and raster
based GIS will be presented in this paper. Only the combination of a
GIS with a hydrological model (soil erosion model and/or nutrient
transport model) provides a decision supporting tool for the analysis of
spatial processes in watersheds.

INTRODUCTION

The increased human impact on soils demands improved prospective in soil

conservation strategies. The instalment therefore of information systems related to this
topic becomes necessary. The final purpose of these systems lies on one hand in the
assessment of the erosion risk (erosivity/erodibility) of soils and on the other hand
increasingly on the qualitative and quantitative determination of non point source
pollution.
Considering the region of interest, such as a complete catchment, to be split into
an arrangement of a certain number of non-overlapping geographical units. Every basic
geographic unit can be characterised by a set of hydrological, environmental, etc.
spatial attributes. Full and complete knowledge about the attributes for each unit is
assumed.
The development of a soil conservation strategy has to be based on a number of
physical characteristics. The suitability of a geographic unit can be defined as a
measure of the physical capacity of a specific location to support a specific land use
(Anderson 1987). However, this is based upon physical factors, describing the
characteristics of the watershed and its land use - slope, soil type, vegetation,
meteorological and hydrological factors, etc. To calculate ratings as functions of these
factors and underlying processes, a number of models have been proposed (Hopkins,
1977; Chapin & Kaiser, 1979; Anderson, 1987; Diamond, 1988; Klaghofer et. al.,
1992 and 1993).
To reach these goals, besides the installment of efficient conservation strategies,
the following steps can be outlined:

77
. The collection, storage and management of soil related data and information to
be used in the determination of suitability for the intended land management use
- usually for an overall watershed consisting of a summary of identified
geographical units with typical classified properties.
. The application of procedures, techniques and models to perform either statistical
based, deterministic or physical based predictions for different scenarios of
management strategies within this watershed. This evaluation and analysis of
geographic units is based on the collected, stored and managed data and
information
. The implementation of detailed conservation strategies for this watershed with its
geographic units, according to ecological conditions as well as legal regulations.
To cvaluatc the soil erosion consequences of different management strategies,
modelling techniques as well as modelling tools are needed to combine available areal
physical information in order to gain the above mentioned results on the loss of
soil/nutrient/pollutant.

GEOGRAPHIC INFORMATION SYSTEMS (GIS)

Soils

Land Cover

Topography

Roads and
Streams

Field
Boundaries

Hiahlv

I;IG. I Spatial information portraying such themes as soils, topography,

land cover, roads, streams and field boundaries are among those data
layers typically used in a GIS for analysing spatial data and producing
interpretative maps and related tabular data.

Geographic Information Systems (GE) provide such a practical tool/solution for

handling data, varying in space and time. They allow for the assignment of specific
land use activities to the most suitable available land units, based on appropriate
measures of suitability. As distinguished from the widely-studied facility location
models which treat certain locations as points in a plane or along a network, GIS are
computer systems (hardware and software) to record, analyse, represent and describe

78
areal!spatial information in a two dimensional or three dimensional manner. GIS are
so called “hybrid systems”, consisting of a
. software for dealing with x, y (and z co-ordinates) + three dimensional digital
terrain model (3D-TGM).
. database system to manage and show the data as well as the expert information.
Fundamentally different types of GIS technology exists and recent years have seen a
dramatically increase in their development and use for managing area1 information.
This can either be done by vector based GIS (information is defined and manipulated
as a vector) or raster based GIS. The later represents the most straight forward
technology for topics based basically on spatial information - spatial database (Fig. I ).
By classifying physical parameters/properties and assigning them to the
appropriate grid cells - the region of interest is subdivided into a number of a
rectangular grid cells - a grid cell referencing system is developed and layed over the
corresponding map of a certain topic. Because the location of each grid cell is known,
an approximation of natural conditions is gained and can be stored and manipulated as
a rectangular matrix of category values. Managing different maps with certain topics
by the aid of straightforward commands, enables the derivation of relationships
between seemingly unrelated material, from a host of (disparate) sources, all within
the framework of a single system.

LIMITS OF THE GIS AND CHANCES FOR FlIRTHER CISES

Other disparate technologies which are traditionally appropriate in the assessment of

GIS results are expert systems (EXS) - e.g. for the determination of land suitability -
and numerical optimisations - e.g. for the identitication of dominant land allocations
strategies. Very recently, the deficiency of modern GIS technologies to support more
than only very simple data analysis and modelling activities have received
considerable attention. Densham & Goodchild ( 1989) made the following statement:

From another perspective, the difference between GIS and SDSS is that SDSS deals
with a solution space provided by one or more analytical modelling elements as well
as a problem (geographical) space manipulated by the GIS database. Hence, GIS
should be seen as an important component of any SDSS. The role of the GIS in the
proposed implementation environment would be to store, generate and manage all of
the data required for a soil conservation, soil erosion, nutrient export and land use
allocation model. In addition, this then can be linked to an integrated expert system,
which combines management models with natural resources models, gaining an
intelligent GIS. It permits the interpretation of relationships within and among the data
themes of different maps.

79
 GIS RELATED APPLlCATlONS TO SOIL CONSERVATION & EROSION
AND NONPOINT SOURCE POLLUTION STUDIES

‘l‘hc vector based software system AKC/INI;O as well as the grid based GIS IDRISI -
dcsigncd to provide inexpensive access to microcomputer assisted GIS technology
(Eastman, 1990) - was used in combination with different watershed models to
cstimatc the scdimcnt and nutrient cxport from
. a lower alpine catchment of a size of some lOOkm2 (Klaghofer & Summer,
1990);
. a subcatchment of this watershed of a size of approximately 65ha (Klaghofcr ct
al., 1992).
L3ccausc contaminants arc transported by runoff to surface waters as well as by
infiltration and percolation to groundwater, erosion processes are often at the core of
soil conservation strategies, non point source pollution and water quality concerns.
‘these hydrological processes are affected by the spatial variability of soils,
topography, land use and cover, climate and human induced changes and management.
fjence, the basic data for modelling these processes to develop conservation strategies
dcpcnd on spatially distributed attributes of a watershed.

TOPOGRAPHY METEOROLOGY

LAND MANAGEMENT
SLOPEGRADIENTS

SLOPE LENGTH
SOIL TYPES

0
SOIL EROSIONMATRIX 4

EROSIONRATE

FIG. 2 ‘The creation of a geographic database.

80

___ .-- _--..__

 The ma.@ source of data were 3D Elevation Models (3DEM) soil type maps and
maps on land management information (Klaghofer & Summer ’1990). They created
the gcomctric data base (Fig. 2). In the next step physical parameters wcrc assigned to
the geographic information in such a way, that the connection led to a set of grouped
input pmmetcr for the watershed model EPIC (Williams et al., 1983) which basically
represents a deterministic approach to soil erosion modelling (Willia’ms et al. 1984).
Its’ results, assigned to the GIS prepared data, assisted in identifying zones of different
erosion potential (Fig. 3). Export models were then used in a spatial and non-spatial
version to estimate the sediment load from the observed watershed.

Upland erosion (l,/ha, a)

0 0.4 - 5.9
G,O - 11,3
z 12,o - 17,3
10,O - 23.9
I 24,o - 20.9
m 30.0 - x,3
I/If not calculated

m boundary

FIG. 3 Distribution of potential soil erosion in the research catchmcnt.

81

---___ --
The non-spatial version sums the soil erosion within the watershed and applies an
export coefficient to the catchment without regard to the erosion distribution. Hence,
an average delivery ratio, depending on the size of the catchment, was applied to the
catchment of interest as a whole (Klaghofer & Summer, 1990; Klaghofer et al.; 1992
and 1993). The spatial version breaks the watershed into several homogeneous erosion
regions. The export from each region is then calculated with an export coefficient,
specific to that region. The delivery ration is calculated using the distance, water must
travel over land, before reaching a river or stream channel. A GIS is well suited to do
this type ofanalysis.
A different approach to the estimation of the sediment and/or nutrient yield of a
watcrshcd can be done by routing the non point source pollutants in a stepwise fashion
from the headwater of the catchment to its’ outlet. For this method the AGNPS Model
(Young et al., 1987) was used only for the small subcatchment, because the model
works on a cell basis that divides the watershed into a limited number of areas of
uniform squares. A specialised database management system organises the spatially
distributed data. However, it becomes possible to analyse any area in the watershed,
covered by a grid box (Fig. 4).

FIG. 4 AGNPS-grid based data: (a) erosion rates [t/ha] and (b) sediment
yield of the cells [t] (Klaghofer et al., 1992).

CONCLUSlONS

Improvements can be reached in the analysis of distributed physical processes, such as

soil erosion processes as well as non point source pollution, by the integration or
linkage of the spatial data handling capabilities of modern GIS technolobv with
hydrological models. Such a combination offers the advantages associated with
utilising the full information content of the spatially distributed data to analyse the
hydrological processes and their controlling factors/parameters. The used GIS were
capable of manipulating both, the input and the output parameters required as well as
gained by the distributed process models. Decisions for soil conservation strategies,
finally, could be derived only from the results gained by the hydrological model in
combination with the GIS.

82
 Dealing with spatial data and information, which, at one stage, have to be
classified in such a manner to reduce their variability to a practical/reasonable size,
raster based GIS technology seems better suited than vector based GIS. Every pixel of
the geographic information can easily be assigned with a physical, chemical,
biological, environmental, etc. related catchment attribute, describing its’ properties.
The matrix structure of the information also allows easy pre- and postprocessing of
data to prepare input as well as output information in order to link a raster based GIS
with a hydrological watershed model.

REFERENCES

Anderson, L.T. (1987) Five Methods for Calculating Capability Suitability. in E.

Crane, ed., Papers from the 1987 Annual Conference of the Urban and Regional
Information System Association, Fort Lauderdale, Urban and Regional
Information System Association, Washington, pp. 125 139.
Chapin, F.S. & Kaiser E.J. (1979) Urban Land Use Planning. U. Illinois Press, Urbana,
IL., pp.289-3 15.
Densham P.J. & M.F. Goodchild (1989) Spatial Decision Support Systems: A
Research Agenda. Proc.: GIS/LIS ‘89, vol. 2, Orlando, Florida, pp.707-716.
Diamond, J.T. (1988) A Multiobiective Discrete Optimisation Model for Land Use
Planning. Doctoral Dissertation, School of Civil Eng, Purdue Univ., 179 p.
Eastman, J.R. (1990) IDRISI - A Grid Based Geographic Analysis System. Clark
University, Graduate School of Geography, Worcester, MA 0 16 10, USA.
Hopkins, L.D. (1977) Methods for Generating Land Suitability Map: A Comparative
Evaluation. J. American Inst. of Planners -/3(J), pp.386-400.
Klaghofer E. & Summer W. (1990) Estimation of Soil Erosion from a Lower Alpine
Catchment. IAfIS Publ. no. 194, pp.67-74.
Klaghofer E., Summer W. & Villeneuve J.P. (1992) Some Remarks on the
Determination of the Sediment Delivery Ratio. IAHS Publ. no. 209, pp. 113-l 18-
Klaghofer E., Birnbaum W. & Summer W. (1993) Linking Sediment and Nutrient
Export Models with a Geographic Information System. IAHS Publ. no. 211,
pp.50 l-506.
Williams, J.R., Dyke, P.T. & Jones, C.A. (1983) EPIC - A Model for Assessing the
Effects of Erosion on Soil Productivity. Analysis of Ecological Systems: State-of-
the-Art in Ecological Modelling. Developments in Environmental Modelling 5,
Elsevier Scientific Publishing Company, Amsterdam - Oxford - New York, pp.
553-572.
Williams, J.R., Jones, C.A. & Dyke, P.T. (1984) A Modelling Approach to
Determining the Relationship Between Erosion and Soil Productivity. Transact.
Amer. Sot. Eng. 27, pp. 129-144.
Young, R.A., Onstad, C.A., Bosch, D.D. & Anderson, W.P. (1987) AGNPS: i
Agriculture Non Point Source Pollution Model: A Watershed Analysis Tool.
USDA-AR& Conservation Research Report, 35.

83
Erosion and sediment yield on plains in the
temperate zone
Dr. Dedkov A.P., Dr. Mozzherin W.I.

1. INTRODUCTION
Sediment yield analysis is one of the most objective methods for the assessment of
erosion intensity in a river bed and on the surface of a river basin. The volume of
sediment yield transported by the river is not an accurate measure of all the products of
erosion in the river basin, A considerable part of these products is accumulated in the
basin and is not trasported beyond its boundaries. Rough estimations show, that on the
plains of the temperate zone sediment yield within the limits of the drainage basin
different natural and anthropogenic conditions forms over a half of all the products of
erosion. Nevertheless, the sediment yield directly depends on erosion and the whole
mechanical denudation in the drainage basin and can be used for assessment of the
intensity of these processes on different areas under different conditions.
This article is an attempt to analyse the evolution of erosion on the plains in the
temperate zone on the basis of data on suspended sediment yield obtained from 1580
river basins and on bed load data from 387 basins. The temperate zone covers the taiga
(coniferous forest), mixed woods, broad leave1 forests, forest-steppes and steppes.
Natural landscapes in this zone are changed to some extent by man’s activity.

2. TERRITORIAL ANALYSIS OF SEDIMENT YIELD

2.1 Influence of the economic development of the landscape

According to the level of the economic development of the landscape all the basins are
divided into three categories. Criterion of this division and mean specific suspended
sediment yield are shown in Table 2.1. Division of the basins into two groups according
to their area is caused by the necessity of some reduction of the dependence of the
values of specific suspended sediment yield on the scale of the drainage area.

84
Table 2.1.Suspendedsedimentyield (r, t/sq.km per year) in the basins of different categoriesof
antropogenicallychangedlandscapes(N - the number of basins)

Degree of
development
Forests
coverage(%)
Cultivated
land (%)
I basinsSmall
N
<50(
river
1 sq.km
r
r Large river
basins >5000 sqkm
N r
I low > 70 < 30 166 9.2 172 6.5

II medium 30-70 30-70 268 34 159 14

III high < 30 > 70 501 107 162 58

Total 935 69 493 26

The role of landscapes

Values or specific suspended sediment yield vary within different landscapes (Fig.2.1 B.
Table 2.2) due to the difference in precipitation and runoff as well as due to different
degree of economic development of the landscapes (Fig. 1, Table 2.2). Median diameter
of the suspended and transported material in different landscapes is also variable (Fig. 1
c>.

85

-. _-.-
 %% A
100

7 A=3

80-

60- 1

A=.2
40-
I

3
A=1

2a-
1
L
0, I I I 1

B .

4u-

1
---c-I----- -------_-- --
0 I I I 1
mm
c

I I
I---
S b s b s b s b S b
O.Ql I
i
Mixed Broad- FClrati- Steppe
Taiga
woods leaved sWve
forest

Fig. 1 A) structure of basins with different types of erosion system ,functioning in different
landscapes of the temperate zone (A = 1 - natural, A = 2 - natural-anthropogenic, A = 3 -
anthropogenic types);

B) suspended sediment yield (r~ tikm’ year) in modern (solid line) and in natural preagricultural
(dashed line) conditions;

(3 average size (median diameter, mm) of suspended sediment (s) and bed load (b)in the
temperate zone.

The above mentioned data test@ that the main difference between erosion
intensity and sediment yield in the temperate zone is due to the level of economic
development of the basins, i. e. population density and the area of farmlands there. As

86
for the difference between erosion intensity and sediment yield within various
landscapes, it is evident that in equally developed basins the differences are,
comparatively small, not more then 2 or 3 times. But irrespective of the level of
economic development in small and large basins, the most intensive erosion is in broad-
leaved forests and forests-steppes, noted for great precipitation amount and
considerable irregularity of its fall and runoff.

Table 2.2. Suspendedsedimentyield (t/sq.km per year) in different landscapesand categoriesof

anthropogenicallychangedlandscapes(in par&thesis - the numberof basins)

Zones Basins Category 0 economic evelopment Total

I II III
Taiga small 6.4 (81) 7.4 (34) 57 (15) 12 (130)

large 10.1 (141) 11 (20) 25 (15) 11 (176)

Mixed small 4.0 (23) 8.2 (95) 63 (20) 15 (138)

woods large 8.4 (86) 9.0 (58) 36 (18) 12 (160)

Broad-leaved small 7.4 (20) 48 (64) 140 (218) 110 (308 )

forests large 1.8 (1) 21 (31) 94 (58) 67 (92)

Forest-steppe small 17 (3) 51 (30) 38 (64) 74 (97)

large 1.2 (1) 17 (22) 50 (27) 35 (50)

Steppe small 4.4 (4) 29 (46) 100 (116)

large 12 (24) 36 (59)

As for erosion in the basins with low economic development (category I) they
are conditionally considered as close to natural (preagricultural) condition. Data on
these basins indicate that in the temperate zone, preagricultural erosion is weak and
specific suspended sediment yield in this zone doesn’t exceed 10 - 20 t/sq.km per year.
It can be supposed to be even lower in reality, for if 30% of the area is cultivated and is
clear cut it makes a landscape produced by man’s activity. Even within this interval
(category I), considerable variations in specific suspended sediment yield can be
observed.

Sector differences

Within various landscapes in temperate zone there are also sector differences
(Table2.3). Two sector maximums of sediment yield are clearly seen in the table. The
first one is characteristic of North America and Western Europe, where a considerably
low sediment yield is combined with the highest level of economic development.

87
Table 2.3. Specific suspendedsedimentyield of small rivers in different sectorsof the forest zonesof
the temperatebelt

Zone Sector N M A r rl
Taiga North America 93 10 2.8 29 7.3
Western Europe 34 8.4 2.1 52 4.2
Eastern Eurone 158 7.8 1.6 23 4.4
Mixed woods Siberia 45 6.1 1.3 10 5.7
Far East 17 21 1.2 65 49
77 10 2.8 77 49
153 7.2 2.7 150 62
69 4.1 2.5 77 34
1 4.0 1.0 50 30
5 12 1.6 31 25

M - average modules of runoff

A - average level of economic development
r - general average specific suspended sediment yield (t/sq.km per year)
r1 - average specific suspended sediment yield (t/sq.km per year)in the category of
anthropogenic change 1.
The second sector, namely the Far Eastern sector, is noted for the highest runoff
and low economic development of the landscape. Between the two maximums there is a
Siberian - East European minimum characterised by the greatest forest coverage and
the lowest runoff (Table 2.3). If we consider only low anthropogenised basins (category
I), we’ll see that american-westeuropean maximum disappears and only the Far East
maximum, caused by a high runoff due to summer monsoons rainfalls is observed. Here
the maximum erosion rate is observed, which is the greatest in natural landscapes of all
the plains in the temperate zone of the Earth.

3. FUNCTIONING OF THE EROSIONAL SYSTEM

3.1 General
As the level of economic development of the area increases, not only the intensity of
erosion changes but also the functioning of the whole erosional system, defined first of
all by the correspondence between surface and channel erosion. According to the
character of the erosional system, all drainage basins in the temperate zone can be
divided into 3 types.
 3.2 Natural type of functioning
Natural type is characteristic of law modified landscape basins of the first category.
Landscapes of this kind predominate in the taiga, they are fewer in mixed woods. very
few in broad-leaved forests and are an exception in forest-steppes and in steppes. About
23% of all available data on sediment yield from the basins in zone refer to this type.
They are characterised by the most even streamflow distribution during a year, their
underground component being 25 - 30%.
Rivers of this type are characterised by the lowest suspended sediment yield
(Tables 2.1 - 2.3) which don’t usually exceed 10 t/sq.km per year; during monsoon
period in the Far East, however, it reaches 40 - 50 t/sq.km per year. The channel
erosion is usually higher than the surface erosion. The method of rough estimation of
the balance between channel and surface erosion (developed by the authors of this
article and on the analysis of the dependence of suspended sediment delivery on
discharge during low-water periods) indicates that the channel erosion forms 65 - 90 %
of the transported suspended sediments.
Characteristic feature of the basins under consideration is a direct dependence of
specific suspended sediment yield on the scale of drainage areas (Table 3.1).

Table 3.1. The degreeof anthropogenicchangeof landscapeand direct dependenceof suspended

sedimentyield on the areaof basinsin the temperatezone of Eurasia

Categories of Total sum Direct Absent Reverse

anthropogenic change dependence dependence
I 115(100%) 76(66.1%) 3 (26%) 36(3 1.1%)

II 120( 100%) 64(53.3%) 3(25%) 53(44.2%)

III 117(100%) 28(23.9%) 3(26%) 86(49.7%)

Total sum 352(100%) 168(47.7%) 9(26%) 175(49.7%)

With the increase of the areas discharge of the rivers rises and sediment yield rises
too in proportion to its square. Sediment yield delivery is considerable as well, for about
40 - 50 % of all the products of erosion are transported beyond the basins. Suspended
sediments and bed load are well sorted, their index of relative entropy varies from 0.6
to 0.7. Median diameter of suspended particles (48 points) is about 0.07 mm. Bed load
yield composes about 20% of suspended sediments. In the structure of the alluvium the
channel and surface fasies play a proportional role (‘normal’ alluvium, according to E.V.
Shancer)

3.3 Anthropogenic type of functioning

Anthropogenic type of functioning of the erosional system is characteristic of the
basins with strongly changed natural landscapes (category III). Nearly half (40 %) of
the river basins in the temperate zone refer to this type, mainly in broad-leaved forests,

89

--- .__-____---
forest-steppes and steppes. Here, streamflow distribution is most uneven with the
underground component of 10 - 30% only,
Surface erosion dominates over channel erosion and forms more than 50 % of the
sediment yield. All basin erosion elements, namely soil erosion, gully, piping and
raindrop destruction are clearly expressed on the farmlands. Among these elements soil
erosion is the most wide-spread one. Intensive gully erosion is characteristic of the
areas with relatively continental climate (steppes, forest-steppes within the continental
sector of broad-leaved forests).
In the basins under consideration a severe erosion is observed and suspended
sediment yield is measured by tens and hundreds t/sq.km per year. The strongest
erosion is observed in broad-leaved forests and in the north of the forest-steppes of the
east of the Russian Plain, where the suspended sediment yield in small rivers reaches
300-500 t/sq.km per year. Here an reverse dependence appears between specific
suspended sediment yield and the area of the basins (Table 3. l), which is formed during
high - water periods when many products of erosion are found in small rivers. These
small rivers cannot transport huge masses of sediment, therefore, a considerable amount
of sediment are accumulated there as alluvium, delluvium and prolluvium. Hence down-
stream turbidity of the rivers and specific suspended sediment yield are reduced. During
the water low periods only channel erosion occurs and relationship between specific
sediment yield and the area basins becomes direct.
Erosion products delivery to the basins of the third category decreases greatly.
The major part of these product (about 80 - 90%) are accumulated on the slopes and in
the upper parts of the river network. Median diameter of suspended particles is reduced
to 0.03 - 0.04 mm (74 points) and in channels - to 0.3 - 0.5 mm (50 points).
In the composition of the river sediment yield the products of basin erosion
prevail. Rise of high waters and increase of amount of sediment yield result in the rise of
the floodplain surface marks and greater depth of the floodplain facies.
The sorting of sediments (an indicator of relative entropy 0.8 - 0.9) is worse. Bed
load yield makes no more than 10% of the suspended sediment yield. The role of the
suspended sediment in the construction of alluvium increases greatly.

Natural-anthropogenic type of Functioning

Natural-anthropogenic type of functioning of the erosional system is characteristic of

the basins with medium rate of anthropogenic change (category III). In all indices these
basins occupy an intermediate position between the basins of the first and second
categories. Predominance of basin or channel erosion, of direct or reverse dependence
of specific sediment yield on the area of the basins is clearly expressed here (Table 3.1).
Indices of size and assortment sediment yield, proportion of bed load and
suspended sediment yield are of intermediate value. Expression of all types of
functioning of erosional system depends not only on the level of economic development
of the area, but also on climate - landscape and geo - geomorphological conditions.
Broken relief and clay soils are favourable to the third type of functioning and on
the contrary cliff rocks and smooth relief are favourable to the first type.

90
 Hydrometeorologically conditioned extreme manifestations of erosion are best
expressed in the basins of the third category, where they often have catastrophic
consequences. In preagricultural time all over the temperate zone (with the exception of
semideserts) channel erosion with small suspended sediment yield and direct
dependence of their modules on the area of the drainage basins prevailed.
As the forest coverage is reduced and more lands are cultivated, the basin erosion
becomes more important and a radical reconstruction of functioning of the whole
erosional system occurs, At present the natural type of functioning remains only on the
quarter of the vast area of the temperate zone manly in the forests of North Europe,
Siberia and Canada.

91
Improved methodology for the
computation of normal annual yield of
suspended sediments from rivers
N.N.Bobrovitskaya & K.M.Zubkova
State Hydrological Institute, St.Petersburg,Russia

Abstract.
Methodology for the computation of normal annual yield of suspendedsedimentsfrom
the rivers of Russia and FSU has been analysed.It is shown, that at the availabletrends
in the formation of sedimentsconcentration and sediment yield, their norm should be
computed for homogeneousobservation periods. Trends are often resulted from joint
effect of climate characteristicsand anthropogenic factors. Formulas of transporting
capacity of the flow applied for the computation of sediment yield are analysed. A
simple equation is proposed for a computation of suspendedsedimentyield where flow
velocity, river depth and a correction coefficient dependent on the trend in sediment
yield changesfor a long-term period and on the streamflow distribution during the study
year are taken into account.

Data sources
Data from 2,000 stationsfor sedimentyield measurementsand data from 3,000 stations
for water discharge measurementshave been used. The duration of the observation
series on sediments concentrations varies within 1 - 52 years; the duration of water
discharge series equals 100 years and longer. The author has created a databaseon
annualsedimentyield and on the factors which determinethis yield.

Analysis of methods for the computation of normal annual yield

of suspendedsediments
In general,when the observationperiod is long enough (longer than 15 years) the norm
of suspendedsedimentyield is computed as a mean arithmetic value out of a series of
the observedvalues. For gauging siteswith short observationserieslocated on the same
river or on the tributaries of that river the norm is computed from the rating curves of

92
 sedimentyield at the noted gaugingsites and at siteswith longer observationseries.The
river reachesupstreamthe comparedsitesshould satisfy the conditionsof analogy.
Normal yield of suspendedsedimentsfor ungauged rivers is computed from
equationswhere basicphysiographicfactors (climate, soils and top cover) are taken into
account(Poliakov, 1935; Lopatin, 1939; Shamov, 1949; Lisitsyna, 1972; et al.).
Therefore, changesin thesefactors observedfrom north to south of Russiacause
changesin long-term characteristicsof sedimentconcentrationand yield, which proves
the geographicalzonality of thesefactors. Topography and geology are most important
geomorphologicalfactors. Thesefactors are azonal.Differencesin the soils and subsoils
structureand rugged topography often causedisturbancesin the zonal concentrationand
yield of sediments.
Since it is diEcult to establisha theoretical dependencebetween water erosion
factors and sedimentyield characteristicsan empirical method has been developedat the
State Hydrological Institute for the computation of sediment yield from ungauged
rivers. This method is basedon the establishmentof relations between mean long-term
water dischargesand sediment yield or between specific sediment yield (Ms) and
specific water discharge(Mg), respectively,for rivers running on comparatively large
and physiographicallyhomogeneousterrain.
More precised dependenciesbetween the above values are obtained with the
account of the topography nature and introduction of the so -called erosion coefficient.
A dependencefor this coefficient determination was first proposed by V.G.Glushkov
and it was later moditied by B.V.Pohakov in 1946. This method was very popular.
Table 1 shows most typical variants to determine erosion coefficients. During 1960-
1975 this method was applied for erosion regionahzation of the FSU territory in
“SurfaceWater Resourcesof the URSS”, 41 volumes.
Normal suspendedsedimentyield for ungaugedrivers is often computed with the
use of mean long-term sedimentconcentrationdeterminedfrom the maps of sediment
concentrationin rivers.
The first map of sediment concentrationin the USSR was prepared in 1939 by
G.V.Lopatin and it was improved severaltimes later (G.I.Shamov, 1949; K.N.Lisitsyna,
1972; et al.).
Analysis of long-term variations in sediment concentration and yield in rivers
made during recent years at the Laboratory shows, that sedimentyield has been greatly
changedsince 195Os’-1970s’ in most of the FSU rivers. Moreover, both increase and
decrease of mean long-term sediment yield characteristics are observed
(Bobrovitskaya,l991,1994).
Time variations of hydrometeorological factors are characterisedby a cyclicity
which is manifestedby subsequentalteration of a number of years with dserent water
availability and different duration. In general,statisticallysignificant differencesin mean
valuesand in variances,(e.g. water discharge)for complete cycles of variations, are not
available.
Trends which appear in the long-term variations of sediment yield and
concentrationmost often result from a joint impact of climate factors variations and
changes in the characteristics of anthropogenic factors effect. Several curves of
sedimenttransportation can be usually plotted for such a river reach from observation
data. Moreover, statistically significant differences in statistic characteristicsfor the
periods correspondingto various water availability most often occur. These differences
are proved by river channel changesas well. More than 24 factors of anthropogemc

93

-__ - ~.---I_
Table 1 - Equationsfor Erosion CoefftcientsDetermination

Equation Developed by: Territiry

S S V.G.Glushkov & European USSR
a = 104J7 B.V.Poliakov (1946)

P,=Kl”Q” R.S.Chalov European USSR

N.I.Makkaveev (1986)
ac=- 4
N'.22 V.P. Svetitsky ( 1962) Amu-Darya

p,(l+P)
a’ = &Ql.SI0.75 A.N.Agaronian (1967) Armenia

s-50 D.Pechinov (1967) Bulgaria

QrPL=
KApLz;;5

Qm - Qpr
P, = f
c
Qi, At N.N.Bobrovitskaya (1967) Valdai Hills

Ms = KsM;i N.N.Bobrovitskaya (1972) European USSR

Ms = KJ,
K.N.Lisitsyna USSR territory
Ms = 1.2M, +O.l66h-4.91 (1960, 1972)

Legend: a, ac, al, apl, ks, kl - erosion coefficients; S - mean annual concentration of
sediments, g/cu.m; Ps, Ms - mean annual yield and specific yield of suspended
sediments,kg/s, t/km2 per year, respectively; N - energy characteristic; I - slope of the
channel; Ipl - slope of ploughed lands; Q - mean annual water discharge, m3/s; K -
erosion coefficient in R.S.Chalov’s and N.I.Makkaveev’s formula, dependent upon
nonuniformity of flow, nature of bed particles and mechanicalcomposition of sediments
from tributaries and upon soils and subsoilsof the basin; Qmax - maximum mean daily
water dischargefor the snowmelt spring flood and Qpr - mean daily water dischargefor
the period prior to the beginning of the snowmelt flood; Dt - number of days from the
beginning of the snowmelt flood up to the day of flood peak; Af - forested area; Apl-
ploughed area in the basin; e - index of soil erosion rate; b - characteristic of the basin
state; B - characterisesthe soil-subsoil composition.
influence on sediment yield have been discovered. Most important factors are:
construction of hydraulic structuresin rivers, ploughing of slopes,forest cut and plant,
excavationof mineral deposits,irrigation, etc.
Changesin sediment concentrationand yield are shown in the Polomet river at
Yazhelbitsy (A = 631 km’). The Polomet river flows down the Valdai Hills (Lake
Ladoga basin). Water dischargeand sedimentyield observationswere started in 1953.
The longestobservationserieson water dischargeswere also used,i.e. for the Tikhvinka
river at Gorelukha (A = 2700 km*) - 104-yearobservationperiod and the Neva river at
Novosaratovka(A = 281000 km*) - 127- year observationperiod, etc.
Phasesof higher and lower water dischargescoincide for all the studied rivers
(Fig. 1). Trends for a long-term period of water discharge observations are not
discovered(at the level of sign&canceof 0.05%). Annual sedimentyield in the Polomet
river at Yazhelbitsyfor 1953-1988changedalmost concurrentlywith annualwater flow
changes.Since 1963, however, the amplitude of sedimentyield variations tended to a
gradual decrease(Fig. 1 ); another curve (Fig. 2) may be plotted on the curves of
sedimenttransport since 1971. On the basisof analysisof sedimentyield integral curve
and sedimenttransportationcurvestwo long-term periods have been selected,i.e. 1953-
1970 and 1971-1986 (Fig. 2). Mean annual values of specific sedimentyield for these
periods are different (at the level of significanceof 0.05%) and are equal to 150 t/km’
and 92 t/km*, while sedimentconcentrationvaried as follows: 270 g/m3 and 150 g/m3,
respectively. Mean specific annual water discharge for these periods was practically
unchanged(statistically signiscantvalues at the level of significanceof 0.05% were not
discovered)and it equals 10.5 l/s per km’. Similar changesin sedimentyield have been
discoveredin other rivers of Lake Ladoga basin.
Thus, in the conditions of streamflow and sediment yield formation in rivers
running in Lake Ladoga basin it is possibleto selecttwo periods since 1950, i.e. a wet
period of 1954-1960and a dry period of 1961-1983with deviationsof 1 or 2 years. In
general, after a wet-year period when much loose particles are eroded and washed
away, sediment concentration and yield tend to decrease.Besides, during the second
period the ploughed lands were about twice less (14.8% instead of 25.8%) and an
intensivesandextraction from the river channelfor constructionpurposeswas observed.
A combinationof climate and anthropogenicfactors probably explainedthe decreaseof
sedimentyield from 1966-1970up to the presenttime.
Analysis of trends in suspendedsediment yield changeshas been made for all
rivers of Russia and FSU (Bobrovitskaya, 1994). The following distribution of series
with variableregimesof sedimentyield has been discovered:
up to 100% on the watersheds of Western Dvina and Araks rivers and in
Turkmenia;
71-98% - in the Ob, Irtysh and Ishim rivers; left-bank area of the Selengariver, in
the upper reachesof the Dnestr river, in middle and lower reachesof the Dnieper and
Don rivers, in the right-bank area of the Volga river basin, in the Kura, Kuma and
Kuban rivers, in the rivers of Georgia, Azerbaijan, Crimea and in the watershedarea of
Lake Balkhash;
50-70% - in the Syr-Darya river, in the upper, middle and left-bank areasof the
Volga river basin, in the upper and middle reaches of the Yenisei, and Amu-Darya
rivers, in the lower and middle reachesof the Lena river basin and in Kamchatka;
25-50% - in the upper and middle areas of the Onega, Northern Dvina and
Pechorariver basins;in the rivers of SakhalinIsland, in the lower reachesof the Neman

95
 140
120
M, lfskm'

10.0

!
8.0
6.0

4.0 ! I
1870
I I
1830
I 1
1910
I 1
19M
I I
1950
I 1
1970
I
Igo0
1

400- 4

m- R
&
a -
, I ,k(
I I I I I I I
llldo 1870 18@2 1910 19X 1960 1970 19Kl

FIN. I. Chronological variations of mean annual water dischargeLy: I - the Neva river at
?i’ovosaratovka 2 - the Tikhvinka river at Gorelukha 3 - the Polomet river at Yazhelbitsy. 1 - mean
annual sediment yieldporn the Polomet river at Yazhelbitsy.

96
 PSkg/s

7.0 -

6.0 -

5.0 -

4.0 -

3.0 -

2.0 -

1.0 -

0.0 2.0
I I
4.0
I I
6.0
I I
8.0
r

10.0
I I I I
Qr m3/s
12.0 14.0

Fzg. 2 Sediment transportation curves: for the Polomet rwer at Yazhelbitsy: I - i9.i3-1972; II - 19-3-
19r.

97

___ - __..-_ __
river basin,in the left-bank area and in lower reachesof the right-bank area of the Amur
river basin, in the lower reachesof the Oka river basin, in small and mid-size rivers in
tundra;
lessthan 25% -in the upper reachesof the Kama river basin, in the upper reaches
of the Chu and Talas rivers and in the watershedarea of Lake Issyk-Kul.
It should be noted that a trend in lower sedimentyield is observedin the Northern
Dvina river basin; trends to sedimentyield increaseand decreaseare observedin other
basins.
In general, normal yield of suspendedsedimentsfrom many rivers should be
computed for a relatively short period homogeneousby the conditions of formation of
suspendedsedimentyield. In some casesthe computation of suspendedsedimentyield
would require the use of equations of the transporting capacity of the flow. Let us
considersome casesof theseequationsapplicationwhen undirectedchangesin sediment
concentrationsand suspendedsedimentyield are observedin river basins.

Application of equations of the transporting flow capacity for

the compution of suspendedsediment yield from rivers
Most of the empirical and theoretical formulas have been developedfor determining
mean sediment concentrationin rivers (Table 2). Hydraulic characteristicsof streams
averagedat the gauging stations over cross-sectionarea are usually applied. Formulas
resulted from multipurpose theoretical and experimental studies and discovering the
physical essenceof sedimentstransportation to a certain extent are most interesting.
These are the formulas of V.M.Makkaveev (1940) M.A.Velikanov (1949)
V.N.Goncharov (1954) A.N.Gostunsky (1954) I.V.Egiazarov (1964) J.F.Karasseff
(1965) A.V.Karaushev (195 1,1960),K.I.Rossinsky and I.A.Kuzmin (1964).
On the basis of observation data received from the network of the
Hydrometeorological Service of the FSU, K.V.Razumikhina applied formulas for the
computation of sedimentstransport to rivers rumring in different phisiographic zones
greatly differing in hydrological regime and conditions of sediment yield formation.
Hence, empirical formulas (Gostunsky, Zamarin, Khachatryan) derived from the
observationdata from rivers and channelsin Central Asia, were appliedto the rivers of
the Volga-Kama basin (Belaya, Viatka, Kama and other rivers) and overestimated(in
50-100 times) values of the computed sediment concentrations were obtained if
compared with the observed ones. Nevertheless,these formulas applied to the rivers
with fine-grainedsediments,gave quite satisfactoryresults.
In case of lowland rivers, most reliable results have been obtained from the
theoretical method of Karaushev for the computation of fractional composition of
bedload.It has been establishedhowever, that this method developedat the WI, should
be restricted dependingon the type of river hydrograph and size of the alluvium in the
channelwhich forms beds in theserivers (Razumichina,1966). The analysisof research
results shows, that in rivers with evident spring snowmelt floods and bed composedof
sandand gravel the computedvalues of sedimentconcentrationare observedduring the
falling limb of the hydrograph while in case of the sandy and silty bed - on the rising
limb. In rivers with multipeaked regimes (and more uniform streamflow) and sandsilty
bed the computed values of the flow transporting capacity are close to the observed
valuesof sedimentconcentrations.

98
Table 2 Formulas for Computatiion of the Transporting Capacity of Water Flows

Equation Developed by: Territiry

Empirical
S = 5.J.1()4v5+u-32 E.A.Zamarin (195 1) CentralAsia

S = 3.3403v3H-‘U-‘C3 A.N.Gostunsky (1954) Canalsof

CentralAsia
S = 0.36 - 1O-3Nv~H-~(v/U)~~‘~ A.V.Karaushev (1960) Small and mid-
sizedrivers

S = 0.024 .v3H-‘U-’ K.I.Rossinsky & Riverswith high

I.A.Kuzmin (1964) sediment
concentrations

Theoretical

S = S,,,, T;L Sroir= 0.15K2Nv2H-’ A,V.Karaushev, ( 1961)

S=0.24.C2v3g-32Bq~. J.F.Karasseff,(1965)
-1
0.75v3g l2 +0.5x
>

Legend: v, H - mean velocity and depth of water flow; U - average hydraulic fall of
suspendedsediments; C - coefficient of Chezy ; N - characteristic number of the
turbulent flow dependent upon Chezy coefficient C; K - parameter which takes into
account the ratio of water bottom velocity to mean flow velocity; f - hydromechanical
parameter of sediments determined from tables (Karaushev, 1977) depending on v, C
and granulometric composition of suspendedsedimentsand bedload; g - acceleration of
gravity; B - parameter considering the ratio of bottom sediment concentration to mean
sediment concentration; h - probability of bedload roiling up from the bottom; e -
probability ofthese particles suspensionin the water flow; &,;l - sediment concentration
when sedimentsroil up from the bottom.

99
 Further investigationson the selection of most optimal formulas or methods for
the computation of sedimentstransportationby flows con&med the necessityof their
testingunder field conditionsrelative to the study gauge-line,reach or river basin.
The computationpractice showsthat in caseof particular river reacheswith rather
permanent size of bedload and suspendedsedimentsand watercourse slope, simple
formulas may be applied, e.g., the formula of A.V.Karaushev (1961). Verification of
this formula under field conditions made by K.M.Zubkova (1982) on the basis of field
measurementsin the Karakum Canal, showed that coefficient K in this formula
dependedon the size of bedload composing the bottom and the banks of the Canal
within the study reach. This is probably explainedby the fact that water dischargefrom
the Canal varied within small limits while the size of bedload over the Canal varied
within great limits.
A comparison of the Karaushev method (196 1) for sedimentyield computation
with the method of Rossinsky& Kuzmin (1964) basedon the data of measurementsof
sedimentsconcentrationand yield in the litysh River at Omsk (Zubkova, 1986) shows
that thesemethodsprovide similar results. Computationsof the transportingcapacity of
the river were made at constantwater dischargesat 60 cross-sectionsalong a reach of
about 100 km long. It was establishedthat coefficient K dependedon the shapeof the
cross-sectionof the river.
Investigations made by L.P.Alexeev, Sh.R.Pozdniakov et al. (1991) on the
selectionof designdependencesto determinemean sedimentconcentrationin the Katun
river (Ob river basin) in two gauge-lines( 186 km and 359 km far from the mouth)
proved once againthat mean sedimentconcentrationof the flow might be computed by
any of the formulas from Table 2, if the design dependenceswere limited by
coefficients.
Analysis of the computationresultsmade by K.V.Razumikhina ( 1966) also shows
that it is not reasonableto use characteristicsof granulometriccomposition of sediments
for computationsbecausedetailedmeasurementsare usually not numerous and, besides,
this value appearsto be permanent and independentof hydraulic parameterswithin
particularbasins.
After generalizingcomputation results by formulas from Table 2 it is possibleto
proposeone more simple formula for the assessmentof mean sedimentconcentrationin
the flow:

S=&V3iH (1)
Peculiaritiesof the formation of sedimentconcentrationin the flow dependingon
the type of the hydrograph, the shapeof the cross -section, the state of the underlying
surface of river basin, the rate of the anthropogeniceffect, etc., may be approximately
taken into account with the help of correction coefficient As variable in time. The
coefficient As dependenton the trend in sedimentyield changesfor along-term period
and on the streamflow distribution during the study year are taken into account.
Let us take the results of computationsby formula (1) for the Polomet river at
Yazhelbitsy as a case-study.Detailed measurementsof the hydraulic characteristics,
sediment concentration and granulometric composition of sedimentsfor 1957-1979
were used. It was discoveredthat As coefficients were about twice higher before the
beginningof the 1970s’if comparedwith the period of 1971-1979.This is evident from
Fig. 3 where data for 1968 and 1976 are given with approximately similar types of the
hydrographs.Besides,the values of As on the rising limbs and on peaks of the snowmelt

100
 gk c
,0
8 *
+-.q
--m-- c
a
- -- - --- __
*--
and rainfall floods are higher than on the falling limbs of the hydrographs,during low-
water periods the values of As tend to be one order lower. The increase of As
coefficientswith the increaseof water dischargesis also observedbut this dependenceis
not regular. Limits of As valuesvariations are rather stable,i.e. 1.0-I. 5 for the rise and
0.5-1.0 for the fall. Limits of As variationsfor the rainfall floods are similar to those for
snowmeltfloods.
It is quite probable that decreaseof As coefficients in 1971-1979 according to
formula (1) is connectedwith the lower transporting capacity of the flow (decreaseof
maximum water dischargesat relatively permanentmean annualvalues), and with less
scour from the basin due to less ploughed areas.As mentioned above, this particular
case shows a joint effect of climate and anthropogenic factors on the regime of
sedimentsconcentrationin the flow.
Thus, it is quite possibleto make a simple diagram of sedimentyield computation
for particular gauge-lineor river reach. In case of long-term observationsof sediment
concentrationsand state of river basins (ploughed area, forested area, excavation of
mineral deposits,etc.) it is possibleto classify correction coefficientsAs in the formula
of the transportingcapacity of the flow.
In conclusionit should be noted that the discussedmethods include investigations
and computations of sediment yield and concentrations in many links of the
hydrographicnetwork, i.e. in temporary streams,and in small, mid-size or large rivers.
A real opportunity also appearsfor a multipurpose use of air photographs and field
observationdata for a long-term period. These data may serve as the basis to predict
water erosion evolution and to normalise anthropogenicload. The authors feel strongly
that the multipurpose approach can be applied for other regions of the world to get
characteristicsof the erosionrate under extreme conditionsof floods formation.

REFERENCES
Alexeev, L.P.; Meerovich L.N; Pozdniakov Sh.R and Poliakov V.Yu. (1991)
‘Investigations of suspendedsediments transport characteristicsin the Katun
river’. Meteorologiya i gidrologiya, No. 5, p.SO-88.
Bobrovitskaya,N.N. (1972) ‘Dependenceof mean long-term suspendedsedimentyield
from the rivers of the EuropeanUSSR upon physiographicfactors’.- Trudy GGI,
vyp. 191, p.68-84.
Bobrovitskaya,N.N.( 1994) ‘Assesmentof trends to sedimentdischargevariationsin the
rivers of the Former Soviet Union (FSU)‘. In: Proceedingsof the International
Symposium, East-West,North-SouthEncounter on the State-of-the-Art in River
Engineeringand Design Philosophies,State Hydrological Institute, St.Petersburg,
p.32-39.
Egiazarov, I.V.( 1964) ‘Comparison of methods for the computation of sedimentyield
with field measurements’.- In: “Metody izmerenia i ispolzovania vodnykh
resursov”,Moscow, ” Nauka”, p.5-23.
Glushkov, V.G. (1961) Theoretical problems and hydrological research methods. Izd.
AN SSSR,414 p.
Gostunsky, A.N. (1954) ‘Suspendedcapacity‘.“Izv. Akad. Nauk Uzbekskoj SSR”,
No.3 p 59-68.

102
Karaushev, A.V.(1960). ‘Problems of dynamics of undisturbad water flows’.
Leningrad, Gidrometeoizdat,392 p.
Lisitsyna, K.N.and Aleksandrova V.I. (1972). ‘Sedimemt yield from the rivers of the
EuropeanUSSR. Trudy GGI, vyp. 191, p.23-5 1.
Lopatin, G.V.(1939) ‘On the problem of solid dischargefrom the USSR rivers’. ”
Meteorologia i gidrologia”, N 1, 1939,p. 106- 108.
Lopatin, G.V.( 1952) Sediments in the rivers of the USSR. Moscow, Geografgiz, 366 p.
Makkaveev, V.N. (1931) ‘On the theory of turbulent regime and suspensionof
sediments‘.“Izvestia GGI”, No. 32, p 5 - 26.
Makkaveev, V.N. (1955) River channel and erosion in it basin. Moscow, Izd. AN
SSSR,347 p,
Makkaveev, N.I.( 1984). ‘Watercoursesactivity as a denudationagent’. In: “Erosionnye
protsessy” (Erosion processeses”.Eds. by N.I.Makkaveev & R.S.Chalov.
Moscow, “My,” p. 5 - 30.
Poliakov B.V. (1935) Investigations of suspended sediment & bedload yield.
Leningrad, 129 p.
Poliakov B.V.( 1946) Hydrological analysis and the computations, Leningrad,
Gidrometeoizdat,1946, 480 p.
Rossinsky, K.I. and Kuzmin LA. (1964) ‘Balance method for the computation of
botton deformations’.Trudy Gidroproekta, sb.12,265 - 271.
Shamov, G.I.( 1949) ‘Suspendedsediment yield f?om the USSR rivers’. Trydy GGI,
vyp. 20 (74).
Shamov, G.L(l954) River Sediment . (River channel and erosion in it basirz).
Leningrad, Gidrometeoizdat,347 p.
Sediment yield its study and geographic distribution, (1977). Leningrad,
Gidrometeoizdat,240 p.
Razumikhma,K.V. (1966). ‘Problems of applicacy of methods for the computation of
sedimentstransport to fluvial flows’. Trudy GGI, vol. 132, p. 18-45 .
Velikanov, M.A.( 1949) Dynamics of channelflows. Leningrad, Gidrometeoizdat, 475
P.
Zamarin, E.A. (195 I). Transporting capacity and admissible rares offlows in canals.
Moscow - Leningrad, Gostransizdat,82 p.
Zubkova, K.M. (1982) ‘Computation of sedimentstransportationin big canals’. Trudy
GGI, vol. 283, p.41-51.

103

___~ - __-__ --- __-

 Erosion of cohesive materials

Ts.E.Mirtskhoulava, Professordirector, ResearchInstitute of Water Managementand

Engineering Ecology Georgia, 380062 Tbilisi

Introduction
Great soil losses caused by erosion in rivers, canals, reservoirs and irrigation
canals require appropriate protection measures.Determination of these measuresto
reduce negative ecological consequencesin the utilization of natural resourcesrequires
a profound knowledge of the erosion mechanisms. It is obvious that protection
measuresand technologies cannot be found without thorough investigations. Erosion
processesshould be studied for a development of rational standardsfor permissible
non-eroding velocities in rivers and canals(the amount of which increasesconstantly).
The knowledge of erosion is crucial for investigating all hydraulic parameters:
nonchannel flows in the form of overland flow, lateral erosion, linear erosion
permanentriver-bed phenomena,etc.
The knowledge of erosion in important not only soil on vast areas, along with
prevention of negative affects in river, canals etc., but for a number of geologic,
oceanographic and exploration problems. For instance, oceanographersand coastal
engineersare interested in the velocity of flow required for a scour of a certain water
pocket or deposition of sedimentsin tideless sea navigation turbulent. The knowledge
of erosion is also essentialto forecast geologic formations, flotation, etc.
Investigation of erosion should bring about progress in the following areas:
relationships between flow parametersand physical and mechanical soil characteristics
in erosion processes;minimum threshold (non-erodible) rate for soils, with specified
properties; duration of erosion for the specified flow and soil parameters,etc. Results
of erosion studies provide a background parameters,to predict result horizontal erosion
and to determine protection measures.These results also help to provide the erosion
control.
It should be noted that argillaceous rock often occur among; sedimentary rocks.
Different authors show that argillaceous rock constitutes 65-82% of the overall
sedimentary mantle of the Earths crust (Mirtskhoulava, 1967; Mirtskhoulava, 1987)
and many differ in age.

104

-~- -.--___ -
 All rocks consist of crystalline and mineral coherent particles, linked together by
the structure. This linkage may be rigid as in igneous and metamorphic crystalline rocks
and in cohesive sedimentary rocks. On the other hand, this linkage may be loose, as in
clay and loess cohesive soils and in coarse sedimentary and sandy cohesion less soils.
Argillaceous rock is a polydispersive system comprising particles of various size.
Grains of sands and silty particles occur in it. The most active role is played by clay
particles with diameters below 0.002 mm, the portion of which exceeds 30%.
Argillaceous rocks are cohesive and plastic both in natural condition and when
moistened; they preserve their shapes in dry. All these properties are most characteristic
and can be easily recognized.

General properties of clay soil.

The structure of soil is one of the primary properties determining soil resistance to
erosion.
This mechanical characteristic of clay soil rock is related to orientation and
distribution of clay particles. The structure depends on the schematic composition of
water, mineralogical composition of clay and generation of rocks in the sedimentary
process. It should be noted that the structure and structural links determine physical and
mechanical properties of soil. These properties also play a very significant role in
erosion.
The structure concerns the dimension, shape and nature of the surface of the
structural elements, while the texture concerns configuration of the above elements in
space. Texture determines the heterogeneity of rock and orientation of its structural
elements.
Cohesion is an important property of cohesive (clay) soil. This cohesion may be
caused by forces of different origin. An important role is played by the attractive Van
der Waals forces of electro-chemical nature.
Along with Van der Waals forces and the attraction of positively and negatively
charged extremities, the particles may be linked by other mechanisms. Cohesion can be
controlled by hydrogen.
Cohesion also depends on cationic links.
An important role in the cohesion of clay is played by chemical cementation of
particles with an other substance. For instance, strong links between particles are
general by iron-oxide.
Investigations by various authors have displayed the role of physical and
mechanical properties of clay in resistance due to different modes of deformations.
Mechanical properties of soil depend on soil composition and structure, i.e. on the
percentage of minerals, the cryptalline configuration, nature of aggregation,
macrostructure and textures. Mechanical properties of clay soil under the effect of
extremum forces are manifested primarily in resistance against compression and shear;
these properties are important for the assessmentof strength and stability.

105
3. Short summary of erosion in cohesive soil.
If we want to solve any problem it may be useful to turn to the past. The problem of
erosion in channels composed of cohesive soil have been studied since the 19th
century. A great number of investigations have been made, many studies being purely
historical.
The first study of erosion seems to be that by Dubois (Lane, 1953) who
published the table in 18 16 of critical erosion rates for some types of cohesive soils.
Echevery (Lane, 1953) published recomendations in 1916 on non-erodible
velocities. At the end of the 19th century the so-called regime theory was developed to
calculate the channel stability. The first attempt was made by J.F.Kennedy. The
formulae were later used together with depth and other parameters to describe the
cross-section of the channel. Considerable works on data generalization to predict
erosion were carried out by For-tier and Scobey (1923) who observed that the Kennedy
relationship ought about some losses in the computation of channel erosion although
the results were acceptable for the computation of non-silted channels.
Since 1930 a lot of attention has been drawn to the tractive force; Lane (Lane,
1953) carried out many investigations on the criterion of erosion stability in channels.
Together with the data of Echevery, For-tier and Scobey, he also used the USSR data
(standards of 1936) (Lane, 1953; Mirtskhoulava, 1988).
Unlike the US manuals, the recomendations proposed in the USSR (Fottier,
Scobey, 1923; Mirtskhoulava, 1988) were based not only on soil types, but they
included the parameter of clay density. It should be noted that the USSR manual
contained slightly overestimated values, if compared with those in the US
recommendations. Physical and mechanical properties, which characterize the erosion
resistance were considered in the US and USSR manuals.
The importance of the permissible (non-erodible) velocities for cohesive soils in
the solution of practical problems is doubtless; therefore field and laboratory
investigations have been undertaken recently in many countries.
The above analysis indicates that until now neither general nor particular erosion
theories exist which would provide solutions of practical problems. Moreover, widely
accepted laboratory tests are not available at present, although a lot of efforts are being
made.
Our opportunities to describe the erosion of cohesive materials reliably in widely-
known terms are so limited that an accurate theoretical solution of this problem cannot
be anticipated in the near future. Therefore we shall not dwell on the proof that
experimental investigations are necessary. Theoretical derivation of a formula of non-
erodible velocity as a function of physical and mechanical properties of eroded material,
parameters, working conditions of a conduit, geometrical characteristics of a channel,
etc. required numerous experiments involving a great variety of grain size distribution,
properties, origin and geographic distribution of soils (Mirtskhoulava, 1966; 1966;
1967; 1970; 1970; 1987; 1988; 1989).
The investigation of shear resistance in cohesive materials was conducted mostly
in the Caucasian Republics of the FSU where the soils are represented by almost all
genetic types: from podzolic soil in humid climate to grey semi-desert soil and from
mountainous meadows in cold climate to terra rossa in humid subtropical zones. We
also investigated some typical soils of the northern Caucasus, Moscow region, the Urals

106
and Tadjikistan. Such experiments with cohesive materials of natural structure have
never been conducted in the world.
Because of many difficulties associated with field investigations of cohesive
materials in the channel we conducted our experiments in hydraulic flumes. Natural
samples having different composition, properties and origin were investigated in
laboratory, with subsequent verification of results in typical channels composed of
characteristic soils. The following laboratory investigations were made:
- laboratory investigations with undisturbed and disturbed samples. with
measurements of basic physical, mechanical, chemical and petrographic
characteristic and hydraulic parameters of flow:
- laboratory investigations in hydraulic flumes with disturbed and undisturbed
soils at different hydraulic parameters (depth, turbulence. quality and
quantity of suspended matters and bedload, etc.
- erosion resistance tests in field conditions for various composition of the
channel bed and hydraulic parameters,
The erosion tests were conducted in the laboratory in glass-walled flumes having
rectangular cross-sections of 50x30 sm. and 1150 sm. long, A high-speed microcamera
was used to observe erosion.
Description of erosion resistance by the above soil types was proved inadequate
during the investigations made in 1950’s soils (Mirtskhoulava, 1966; 1966; 1967; 1970;
1970; 1987; 1988; 1989). Basic findings were later confirmed in soils (Mirtskhoulava.
1966; 1966; 1967; 1970; 1970; 1987; 1988; 1989).
Erosion investigations, both visual and microscopic, together with measurements
by high-speed microcameras, have shed light on the erosion of cohesive materials,
which in fact varies with the composition, structure, properties, moisture content and
degree of cohesion.
The growing erosion rate (including the nature of clay and its interaction with
water) seems to depend on water resistance of both hydrocollodial and cohesion links,
water resistance of aggregates and the materials filling the pores between aggregates;
dimensions of macro- and micro- cracks are also important as they control the
penetration of water into cracks and the generation of the so-called wedge pressure.
The erosion process in clay of natural structure (water saturated) is a continuous
process of scour which grows in time and depends considerably on properties of soil
and characteristic of turbulent flow. The erosion process may be splitted up into a few
stages.
Zt should be assumed that generation of the erosion surface is associated with
heterogeneity of cohesion forces. A certain part of soil may be more resistance to
erosion due to this heterogeneity, so that less cohesive aggregates are detached faster
than the stronger ones, and a certain roughness appears.
The erosion surface is difficult to be described mathematically, so that
phenomenological approaches seem to be appropriate.
Character of flow is determined by active forces while properties of soil control
passive forces. If resistance of incohesive materials depends on their density and on the
density of water, the resistance forces in cohesive materials are much more complicated.
The stresses due to turbulent flow on the surface of heterogeneous soils are
obviously heterogeneous too and they are concentrated about any defect so that a

107
damage or failure is initiated. This has a significant role for the fatigue of a detached
aggregate.
Generation and growth of microscopic cracks is an important issue for the failure
mchanism in cohesive materials. The erosion resistance of spil is very sensitive to
surface deformations. Cracks reach sizes in which water may move freely and pressure
reaches significant magnitudes. Investigations of erosion permit assumption that failure
occurs if cracks are open and closed as a result of separated dynamic action of water.
Cracks grow eventually as a result of these multiple deformations, their links become
weak and the aggregate is detached.
Fatigue of links is referred to as failure due to concentration of irreversible
mechanical and chemical factors upon prolonged dynamic action of the turbulent flow.
The term “fatigue” is usually understood as a certain type of damage due to multiple
cycles of stresses with amplitudes below the limit of material fatigue. Determination of
the number of cycles which bring about damage makes it possible to postulate the
temporal behavior of erosion, i.e. intensity of erosion as a mnction of velocity
fluctuations and erosion resistance of soil soils (Mirtskhoulava, 1967). The term
“fatigue” has been taken from strength of materials science soils (Mirtskhoulava, 1967;
1988; 1989). The term “fatigue failure erosion” is not clearly defined. A fatigue crack is
difficult to be found in soil, as immediately after the fatigue occurs, the roughness
element becomes detached. Therefore, more objective criteria should be developed,
perhaps with the appearance of cracks on the eroded surface.
Analysis of test data shows that cohesive soil is capable to with stand a certain
number of loading cycles soils (Mirtskhoulava, 1966). Erosion stability depends on level
of the highest velocity fluctuations, The higher the level the lower the number of cycles.
Together with the loading level, the duration of velocity fluctuations and frequency of
dynamic action also control the erosion stability. The latter may increase with growing
frequency as for the same velocity level and high frequencies soil is resistant to a greater
number of cycles than at low frequencies
The above approach may be very useful to explain many properties of erosion of
cohesive soils soils (Mirtskhoulava, 1966; 1966; 1967; 1970; 1970; 1987; 1988; 1989).
in particular, for the latent period prior to initial erosion. The author proposed a fatigue
theory of erosion in cohesive soil as early as in the second half of the 1950s
(Mirtskhoulava, 1988). This approach seems to be applied later in corrosion of metal
and other resistant materials soils (Mirtskhoulava, 1988). The fatigue theory of wear
appeared even later (Mirtskhoulava, 1988).
Analysis of experimental data has proved that the non-erodible velocities depend
primarily on the size and form of sediments. cohesion between particles, physical and
mechanical properties of materials and water, depth of flow, slopes of bed and free
surface, longitudinal and lateral velocities of flow, turbulence, infiltration capacity of
soils, surface tension between grains, form of bed, angle of slope and geometry of flow
in the channel, time of the action of flow, and variability of all these parameters. This
simple list indicates that an accurate analytical solution for non-erodible velocities is a
very complicated task and, therefore, the appropriate laboratory and field methods
should be improved.

108
Non-erodible velocity of water in cohesive (clay) soils.

The absence of manuals for non-eroding velocity in channels with cohesive bed often
result in overestimation and overdesign of channels and erosion and damages
(Mirtskhoulava, 1966; 1988; 1989).
In the analysis, all soils are replaced by idealized media characterized by
homogeneity and continuity or some other properties discussed earlier.
Erosion is the detachment of aggregates and units linked by cohesion. The
characteristic dimension is taken as the mean diameter of the sphere equivalent by
volume to the detaching units. The latter is immediately removed by flow. The duration
of time required for a detachment depends on flow parameters and properties of soil.
Erosion depends on the quantity of loading cycles caused by near-bed velocity
fluctuations,
Analysis of the process shows that a clear relationship exists between erosion
velocity and the detachment resistance (fracture fatigue strength).
Non-eroding velocity is highest value of mean flow velocity without channel
scour (destruction), inadmissible for normal operation.
The presented data for factors causing the scour of cohesive soils and
contemporary knowledge of channel flow dynamics make it possible to suggest the
method for a determination of permissible (non-scouring) flow velocities developed on
a sufficiently strict scientific basis. This method can guarantee the normal operation of a
structure at maximum reduction of its cost.
Without a detailed calculation given in (Mirtskhoulava, 1966; 1988; 1989)
equation for permissible non-eroding mean velocities Clri.perand near-bed velocities
V .ln.,jerin cohesive soil under plane turbulent flow will read:

(1)

Introduction of the factor fi as a ratio of the eroding and permissible non-

eroding velocities is caused by a necessity of providing a sufficient erosion safety.
At present the inclusion of the swelling stress ooh and particularly its numerical
value necessary for practical computation is difficult. Therefore, the relationship for
permissible non-eroding near-bed and mean velocities are taken as follows
(Mirtskhoulava, 1967; 1988; 1989).

VA,,. = 1.25 --[(ps

2m - p,,)gd + 1.25(‘;k]
d 1.3P,,H

Vanper = 1.25 --[(p,

2m - p,,)p’ + 1.25+] (4)
2.6p,,rz

(5)

109
 In the derivation of Eqs.(3) - (5) one assumed a linear elasticity of the units
detaching during erosion. In general, however, the clay soil is not ideally elastic. Full
deformation prior to erosion is a sum of elastic and a certain additional plastic
deformations.
Inclusion of plastic properties is somehow difficult, so that a certain simplification
of the solution seems to be appropriate. One of the methods of approximate analysis of
plastic and other properties of soil consists in finding erosion resistance characteristics
via so-called elasto-plastic tensile fatigue strength.
The quantity depends on the number of factors. of which chief ones are cohesion
plasticity, water strength of bonds. strain modules, porosity. breaking and conditions of
the generation of crack, and swelling.
Adding to these expressions the near-bed and permissible (non-erodible)
velocities and erodible near-bed velocities, ones obtains

(7)

2m
VAsc = 1.25 [(I% - P,,)d + 1.q] (8)
0.72p,n

Upon comparison of Eq.(7) with inclusion of plasticity versus Eq.(3) and Eq.(4)
one may see that the inclusion of plasticity makes it possible to increase the permissible
velocities by the factor of 1.3
Analysis of tabular data provided in (Mirtskhoulava, 1967) with part of
numerous results only, indicates that the design scheme proposed and the formula
obtained are adequate for a description of erosion in a wide range of input parameters.
Mean values of the ten practical values observed are grouped around the design value
with probability of not less than 90%. Recalling the stochastic nature of erosion. one
may accept this scatter.
Without detailed calculations given in (5) permissible (non-eroding). Velocities of
water in cohesive soils in terms of reliability theory can be determined as follows:
2m
VAnper = 1.25 ___ (9)
2 Won
K Ps - P,,)Pf + 1.25(‘s]

Computation by these formulae requires the knowledge of Go ,,, , oHd. Heaving

data on cohesion forces in the condition of full saturation one may easily find oKd. As
for or. , being standard deviation for z’,~depending on the square of velocity
fluctuation in the near-bed layer, it is computed by data measured for velocity
fluctuations of flows with similar conditions. This parameter may be found on the basis

110
of various correlations for systematic variances and correlation functions of stationary
random processes measured and recorded by well-known methods.
Using the above formulae and the characteristic reliability function one may
determine permissible (non-erodible) velocities of flow which secure failureless
operation of the eroded bed with a given reliability determined by the probability Y. The
formulae make it possible to account for the duration of erodible flow t, number of
outbursts, variance of detachment stresses and erosion resistance of aggregates (which
depend respectively on variability of cohesion force and near-bed velocity fluctuations).

Land erosion on erosion model.

Erosion is one of the most prevailing and serious diseases of the soil cover. Erosion and
particularly after-effect of erosional processes (such as intensive wash-off and scour of
soil, bogging and some other processes) are the main reasons of ecological
deformations of agrosystems and failure of many lands.
Non-eroding velocity of overland flow, hydraulic resistances, influence of
humidity, man’s farming activity, etc., together with traditional indices are taken into
account in the suggested relation for slope erosion forecast. Joining the description of
erosion to exact sciences, when analyzing the procedure, allows us to use the
achievements of exact sciences that were impossible until now,
The quantity of wash-off soil from a plot (having unit width and length) in a
watershed to the end of the erodible part of slope can be determined by formulae. For
the determination of the total wash-off amount. the obtained result is multiplied by the
slope area (cab), where his the slope width: (Mirtskhoulava, 1970; 1987: 198 1)

9 x zT= 11. lo-“#$wd

I
L
308(an,)” ‘in ‘rn: ‘I(“x~~ +13. 10-6v;;,I
Kpfr ((7 n,,)i’ “‘mf ”
Observation and experiments have shown that after rain water fills all
(12)

microcavities of the relief and it initiates the sheet flow that is discontinuous and there is
a large number of small streams that flow, then. together due to their motion down the
slope and form a brook.
Number and power of these streams depend on the shape of lend surface and
character of soil cultivation. The more uniform the soil cultivation the more streams
come per unit of slope width and the less the value m is. the value m is less at cross
ploughing than at longitudinal one etc. For approximate calculations we can take
m=1.2+3.0 according to Kostjakov A.N. (Mirtskhoulava, 1987)

Maximum allowable erosion rate

Scientific and engineering projects provide soil protection against erosion (including
catastrophic events) but the cost of the protection is still very high (Mirtskhoulava,
1970). Therefore, at present we are reconciled to soil erosion observation until erosion
becomes critical (Mirtskhoulava, 1970; 1987; 1989).

111
 In our opinion, there is a need for quantitative limits of permissible erosion rate
using such indices as service life and effective thickness of the soil cover, i.e. the
difference between full initial thickness of the soil cover and the maximum permissible
one (H-HP,,); here the maximum permissible thickness means enough soil to create the

mmuy
conditions both for plant growth and for agricultural machines operation. In this case
the decrease (due to erosion) of soil thickness below the permissible level leads to
economic consequences.
&-Hperr taking into account the permissible erosion rate can be
expressed by the following relation
H,, - Hper H,,.
Ipcr = (13)
c =- ( ',-
where Hclcy is effective thickness of the soil cover (mm), C’,.is service life of the
soil cover, per year.
When the erosion rate is reduced during some period of time, the failure
(Mirtskhoulava, 1987) will take place at the thickness of the soil cover H(t) decreasing
gradually during time (I), being equal or less than the permissible H value.
The normal distribution law can be used for the description (Mirtskhoulava.
1987)

(14)

Equation (14) can be written as equation of bond (nonexceedance equation) when

using the table of normal distribution function

(15)

where: H(t) is initial thickness of the soil cover, HP,,.is permissible maximum fixed
level of the soil cover thickness at which the soil normally functions and below this
value the normal conditions are disturbed (the failure begins).
Thickness of soil H(t) at a moment oft (as a result of the erosion influence) can
be considered as a random quantity the value of which is decreasing in time t. The
thickness may be assumed to decrease according to this law.

mm
year,
t
H(t) r Ho- It (16)
where Ho is initial soil thickness, I is total mean annual rate of erosion, the linear
dimension of which is taken according to values H and t. i.e. if the thickness is in
millimeters the dimensionality I is taken in is time (in years).
Standard deviation O, and o~~~,. are established from observation data on
analogous areas or by the use of Taylor series expansion. When taking the necessary
data for approximate estimates and calculations, the “three-sigma” rule may be used
(Mirtskhoulava, 1987).
We will illustrate the solution of the problem with a hypothetic numerical
example.

112
 The initial thickness of the soil cover H=95 sm. and the maximum allowable
annual erosion rate at a given probability of failure-free operation P=O.975 that
corresponds to value Z=1.96. Time of soil functioning ~600 years.
Using “three-sigma” rule and tolerances (to simplify calculations) equal to 20% of
a mean value of parameter. CT~and oHperare 0 N = 0.2H 0 ; o Nper=0.2HI,,r cr.
H’-2H per+ HfC!.
These values are introduced into eq.( 15): Z’ = ,
0; I& Hp0
After simpification, we obtain quadratic equation.
/ - 0.23 171! + 0.0130 = 0; j = 0.1365 cm.; ZJ = 0.0953 cm respectively
Thus, for soil the initial thickness of which H = 95 cm and maximum allowable
one 25.5 cm at soil service life of 600 years, it may be allowed 1 = 1.365 mm/years
(minimum value from square root of the equation). At such erosion intensity the failure
free soil functioning of probability 0.975 (Z = 1.96 corresponds to P = 0.975) will be
provided for 600 years,

Conclusions
General tendency in the environmental protection exhibits that the problem of erosion
and its forecast various types and stages will be given more and more attention. and
that it requires further theoretical and experimental efforts.
Our investigations were direction toward a refinement of the main physical and
mechanical properties of erosion stability in various soils and materials with the
inclusion of dynamic action of flow and its hydromechanical parameters.
Hydraulic investigations of the problem should go hand in hand with fundamental
investigations by hydromechanical methods and in terms of the mechanics of soils and
materials.
Statistical methods are very important for a description of erosion, because the
present disagreement in various relationships is obviously linked to an insufficient
consideration of the variability of the factors involved.
New prospects to lower the risk in stability analysis of erosion in channels as in
the assessmentof the variability of soil and flow parameters are still latent in the random
of channel changes factors controlling these changes.

Symbols
c&!.s - design specific cohesion of cohesive soil
I: - free fall acceleration
H - depth of flow

m-
I - free surface slope
K - coefficient to characterize probability of deflection of cohesion from mean
value
coefficient of working conditions including the effect of various factors in
channel conditions

113
 - overload coefficient accounting for eroding capacity of flow due to velocity
fluctuations
- roughness coefficient
- probability
- mean permissible non-erodible velocity averaged over cross-section
- roughness height
- density of water
- density of soil grains
- hill slope erosion ( t/ha“ )
- the average frequency of pulsation velocities; it can be established according
to the Strouhal number o = 0.73WH where Y is the mean velocity of the
sheet flow and H is the sheet flow depth; when data are lacking to compute
0, it was assumed to be equal to 10 s-r ; (Mirtskhoulava, 1970; 1987; 1988;
1989)
the mean diameter of entrained aggregates, assumed equal to 0.004 m
- (Mirtskhoulava, 1967)
- a coefficient accounting for the deviation of the sheet flow motion from the
accepted smooth water surface motion
- the roughness coefficient (Mannigs coefficient)
- the length of the plot (m)
- the duration of excessive precipitation (s)

References
Fottier, S.; Scobey, F.C. (1923) Permissible canal velocities. Trans ASCE, Paper 1588.
Lane, E.W. (1953) Progress report on studies of the design of the stable channels by
the Bureau of Reclamation. Proc. ASCE, ~01.79, ’ 280.
Mirtskhoulava, Ts.E. (1966) Studies on permissible velocities for soils and facings,
Modern trends in Hydraulic Engineering Research, Central Water and Power
Research Station, Poona. vol. 1 1.
Mirtskhoulava, Ts.E. (1966) Erosional Stability of Cohesive Soil. J. Hydraulic Rec.,
vol. 4.
Mirtskhoulava, TsE. (1967) Bed scour and estimation methods of channel stability.
Moscow, Kolos, 180 pp. (in Russian)
Mirtskhoulava, TsE. (1970) Engineering methods for design and forecast of water
erosion. Moscow, Kolos, 240 pp.
Mirtskhoulava, Ts.E. (1970) water erosion, factors conditioning it and methods of
prognosis. Proc. Intern. Water Erosion Symp., Prague.
Mirtskhoulava, Ts.E. (1987) Reliability of hydro-reclamation installations.
A.A.Balkema Publishers, Brookfield.
Mirtskhoulava, Ts.E. (1988) Basic physics and mechanics of channel erosion.
Leningrad, Gidrometeoizdat.
Mirtskhoulava, Ts.E. (1989) Basic physics and mechanics of channel erosion. Delft,
The Netherlands.

114
Mirtskhoulava, Ts.E. (198 1) Land erosion, research equipment, forecasting methods
and prospect for their improvement. Erosion and Sediment Transport
Measurement (Proceeding of the Florence Symposium 198 1) IAHS, Pub]. No 133.

115 ,
 2. SEDIMENT TRANSPORT

I_- -__-
Sediment non-uniformity effects on entrainment
and transport
by K.G. Ranga Raju and M.K. Mittal
Department of Civil Engineering, University of Roorkee, India

ABSTRACT
Entrainment and transport characteristics of uniform
sediments are reasonably well understood. These mechanisms are
rather complex in case of sediment mixtures and are, as such, not
fully unders toad. at present. In the case of non-unlform sediments
the finer particles are shielded by the coarser ones and the
latter, in turn, are more exposed to the flow. This results in the
possibility of entrainment of different sizes under different flow
conditions. Also the transport rates of the different sizes would
be different from each other for a given flow condition and some
sizes may not move at all. Consideration of these effects is vital
in computation of reservoir sedimentation rates and degradation
including armouring effects. A thorough review of the state-of-art
related to initiation of motion and transport of non-uniform
sediments is presented.

1.0 INTRODUCTION

Natural sediments are invariably non-uniform. The extent of

non-uniformity of the bed material of a river is dependent on the
distance from the source and the slope apart from several other
factors related to the geomorphology of the catchment. In general,
as one moves along the length of an alluvial stream, one
encounters a reduction of stream slope and a reduction in the size
of the sediment. Analysis of the size distribution of river
sediments has shown (11 that the geometric standard deviation of
the mixture, u increases with increase in the median size, d
g’ 50;
see Fig. 1. Here

1 dS4 d50
(1)
=g =- 2 I-Yg + -1
d16

in which d and d are the sizes for uhi ch 54% and 16% of the
84 16
material are finer. The relaticnship shovn in Fig. 1 shovs that a

116
10.0 I I I I 1 I I I I ,

8 - l Gorda

0.1 0.2 O-4 0.6 1.0 2 4

dso (mm)

Fig .l variation of rs with d50 (1

d (mm)

down straom
of Kosi borrogQ 1
I
Ganga (at Hardwar)

0 I I I ,
-2
6x10
d (mm)

Fig. 2 Size distribution of thQ bad materials of the KoSi

and Congo rivrzrs

117
large spread of sizes may be expected in the case of coarse
sedimcnl.s (1, I?. Ill t.t1c Ilppcr r-c!nctlcn) ) wt~rr~cns I.tlc! nnn-~~n 1 form 1 Ly
I :i IIOL :;o t’1~0110~111’:‘!‘1 Ill L11c lower r~cnctlos of‘ il rlvcr, whor-c LllC
malcr~li~l Is f Inor-. ‘I’ho pronounced non-un I form1 Ly of lho bed
ma Lor- la 1 0 I‘ LI\C G;lrl~;l river al tI;1I~dw;lr‘ (India) in Its upper
I’CilCtlf!!; I !;cIc;rl-ty dlsco~~t~lblr? l’rorn I’lg. 2. On Lhc oltlcr InrId,
pracLlcally all the bed material is wlthin a size range of 0.10 mm
to 1.0 mm for the Kosi river (India) in its lower reaches as may
also be seen from Fig. 2. Further the discharge in a river tends
to increase in the downstream direction, which means the river
generally has a greater capacity to transport sediment in the
lower reaches. These two facts lead to the strong possibility that
practically all the size fractions tend to move (and probably
mostly in suspension) in the lower reaches, whereas the condition
of partial transport, i.e. one in which some sizes do not move, is
likely to be encountered in the uppe.r reaches. Such reaches would
also be characterised by a significant amount of bed load
transport and relatively little suspended transport. Thus
consideration of the effects of sediment non-uniformity on
entrainment and transport appears very necessary in case of
coarser sediments likely to move essentially as bed load.

Mathematical models are in extensive use today for computing

bed level changes in alluvial rivers. The phenomenon of partial
transport as well as that of different size fractions having
different transport rates even for a given flow condition, need to
be accounted for in making meaningful computations in such
practical problems as reservoir sedimentation and degradation
downstream of dams. Referring to Fig. 3, one can see that the
median sizes of the bed material, bed load and suspended load are
all different in the case of a highly non-uniform sediment,
thereby making the use of a single representative size of the bed
material like dSO extremely questionable. Furthermore, Fig. 3
shows that the fractions of a given size in the bed material, bed
load and suspended load, viz. ib, IR, and is respectively, are all
different making it necessary to make sediment load computalions
fractionwise. It may also be seen that there is a size beyond
which the bed material is not moved at a given shear stress and it
is necessary to get information on the relationship bctuecn this
size and the flow conditions. In other words, information is
required on the incipient motion condition for the different
fractions in a sediment mixture.

The objective of this paper is to summarise and critically

evaluate the existing information on the critical tractive stress
of sediment mixtures and on the methods of calculating bed load by
size fractions.

118
 2.0 CRITICAL TRACTIVE STRESS

Although the critical condition, i.e the condition of

incipient motion, has been expressed in terms of the critical
velocity, critical lift or critical tractive stress, it is the
concept of critical tractive stress which has found the greatest
favour amongst engineers. It is generally accepted that T* is a
C
unique function of Rz(see Ref. 1) for uniform sediments. Here

T
C
T’
C
= (21
(P, - af)d

and
U .C d
(3)
R; =
u

in which d is the sediment size, T is the critical shear stress,

C
u is the kinematic viscosity of the flowing liquid, Y, and rf are
the specific weights of sediment and fluid respectively and u,~ =
qFr 9 pf being the mass density of the fluid.

The flow over a bed consisting of sediment of different sizes

is shown in Fig. 4. It is obvious that the coarser fractions
experience a larger force than they would if they were part of a
uniform-sediment bed, on account of the fact that they are more
exposed to the flow in case of nonuniform-sediment beds. On the
other hand, the finer fractions are shielded by the coarser ones
and are consequently subject to smaller fluid forces than in the
case of uniform sediments. In other words, different size
fractions may be under the incipient motion condition,at different
shear stresses. In fact the experimentation as well as any
theoretical analysis for the case of initiation of motion in case
of sediment mixtures is more comPlicated than for the case of
uniform sediments.

2.1 Definition of Initiation of Hotion for Mixtures

The critical shear stress of sediment mixtures has been

defined using two methods, viz. the Largest Grain Method (LGM) and
the Reference Transport Method (RTM). LGM involves determination
of the largest size in a sediment mixture that is moved by a given
shear stress. This shear stress corresponds to the critical shear
stress of that largest size provided coarser sizes are available

119

__.. ____--- - -
 mata
1C

\
/ gad riot
;;
C
.-
c
,
Y

C
0,
LJ
di = AVQ~O~Q of size 2

_I
L

ronga dl to d2
i!

( dds dl di dz (dso)F ’5&

SiZQ

Fig.3 Diagram showing typical 5izQ distribution of

bad-tnotariof load and bQd-mOtQriO1

(a) Uniform sadimant

of siza dl
‘d %

l-q&$-La

dl
(C 1 Non- uniform sudimQnt

(b) uniform sediment

of siza d2

Fig. 4 Illustration of hiding and QxposurQ QffQct:,

in cosQ of non-uniform sadimQnts

120

-____~- -. ---- --__ _.-_-._--

.__. -.-. -.-~
 on the bed. In the RTM transport rates of individual fractions are
measured for a number of discharges and the shear stress
corresponding to a small reference transport is determined (as the
critical shear stress) from a relation fitted between the
dimensionless critical shear stress and the dimensionless
transport of each fraction. Parker and Klingeman (2) recommend
that the dimensionless transport parameter W’ be 0.002 under the
critical condition. Here

qB (G - 1)
w* = (4)
;r, d-ij-- (RbS) 3’2

in which G is the relative denslty of the sediment, qR 1s the bed

load transport rate per unit width, Rb is the hydraulic radius
with respect to the bed, S is the slope and g the gravitational
acceleration. Wilcock and Suthard (3) present a critical appraisal
of the two methods.

2.2 Criteria for Initiation of Motion

The first mathematical treatment of the influence of

non-uniform size distribution on initiation of motion was
developed by Egiazaroff (4). He proposed a method based on the
assumption that the velocity at a distance of 0.63 times the grain
diameter is equal to the fall velocity of the sediment. llis
expression for the dimensionless critical shear stress for any
size d., viz. 7: is
1 l ci ’

in which d a is the arithmetic mean size of the sediment. Ashida

and Michiue (5) proposed a modification of Eq. (5) for di/da less
than 0.4. Their equation is

T* ci dl -1
= 0.85 d (6)
T 1 a 1
*ca

121

-__-
in which T ,ca is the dimensionless critical shear stress for the
arithmetic mean size. Hayashl et al. (6) recommend the following
equations :

K
*ci
= for dl/da < 1.0 (7al
7
*ca

K
t
*ci

*ca
=
log
log 8
8 d/d,
1
2
for di/da 2 1.0 (7b)

All the foregoing relationships are shown in Fig. 5 in which

the experimental data of Misri(7) obtained using LGM are plotted.
The data show good agreement with the relation of Hayashl et al.

Determination of 7ci from Eq. (7) needs knowledge of ~~~~~

Garde and Ranga Raju (1) proposed a value of 0.032 for -caca but
the recent analysis of Ranga Raju and Pate1 (8) shows considerable
variation of T with changes in u . As such, no satisfactory
*ca i.2
method is available for the determination of T +ca at present.

Parker and Klingeman (2) find the dimensionless critical

shear stress for the median size, T to be 0.0876, whereas much
‘50’
lower values are proposed by Gessler (9). Using ~~5~ = 0.0876,
Parker and Klingeman write the critical shear stress formula as

= di -0.982
K* ci 0.0876
I-lda (8)

Yet another form of equation for ~,c~ has been used by some other
investigators like Ashworth and Ferguson (101, Niekerk et al. (11)
and Wilcock and Suthard (3). They write

‘I:

*ci
= Id’ (9)
r*50 1 dso

122
 0.I 0.2 0.4 0.6 1.0 2 4 6
di/da

Fig. 5 Critical tractive stress of

non-uniform materials

t / V.3”
I-- L-1 -

I.50
y
3.00
5.00
10-
I ,~11111 I I I11111 I II-
6 10-2 2 46,$ 2 46 1002 4

Fig.6 Rulationship proposed by Wiberg ond Smith (13)

for criticul tractive stress

123

___ --
The variation of ~~50 and /3 obscr-vcd by thcsc invcsligalor~s 1s
from 0.024 to 0.087 and from -0.65 to -1.06 respectively. Recently
Wilcock (121 used an equation of the form

T ‘*i B,
*ci
= a- (101
=* a I S I
*a

in which S,. and S are the dimensionless critical shear stresses

1 ‘a
(treating the sediment as uniform) for sizes and d
di a
respectively as per Shields (see Ref. 11 and a and 131,are
empirical constants. The variation of a and 0, with changes in the
sediment characteristics still remains to be fully understood.

Following a theoretical analysis and using some experimental

data, Wiberg and Smith (131 proposed a graphical relationship for
-IIl ci in the form of Fig. 6. Here k is the equivalent sand grain
5
roughness of the bed. By making di/ks a parameter in the analysis,
Wiberg and Smith account for both sheltering and exposure effects
on critical tractive stress. Some limitations of the methods of
Wilcock (121 and Wiberg and Smith (131 have been pointed out by
Kuhnle (141 on the basis of his analysis of laboratory and field
data.

2.3 Comments

From the review of the relationships presented it is obvious

that information regarding the initiation of motion of sediment
mixtures is somewhat less conclusive than for uniform sediments,
for which the Shields’ type of formulation is generally considered
satisfactory.

3.0 BED LOAD TRANSPORT

Eversince the pioneering work of Einstein (151, it has been

recognised that calculation of sediment transport rates of highly
non-uniform sediments using a single representative size of the
bed material would not be logical. It is obvious that the median
size of the transported material would be a function of the
discharge, particularly when all fractions of the bed material are
not moving. Figure 7 is a typical bed load plot based on
laboratory data showing clearly the hiding effect for the finer
fractions and the exposure effect for the coarser fractions of the
bed material. These effects are usually accounted for with the

124
 2.01 J ’ 1 J I,,, I I I Illll~ J , ,,,,,I I I 111111
di(rnm\

I
I
I I I I , , I, I
0.01 . I ,111111 I I 111111 I , I I,,,,

lo2 10’ loo 10’ IO2

*B

Fig.7 Variation of 0B with c’o for individual fractions

in a mixtura AXsdi

10’ I I I lllll 1 I I lllll I I I I Illl( 1 I I111111

0 Clear Watar river

4 Snake river Field
data
n Baar river
+ East-Fork river

0 Einstein
A Misri Fltime
0 HRS data
x Samaga

io2 I I !llllS I I I I III1 I I I I1111 I I I lllll

lo2 12 IO0 10’ ,02

KB Tk/(A1(sdi)

Fig.0 variation ot LB ?B
with g, -C; / (AYsd I (25)
125
help of a hiding-cum-exposure coefficient,<R, defined as ~~~~~~~~
where T = the effective shear stress which glves the transport
eff
rate of size d equal to that of the uniform material of the same
i
size as per the relation between r: and #R. Here #R is the
dimensionless transport parameter for an individual size di and is
defined as (i,q,l/(ib~sdil d (af/(Aasl g di and -c;=To/(Axsdi 1;
igqg is the rate of bed load transport for size d
i’ AT,= rs- ;rf’
and ho= 7f R’S,, R’ being the hydraulic radius corresponding to
grain roughness. Computation of bed load transport rates
fractionwise really boils down to the development of a method for
the determination of CR.

3.1 Semi-Analytical Approaches

Amongst the semi-analytical approaches to the problem of

calculations of bed load transport rate fractionwise, no reference
is made to the method of Einstein (15) herein, because it is known
to ignore the exposure effect totally and is found to result (16)
in poor predictions for the coarser fractions.

Misri et al. (16) proposed a conceptual model for the

determination of <, based on the following premises :

(i) Lift is the predominant force on particles finer than the

arithmetic mean size, whereas drag is the predominant force
for the coarser particles.

ii 1 The pressure in the wake behind partic les of. size da is

constant.

(iii) The velocity distribution may be described by the

logarithmic law uith the datum located at 0.5 da below the
top of the grains.

(iv) The effective shear stress for particles of size d a is the

grain shear stress, TV’.

Their relationships for <, may be expressed as follows :

d
i di
eB = 7 for --d--- < 1.0 (11)
a a

126
and

cn = 0.72 log 30.2 [(d/d,) - 0.501

di
for --d-- t 1.0 (12)
a

While experimental data indicate qualitative agreement with Eqs.

(111 and (121, the quantitative agreement is not satisfactory.

The analytical model for bed load transport of sediment

mixtures proposed by Niekerk et al. (11) has the following
features :

(i) It uses a critical tractive stress relationship of the type

given by Eq. (9).

(ii) The bed shear stress is assumed to be fluctuating with time

and the transport rate for infinitesimal time increments is
calculated considering this variation and these rates summed
up to get the total transport rate for the given average
shear stress.

(iii) A modified Bagnold equation in which the transport rate is

related to the difference of the bed shear stress and the
critical shear stress is used for fractionwise calculation
of transport.

Al though the model was validated using some field and flume
data, further check on the model is required particularly in view
of the uncertainties associated with the estimation of critical
tractive stress from Eq. (9).

3.2 Empirical Approaches

Empirical .relations for 5, or dlrec tlY for the transport

rates of individual fractions have been proposed by a number of
investigators like Einstein (151, Einstein and Chien (171,
Pemberton (181, Samaga et al. (191, Proffitt and Sutherland (201,
Ashida and Michiue (21) and Samaga et al. (22). The limitations of
these methods as indicated by an analysis of a large amount of
flume and field data have been brought out by Misri et al. (161,
Samaga (23 1 and Ranga Raju et al. (241. In view of the
shortcomings of most of the foregoing methods, Mittal et al. (25)
suggested a modification to the method of Samaga et al. (22) by
reanalysis of all the available data. The modified transport law
may be summarised as below :

127
 There is a unique relation between 4, and <, T: given by
Table 1.

TABLE 1

RELATION BETWEEN #B AND cB’;

E, ‘: 0.02 0.035 0.052 0.105 0.23 0.42 1.40 5.0

lo+ 1o-4 4x10 -3 10-l 1.0 4.0 25.0 94.0

@B

The hiding-cum-exposure parameter,,EB, may be obtained from Fig.

8, in which LB is a function of Kramer’s Uniformity Coefficient,
M (see Ref. 11 as per Table 2 and KB is related to ~o/~c as shown
in Fig. 9.Here xc is the critical shear stress (as per Shields)
for the arithmetic mean size of the mixture.

TABLE 2

RELATION BETWEEN LB AND II

M 5 0.25 0.30 0.35 2 0.40

0.625 0.769 0.909 1.0

LB

A comparison of the transport rates computed using Tables 1

and 2 and Figs. 8 and 9 with the measured rates for a vast amount
of data is shown in Fig. 10. While the agreement for these data is
satisfactory, a recent check on the method (261 using the data of
Pate1 and Ranga Raju (261, Wilcock and Mcardell (271 has not
yielded satisfactory results; see Fig. 11. It is worth pointing
out that the value of M for the data of Wilcock and Mcardell was
0.073. The lack of agreement of these data with the method of
Mittal et al., thus raises doubts about the constancy of LB for
values of M less than 0.25 as given in Table 2.

128
 Fig.9 RQlation batwaan Ke and Ii/‘cc (25)

129

~ ~-. .-...--- - .
 Computed bad load, N/ms

Linu ot agraement

10C

102 0
0
0
0
+

o HRS 1
0 0
0 V Einstein
0 Lab dato
0 0 a Misri
0
0 x Samaga 1
IO3 0 Clear Water riw r
/ Snaka riwr
Fiald data
O\\’ + East Fork river
l Baar rivar I -

I/ , I III I I II I I III I I III I 111

-5 -4
10 10 Ii3 102 G’
Computad bad load, N/mS ~

Fig. 10 Computad and o b sarvad bed load transport rates

tor individual fractions using Mittal at al.‘s method(25)

130

-------
 Ralation as per /
Tabla
o POtUl and Ranga Raju (26) 0

v wilcock and MCOrdQli (27)

0
0 0
0 OO&
0
0 V
0. 0 0
/
vv
/ o” v v V
0
O 0
OO vdJ v
0 ovv
%I
V v 7
WV
vvv

0.01 -6 , I I I I II, I , !,I,,, I I,,1111 I I I,,,,, I I!IItll I I 1,111, t , ,‘!,,L

10 105 10
-4 103 1.0 10

Fig. 11 Chack on 5e ‘c; /( hYsdiI versus !&I relation

131
4.0 CONCLUSIONS

The review of the state-of-art on entrainment and bed load

transport of coarse non-uniform sediment presented in the paper
has clearly brought out divergences in the different approaches to
the problem. While Fig. 5 is a reasonable indicator of the
relative mobility of the different fractions, its use in actual
computations is hindered by the lack of a suitable predictor for
T So far as Eq. (101 is concerned, its use in critical
*ca’
tractive stress studies is limited by the uncertainties associated
with a and p
1’

The method of Mittal et al. (25) appears to be better than

several other methods in terms of accuracy so far a’s the
computation of bed load transport rates of individual fractions is
concerned. But its inability to yield good results in cases of
extreme non-uniformity (i.e. low M values) makes it desirable to
develop an analytical model for the variation of <B by making
appropriate modifications in the approaches of Misri et al. (161
and Niekerk et al. (11).

5.0 REFERENCES

1. Garde, R. J. and Ranga Raju, K.G., Mechanics of Sediment

Transportation and Alluvial Stream Problems, Second Edition
by Wiley Eastern Ltd., New Delhi 1985.

2. Parker, G. and Kingeman, P.C. , On Why Gravel Bed Streams are

Paved, Water Resources Research, Vol. 18, No. 5,October 1982.

3. Wilcock, P.R. and Suthard, J.B., Experimental Study of

Incipient Motion In Mixed Size Sediments, Water Resources
Research, Vol. 24, No. 7, 1988.

4. Egiazaroff, I.V., Calculation of Nonuniform Sediment

Concentrations, Journal of Hydraulics Division, Proc. ASCE,
Vol. 91, No. HY-4, 1965.

5. Ashida, K. and Michiue, M., An Investigation of River Bed

Degradation Downstream of a Dam, Proc. 14th Congress, JAHR,
Vol. 3. 1971.

6. Hayashi T, Ozaki, S. and Ichibashi, T., Study on Bed Load

Transport of Sediment Mixture, Proc. of 24th Japanese
Conference on Hydraulics, 1980.

7. Misri, R.L., Partial Bed Load Transport of Coarse Nonuniform

Sediment, Ph.D. Thesis, University of Roorkee, Roorkee,
India, 1984.

132
 8. Ranga Raju, K.G. and Patel, P.L., Erosion and Deposition of
Nonuniform.Sediments, Report Submitted to CSIR, New Delhi,
India, Oct. 1993.

9. Gessler, J., The Beginning of Bed Load Movement of Mixtures

Investigated as Natural Armouring in Channels,’ Report No.
69, Swiss Federal Institute of Technology, Zurich, Dec.
1965 (Translation).

10. Ashworth, P. J. and Ferguson, R. I. , Size Selective

Entrainment of Bed Load in Gravel Bed Stream, Water
Resources Research, Vol. 25, No. 4, April 1985.

11. Niekerk, A.V., Vogel, K.R., Slingerland, R.L. and Bridge,

J.S., Routing of Heterogeneous Sediments over Movable Bed :
Model Development, Jour. of Hyd. Engg., Proc. ASCE, Vol.
118, No. 2, Feb. 1992.

12. Wilcock, P.R., The Critical Shear Stress of Natural

Sediments, Jour. of Hyd. Engg., Proc. ASCE, Vol. 119, No.
4, April 1993.

13. Wiberg, P.L. and Smith, J.D., Calculations of the Critical

Shear Stress for Motion of Uniform and Heterogeneous
Sediments,Water Resources Research, Vol. 23, No. 8, Aug.
1987.

14. Kuhnle, R.A., Incipient Motion of Sand-Gravel Sediment

Mixtures, Jour. of Hyd. Engg., Proc. ASCE, Vol. 119, No.
12, Dec. 1993.

15. Einstein, H.A., The Bed Load Function for Sediment

Transportation in Open Channel Flows, United States
Department of Agriculture, Tech. Bull. No. 1026, 1950.

16. Misri. R-L., Garde, R.J. and Ranga Aaju, K.G., Bed Load
Transport of Coarse Nonuniform Sediment, Jour. of Hyd.
Div., Proc. ASCE, Vol. 110, No. 3, March 1984.

17. Einstein, H.A. and Ning Chien, Effect of Sediment

Concentration near the Bed on the Velocity and Sediment
Distribution, Institute of Engineering Research, University
of California, Berkeley, Series No. 33, 1955.

18. Pemberton, E.L., Einstein’s Bed Load Function Applied to

Channel Design and Degradation, Sedimentation Symposium to
Honour Prof. H.A. Einstein, Fort Collins, Colorado, U.S.A.,
1972.

19. Samaga, B.R., Ranga Raju, K.G. and Garde, R.J., Some Aspects
of Transport of Non-uniform Sediment, Proc. of the Second
International Workshop on Alluvial River Problems, Roorkee,
India, 1985.

133

\ -
20. Proffitt, G.J., and Sutherland,A.J., Transport of Non-uniform
Sediment, Jour. of Hyd. Res., IAHR, Vol. 21, No. 1, 1983.

21. Ashida, K. and Michiue, M., Studies on Bed Load Transport

Rate in Open Channel Flows, Proc. of the International
Symposium on River Mechanics, Bangkok, 1973.

22. Samaga, B.R., Ranga Raju, K.G. and Garde, R.J., Bed Load
Transport of Sediment Mixtures, Jour. of Hyd. Engg. . Proc.
ASCE, Vol. 112, No. 11, Nov. 1986.

23. Samaga, B.R., Total Load Transport of Sediment Mixtures,

Ph.D. Thesis, University of Roorkee, Roorkee, India, 1984.

24. Ranga Raju, K.G., Mittal, M.K. and Porey, P.D., Hiding and
Exposure Effects in the Bed Load Transport of Nonuniform
Sediments, Proc. of the International Grain Sorting Seminar,
Ascona, Switzerland, 1991.

25. Mittal, M.K., Porey, P.D., and Ranga Raju, K.G., Bed Load
Transport of Non-uniform Sediments, Proc. of the Euromech
262 Colloquium on Sand Transport in Rivers, Estuaries and
the Sea, Wallingford, U.K., 1990.

26. Patel, P.L. and Ranga Raju, K.G., An Evaluation of the

Methods of Fractionwise Calculation of Bed Load, Proc. of
the National Symposium on Recent Trends in Design of
Hydraulic Structures, Roorkee, 1994.

27. Wilcock, P.R. and Mcardell, B.W., Surface Based Fractional

Transport Rates, Report of Dept. of Geology and
Environmental Engg. , The Johns Hopkins University,
Baltimore, Maryland, U.S.A., 1992.

134
Sediment transport patterns observed
in Southern African rivers
Prof A Rooseboom
Dept of Civil Engineering, University of Stellenbosch

INTRODUCTION
After more than 20 years involvement with sediment load data for southern African rivers, the
main impression which remains is the variability thereof.
Not only are daily loads of rivers extremely variable but even on an annual basis the
variability is such that longterm records are required in order to establish accurateestimatesof
averageannual loads.
It is therefore extremely risky to draw conclusions from limited records.

VARIABILITY OF DAILY SEDIMENT LOADS

The sediment loads of inland rivers in southern Africa consist mainly of particles smaller than
0,2 mm in diameter. These loads are carried in near-homogeneous suspension by flood
dischargesin rivers.
Sediment particles are initially entrained either by overland sheet flow or minor streamswhere
the carrying capacity becomes sufficient to transport the particles. From this point the
carrying capacity per unit volume of fluid tends to increase downstream except where
significant retardation takes place. By the time the transported sediments reach main river
courses, the carrying capacity tends to be far greater than that which is required to transport
the sediment load which is being fed in to the courses so that the availability of sediment for
transport is the limiting factor in determining the actual load where mainly fine sediments are
being transported. This is accentuatedby the fact that river beds consist mainly of coarser
materials which are not brought into suspensionas readily as finer particles.
Figure 1 depicts the scatter observed when recorded daily sediment concentrations are plotted
against discharge for typical rivers which carry fine sediments. It does not make much sense
to draw a single line through these points in order to obtain a so-called rating curve. Double
mass curves, which will be described further on, provide more meaningful relationships for
determining average sediment loads where these loads are carried mainly in suspension.

135
 CjO
2.60 0
vo '
0 0
2.60- 0
0
0 0 0 0
2.40- 0 n A
0 0 9
2.20 - 0 Q
0
2.00
1.80

1.60
1.40
1.20

0.80
0.60
P- I - I - I I
0.40 u
0
k!Do% 0 -0 0 0
0.20

0.00
50 100 150 200 250 300 350 400

DISCHARGE (m3/s)

FIGURE 1: CALEDON RIVER AT JAMMERSDRIFT : MAY 1969 - MAY 1976 : MEASURED

SUSPENDED SEDIMENT CONCENTRATION : DISCHARGE
VARIATIONS IN ANNUAL SEDIMENT LOADS
Even on an annual basisthe sedimentloads carried by rivers tend to be highly variable.
The most comprehensivelong term record of the load being carried by a river on an annual
basis in southern Africa is the combined record for the gauging stations at Prieska and
Upington on the lower Orange River. Although these stations are some distance apart, little
sediment is fed into the Orange River between them. Their sediment load records may
therefore be combined to form a continuousrecord for the period 1929 - 1969.
Thesedata have been plotted in the form of cumulative sedimentdischargevs time (Figure 2)
as well as 10 year moving averagedvalues vs time (Figure 3).
The following observationscan be made from thesefigures:
(i.) Annual loads vary greatly (Figure 2). The load carried during the hydrological year
1933 - 1934 of 282,4 million tonne was for instance25 times greater than that of 11,2
million tonne for the previous hydrological year. According to the original records the
load carried during a specific day during 1933 - 1934 was in excessof 25 million tonne
i.e. more than would be carried in many a year.
(ii.) Even when the 10 year moving averageis plotted (Figure 3) a large degreeof variability
is evident.
(iii.)Both Figures 2 and 3 indicate that the average sediment load decreasedbymore than
50% during the period 1929 - 1969. This decreasehas been attributed mainly to a
decrease in the availability of erodible soils (Rooseboom and Harmse, 1979).
Significant changesin averageloads have also been deducedfor other catchmentsfrom
reservoir surveys,but thesecannot be quantified with the samedegreeof accuracy.

DETERMINATION OF AVERAGE ANNUAL SEDIMENT LOADS FROM STREAM

SAMPLING RECORDS
Where continuous sediment sampling has been undertaken for a very long period it is
relatively easy to obtain representativeaverage yield values for different periods and to
identify periods during which averagedloads tendedto be constant(Figures 2 and 3).
Typically however one is forced to draw conclusions regarding average loads from much
shorter records.
It is common practice to draw a rating curve of sediment concentration as a function of
dischargebasedon short term sampling and to use this rating curve in conjunction with longer
term discharge records. A rating curve based on limited data may only be used where a
singular relationship does exist betweensedimentconcentrationand discharge. Rating curves
based on scattereddata may only be used where long, continuous records exist and should
only be used conjunction with flow records which cover the same period as the recorded
sediment concentrations. In the latter case, the preparation of a rating curve is really
meaninglessand alternative calculation methodsmay as well be used.

137
9 1
m
2 1

0k
c.3
a
a 1
&
cn
l-l 1

w
>

HYDROLOGICAL YEARS

FIGURE 2: CUMULATIVE SEDIMENT DISCHARGE

ORANGE RIVER BASIN : 1929 - 1969

138
 YEARS

FIGURE 3: 10 YEAR MOVING AVERAGES OF SEDIMENT DISCHARGE FOR THE

ORANGE RIVER CATCHMENT : COMBINED DATA FOR PRIESKA AND UPINGTON
Sediment concentrationstend to be high after long periods of low run-off from catchments
and vice versa. This explains the stable progression of double-masscurves of cumulative
sedimentdischargeversuscumulative water discharge(Figure 4).
The averaged slope of a double mass curve as depicted in Figure 4 representsthe average
sedimentconcentration. This value can be multiplied by the long term averageannual run-off
to obtain the averageannual sedimentload.
By determining the average slope of the best fitting line through points equidistant on the
double mass curve this averageconcentrationcan be determined. It is typically found that the
slope after say five years of record changesvery little.
Even when the averageannual dischargeduring the record period differs from the long-term
average, the average slope may be used together with the long term average discharge to
obtain an accurateestimateof the averagesedimentload.
It is thus possible to obtain an accurateestimateof the averagesedimentload from continuous
(daily) sampling recordsof 6 years and even less. This representsan enormoussaving in time
and costs relative to if say the averagedload is taken as the arithmetic mean of the individual
annual loads in which case longer sedimentrecords are required to obtain meaningful results
(compareFigure 3).

DETERMINATION OF AVERAGE SEDIMENT LOADS FROM RE-SURVEYS OF

RESERVOIR SEDIMENT DEPOSITS
Very little continuous (daily) sampling of sediment loads in rivers has been undertaken in
South Africa since 1971. This can be attributed to the facts that
(i.) the collection and processingof daily samplesis costly and cumbersome.
(ii.) regular resurveying of existing reservoirs is being done very efficiently by the
Department of Water Affairs and the number of reservoirs being covered provides a
wide spectrumof catchmentsfor which sedimentyields can be determinedwith relative
easeand at limited cost.
When sediment loads and yields are to be derived from reservoir re-surveys care should be
taken:
(i.) not to use depositsyounger than 8 - 10 years.
(ii.) not to use information for smaller reservoirswith ill-defined trap efficiencies.
A critical component of the conversion of sedimentdeposit volume into mass is the variable
density of sedimentdeposits.
In order to overcome the uncertainties which are involved, an indirect method has been
developed for converting volume into mass which circumvents the practical problems
involved in estimating the averagedensity of the deposit at a given stage.
Assuming that a logarithmic relationship exists between sediment deposit volume and time,
(Rooseboom, 1975) it was found that for a number of South African and USA reservoirs it

140
0
095 I,b 1,; LO 235 3:0

CUMULATIVE WATER DISCHARGE (10gm3)

FIGURE 4: CUMULATIVE SEDIMENT DISCHARGE VS ‘CUMULATIVE WATER

BETHULIE BRIDGE OCT 1964 - SEPT 1969
was possible to expressthe volume of a sedimentdeposit after t years (Vt) as a fraction of the
volume after 50 years (V50)by meansof the relationship:

v,
-=0,3761n&
50 3

IMPACTS DOWNSTREAM OF DAMS

As a result of a storagereservoir trapping a major proportion of the incoming sedimentsand
with only a few percent normally passing through, the sediment equilibrium in the river
downstream is disrupted. The few percent that typically passesthrough consists mainly of
very small clay particles carried in colloidal suspension.
With flows that leave the reservoir carrying very little sediment,the bed and banks of the river
directly downstream of the reservoir tend to be eroded. The materials forming the bed and
banks also become progressively coarser as the finer sediments are carried away first.
Sediment loads which are carried downstream therefore consist of increasing particle sizes
with only the colloidal componentfrom upstreamof the dam remaining constant. When flood
plains downstreamare inundated under flood conditions, the depositedsedimentsinclude the
coarser material which is deposited rapidly where flow velocities decreaseon the one hand
and colloidal material which is only depositedwhere long term ponding occurs on the other
hand.
With the bed and bank materials becoming coarser,the changing habitat leads to ecological
changes which can have far-reaching consequences. An example is increased blackfly
(Simulium chutteri) populations leading to increased stock losses in neighbouring farming
areas.
Reservoirs do not only trap sediments but also cause flood peaks downstream to become
much lower (Flood attenuation) except when very big floods occur. Whereas scour of river
beds and banks takes place directly downstream of dams, river channels tend to become
shallower and narrower further downstream.
For more detailed southern African information, the reader is referred to (Rooseboom 1992)
and the extensivebibliography included in this publication
REFERENCES
Rooseboom, A., 1975: “Sedimentafvoer in riviere en damkomme.” DSc Ing Thesis,
University of Pretoria, 1975
Rooseboom,A and Harmse, H.J.von M., 1979: Changesin the sediment load of the Orange
River during the period 1929-1969. Proc. Symp. on the hydrology of areas of low
precipitation. CanberraIAHS. Publ. no. 128,459-479.
Rooseboom, A., 1992: Sediment transport in rivers and reservoirs - a southern African
perspective. Water Research Commission Publication. No. 297/l/92. Pretoria,
South Africa.

142
Theoretical Premises for Determining of
River Bed Scour and Accretion Areas
Alexander N. Butakov
EngineeringFaculty, FriendshipUniversity of Russia

Stepped shape of longitudinal river-bed profile

Usually it is assumedthat the normal longitudinal profile of free-water surface and
averageprofile of a river bed are describedby concave curves. These curves indicate to
the regular and gradual decreaseof slopes from the river source towards its mouth.
Under the circumstancesthe scouring process prevails in the headwaters, repeated
sedimentationtakes place in the middle part of river, and the process of sand and silt
sedimentationbecomespredominant in the downstreampart.
However, a smooth longitudinal profile is just the first approximation. Actual
profile of free surface always has the stepped shape. Steps overlapping the normal
profile have different size and shapeand some of them are often included in others. The
largest onesare formed along the river confluents. Free water surface slope is decreased
just upstream of cotiuents and considerablyincreaseddownstream of them. Depending
upon dischargesof confluents and phaseof hydrologic year the length of river part with
considerablechangesof free surface slope varies within a wide range - from 2-3 up to
tens and evenhundreds of channelwidth.
The stepped shape of longitudinal profile of free surface is formed also along a
constricting and widening of a river valley, along local sedimentationbasesetc.
As free-water surface slope varies non-uniformly along a river bed therefore the
natural river flow is non-uniform too for all water levels. Usually in analysisof channel
deformations the considered river-bed length is divided into separate sections with
roughly equal surface slopes and river-bed parameters (width, depth, bed roughness)
along each of them. Stream flow in these zones is assumed as quasi-uniform.
Unfortunately, for this traditional scheme the conjugation zone between two sections
has to be reduced to a single transition cross-sectionwith an artticial turning point of a
free-water surface curve on it. Theoretically, discontinuities or drops of free-water
surfaceare possiblein the transition cross-section.
Water level drop varies with changesof water levels and river discharges.Thus,
the width of a river valley affects water levels during the flood, when river flow
impounds flood-plains. Along a narrow valley the flood water level rises higher then
within the other sections,where the valley is broader. Within these sectionsthere is a
backwater level or a positive water level drop accordingto their sequence.

143
 A backwater level is also formed under artificial flow narrowing by cofferdams
protecting pits of hydraulic structure, bridge abutmentsand alignmentstructures.There
is a water level drop inside a cross-sectionof those structures.A water level is lowered
by a dredgecut etc.

Behaviour of free-water surface slope along the conjugation zone

Becausediscontinuity or turning point of free-water surfacecurve can not be realisedin
natural smooth flows, the two sectionsare separatedby a distanceto form a conjugation
area or a conjugation zone, consistingof upstream and downstreamparts. Backwater
and dropping curves are formed within the conjugation zone and can have different
positions relative to each other. Uniform free surface slope along the adjacent sections
becomes to non-uniform one within the conjugation zone. Changesin a free-water
surface slope lead in to a non-uniform distribution of stream velocities along a
conjugationzone to causedeformationsof the channelbed.
In a prismatic bed the dropping curve indicatesto depth reduction downstream
and the backwater curve indicates to growth of depth. In a non-prismatic bed the
changes in depth can not correspond to changes in slope. Hence, the curve
correspondingto slope increasingdownstreamwill be referred to as the dropping curve
and the curve correspondingto slope decreasingdownstreamwill be referred to as the
backwatercurve, regardlessof depth change.
In general,free-water surface slope is determinedby hydraulic parametersof the
streamand the bed, i.e. bed shape,roughness,and streamvelocity. However, in erodible
beds stream forms its bed itself, so that the bed is best fitted to the conditions of bed -
streaminteraction (transportationof the certain amount of sedimentsat the given store
of potential energy). Therefore, a distribution of slope along the zone of nonuniform
flow is determinedmore by conjugationconditions i.e. by the free-water surface slopes
of the upstream and downstream sections and water level drop at conjugation cross-
section than local changesin a bed shape.Taking into account this concept the law of
slopechangesalong the conjugationzone will be established.
Let stream flow is close to uniform in a far distancefrom the conjugation cross-
section (i.e. the conjugation cross-sectionof free-water surface curves, which possibly
have different signsof curvature). There the free-water surfaceslope is equal to I,, . As
well as in a far distancedownstream stream flow is close to uniform, and the slope is
equal I,, . All casesare possible: I,, > I,, , IO1= I,, , and I,, < I,, .
The conjugation zone within which the slope I,, becomes to the slope IO2 is
divided into the two parts (see Fig. 1). The first part has length I,, and the second one
has lo2 . The total free-water surface drop ZW,, withm the conjugation zone loI + I,, is
divided into the free-water level drop Z,,,,, within the first part and ZW,within the
second part. The free-water slope ranges from I,, at II = 0 to I, at the conjugation
cross-section( 1, = l,,r) for the upstreampart and from I,, at I, = 0 to I,, at the end of
downstreampart ( l2 = Zo2 ).
Let us considerthe three possibleschemesof a conjugation.
Let us assume that a valley width is less along the upstream part than
downstream. The water level drop is positive if a stream flow is uniform at both

144
 Fig. 1

145

-
-
conjugated sections (see Fig. la). In this case the conjugation curve consists of the
dropping curve within the upstreampart and the backwater one within the downstream
part. Such water slope distributions are observedin a river-mouth and during a flood,
where the flow leavesa contractedvalley for a broad one. The free-water surface slope
of reservoir is equal zero in the first case.
During a low water the water level drop can became negative at a conjugation
cross-section,becausea free-water surface co-ordinatesdecreasemore along a narrow
valley than along an expandedone. In this casethe backwater curve extents within an
upstreampart (see Fig. lb). The similar situation in the slope changesis observedfor
conjugation sections of channelshaving different areas and shapesof a cross-section.
This and relative problems concerningartificial changesin bed width is the subject of a
specialresearch.
If a river valley narrows the backwater curve is formed just before a narrowing
and then becomesto the dropping curve of slope downstream(seeFig. lc).
The preliminary research gave the following results. Any curve of free-water
surfaceslope can be approximatedas an exponentialcurve by the chooseof exponential
factor. So, the following equation is true for the free-water surface slope within the
upstreampart:
I,, = I,, + I,, a; e-‘~C~ (%I -i;)
(1)
where a; is a coefficient deter-mining a value and a sign of free-water surface drop
within the upstream part, C; is a coefficient determining a curvature of a free-water
surfacealong this part and dependedon a valley configuration,particularly on its width
and flood bed sizes, k, = lTBOI / H,, , A r is a coefficient of hydraulic fktion, B,, and
H,,, are a flow width and depth in an upper cross-section,consequently,< = 1, / B,, .
If the free-water surface slope is growing along an upstream part (the f3st
scheme)than a; > 0, ifit is decreasing(the secondscheme)than a,l < 0 . ai = 0 means
the constantslopewithin the upstreampart.
The following equation describesfree-water surface slope along a downstream
part:

(2)
where a,l' and C;’ have the samemeaning as a; and Cl and describinga downstream
part. a;' < 0 indicatesto the backwater curve, a;' > 0 indicatesto the dropping curve.
a; = 0 meansthe constantslopewithin a downstreampart.
The numerical value of the coefficients ai and aj’depend on a value of free-
water surface drop within a conjugation zone. Moreover, they depend on each other.
The free-water surface has to change smoothly, without gap, at a conjugation cross-
sectionfrom an upstreampart to a downstreamone. Therefore, the following condition
is true:

d-c, dZw2
-I I
dl,
'I='a
=- 4 l*=O

146
which is equalto

The slope value II, is determinedfrom the equation (1) at < = z, and the slope
valueII, is determinedfrom the equation (2) at z = 0. We have the following equation
after substitutionsby the founded expressionsinto (3) :

I
al’= 02-- 1
1 (4)
IO, l+a,

Becausethe value of a free-water surface drop has to be constantand taking into

accountthat Zw, is changedfrom Zw,, up to Zw, during II is changedfrom 0 to Z, and
taking into considerationthat 1,= BO,<, and -dZ,, = I,,dl, , after substitution I,, from
the equation(1) integration and sometransformationwe have:

Making the similar transformationfor the equationdeterminingthe slope and free-

water surfacedrop within the downstreampart, the following equationis found:

a;‘=- AZwm
kTc,, = _ “TX kT c; (6)
Io,Bo, k IOPOI1+ a;
BecauseAZw, = AZ,,,,,+ AZw, , insteadof (5) we have:

(7)

This equationis the starting equationfor the coefficient a,l determination.

Changesof flow velocity and bottom co-ordinate of a river bed

along the conjugation zone
An equationdetermining a velocity alteration along a conjugationzone is deducedfrom
the following equation(Butakov, 1971):

J72= ~;,+-@t’) 1+ Ak, j Je-‘+(‘)dj-

0
1 (8)
147
 md A = 2gJOl BOl/(kTvi2)

where H is the averagedepth of the stream in a present cross-section,a depth is a

variablein the time, V, is the streamvelocity in the first cross-section,g is the gravity
acceleration,m , = 4/3 if Chezy’s coefficient was defined according with Manning’s
equation,A = 1 in the caseof an uniform flow in the first cross-section.
The equation(1) for the relative slope J1 = 1,/I,, is rewritten as:

Putting the value of J, into the equation(8) leadsto the following expressionfor
a squareof a relative velocity of a stream U = Y2/Vo2 along an upstream(upper than a
conjugationcross-section)zone of a river bed:

i;
k,aj ’ b[Q(i;)+C;‘;ldj-
l+ kTjekTeii;)dT+---&e (9)
0 e 0
1

A=1 is acceptedbecauseof a streamflow is closeto uniform in a distanceI,, upstream.

Free-watersurfaceslopealong a downstream(below a conjugationcross-section)
zoneis definedby the equation(2), which is rewritten for the relative slope:

J = l+a~~-ate-W~‘h
2 I I

The expressionfor the U alteration along the secondzone is also obtainedfrom

the equation(8). Becausethe relative flow velocity in the conjugationcross-sectionis
not equalone, but is definedfrom the first zone calculation,we have:

where
1+ 4kT(’+o;?qek’8(~@
_ 4kTa~,~ek,[e(~,-c~~]d~
0 0 1 (10)

The coefficient A, = 2glo130,/(V~kT), where I,, and V, are the free-water

surfaceslopeand flow velocity at the conjugationcross-section,i.e. if < = t, or Z2= 0
consequently.Becausethe coefficientis

4=

148
 Relative distance < has been changedby relative distance T, which has the origin
point at the beginning of a conjugation zone. Consequently T = F - T1. Because the
01 i; 7 z,
integral could be written as )=i
0 0
+
0
f , we have 1 f f
0
=
0
-
0
. Also the obvious

equationsare:

8 (‘;)=8 (q-6 (e,)

After straightforward transformationsthe following equation for U” alteration

along the secondzone has been obtained:

(12)

A flow velocity is definedby the expression:

v=v,tliT (13)

The concluded equation (9) and (12) allow to calculate a flow velocity along a
whole conjugate zone under any combinations of backwater and dropping curves
showed in Fig. 1. The investigation of processesof river bottom evolution involves
bottom deformations and flow velocity alterationscausedby the deformationspassing.
This problem should be performed by computer aid.
For this aim achievementa period of the time T is divided into severalintervals
At = T/n and a zone length I is divided on elemental portions AZ = l/i. Then the
equationfor the function U could be written as:

149
 UnIi = (Uie -krF(b) +kTAL(l+aj)[(l+a$(~ekTuF -iekTuF) -
0 0

map WW
(15)

where Ud = cCJ~,~,F = ) L, =c,, I=-$

Relative co-ordinate of a river bottom (i.e. co-ordinate is divided on an average

depth Ho, at the first cross-section)is defined by the following equation in the caseof a
initial constantdepth of a river bed (Butakov, 1990) :

where E = q&t where qso is a river sedimentsdischargein the first

( 1 --E>B,,H,,AL ’
cross-sectionin volume units, E is a coefficient of porous bed sediments.

Actual co-ordinate of a river bottom is Z,,i = z”,i Ho, .

Solution methods
Let us examine the following examples. The formation of a longitudinal profile of a
river bed within areasof valley width alterationsis the theme of this part.
Let a valley is narrow along an upstream part and is broad downstream. As
mentioned above, a flood level is considerablyhigher within a narrow part than within a
broad one. An additional drop of co-ordinates of free-water surface arising causesthe
redistribution of free-water surface slopes.Along a upstreampart the free-water surface
slope gradually grows, attained the maximum at its end. Along a downstream part it
gradually decreasesfrom the maximum at a conjugation cross-sectionto approachthe
slope of free-water surface deter-miningan uniform flow along that part. This case is
shown on Fig. la.
For the determination of scouring and accretion areas of channel bed the very
large (disastrous) flood should be taken into account, when those areas are formed.
During such floods a valley is submerged given a large depth accompanying stable
continues streams.The value of C, coefficient should be set by comparing an actual
and a calculated depth and width of a continues stream under present river bed
conditions. The usual value is less then one ( C, = 1 in the case of inner bar or anti-bar
formation). An actual width includes parts of a river valley close-fitting to a stream
course, where stable one-directional streams are, and a width of river arms, if they
present.

150
 To simplify calculationlet an averagebed depthsare approximatelyequal within a
whole conjugatezone at initial time point. Then the equation (9) and (12) for the U’
and U” are:
u’=l+af
1 + c;
-k,Ci(G, -‘;) _ ,-kr(C& +i;)

1 (17)
u” = (u; _ o,‘)e-kd%) _ D;e-“+@$, ‘D,r, (18)
where Di’= (l+a;)(l+a!‘), Dil= (l+a;)x Dj’= D,“- 0,’ , U; = U&,
1+c;”
The equation(13) determentsan averageflow velocity within a cross-section.
There are two casesfor a calculationof river bottom deformation, which limit the
whole spectraof an actual longitudinal profile of a river bottom.
The first one is based on the constantriver flow width during rebuilding of river
bottom relief, which is defined as:
B=Q/(vH) (19)

where the velocity V and the streamdepthH are given for the initial stage.
Flow velocity is
V=Q/(BH) (20)

during any subsequentperiod of the time, where B is a width of a continuous stream

given for the initial stage.
An average depth of a continues stream is calculated with an influence of the
bottom deformation
Hit,i = (l+ 'n,ilHOI (21)

where an origin point for the co-ordinate z?,i is equal the initial bottom and is defined
by the equation(16).
The alteration of a flow velocity, which is determinedby evolving deformations
accompany a free-water Surfacealteration. During any period of longitudinal bottom
profile developmentfree-water surfaceslopeis definedby the following expression:

(22)

The secondversion of a solution is based on a changeableflow width, which is

followed by the equation (19). However a velocity V and a depth H is determinedby a
consideredperiod of a relief development.Therefore a velocity function is calculated
accordingto the equations(14) and (15). Thus a flow velocity is growing along a bed
erosion zone and is decreasingalong a zone of a sedimentsaccretion. In theory (if a
sediments composition is homogeneous) a depth of erosion is growing infinitely.
However, a limit exists. It is determined by a backwater influence resulting from an
accretion zone. The total free-water surface drop A-Z,,,,is decreasedwithin an scoring
zone leads to the decreasingof the slope amplitude a,) by the equation (7). An
increasingof a flow depth is favoured by a backwaterleadingto velocity decreasing.

151
 Free-water surfaceslope is growing up within the zone of sedimentsaccretion.At
first glance it would seem that the decreasingof flow width within a scouring area is
u.njustiEed.However, that is the case. A flow velocity with width is non-uniformly
distributed. It is always more on a midstream where the major bottom scouring is in
progress.A water dischargeconcentrateson a channelline and zones of a tranquil flow
accompanying sometimes with backward jets are formed. As observational data
indicated such zones silt up or overgrow by vegetation with time. This process is
favourablefor decreasingof an actualbed width.
The increasingof a bed width takesplace within an accretionzone.
The noted two types of calculationsgive similar deformation outlines. In the case
of homogenouscomposition of sedimentsand the fixed initial flow depth the maximum
of deformations is noted at a cross-sectionwhere the derivative is maximum too.
During deformations evolution the zone of maximum depth moves towards a cross-
sectionof maximum velocity and when a large depthhas been formed a cross-sectionof
maximum depth is closely marched by that of maximum velocity. The calculation
resultsfor the secondmethod is depictedin Fig. 2.
The sedimentstransported by flow from an scouring zone accumulate along a
zone of velocity decrease.The longitudinal profile becomeswavelike. During an initial
period of evolution frontal slopes of erosion and accretion zones are steeperthan rear
ones. The slopes steepenthrough time, however rear slopes became steeper quicker.
The velocity of vertical deformation decreasesgradually and the accretion crest shifts
downstreamregularly.
However, the mentioned methods of calculation lead to the deformation value
which significantly distinguishesfrom each other. The difference may be ascribedthat
the first method gives the conceivable minimum of the longitudinal bottom
transformation and the second one gives the maximum. The actual profile of river
bottom is arranged within the boundaries given by the noted methods and could
approximateto either one or another bounding curves. Regular variations are dictated
by a local peculiaritiesof a valley structure and soil conditions.
The local backwater resultsfrom a contractedvalley, below that a dropping curve
of a free surface arises. The distribution of relative slopes is given by the equations
mentionedabove.However, the amplitudes a,l and aj’ reservetheir signs,so that a;
becomesnegative and a;’ >O. The same equationsare still valid for the velocity V, the
depth H, and the bottom co-ordinate Z,, which has been used for the calculation of
longitudinal profile of a river bottom within an expandedzone of valley. The earlier
operatingsequenceholds.
For this case the two variants of solution can be also considered, which is
determined by the borderline curves of a longitudinal profile, describing the process
evolution in time.
At the beginning the bed scouring is found in an upstream part of contracted
section of a valley. A scouring area progressivelygrows in depth and length and the
accretion zone runs just before the narrowing within a backwater zone. The
deformationvelocity is reducedwith the passageof time.
The second method displays the formation of a new accretion zone, which is
found downstream of a channel scouring zone, where free surface slope is enhanced
(seeFig. 3). This phenomenais explainedby the following reasons.
On account of the significant increaseof flow velocity within an upstreampart of
contracted zone, where a channel scouring is observed, the sediments discharge

152
Fig. 2

Fig. 3
dramatically increasesdownstream. However the actual velocity can not support the
sedimenttransport derived from intensivescouring downstreamof scouring zone. The
partial sedimentationhappens.This process leads to an appearanceof a downstream
accretionzone in spite of a river runs in a narrow valley. It should be pointed out that
the zones of sedimentsaccretion coming about in that case have been described by
Makkaveev ( 1955). He intimated that in natural rivers there were zones, where a bed
had a narrow valley and yet is f?lledby alluvial depositsfor the great depth. It could be
found stablechannelscoring areasat a very long extent. In plan the river bed is usually
slightly bent and curvature of the bed is closeto curvature of the river valley. Formation
of stable channelscouring areasis favoured by low volume of entering sedimentsand
difficult-to-scour soil. Combination of the noted conditions is aided to formation the
most beautiful place at river Volga namely Ples glorified by the famous Russianartists.
Unfortunately now Pleshad been submergedby the water storagereservoir.
The formation of hydraulic structures appearsonly slightly in the scoring areas
and usually microforms could be observed. Sand dunes or sometimes shoals are
relatively rare. The shoals form gently sloping channelbar without well-defined crest
and are named by the special term in navigational practice. Any kinds of hydraulic
structurescan evolve within accretionzones.

Conclusions
By this means the set up theoretical prerequisites and deduced equations allow to
resolve the very sophisticated problem within the framework of one-dimensional
hydraulics and to derive the answer about , determining steppedshapeof longitudinal
river-bed profile. Unfortunately the one-dimensionalapproachdoesnot allow to study a
specialfeature of developmentof transversalbed profile, which is formed distinctly in
the channelscouring zones.This problem could be solved by switching to hydraulics of
two dimensionalflows.

Bibliography
Butakov, A. N. (1971) ‘Plane currents’. Trudy LIVT, Vol. 129, p. 10-15. (in Russian).
Butakov, A. N. (1990) ‘Theoretical models of channel forms development’.
Proceedingsof the Fifth All-Union Hydrological Congress,Vol. 8, Leningrad. (in
Russian).
Makkaveev, N. I. (1955) River channel and erosion in its watershed, AN SSSR,
Moscow, 346 p. (in Russian).

154
Sediment associated transport of
Chernobyl radionuclides in the Pripyat
River, Ukraine
0. V. Voitsekhovitch, V. V. Kanivets, V. I. Vishnevsky
Ukrainian Hydrometeorological Institute, Kiev

Introduction
On April 26, 1986 the nuclear accident on the reactor N-4 of the Chernobyl NPP took
place. As a result about 4% of fission products were releasedto the atmosphere.Most
of the released radioactive materials due to wind influence were deposited on the
watersheds areas of the Dnieper River (as well as the Pripyat catchment) heavily
polluting a large territory (Vakulovsky et. al., 1991, 1994). The flood plain areasin the
vicinity of the Chernobyl NPP were extremely contaminated by the radionuclides as
well (Voitsekhovitch et. al., 1991).
This and adjacent drainagebasinswill remain the main potential sources of the
Dnieper cascadecontamination by the Cs-137 and Sr-90 due to run-off and erosion
processes,which carry radionuclides downstream through the Pripyat River and
Dnieper cascadeto the Black sea(Voitsekhovitch et. al., 1993).
The highest level of radioactivity in the water of the Pripyat River was observed
during the initial period after the accident. Some time radionuclide contents exceeded
a sanitarypermissiblelevel for some particular radionuclidesin the natural waters. From
1986 to 1994 radionuclides contents in the river were gradually decreasedin general,
but during eachhigh water period its concentrationin water increasedunder the wash-
off and erosion phenomena, creating a secondary contamination of the water masses.
Therefore, the ratio between Sr-90 and Cs-137 in stream changedsignificantly as well
as the ratio between its suspendedand soluble part in the water. These differenceshave
to be determined by many geochemical and hydrological species taken from the
contaminatedareas. Some aspectsof the “solid -liquid” interaction between suspended
and solubletransport of Cs-137 in the river and reservoir are consideredbelow.

Experimental methods
The main objectives of the carried-out experimental work were to evaluate a total
radionuclide transport by the river flow from the contaminated areas and to study the
fate of radionuclidesassociatedwith suspendedparticles and transportedin soluble form

155
in the river-reservoir system. Experimental data were collected in the Ukrainian
Hydrometeorological Institute within the framework of different national monitoring
programs.Besides,the goal of investigationswas to obtain parameters,necessaryfor a
predictive modelling of the radionuclidestransport by the water flow. The samples of
water and suspendedparticles were taken at severaltransects,covering 150 km of the
Pripyat River, the Kiev Reservoir and other reservoirs of the Dnieper cascade(system
of reservoirs).The suspendedmaterial was sampledusing a tangential filtration system
and by standard devices. In order to evaluatethe origin of suspendedmaterial and to
quantify the relative importance of the factors, influencing solid-liquid partitioning of
radionuclides,the following characteristicswere measured:particles size distribution,
mineralogical composition, dissolved oxygen, pH, temperature of water, amount of
organic matters, ammonium etc. The sampling of water with suspendedmatters for
radionuclidecontaminationanalysiswas carried out by the specially designedtechnique
MIDI4 (Manual..., 1990), which consistedof three linked sections (pump with water
flow meter, filtration system and absorberbox). Usually, 100-500 litres of water were
taken for each sample.Samplingsystem of ANPA-DISP (Italy) (Sansoneet. al., 1994)
was also used. This samplingsystemhas been designedto gather different fractions of
suspendedmaterialsusing nylon cartridge filter with 40, 10, 0.45 pm pore size and with
diameter of 60 pm and a filtration area 0.49 sq.m. The water filtered ranged from 500
to 2000 litres.
DifIerent sorts of adsorbent materials on the basis of ammonium
exocyanocobaltferrateresinswere usedto fix cesiumdissolvedin water (Sansoneet. al.,
1994; Remes, 1995). To determinethe amount of suspendedmaterial, the filters were
dried and weighted before and after the filtration of water. All filters and resins were
analysedby gamma-spectrometry.Gamma-ray spectrometryanalysisof the water and
suspendedsediments were performed using Ge(Li) and HPGe detectors.The samples
were counted from 5 to 20 h giving a standarddeviation lo-15%. For Sr-90 analyses
the ordinary radiochemicalapproachhasbeenused (Manual..., 1990).
The radionuclidestransport by the river flow in dissolved and suspendedforms
was calculated on the basis of routine hydrological flow calculation procedure by the
daily (for flow) and ten-day or monthly averageddata (for radionuclidecontents).

Radionuclide transport by the river flow

During the period after the accidentthe hydrological characteristicsof the Pripyat River
near Chernobyl were obtained. The averagelong-term river runoff is equal 400 m3/s,
average maximum discharge during the spring flood - 1500 m3/s. In 1987-1994the
annualrunoff was near norm (Fig. 1). Using speciallyorganised measurements,data on
suspendedsedimenttransport of the Pripyat River near Chernobyl were obtained. It
varied from 2 kg/s (when water dischargeswere lessthan 150 m3/s)to 200-250 kg/s
during the floods. Averaged data for 10 days, for months and all years after the
accidentwere also obtained. The sedimenttransport by the Pripyat River into the Kiev
reservoir for 1987-1994 varied from 260 to 1280 thousand tons. Total amount of
suspendedsediment inflow to the reservoir by the Pripyat River for this period was
about 5 mln tons (Vishnevsky and Voitsekhovitch, 1994).
The average size of suspendedparticlesvaried from 0.02 mm during low water
period up to 0.25 mm during floods. Using obtaineddata, the amounts of suspended

156
particles of different size for every year and for the period of 1987-1994 were
calculated. The amount of suspendedsedimentsin the river flow with size particles less
than 0.05 mm was about 55-60 % of the total.
On the contrast, the bed sedimentsin the Pripyat River mainly consist of sands
with very little silt and almost no clay size sediments.
It was found, that large part of ‘37Cs, as well as some other radionuclides,were
associatedwith suspendedparticles. Ratio of 137Csbetween solid and dissolvedpart in
river flow for different observationseasonsis presentedin Fig. 1. Moreover, from year
to year this part has a tendency to increasingdue to erosion processesremaining to be
rather intensive on the basis of decreasingof Cs-137 contents in mobile forms in the
upper surface layer of catchment landscape, because of its furation and vertical
migration in soils. Also the monthly averagedCs-137 and Sr-90 concentrationsin the
inlet and outlet water flow of the different reservoirs were considered(Fig. 2). These
data have confirmed that the reservoirs can be consideredas a temporary sink of the
material transported by the rivers crossing the areas contaminatedby the Chernobyl
event. The assessmentof annual radionuclide transport by the Pripyat River and the
DnieperRiver is presentedin Table 1.

Table 1. Annually averagedradionuclidesinflow to the Kiev reservoir(Ci/y.)

Year Pripyat River Dnieper River

’ 37cs 137cs 90Sr 13’cs 137cs 9oSr

solub. susp. total solub. susp. total

1986 1524 261 746 737 98 288

1987 244 100 280 261 113 216
1988 160 96 506 182 69 140
1989 98 76 241 157 34 97
1990 62 63 273 100 39 100
1991 39 39 389 29 26 122
1992 25 22 92 22 13 18
1993 54 40 371 15 7 25

The results presented in Fig. 1, Fig. 2 and Table 1 also have defined a very
important role of the suspendedparticles as well as the physical and chemical
parametersof water bodies on the adsorbtion-desorbtionprocessesfor radionuclideson
suspendedmaterials. For instance,on the basis of the presenteddata it is evident that
only l-2 % of total inlet of Cs-137 to the Kiev Reservoircould reach the estuary of the
Dnieper River. Most part of Cs-137 as a result of absorbtionon suspendedparticles and
due to direct uptake by the sedimentwere accumulatedin the Dnieper cascadebottom,
and were mostly tied firm on the solids. Some aspectsof Cs-137 transfer on the
suspendedparticlesto be transportedfrom catchmentareasto the river-reservoir system
are presentedbelow.

157
 a:>:.:
....
~-:s.:.:.:x..;... .
gg$w<: .... . . .,..~.....,.......................

~~~
 Kiev Kanev

40 7
35
30 Sr-90
25 1

01.87 01.89 01.91 01.93 01.87 01.89 01.91 01 93

Wet q Outlet l nlet ClOutlet

50

40 i I cs-137
1

iih
01.87 01.89 01.91 01.93 01.87 01.89 01.91 01.93
n Inlet q Otitlet Wnlet q Outlet
i

Krementchug Dneprodzerdzinsk
1
40 7 40
35 35
30 1 30
!
25
20
15 1

0 ! I
01.87 01.89 01.91 01.93 01.89 01.91 01.93
b7let q Outiet n inlet q Otilet
J J

4 cs-137
1

01.89 01.91 01.87 01.89 01.91

hlet OOutlet n inlet OOutlet
!

Fig. 2 Monthly-averaged radionuclide concentration in water at the inlet and outlet j’orn the
reservoirs in yCii1 (1987- 1993 y.)

159
The fate of radioactive substancesin river-reservoir system
The measurementsshowed, that the level of radioactive contaminationdependson the
size of particles. Usually small particles (especiallyless than 0.05 mm) are much more
polluted than large ones (Fig. 3). Stream flow conditions in the Pripyat River (if
comparedwith the reservoir condition) usually don’t allow the suspendedparticlesto be
deposited. This is the reasonwhy the streambottom sediments,presentedby sands,are
comparably “clean”. The differencesbetween contaminationof silt-clay sedimentsand
sandscan reach severalorders of magnitude.

Radioactlvities of ‘37Cs
1 - Fall 1866
2 7 Sprln 1987
3 -_ Fall 18 67
47Sp’ 1986
5 - Fall?I1 60

Sedlment Diameter (mm)

Fig. 3 Contribution of 137Cstransport, associated with suspended particles of different sizes

This fact has to be taken into account when studying a peculiarity of sediment
radionuclidestransport by current or wave flows (Walters et. al., 1982; Voitsekhovitch
and Zheleznyak, 1994). Besides,this sorption-adsorptionability of the solid particles of
sedimentscan be used for an identification of the origin of different radioactively
contaminated suspendedmatters on the catchment areas with defined species of its
contamination (IAEA 1993). Relatively low erosion rate on the territories inside
Chernobyl close-in zone caused,that from year to year the “far zone”(outside area
with significant level of radioactive contaminationof the Pripyat River basin) becomes
more and more important. A special investigationwas carried out for the evaluation of
its role and its runoff contribution from different parts to the total pollution of the
Pripyat River by suspendedsediments.
It was determined,that the Pripyat River basin could be divided into four zones
with different characteristics of turbidity and modules of the solid flow. The
parametersof annualsedimenttransport (including quantity of solid material of different
size) were calculated for each zone and for the whole basin. As a result, a good
conformity was found between transport of sedimentswith size less than 0.05 mm
(carried by the Pripyat River near Chernobyl) and the total amount, carried by the

160
tributariesfrom the catchmentarea. Thus, it gave an opportunity to find some particular
limited areas inside the Pripyat River catchmentwith a great impact of the secondary
pollution of the river. The comparison of radionuclide contamination speciesof the
suspendedparticles caught near Chernobyl in the river flow can provide the unique
possibility to verify its origin from the catchmentarea. This methodological study is in
progress;future researchresultswill be obtainedlater.
The investigation,carried out in the Kiev Reservoir and in other reservoirs (Fig.
2), shows, that most of the radioactive material, associatedwith sediments, are
depositedonto the bottom. Dnring post-accident period 98 % of total amount of
137Cs were depositedin bottom layers of six reservoirsand more than 50 % of it in the
Kiev Reservoir. The most polluted bottom sedimentsin the Kiev Reservoir occur in the
upper area of the reservoir near the mouth of the Pripyat River and submergedriver-
bed of the Dnieper River. The deposition of relatively clean solid substancesonto the
initially contaminatedbottom sediments during the recent years caused a submergence
of most polluted layer. The disposition depth of the mostly contaminated layer of
bottom sedimentsdated by 1986 for the upper part of the Kiev Reservoir reaches lo-
30 cm deep (Voitsekhovitch, 1993) decreasingto the layer of 2-3 cm deep near the
dam. The investigation shows, that D&per reservoirs serve and would remain the
main receiver of contaminatedwater and radioactive suspendedsediments. The
progressin calculation and forecastingmethods demandsa proper understandingof the
mechanisms,which control the radionuchde adsorbtion-desorbtionprocesses. On the
basis of above mentioned studiesit became clear that severalprocessesare related to
the governing role of the solid material associatedwith radionuclides(as well as some
other toxic materials)during their motion in water under the runoff flow. Main of them
are as follows:
- erosionwash-out from the catchmentareas;
- river-reservoir water/sedimenttransport;
- sedimentationprocessesin the reservoir;
- secondary resuspensionof the contaminated sedimentsunder the wave and
current conditions;
- sorbtion-desorbtion, kinetic mechanisms of mineral components and direct
consumptionof radionuclidesby some aquaticalgaeand phytoplankton;
- diffusion and other mass exchangeprocesses,governing vertical migration of
radionuclide and transfer radionuclidesbetween solid part and pore water of different
layer of sediments.
The first part of the consideredprocessesresponsiblefor radionuclide transport
with suspendedmatter can be simulated on the basis of the well known hydrological
approaches,standarddata collection proceduresand availablemathematicalmodels.
The secondpart of the processesis less studiedfrom the point of view of the laick
of accurate parameters and the absenceof strong recommendation for estimating
distribution parameters.Most significant of them is distribution coefficient Kd, that
determines a ratio of radionuclidesbetween soluble and solid phasesin the water. The
Kd value also dependson the time passed after radionuchde entrance into soil or
bottom sediments (ageing effect). Therefore, a reliable prediction of radionuclide
behaviourin “solid-liquid system”is only possibleif distribution coefficientsfor a given
“suspendedmatter-water”systemat a given time could be assessedon the knowledge of
about m.echanismsand kinetic parametersof sorption-desorptionprocesses.A number
of methods of laboratory and insitu determination of Kd value has been describedin

161
literature (Beneshet. al., 1988; Cremers, 1988; Konoplev et. al., 1992; Valske, 1993;
Sansoneet. al., 1994).
However, analysisof these data shows,that there are no ah-purposeKd valuesto
be used for modelling the behaviour of radionuclidesin any solid-liquid interactions
studies.For example,the variance of Kd values for 13’Csin surfacewater is known to
be 4 orders of magnitude.Moreover, the difference in Kd at the same site for different
layer of bottom sedimentsin the rivers and lakes can be as great as two orders of
magnitude.Also sorption-desorptionmechanismsas a role are not reversible. By this
phenomena,13’Cs ( as well as other substancelike some heavy metals, pesticides,etc.),
which could be firmly fixed on clay mineralsor organic matter of the eroded catchment
in equilibrium condition betweenparticles and pore water of the soil solution can be in
no equilibrium state during its adsorption in natural river flow. As was shown by
recent studies(Konoplev et. al., 1993) the radionuclidesoccur in the solid sediments
phasein physical and chemical forms which differ by their ability to exchangewith the
water phase.The equilibrium of the exchangeableform with the solution is attainedin a
short time ranging from several minutes to tens of minutes, whereas the fixed form
passesinto solution much slower. As a consequence,Kd calculatedfor the total Cs is
growing during severalmonths in solids-liquid suspension.Therefore, the distribution
coefficient can be also defined for the exchangeableform , which usually increasestwo
fold in the first severalten hours and then remains constant. So, exchangeableKd is
better to be used for simulation of fast processes.Therefore the identification of Kd
value for suspendedmaterial in natural polluted water is not a trivial task, because the
value of Kd extremely dependson the size of particles and their mineral or organic
contents, time interaction with water, the chemical characteristic of water, its
temperatureand time passedafter the radionuclideentranceinto the solid liquid system.
For example, recently obtained results (Voitsekhovitch and Kanivets, 1993) showed
some unusual distribution of 137Cson different size fractions of freshwater algae and
phytoplankton. Less sizes of particles were associatedwith less concentrations of
consumed 13’ Cs in the water column, when size of fractions were separated,using
filtration facilities. At the sametime absolutelydifferent resultsfor the samefractions of
organic particles were obtainedwhen sedimentationapproachto size particle separation
was used. So, it is necessaryto be very careful, using experimentaldata on suspended
matter size particles distribution for its applicationstudyingtheir hydraulic and sorption-
desorptionproperties.
The consideration of the Kd value approach determination is an individual
complex problem. In this connection,the following general questions,that outline the
tasks of the study, have been formulated:
What methodology is most acceptableand correct when suspendedparticles of
different nature have to be sampledin river flow and reservoir water column?
What are Kd’s for suspendedparticles,that have been transportedby water flows
from sources(erosion run-off, small stream,river flow and by reservoir flow) ?
How to clarify the difference betweenKd’s for suspendedmatter and that for the
bed sediment?
And, f?nally,how to comparelaboratory valueswith field results?
The above mentionedproblems have been in focus of researchprogram initiated
by Ukrainian Hydrometeorological Institute during recent years. The results of this
study can assistto develop and to improve the knowledge of the role of suspended
particlesin the fate of radionuclidestransportedby the water flows.

162
Bibliography
Benes, P.; M. Kunsova; J. Slovak and P. Lam Ramos (1988) ‘Analysis of the
interaction of radionuclideswith solid phase in surface waters using laboratory
experiments’.J. Radioanal.Nuclear. Chem., Articles 125, p.295-315.
Cremers,A. and P. N. Herion (1985) ‘Radionuclide partitioning in sediments:theory
and practice’. In: Seminar on Behaviour of Radionuclidesin Estuaries,Renesse
(the Netherlands). September,1984. CEC , XII/380/85- EN: l-25.
IAEA ( 1993) ‘Advisory group meeting on use of nuclear technique in studying soil
erosion and siltation’. Report to the Directorate General, Vienna, 26-29, April,
38.
Konoplev, A. V.; A. A. Bulgakov; V. E. Popov and Ts. I. Bobovnikova (1992)
‘Behaviour of long-lived Chernobyl radionuclidesin soil-water system’. Analyst,
117, 1041.
Manual on organisation of environmental monitoring in surrounding the active
nuclear power plants (1990), Hydrometeoizdat,Editor K. P. Makhonko, 294 p.
(in Russian).
Sansone,U.; 0. V. Voitsekhovitch and V. V. Kanivets (1993) ‘Sedimentationstudies
and evaluation of suspended radionuclide transport by rivers and Dnieper
reservoir flow. ’ In: Progress report ( 1992-1993) ‘Modelling and study of the
mechanismsof the transfer of radioactive material from the terrestrial ecosystem
to and in water bodies around Chernobyl,IFE, UK.
Sansone,U. ; 0. V. Voitsekhovitch and V. V. Kanivets ( 1994) ‘Suspended Cs-137
transport fi-om the rivers located in the Chernobyl areato the Kiev Reservoir’. In:
Progress Report (1993-1994) ‘Modelling and study of the mechanismsof the
transfer of radioactive material from the terrestrial ecosystemto and in water
bodies around Chernobyl,ANPA-DISP, Italy, p.59-66.
Valske, E. ( 1993) The behaviour dynamics of radiocesium and radiostrontium in soils
rich in organic matter, Ph. D. Thesis, Katholike Universitet Leuven, Belgium.,
135 p.
Vakulovsky, S. M. ; A. I. Nikitin and 0. Voitsekhovitch ( 1994) ‘Cesium - 137 and
Strontium-90 contamination of water bodies in areas affected by releasesfrom
Chernobylnuclear power plant accident : an overview’. J. Environ. Radioactivity,
23, p. 103-122.
Vishnevsky, V.; 0. Voitsekhovitch (1994) ‘Suspended Sediments Transport in the
Pripyat River System’. In: Proc. Conference ‘Dynamic and Termic of rivers,
reservoirs... ’, Vol. 1 (in Russian).
Voitsekhovitch, O.V.; V. A. Borsilov and A. V. Konoplev (1991) ‘Hydrological
aspects of radionuclides migration in water bodies following the Chernobyl
accident’. Proc. Seminar. Comparative Assessmentof Radionuclides released
during three Major Nuclear Accident: Kyshtym, W indscale and Chernobyl.
Luxemburg, October 1990, Brussels,CEC, p. 527-548.
Voitsekhovitch, 0.; V. Kanivets; G. Laptev and I. Y. Biliy ( 1993) ‘Hydrological
processesand transport by surfacewater pathwaysas appliedto water protection
after the Chernobyl accident’. Proc. Hydrological Considerationsin Ralation to
Nuclear Power Plants,Paris, September,p. 85-105.

163
Voitsekhovitch, O.V.; M. I. Zhelesnyak and Y. On&i (1994) Chernobyl nuclear
accident. Hydrological analysis and emergency evaluation of radionuclide
distribution in the Dnieper River, Ukraine, PNL Report-9980, 96 p.
Walters, W.; R Ecker and Y. Onkhi (1982) Sediment and radionuclide transport in
rivers, PNL Report. U.S. Nuclear Regulatory Commission,Washington,75 p.

164
Fluvial Sediment Transport in the Arid
Regions of Central Asia
Dr. Ismagilov H.A.

Particle flow velocity in mud flow

Introduction
The foothills of Central Asia are composedby loams, sandy loams and clays. During a
heavy shower mud flows with turbidity up 300 kg/m3, and more, are generatedthere.
The topic of this article is about channel relationshipsbetween mud flow in the
Central Asia and particle flow velocity in mud flows. A number and associations
dependenciesare proposed to determine the main parameters of mud flows with
considerationof suspendedsedimentinfluence.

The hydraulic size of sediments.

The main characteristicof the sedimentsmovement is fall velocity, which dependson a
hydraulic size of sediments.The fall velocity at small content of sedimentsin the flow is
termed “free fall”, and with large content of sediments-‘I constrainedfall”.
The presentarticle considersthe fall velocity of sedimentsat constrainedfall. As a
rule, following equation is used to determinethe constrainedfall velocity:
0, = pa (9
where o - velocity of alluviums fall in clear water; p - coefficient, indicating a
decreaseof fall velocity whith much sedimentsin the flow owing to higher velocity and
a density of the mixture.
To determine p, we use formulas, suggestedby A.M.Velikanov, V.N.Goncharov,
D.MMins, O.G.Natishili, A.V.Karaushev, U.RMirzaZode and others. These formulas
however deal with the influence of density of the mixture, only but they do not take into
account a viscosity of the mixture.
On the basis of formulas of V.N.Goncharov and A.V.Karaushev we determined
the following relation between coefficient p and the density of the mixture:
a) for laminar regimesof sediments(d < 0.15 mm):

165
 VAP, -Pc)
P= (2)
VOPC
(Pr - P)
b) for transitional regimes(0.15 < d < 1.5 mm):

x VCP,- PC)x
p=ic) ( VOPC (3)
(PT- PI
c) for turbulence regimes(d > 1.5 mm):

p= P(PT
--PC) (4)
dP&T-Pc)
where: p , pT , pc- density of water, sedimentsand mixtures.

For determination of relations between a coefficient of kinematics viscosity of

clear water and the mixture we used the data of laboratory experiments,and the results
of mixture viscosity measurementswith help of viscosimeter VPM-4. Moreover, the
results by SEMirtshoulava (1967) and E.KRabkova, I.R.Kulesh, M.A.Mostkov
(1960) where used. These data made it possible to plot the appropriate curves for
different diametersof sediments(Fig. 1).

1 - dc0.01 mm (Mirtshulava C.E.)

0.8 2 - dc0.01 mm (Levi 1.1.)
3 - d=0.014 mm (Mostkov M.A.)
0.6
4 - d=0.021 mm (Ismagilov H.A.)
5 - d>0.05 mm (Einstein A.)
6 - d=0.003 mm (Daida A.)
7 - d<O.1 mm (Rabkova E.K.)

S
Fig I Functional relationship v,/ v, = f(s)

On the basisof these curves the following formula was derived:

VB+ 0.033
V sm Fso’ s5 (5)
B

In accordancewith (2) - (4) the formula (5) may be written as:

166
 a) for laminar regimes:

(6)

b) for transitional regimes:

(7)

c) for turbulenceregimes:

(8)

Fig. 2 shows the curves of coarsesedimentsdistribution over depth. In accordancewith

these curves, the hydraulic fall of alluviums is different over the depth depending on
flow velocity and nonuniformity of the deposit. In easy of high velocity and a small
uniformity of sediments,coarse sedimentsare distributed over depth. With increaseof
nonuniformity, fine sedimentsare stored on the top, of the flow and coarse sediments
are found in the bottom.

Y/H v/a:
123456 i 1 1 - 180
1.o.li \ 2 - 240
3 - 500
4 - 900
0.8 5 - 1200
\\ 6 - 1600
0.6

0 0.2 0.4 0.6 0.8 ~Yhl

Fig 2 Change of hydraulic coarse of alluviums

The analysisof Fig. 1.2 shows, that curves may be obtained from the following
equation:

w, = a&,[l-(Y/qz] (9)

167
 where o, is a hydraulic fall of sediments;
z is an index, which dependson the flow velocity and
degreeof uniformity (0 I z I 1)
For high velocity and small uniformity of sedimentsthe value of z is very small.
With the increaseof nonuniformity of sediments,the value of z must be higher.
The coefficient py of density stratification is decreasingtowards the bottom and
its value is not high:

p, = p,Ll+ 0.3(Y/H)x] w-9

where is coefficient of densitystratification of sedimentson the bottom.

168
Channel Realationships of Mud Flow in the Central Asia

Introduction
All the rivers in the Central Asia are subdividedinto two categories:
l the rivers with regular water flow during the year;
l ephemeralmud streams,flowing during heavy showers,in particulary.

The number of ephemeral streams are 42% of the total number of rivers.
Maximum dischargesof ephemeralstreams result from rare but heavy showers and
producing mudflows. Dischargesof mud flows in ephemeralstreamsreach 500-600
m3/sand more, at flow speedup to 5-6 m/s and sedimentsconcentrationup to 400-450
kg/m3. Mudflows passingdown the channelsof ephemeralstreamscausegreat damage,
which can be hardly comparedwith the damageproduced by high floods in river.

General
Mudflow control requires a developmentof design relations between the channel size
and form, and main factors defining the formation flow dependencer,for mountain and
Piedmontrivers in the Central Asia.
Many scientists (Ilyin I.A.,Altunin S.T,Artamonov K.F.) investigated channel
relations between mountain and Piedmont rivers in the Central Asia. But they studied
rivers at the assumtionof constant water flow during a year, that is why not all factors
were taken into account which had a considerableinfluence on forming the mudflow
river beds, such as soil characteristics (rocks) making channel banks subject to
deformation,influence of suspendedsediments,etc.
We here establishedchannel relations for ephemeralmudflow streamsusing the
structural formulae, suggesteedby M.A.Velikanov (1955) for determinationof channel
sizes.
M.A.Velikanov suggestedthe following structural equationsfor the channelwidth
and averagedepth of the water flow:

-=
B
d
K

[ 1- Q
’ d”@
x, (11)

[ 1
$=K2 ~ x2 Q
d25&i
(12)

169
 where Q - water discharge;d - mean diameter of bed deposits;i - channel slope;
B - channel width; H - averagedepth of flow; Kl and K2 - parameters;Xl and XZ -
exponents.
On the basis of the analysisof morphometrical relations we took XI =0.5; XZ
=0.25. K, and KZ - changeable.For revealing the characterof changesin Kl and KI in
mudflow rivers in the foothills of Central Asia, the data on hydraulic parametersof
mudflow rivers received by R.G.Ufin,S.E.Esimov and V.V.Korobov were used. For
foothill mudflow rivers in Central Asia parametersKI and Kz are subjectto considerable
changes(Table 1): K, from 0.33 to 2.87; Kz from 0.24 to 1.1. In spite of this for
mudflow watercourseswith lessstablechannelthe parameterK, - is increased,and KZ -
is decreased,in such a way that cross section of the channelbecomesmore splitted in
comparisonwith other rivers, theseparametersdependon the propertiesof soils making
channelbanks,and on the quantity of transportedsuspendedsediments.

Table 1. Calculation of K1 and & values (X1 = 0.5; X’, = 0.25)

River H KI
1. Nukur 998 0.020 1.45 1.94
2. Firuza 244 0.029 0.170 30 1.30 0.87
3. im.Khudaiberdyev 50 0.020 0.080 7 1.30 0.35
4. Arnov 28 0.018 0.090 5 1.00 0.33
5. Chartaksai 110 0.011 0.030 100 0.45 2.27
6. Chartaksai 111 0.008 0.026 79 0.66 2.04
7. Sasyksai 67 0.020 0.023 60 0.48 1.75
8. Shishaksai 53 0.015 0.023 55 0.36 1.70
9. Irvadan 16 0.016 0.025 20 0.15 1.26
10. Chirvan 93 0.022 0.025 40 0.39 1.12
11. Beshbuz 30 0.040 0.030 15 0.50 0.91
12. Sardobsai 100 0.016 1 0.030 1 40 1.03 1.06

As a result of the analysisfor estimatingthe soil properties for deformation, we

used coefficient of soil solidity suggestedby M.M.Protodiakonov. Loose soils have the
least value of of the soil solidity coefficient, rocky soils - have the greatest one.
According to M.M.Protodiakonov, solidity coefficient for different soils from the
following equations:
for loose soils fx = tgu, (14)
tg(49) + c
for cohesivesoils fx= 9 (15)

for rocky soils

where p - angle of soil inner friction;

C - clutch coefficient;
9 - voltage at which soil resistancefor deformationsis determined;
Rc - stability limit for pressing.

170
 Values of stability coefficient calculated for different soils occured in foothills
rivers in the Central Asia, are given in Table 2 (according to Karpov I.M., 1985),.

Table 2. Valuesof soil stability

Soils, making channelbanks f

1. Sands.sandyloams 0.5 -x0.6
2. Loams, sandyloams with coarsesandsand gravels 0.7 - 0.8
3. Loams. sandyloams and clays with gravels and pebbles 0.9 - 1.4
4. Coarsefragmental soils 1.5 - 2.0
5. Semi-rocks 2.0 - 4.0
6. Rocks 5.0 - 20.0

The analysisof data on hydraulic parametersof mudflow shows that the channel
banks are formed by sandy-loamsand loams with gravels and pebblesand by coarse
fragmentalsoils. For such soils the stability coefficient varieswithin 0.5 - 2.0.
Fig. 3 gives parameter link K with fx and S. It is characteristicthat with the
increase of soil stability coefficient and decreaseof the turbidity the parameter K is
decreasing.The mathematicaldescriptionof this relation has the form:
1.33

that in combinationwith (11) gives the generalexpression:

D /l+s)‘,33 (17)

Fig 3 shows the link of parameter&, fx and S. This link shows that with the
increaseofjr and S, the parameter& tendsto increase.
On the basisof link diagramthe following formula is derived:

Kz = 0.62f&,l+s, (18)
If combinedwith (12), it gives the generalexpression

N= o.02fX(l+s) Q”.25do.375
ki)“.125

171
 KI - /

2.0
0

0 Cl

0
;

B
,,,’

/
/
‘, /
1.0
o ,pc’

,/’

-4”

’ ~
//
J

0 1.0 2.0 1+s

fx
F’ig 3 Relation Kl = dfx.S)

K2

7.0
Fig 4 RelotFon KZ = cfx.S)

The general relation associationdescribing the change of relative channel depth

by using formulae (17) and (19) may be presentedas follows:
H (,62f,2.33 d0.625(gj)0.125

2 = (l+s)0.33 - ~0.25

In summary,theserelationspermitted to determinethe parametersof mud flow for

many rivers of the Central Asia and to develop methodsfor protection againstmud-
flows in the foothills of the Central Asia.

172
Bibliography
Altunin, S.T. (1962) 7he rivers regulation, Moscow, Selhozizdat,352 p, (in Russian)
Artamonov, K.F. (1963) Construction of water diversion structures in predmont
regions, Frunze, Proc. Academy of SciencesRepublic Kirgyhizstan, 265 p. (in
Russian)
Iliyn, LA. (1959) The water resources of Fergana valley, Leningrad, Gidrometeoizdat,
300 p. (in Russian)
Karaushev, A.V. (1977) Theory and methods for mudflow calculation, Leningrad,
Gidrometeoizdat,277 p. (in Russian)
Karpov, I.M. (1976) 7hejlowing’s phenomenon at Uzbekistan, Tashkent,Fan, 134 p.
(in Russian)
Kulesh, N.P. (1960) ‘On the viscosity coefficient for clay mixture’, Leningrad,
ScientificBulletin N4. (in Russian)
Mirtshoulava, C.E. (1967) The river erosion and methods of estimation its stability,
Moscow, Kolos, 179 p. (in Russian)
Mostkov, M.A. (1959).meov of riverprocesses, Moscow, 216 p. (in Russian)
N&tin, I.K. (1980) Complex turbulence flows and processes of heat transfer, Kiev,
237 p. (in Russian)
Rabkova, E.K. (1956) ‘Investigation of mudflows in Fergana Valley’, Tashkent, Fan,
p.87-201. (in Russian)
Ufin, A.P. (1965) Hydromechanization, Moscow, 495 p. (in Russian)
Velikanov, M.A. (1954) Dynamics of riversfZows, GITL, v.1, p.324, v.2, p.384. (in
Russian)

173
3. RIVER CHANNEL DYNAMICS
River response to natural and
man-made change
Stevan Bruk
13, rue du Colonel Oudot
75012 Paris, France

Abstract
Rivers are at the same time- mechanical, ecological and
technological systems which interact between themselves and
determine the response to change. Natural change is omnipresent
and permanent, while man-induced change is - either or both - a
managament goal and a by-product of human activity in the river
basin.

Simulation for predicting river behavior is possible only

when quantitative relationships between the governing factors
are known. If this is not the case, like in ecology,
descriptive models are also useful. Calibration of mathematical
and scale mode1.s by cumulative effects does not prove the
totality of the model, nor each of its parameters; complexity
of models increases data error which reduces the realibility of
predictions.

The recognition is growing that the time span of feasible

predictions of river behavior is limited by the chaotic
response generated by non-linearity of the basic equations,
augmented by input stochasticity.

The management of river response must be flexible,

adaptable to changing objectives and imprecise, fuzzy criteria
and constraints. Avoiding biased approaches, dominated either
by technology or ecology, management policies should accept
that man and his activity are inseparable from the river system
and its environment.

The study of rivers requires an integrated scientific

discipline, known as potamology. While some branches of this
science progress normally, others are in apparent crisis or
even in a rudimentary, pre-paradigm state. Models consist of a
"hard core" based on fundamental laws and the auxilliary
hypotheses and assumptions, adjustable to experimental
evidence. The proof of creativity is the prediction of new
facts, confirmed by subsequent observation.

Progress in understanding river response needs an exchange

of information on river behavior between specialists of
different professions and from different geopolitical regions.

174
Introduction

Omission and simplification help us to understand,

but help us, in many cases, to understand the wrong
thing... (Aldous Huxley: Brave New World Revisited,
1958)

Any analysis is simplification by definition: a question

is extracted out of the immensely complex reality, then
described and presented in some of its facets, like the
elephant of the old Persian tale. The perception of rivers is
no exception: the river appears differently depending upon who,
when and how he looks at it.

The risk of understanding the wrong thing about rivers is

best minimized by the exchange of views from different angles,
between specialists of different professions and from
different regions. The present report is meant to stimulate
such discussions rather than to propose one-sided conclusions,
to contemplate broader perspectives rather than to insist on
calculation procedures. In its present preliminary form, it is
more a personal view than a state-of-the-art review. For
brevity, familiar notions, facts and mathematical expressions
are only hinted upon or commented shortly, as thought
necessary.

The paper points out a number of discussion subjects,

which are for convenience presented in 8 boxes and shortly
commented upon in the subsequent texts.

Box 1 The triad of the river system

1.1 Rivers as mechanical systems
1.2 Rivers as ecological systems
1.3 Rivers as technological systems
1.4 Interactions
1.5 Appreciation of river response to change

The river appears simultaneously as a mechanical,

ecological and technological system, as marked in the Box anid
shown on Fig. 1. The elements of the triad interact an'd
transmit change onto each other.

1.1 The river is a mechanical system which transports

water and solid matter and interacts with fixed or mobile
boundaries. Stochastic hydrological input and turbulence make
the system time dependent and the river channel mobile.

1.2 As the river offers habitat to living organisms, it

is an ecological system, which interacts with its environment
in the fluvial corridor and river basin. Change is omnipresent

175
 S. Bruk: River Response to natural and man-made change

FIGURES

ECOLOGICAL MECHANICAL

SYSTEM
SYSTEM

TECHNOLOGICA

SYSTEM

Figure 1: The river system triad

The three systems interact and influence each other: any change
within one of the systems is transmitted to the other two.

Input error

Simulation

MODEL COMPLEXITY

Figure 3: Reliability of prediction versus model complexity

The predictive capability of the model increases with its complexity, but
cumulative error generated by input data reduces the reliability of the
prediction

176
with ecological successions responding to the transfer of
matter and energy across the mechanical fluvial system. Living
organisms on their turn influence conveyance capacity, flow
velocity, bed mobility, bank stability and the performance of
engineering works.

1.3 Rivers belonging to technological systems are subject

to engineering measures and management, which produce changes
intentionally for utilitarian purposes, or as undesired side
effects of human activity (e.g. pollution of water and
sediment). Engineering works change river mechanics - water
flow, sediment transport and channel forms, which then affect
the aquatic and the adjacent terrestrial environment with a
possible feedback effect on the mechanical system and the
technological system itself.

1.4 The interactions between river mechanics, ecology and

technology are understood in principle, in general lines at
least. However, for mathematical simulation of the fluvial
processes, these have to be expressed also mathematically,
which is still a much needed research objective.

1.5 Very naturally, river response to changes is

appreciated differently by different professions and
scientific disciplines, such as geography, geology, hydrology,
biology, civil and hydraulic engineering, etc. Moreover,
considerable differences in concepts, outlooks and methods
exist between scientists and researchers within the same
profession, but in different parts of the world.
Multidisciplinary and transdisciplinary approach is thus
essential, extending over geographical and geopolitical
regions.

Box 2: Rivers as mechanical systems

2.1 Inherent non-linearity and complexity of river mechanics
2.2 Sediment transport and management
2.3 Self-preservation of rivers and adjustment to change
2.4 Observations and statistics
2.5 Extremal hypotheses

The perception of rivers as mechanical systems is based on

Newtonian mechanics, integrating the uncontested differential
equations within the boundaries of the river'. This approach
offers an insight into river behavior and a foundation for
predictive modeling [de Vries,1993]. However, the complexity of
the phenomena and the need for practical solutions has
instigated researchers to use short-cuts, at one hand based on

1 An excellent analysis of this method is given by E.

Mosselman [Mosselman, 19921.
177
observation of the river as a whole, and at the other hand, by
making use of some generally accepted physical hypotheses.
Several discussion subjects appear in this context.

2.1 River flow displays at one side a random, quasi

chaotic movement of water and sediment particles, and at the
other side, an almost life-like behavior of the river as a
whole. Both phenomena are explained by the inherent complexity
and non-linearity of river mechanics. Even comparatively simple
cases seem to be rather complex. Since ages, people built dikes
along rivers and used composite flood channels; nevertheless,
the flow in such channels still attracts research with new
calculation methods proposed [e.g. Smart, 1992; Ackers, 19931.

2.2 Conceptual difficulties persist in the interpretation

of sediment transport. Sediment laden river flow can be
considered as two-phase flow, marked by solid-fluid
interaction, which is imperfectly understood. Particularly
limited is the understanding of boundary phenomena of bed-load
transport and bed forms. Currently, scores of formulas are in
use, based on different considerations and without real
criteria about their validity [cf. Simons and Senturk, 19921.
Pragmatically, the lack of coherent theory is supplemented by
empirical information, based on physical experiment and
engineering observation.*

2.3 Observers were always impressed by the life-like

behavior of rivers, their amazing capability for self-
adjustment to a variable environment [J.F. Kennedy - many
communications]. For instance, rivers develop bed forms and
modify channel resistance as a response to change of water flow
or sediment input. Each and every part of the river changes in
time, but the identity of the whole remains. Rivers display an
intrinsic similarity in very different conditions (in Russian:
"avtomodel'nost"'): certain proportions observed in alluvial
channels could be detected even in some features of the Moon
[Graf, 19711. Such behavior is characteristic for complex,
highly non-linear systems.

2.4 As certain laws of comportment of complex systems can

be detected even without explicit knowledge of all the relevant
factors, useful information could be extracted from the
observation of river forms, generalized, for instance, as the
well-known "regime formulae" [e.g. Blench, 19691, or
morphometric relationships [Velikanov, 19551. Whereas, this
approach is justly criticized as lacking scientific foundation
[de Vries 19931, river engineers in many countries find it
useful for design. More information could be extracted from the

2
The low level of understanding of two-phase flow is
reflected in the'lack of unity in the concepts of such closely
related phenomena as alluvial river flow, sediment transport in
rigid beds and hydraulic transport in pipes [cf. Nalluri and
Kitshri, 19921.

178
observation of river forms by going beyond correlating mean
values, into analyzing scatter, statistical moments, outlayers,
autocorrelation, spectral properties, etc.

2.5 Controversy exists about the applicability of

extremal hypotheses to explain river behavior, such as the
principle of minimum energy, stream power and alike. Such
approach is criticized as teleological and thus philosophically
incorrect [Mosselman 19921, as contradictory to observations
and as inconsistent with nature which prefers redundancy [de
Vries 19931. Nevertheless, the approach has been applied to
explain fluvial phenomena [Leopold, Langbein, Bagnold] and is
used to develop practical calculation methods [Yang and Kong,
19911.

Box 3. Rivers as ecological systems

3.1 Descriptive models
3.2 Fuzzy criteria and constraints
3.3 Ecotones

The main elements of the fluvial hydrosystem - the river

basin, river channel and flood plain - are bound together by
three-dimensional, time dependent exchange of matter and
energy, which determine the ecological system with its
interactions and successions [Amoros and Petts, 19931.

3.1 Considerable progress has been achieved in explaining

the fluvial eco-system, leading to descriptive models which
eventually may be developed into predictive mathematical
models, provided the relationships between the main parameters
can be expressed in quantitative terms. This is still a
research task of the much needed cooperation between
ecologists, hydrologists, hydraulic engineers and other
specialists concerned with rivers.'

3.2 It is important to note, that criteria and

constraints of fluvial ecology are necessarily fuzzy and
imprecise and thus have to be approached by the appropriate
techniques. This offers a wide and most rewarding field of
activity for present and future research.

3.3 Special attention is to be paid to the contact

between aquatic and terrestrial environment, the ecotones
[Naiman and Decamps, 19901. These are of particular interest to
river engineering as well, since most of the engineering
structures affect and change the contact between the river and
the land (embankments, dikes, river structures, etc).
Apparently, in their design, which was so far based almost
exclusively on mechanical criteria, ecological considerations
will gain importance.

179
 Box 4: Rivers as technological systems
4.1 Objectives
4.2 Technological means
4.3 Interactions between objectives and means
4.4 Concepts and design philosophies

Since the dawn of civilization, rivers have been the

favorite areas for human settlement and activity: hence, rivers
are inseparable from human technology and with few exceptions
in remote areas, no river is in its "natural state" any more:
they are technological systems on their own and parts of larger
systems at the same time.

4.1 By applying technological means, rivers are forced to

respond to management objectives, which originate from very
different sources: from satisfying elementary needs of
protecting river banks against erosion and land against
flooding, to the complete transformation of the river into part
of a sophisticated multipurpose water development project or
integrating it into an urban environment.
The objectives are based on social needs founded on an
ethical background which transforms into political will: shifts
in social ethics are reflected on the objectives. The best
example is the change of objectives in the last decades with
the assertion of environmental awareness and ethics, which are
becoming predominant in defining river management objectives.

4.2 The size of rivers by itself limits the choice of

technological means: they must be simple, inexpensive per unit
length and resistant to adverse influences. River engineering
works are the typical example: changes in the economy of
construction modifies these through the ages. It suffices to
open any old textbook on river engineering and compare it with
present day's practices. The balance between capital
investments and costs of maintenance is a major consideration
in selecting the appropriate technology: it varies from place
to place and changes in time, depending on the socio-economic
and political circumstances.

4.3 The objectives and technological means strongly

interact between themselves: the best example iS the
development of navigable waterways on rivers. For instqnce,
navigation requirements grow as the technological means become
more powerful, like mass dredging, possibility of large scale
river works, construction of dams and barrages. This is an
endless process, controlled only by external considerations, of
which environmental awareness and the respect of fluvial
ecology seem to be the most powerful in these days.
Environmental ethics limits the increasing "de-
naturalization" of the fluvial corridor, and even incites
technology to develop means in order to "re-naturalize" them.
The latter are by no means simple nor cheap: it is not easy to

180
undo what has been done to the river during the ages of
technological interventions [Larsen, 19921

4.4 Design concepts and philosophies reflect both the

objectives and the means of river management and technology,
but are influenced also by the tradition of the engineering
profession in different regions of the world. In general,
concepts change slowly and attitudes are rather conservative.
The reason is probably in the mere size of river works at one
side, and in the uncertain criteria at the other side.
Experience about the performance of large scale river works is
often spurious, based -on impressions rather than on stringent
analysis3

Prediction of river response to change is still rather an

auxiliary tool of design than its backbone, like in other
branches of technology. The reason is probably the high cost of
prediction which often requires expensive field investigations,
sophisticated models (mathematical and/or scale) and often
lacks confidence. This situation gradually changes as
prediction methods advance, become less expensive and gain
confidence.

Box 5: The nature of change affecting river

behaviour
5.1 Change as a natural phenomenon
5.2 Change as a by-product of human activity
5.3 Change as a management goal
5.4 Change in perception

Two kinds of changes affect our lives more than probably

ever before: (a) the change of the world around us and (b) the
change of how we see and understand the world. These changes
are present in all aspects of life and science and the subject
of this paper is no exception to that.

5.1 Change is a permanent and inevitable natural process.

The independent variables of river mechanics are subject to
change: climate, hydrology, soil surface, water temperature,
chemistry, flora and fauna, etc. Extreme events, such as
exceptional floods, earthquakes, landslide, etc, produce abrupt
changes, modifying river channels sometimes in a decisive way.

3 The writer recalls a case when owing to different

traditions in two neighboring countries, the relative elevation
of river training structures (as regards to the river stage) on
the same navigable river differed by one meter, so that
transition was necessary at the border where the two systems
met. Neither side wanted to change its practice, since the
river engineering structures "performed well" in both
countries.
181
 5.2 Human activities in the river and its basin, such as
land use, urbanization, roads and railways, water resources
development projects, all affect hydrology, sediment input,
water temperature and chemistry and indirectly, successions of
flora and fauna.

5.3 Change is also a goal of physical, regional and urban

planning, aimed to satisfy engineering requirements to modify
rivers, generally out of their "natural" state (dams, dikes,
bank stabilization, river training, intakes, etc). In recent
years, river engineering measures are being applied by
environmental considerations, to bring the rivers if not back,
but at least closer to their erstwhile state ("re-
naturalization" of regulated water courses and river corridors
[Larsen, 19921).

5.4 An all important change in perception is the growing

recognition of the inherent impossibility of long-term
predictions for highly non-linear systems [Davies & Gribbin,
19921:

. . . because no system can in principle be described with

perfection, completely accurate long-term weather
forecasting can never be achieved - nor can accurate
forecasting of any other chaotic system. We stress that
this is not just a human limitation. The Universe itself
cannot 'know' its own working with absolute precision, and
therefore cannot 'predict' what will happen next, in every
detail. "4.

This could possibly explain the difficulty in simulating

movable bed rivers, which are complex non-linear systems with
stochastic input: the incapacity of predicting long-term
evolution of rivers is probably not just a human deficiency due
to imperfect knowledge, but is caused by inherent properties of
the river itself as a highly non-linear system, with stochastic
information input. There is an intrinsic uncertainty of the
responses of the river system to the genuinely random inputs.

Box 6: Simulation and modeling of river response

6.1 Scope of modeling fluvial processes
6.2 Morphological models
6.3 Calibration and evaluation
6.4 Model complexity and data requirements
6.5 Is long-term prediction possible?

4
Paraphrasing the above statement, one could say:
"It is not just that we don't know what will happen next to a
river: the river itself cannot know that".

182
 The comments below point out some difficulties and
limitations of modeling rivers for engineering purposes. By no
means are they intended to challenge the usefulness of models:
the ultimate criterion is the support given by the model to
reach the best decisions and it is far better to base decisions
on even imperfect models than on no models at all. The
interpretation of model results is then a joint task of both
the modeler and the user of the model. From the pragmatic,
utilitarian point of view the main criterion of prediction is
how well it serves management purposes. The position of
modeling in the project is shown on Figure 2.

6.1 Simulation needs a good (if even imperfect)

understanding of the phenomena, reliable descriptive models,
availability of basic transformation equations, parameters and
constants, quantitative information on boundary and initial
conditions, etc. Simulation is current for problems in
hydraulics and channel morphology, but also increasingly for
the prediction of pollutant transport and transformation
processes [Jolankay, 1992; Olesen and Larsen, 19921. Ecological
responses are still estimated mainly by experience and
analogies.

6.2 Morphological river modeling has made considerable

progress both as regards numerical and scale models [Cunge,
1989; de Vries et al, 1991; de Vries, 19931. These methods are
very far from the direct numerical solution of the fundamental
equations of hydromechanics, which is possible for much less
complex situations only5.

River models comprise numerous simplifications, hypotheses

and assumptions, sometimes widely divergent. Thanks to an
extraordinary effort, the comparison of 12 selected models in
the USA resulted in a good insight into their performance6.
While the work still goes on, the objective of developing
criteria for selecting the best models for given situations has
not yet been attained [Fan, 1988; Fan and Ben, 1991.

6.3 Calibration is indispensable both for mathematical

and scale models. The comparison between model and prototype is
based on measurable quantities such as water levels, channel
forms, etc, which are in fact cumulative effects of the
modelled fluvial processes. Now, similar cumulative effects can
be obtained by a multitude of combinations of the relationships
and parameters built into the model. Hence, corroboration of
predicted and observed cumulative effects does not prove each

5
The difference can be appreciated when comparing with the
efforts needed for the direct numerical simulation of
turbulent flow over a backward-facing step [Le et al, 19931 and
the precision of flume experiments by which the results of
simulation could be confirmed [Jovic, 19931.
6 The effort is carried out by the Bilateral Working Group,
coordinated by Dr. Shou-shan Fan from the US Federal Energy
Regulatory Commission, Washington, D.C.

183
 S. BRUK: RIVER RESPONSE TO NATURAL AND MAN-MADE CHANGE
1
IDENTIFICATION
I
DESCRIPTION
HISTORY
PROJECT
OUTLINE

3 4
FORMULATION INFORMATION CONCEPTUALIZATION
OBJECTIVES DATA MODEL CONCEPT
CRITERIA FIELD RESEARCH INPUT SIMULATION
CONSTRAINTS ESTIMATION COMPUTATION
I I

INTERPRETATION
SENSITIVITY
RELIABILITY
JUDGEMENTS

DESIGN
DETAILS
ECONOMY
ENVIRONMENT
TIMING

I VERIFICATION
REVIEWING
PUBLIC HEARINGS
FINANCES
EVALUATION

IMPLEMENTATION
CONTRACTS
CONSTRUCTION
OPERATION
MAINTENANCE

Figure 2: Flow chart of project activities

Simulation and modeling are in step 4 of the activities

leading from project identification to implementation

184
and every parameter or relationship: many of these can be just
wrong, the errors mutually canceling out each other'.

Limitations arise from the lack of reliable historical

data, and from restricted time and money. Calibration and
sensitivity analysis lege artis needs extraordinary input of
means, as shown in the case of the Danube River [Belleudy,
1992; Loy et al, 19921 Hence, predictive calculations are often
carried out by incompletely calibrated and non-validated
models, in good faith that the results will not be far from
reality. Generally, predictions are the better, the stronger
was the change of the natural regime of the river and are the
worst when comparing two "natural" states of the river, which
differ only slightly.

6.4 The capability of a model to simulate reality can be

improved by better mathematical description of the phenomena
and by including more parameters into the analysis. However,
higher complexity of the model requires more input data, and
each data carries a certain amount of error: the more
parameters are used in the model, the larger is the cumulative
effect of data error (Figure 3).

Hence, for a given case the best is to select the simplest

code, with lowest data requirements, but still able to support
efficiently the required engineering or management decisions.
The complexity of the model could then be increased for higher
precision, if this is both necessary and possible.

Large rivers like the Amazon, Brahmaputra, Mississippi,

Volga, Danube and others merit to be explored and described
individually, one-by-one, finding out the best, site-specific
means and methods of simulation. Medium and small rivers could
be classified and interpreted by geographic regions. The best
suited models should than be evaluated without prejudice,
taking into account the world wide experience of researchers
and engineers. This is an immense task which needs a joint
effort of the international community.

6.5 The intrinsic uncertainty in the responses of the

river system to the random inputs probably makes impossible
long-term predictions of river behavior. If this were true,
than two identical prototype-size rivers (scale models) with
identical inputs would evolve differently. The question was
raised also by E. Mosselman [1992]: "Chaos in river morphology
is still a hardly explored subject..." and further: "Chaotic
behavior of rivers would imply that, beyond a certain limit,

7
For instance, in the writers personal experience,
underestimated trap efficiency of the Iron Gates Reservoir on
the Danube was compensated by overestimated sediment input so
that the prediction of reservoir sedimentation was basically
correct, in spite of the imperfections of the model [Varga et
al, 19891.

185
long-term predictions of bed topography patterns or river
platforms cannot be made more accurate by, for instance,
improving the sediment transport formula or the bank erosion
equation. Research should then be devoted to delimiting the
range of possible outcomes, related to the envelope of the
chaotic attractor in the phase space."

The impact of this on simulation of fluvial processes has

yet to be evaluated. No simulation can be perfect, in
principle; and a mathematical model which would simulate a real
river near to perfection, should manifest the same
indeterminism as the river itself: repeated tests would result
in different evolutions of the river, the difference increasing
with the increase of time-span of predictions. The definition of
the time-scale of acceptable prediction accuracy is thus an
important research task of simulation.

~Box 7: Management of river response

7.1 The "technologistic" approach: "Clever engineers and dumb
nature"

17.2 The "ecologistics" approach: "Nature knows it best", or

I "Dumb engineers and clever nature"

7.3 The "cybernetic or management" approach: "Man is part of

the river system"

Management of river response is an essential part of the

more general concept of river management. It concerns only the
adjustment of the river channel (including the floodway) to
river management goals and also the attainment of ecological
management objectives.

7.1 The technologistic approach to river response

management consists in the adjustment of the river channel to
pre-set management objectives. Typical examples are manifold:
river bank protection against erosion; protection of land and
property from floods; increasing the channel conveyance
capacity for the passage of flood discharges; adjustment of
channel depths, widths and alignment to navigation
requirements; restricting the river channel for urban
development, etc. The natural trends of the river were
considered only in so far as they could make the river
engineering works cheaper and safer. Examples of total neglect

8
The writer does not feel qualified to propose an opinion
on this question, which might or might not be of great
importance for river simulation. It is sure, however, that the
matter needs a deeper examination.
186
of environmental or aesthetic considerations or criteria are
abundant in all parts of the world.

7.2 The ecologistic approach is a reaction to the former

one. In extremis, it assumes a rather negative position to any
engineering intervention in the "natural" river system. Typical
of this attitude is the report of the Equipe Cousteau [1993],
which rejects almost all what has been done and what is planned
of engineering works on the Danube river and its system. The
fallacy of this approach is in disregarding that what was
"good" for nature without man is not feasible any more when man
is present. The goal of a natural river could be attained only
in the improbable case of by removing people from its system.

7.3 The compromise between these two extremes is here

named the "cybernetic or management" approach, following the
terminology used in water resources management [Djordjevic,
19931: it accepts that the river is at the same time both a
natural and technological system.
Man is part of nature in general, and of the river system
in particular. In this concept, the river manager / engineer is
not external to the river but belongs to it, as its constituent
part. His presence is thus permanent, his activity
uninterrupted and inseparable from the life of the river. In
this way, the river system which comprises the engineer/
manager, is 'able to react not only to natural influences, but
also to human interference.
The main difficulty arises from the definition of
management objectives, which by their very nature are fuzzy,
imprecise and time-dependent. The management thus needs to be
flexible and permanent, reducing irreversible interventions to
the bare minimum.

8. The scientific study of rivers

8.1 Potamology - the integrated science on rivers
8.2 The present status of potamology
8.3 Lack of consensus
8.4 The hard core and protective belt
18.5 Criteria of a proqressive science

It has been recognized since long, that cooperation of

specialists of various backgrounds is necessary to deal with
rivers efficiently. It appears, however, that the barrier
between the various disciplines is too high to be overcome by
"ad hoc" multi-disciplinary teams dealing with particular case
studies. As stated at the IAHR Workshop [IAHR, 19911:
"Hydraulicians, hydrologists and biologists are largely
ignorant of each other's science". It suffices to compare the
references in books which deal with similar subjects, but

187
looked at from different anglesg. Specialists having a common
background should thus be educated through special academic
programmes for the scientific study of rivers.

8.1 Integrated scientific branches are frequently denoted

by freshly coined terms such as, for instance, "eco-hydrology"
[Nachtnebel et al, 19921. Avoiding the creation of another new
term, the integrated science, which would extend over the whole
range of specialties needed for the study of rivers, could
probably be called as potamologyl'.

Potamology as the scientific study of rivers reflects the

complexity of the river system: it integrates the respective
parts of at least three categories of scientific disciplines
which deal with rivers:
* natural sciences such as hydromechanics, hydrology,
limnology, biology, ecology, etc.
* engineering sciences - hydraulic engineering,
geotechnics, civil engineering, bio-engineering, etc
* social sciences, dealing river management, economy,
decision making, etc.

In its fundamental side, potamology describes and explains

rivers, and in its applied side, supports management decisions
related to rivers.

8.2 At present, no distinct scientific community

specialized in rivers can be identified, which would share
common beliefs and uncontested pre-conceptions - or paradigmsll
[Kuhn, 19701, or use common models for the study of rivers.
Hence, potamology -if accepted as a separate scientific
discipline - is in the early, "pre-paradigm" stage of its
development. Certain important areas of potamology are still
dominated by descriptive models, lacking the quantitative
relationships needed for predictive mathematical models.

9
For instance, among the 450 references in [Amoros and
Petts, 19931, and the 120 references in [IAHR, 19911, only 12
overlap, in spite the related subject.
lo Potamology denotes the multi-disciplinary scientific study
of rivers, in general, but became used for the biological study
of rivers mainly [Encyclopedia Universalis France S.A., Corpus
18, p.811, 19891. This term sounds definitely better than
"fluviology", of mixed greco-roman extraction.
l1 The applied matters related to rivers which hardly can be
called as scientific theories, are put here in the light of
some concepts of philosophy of science which concern subjects
such as the theory of relativity, quantum physics, etc. While
risking the ridicule in doing this, it is believed that these
concepts may inspire thinking in the right direction.
188
 Some branches of potamology are already in the more
advanced stage of "normal science" marked by steady evolution,
where new knowledge is added to the existing one. For instance,
the hydraulic simulation of river flow by 2D and 3D models is
fairly successful, firmly based on the principles of fluid
mechanics and more particularly, on the uncontested Navier-
Stokes and Reynolds equations. The occurrence of chaos
generated by non-linearity does not refute the validity of the
approach, just requires a better definition of criteria for the
time span of feasible predictions.

At the other hand, some branches display the visible signs

of crisis, marked by the proliferation of competing theories,
professional insecurity, increasing interest in philosophical
questions, as pointed out with regards to hydrology [Beven,
19871. The search for an acceptable paradigm is obvious in some
areas, where so far all attempts have failed to find a reliable
concept, e.g. in sediment transport mechanics, and in
particular, for bed-load prediction [Bruk, 19921.

8.3 Accepting that potamology is not yet a "normal

science", it would be vain to insist on a consensus about its
content, methods and models. The groups who deal with rivers
are still to disparate to make such an effort productive. The
main obstacle in developing a unified outlook is the dissimilar
academic education and professional background of those who
deal with rivers. The lack of basic knowledge of each others
disciplines often frustrates efficient cooperation.

In the "pre-paradigm" state of a science, brusque changes

and promotion of new "revolutionary" ideas, which disregard
what already has been achieved, can hardly result in
significant improvements, either. Progress can be expected from
a patient exchange of views between professions and experts
from different geopolitical regions, maintaining a critical
spirit though, doubting concepts, however attractive they be,
and double-checking data which seem to corroborate them.

8.4 A clue for validating theories and models could

possibly be found in the concepts of I. Lakatos [1970]: a
mode112 consists of its "hard core" and a "protective belt"
around it. The hard core is deducted from theory and cannot be
modified without major changes in scientific perception. The
"protective belt" is shaped around the "hard core" in a
hierarchy of auxiliary hypotheses, assumptions, parameters and
constants. All elements of the protective belt can be adjusted
in the light of experiments and field observations, without
affecting the hard core and without contesting the validity of
the model (Figure 4).

'* Rather than to scientific theories, I. Lakatos refers to

"research programmes", such as Newtonian mechanics, quantum
physics, etc. His ideas are freely used here to assess
simulation models in an applied branch of science.

189

- -
 S. BRUK: RIVER RESPONSE TO NATURAL AND MAN-MADE CHANGE

THE HARD CORE

ASIC CONCEPTS AND EQUATIONS

THE PROTECTIVE BELT

AUXILLIARY HYPOTHESES, ASSUMPTIONS

PARAMETERS, COEFFICIENTS, CONSTANTS

TEST

EXPERIMENT, FIELD

SENSITIVI; ANALYSIS1

I VALID

Figure 4: Validation of a model

In the processof verification, the hard core stays immutable,while the protective belt is adjustedto the results
of experiments and / or field investigations ( the terms “hard core” and “protective belt” are borrowed Ii-om
I. LAKATOS [ 19701)

190
 Following this concept, the hard core of river flow
simulation consists of the fundamental equations of hydro-
mechanics, while all the additional hypotheses, assumptions,
semi-empirical or empirical relation,ships (e.g. sediment
transport equations) are included in the protective belt.
Whatever the results of testing were, no one would ever
challenge the hard core of the model, while the model
developers would not hesitate to change any element of the
protective belt in order to obtain a good fit between
predictions and measurements13. Putting speculative arguments
and non-substantiated hypotheses into the "hard core" risks to
devaluate the model and deprive it from possible progress.

8.5 It should be recalled, that any model or theory which

contains several hypotheses and many parameters can be adjusted
to interpret known facts. A theory is creative (or progressive,
in the terms of Imre Lakatos) only if it predicts new facts,
which then are proved by experiments14.

Closing remark:
To close the report, the quotation from the front page is
recalled: when simplifying the complexity of the real
river in order to analyze and understand it, the risk of
understanding the wrong thing should never be overlooked.
With this in mind, the search would become more hesitant
perhaps, but will stick safer by the right track.

l3 Typically, modelers of river morphology are indifferent to

applying this or that sediment transport equation in their
models [see Fan, 1988 and Fan & Yen, 19921.
l4 For instance, the creativity of river response simulation
would be confirmed if the predicted chaotic behavior could be
proved by numerical or scale modeling.
191
References

ACKERS, P. (1993) : Flow formulae for straight two-stage

channels, Journal of Hydraulic Research, Vol. 31, No. 4

AMOROS, C. and G.E. PETTS (1993): Hydrosystemes fluviaux,

Masson, Paris Milan Barcelone Bonn

BAGNOLD, R.A. (1966): An approach to the sediment transport

problem from general physics, US Geological Survey,
Professional Paper 422-J

BELLEUDY, Ph. (1992): Part II: Comparison of field data with

modeling results. In: Modeling of the Danube and Isar Rivers
Morphological Evolution. Proceedings, 5th Int. Symp. on River
Sedimentation. Karlsruhe. Germanv

BEVEN,K. (1987): Towards a new paradigm in hydrology

Water for the Future: Hydrology in Perspective Proceedings of
the Rome Svmnosium IAHS Publ. no.164

BLENCH, T. (1969): Mobile-bed Fluviology, University of Alberta

Press, Edmonton, Canada

BRUK, S. (1992): The management challenge in sediment research,

Proceedings, 5th Int. Symp. on River Sedimentation, Karlsruhe,
Germany

CHANDRA NALLURI and M.M.A.U. KITHSRI (1992): Extended data on

sediment transport in rigid bed rectangular channels, Journal
of Hydraulic Research, Vol. 30, No.6

CUNGE,J. (1989): Review of recent developments in river

modelling Int. Conf. on Hydraulic and Environmental Modelling
of Coastal, Estuarine and River Waters, Bradford, U.K.

DAVIES, P. and GRIBBIN, J. (1992): The Matter Myth - Beyond

Chaos and Complexity, Penguin Books

DJORDJEVIC, B. (1993): The Cybernetics of Water Resources

Management, Water Resources Publications, Fort Collins, U.S.A.

EQUIPE COUSTEAU (1993): The Danube...For Whom and For. What?

Final Report, European bank for Reconstruction and Development
Agreement, Edition Equipe Cousteau, Paris, France

FAN, Shou-Shan (1988): Twelve selected computer stream

sedimentation models developed in the United States,
Subcommittee on Sedimentation,- Interagency Advisory Committee
on Water Data, Publ. by Federal Energy Regulatory Commission

192
FAN, Shou-Shan and Ben C. YEN, editors (1992): Report of the
Workshop on Understanding Sedimentation Processes and Model
Evaluation, Washington, December 16-18, 1991, Federal Energy
Regulatory Commission, Washington

GRAF, W.H.(1971): Hydraulics of sediment transport, McGraw-Hill

IAHR [1991]: Hydraulics and the Environment, Publication of the

IAHR Workshop on Matching Hydraulics and Ecology in Water
Systems, Journal of Hydraulics Research, Volume 29.

JOLANKAY, G. (1992): Hydrological, chemical and bioloqical

processes of contaminant transformation and transport in river
and lake systems, Technical Documents in Hydrology, UNESCO,
Paris

JOVIC, S. (1993): An experimental study on the recovery of a

turbulent boundary layer downstream of the reattachment,
Engineering Turbulence Modeling and Experiments 2, Editors W.
Rodi and F. Martelli, Elsevier Science Publishers B.V.

KUHN, T.S. (1962): The Structure of Scientific Revolutions, The

University of Chicago Press, Second Edition - 1970

LAKATOS, I. (1970): Methodology of Scientific Research

Programmes. Cambridge University Press, ed. 1978

LARSEN, P. (1992): Restoration of river corridors, Chapter 5.6

in River Engineering Handbook, Vo1.2, Ed. G.Petts, J.Wiley &
Sons

LE, H., P. Moin and J. Kim (1993): Direct numerical simulation

of turbulent flow over a backward facing step, 9th Symposium on
Turbulent Shear Flow, Tokyo

LOY, G., Ph. BELLEUDY, J. KELLERMAN, B. SOEHNGEN (1992): Part

III: Range of uncertainty caused by scattered field data with
special respect to grain sorting. In: Modeling of the Danube
and Isar Rivers Morphological Evolution. Proceedings, 5th Int.
Symp. on River Sedimentation, Karlsruhe, Germany

MOSSELMAN, E.(1992): Mathematical modelling of morpholoqical

p recesses in rivers with erodible cohesive banks,
Communications on Hydraulic and Geotechnical Engineering,
Report No. 92-3, Faculty of Engineering, Delft University of
Technology, Delft, The Netherlands.

NACHTNEBEL, H.P., K. KOVAR and Z. ZUIDEMA (1993): Hydrological

bases of ecologically sound management of soil and groundwater,
Technical Documents in Hydrology, UNESCO, Paris

NAIMAN R.J. and DECAMPS, H. (1990): The Ecology and Management

of Aquatic-Terrestrial Ecotones, Parthenon Publishing Group,
Man-and-Biosphere series vol.4

193
OLESEN, K.W. and LARSEN, H. (1992): Modelling Cohesive Sediments
and its Influence on the Oxygen Conditions, Proceedings, 5th
Int. Symp. on River Sedimentation, Karlsruhe, Germany

SIMONS, D.B. and SENTURK, F. (1992): Sediment transport

technology, Water resources publications, Littleton, Co.,
U.S.A.

SMART, G.M. (1992): Stage-discharge discontinuity in composite

flood channels, Journal of Hydraulic Research, Vo130, No.6
SOEHNGEN, B., J. KELLERMAN, G. LOY (1992): Part I: Measurements
and Formulation. In: Modeling of the Danube and Isar Rivers
Morphological Evolution. Proceedings, 5th Int. Symp. on River
Sedimentation, Karlsruhe, Germany

VARGA,S., BRUK, S. and BABIC-MLADENOVIC, M-(1989):

Sedimentation in the Iron Gates Reservoir on the Danube River,
Proceedinqs, 4th Symposium on River Sedimentation, Beijing,
China

VELIKANOV, M.A. (1955): Dynamics of channel flow, (in Russian),

Gostekhizdat, Moscow

de VRIES, M., (1993): Use of models for river problems, UNESCO,

Paris

de VRIES, M., KLAASEN, G.J and STRUIKSMA,N. (1991): On the use

of movable-bed models for river problems: a state-of-the-art,
International Journal for Sediment Research, Vo1.5, No.1
Beijing, China

de VRIES, M., and van der ZWAARD, J.J. (1975): Movable-bed

river models, Proc.ASCE Symposium Modelling '75, San Francisco,
U.S.A.

YANG, C.T. and KONG, X. (1991): Energy dissipation rate and

sediment transport, Journal of Hydraulic Research, Vol. 29,
No.4

194
The relation between river channel
dimensions and discharge of water
K. V. Grishanin
St. PetersburgUniversity of Water Communications

The quasiuniform motion of water in river pools

The long river (L >> B) with noninundated flood plain and negligibly small lateral
influx is considered below. Variation of water discharge is supposed so slow, that
absolutevalues of local accelerationsalways turn out to be small as compared with the
longitudinal component of gravity acceleration:

p+I
dt (1)

By this reasonrating curves are in practice singlevalued:

Q = Q(z) (2)
Let us term a long river reach statistically stable,if at each dischargeof water the
distributions of basic elementsof the flow - widths, mean depths,heights of the bottom
along the fairway - satis@ conditions of stationary random functions. As it is known
these conditions include invariability of mean values, invariability of dispersionsand
requirement that the coefficient of correlation between values z(l) and x(2,A I) of the
function x dependsupon 1 only. The statisticalstability of a river reach is conditioned by
stationary of dischargeprobabilities.
The statistical stability is a temporal state of a river reach. It is conserved until
climatic and geomorphologic conditions of the basin do not change and until human
activity does not intervene in the life of the river. However, it is not anything
exceptional.The long life of a large or middle size river is the successionof equilibrium
state and lessprolonged intermediate statesof transition.
Deep and shallow section - pools and bars - take turns along the river. Depth of
the flow at the upper part of a pool, dividing two adjacentbars rises in the downstream
direction. The longitudinal profile of the water Surfacehas here the form of backwater
curve. Depth at the lower part of a pool dimimshesin the downstream direction and one
has there the drop-down curve. The motion of water at the contact of backwater drop-
down curves, provided that discharge of water does not change, is uniform. If the

195
dischargefluctuates,the motion becomes quasiuniform: the velocity of water changes
becauseof the changeof the discharge:

dU --- 1G’Q
dl-oal (3)
The length of the sectionwith uniform or quasiuniform motion of water depends
upon the length of a pool. Long pools may have such sectionsas long, as the distance
betweentops of oppositebanks, and longer. These sectionsbelong to least deformable
parts of a channeland therefore they often are chosenfor a gaugingstation location.
It is clear by intuition that cross section dimension of a statistically stable river
channelhave regular relationswith the dischargeof water and with the bottom ground.
To reveal theserelationslet us apply to the systemof Saint-Venantequations:

($!!+wdU+d” =0 (5)
al al at
Excluding the derivative d U/d I from the equation of motion (4) with the aid of
the continuity equation(5), one obtains:

Following conditions are satisfied at the contact of backwater and drop-down

curves:

afa 0 (7)
z=

I,-r=o (8)

The equationof motion correspondingto theseconditionsis purely kinematicsone:

ido iau (9)

at uat
CL)
The integrationyields:

o=N2U (10)

where N is an arbitrary function of the spacevariable: N=N(I). Equation (10) may be

rewritten in the form:

BH = NQ"' (11)

196
The value of N in a given cross section of the quasiuniformflow is constant.Following
considerationsw-ill help to expressN through elementsof the flow. Cross-sectionsof
river pools, as most cross-sectionsof natural channels,have large relative width:

B
>> 1
w
River pools, besidesthat, have steepbanks. The last property may be expressedby the
following relation for the angle of banks slope:

ctga = --
2dt 6)
1dB dZ -i
-
t
= O(1)

If conditions (12), (13) are valid, the width of the river stream till flood plain
inundation, may be consideredconstant. Normalisation of the arbitrary function N by
meansof parametersB and g yields:

N = MB3’4g-“4 (14)
where A4 is a dimensionlessquantity - new arbitrary function. Substitution of the
expression(14) into equation (11) leads to the following formula of the mean depth of
quasiuniformflow:

H=M---- Q’”
WY4
The quantity A4 after simple transformation of the equation (15) may be expressedas a
function of two criteria of similarity - geometrical and dynamic ones, namely the
relative width of the flow and Froude number:
-l/4

Values of both criteria at a given cross-sectionof a quasiuniform flow, at water stage

fluctuations, change,but the value of A4 does not changeand consequentlyit plays the
role of the invariant of similarity transformations.
Measurementsshow that in the broad domain, which embracesall quasiuniform
flows with sandbottom, values of invariant A4 fluctuate accidentallyaround some mean
value, approximately equal to 0.9. More precisevalues of A4, obtained in different sets
of experimentswill be cited below. The graph of equation (15) plotted in Fig. 1 is
based on measurementsmade at 35 gauging stations, located on 25 rivers in East
Europe and Siberia. All these rivers have sand bottoms. Three dischargesmeasuredat
eachstation in 1962 - maximal, minimal and close to the averageone were taken. Data
related to the sites of measurementsare given in the author’s book (1979). The
measurementsgave the mean value of the invariant A4 equal to 0.92&O.12. These data
may be supplementedby results publishedby Los ( 1973) in relation to some rivers of
Poland. Measurementsmade at 28 gaugingstationson polish rivers gave the mean value

197
Fig. 1

of A4 equal to 1.13. After the exclusion of measurements fulfilled at 13 stations with

cat&rent areasless than 5000 km’, the mean value of A4 diminkhed to 0.92 (river
with small catchmentsmay have equatic vegetation in their channels). Some more
independentevidenceabout approximate invariability of M in sandy channelscontains
in the paper by Chernov (1960). He derived, on the base of 40 measurementsmade at
sometributaries of the CaspianSea,the following formula of mean velocity:

(17)

where k = 1. From the table in this paper can see that the mean value of the relation
hmx/H in the experimentswas equal to 1.57. Substitution of this value in equation
( 17) yields:

U = 1.4

that is the formula (17) at M=O.85.

The arbitrary function A4, beyond the domain of invariability, slowly diminkhes
with the decreaseof the mobility factor U/&& . The empirical graph of the relation
of A4 to the mobility factor q&c is shown in Fig. 2. The graph is based on
measurementsmade in rivers of Kazakhstanwith pebble bottoms. The averagingline
has the equation
u
A4 = 0.45+ 0.09___ (19)
e40
Ar-
Values of A4 equal to 0.5 t 0.6 are typical of pools with pebble and bouldersbottoms.
Two graphs of equation (15) are show on Fig. 3 and Fig. 4. First one relates to
the cross-sectionof the Yenisei River at Shit, where the bottom sedimentis pebbled and

198
M=O.64. The second relates to the river Mississippi at S. Louis, where the bottom
sedimentis sand and A4=1.04.The graphsshow high exactnessof the local relations of
the quasiunifornrflow in river - dispersionof point is very small.

The motion of water in stable unlined canals with sand bottoms

The quasiuniform flow securesthe statisticalstability of a channelwith movable bed.
Small accidental deformations of the channel may be corrected with the aid of
dredging. It may be establishedon the basis of measurementsmade in rivers that a
cylindrical channelwith sandbottom will be statisticallystableif values of the similarity
invariantA4 are in rounded off limit:

0.75<M<l.O (20)

M
10
r

0 1 v 4

A2 0 5

n 3 X 6

I -.II-
0 2 4 6

Fig. 2 1 - Kolwa River at Petresora; 2 - Chu River at Tash-Utkul; 3 - Assa River at Maimak:
4 - Bcukal River at Energia; 5 - Talgar River at Talgar; 6 - Koktal River at Aral-Tyuhe

H, m

0 10 20
feet 3o 4o 50(9w4
Fig. 3 Fig. 4
The artificial origin of canals necessitatesa small correction in the inequality (20).
Builders of canalsdesiringto diminish volumes of ground which are subjectto removal,
make canals cross sectionsdeeper and narrower than the section of river pools. At a
given conveyancefactor of the channel K = Ix R5’3, cross-section with lesservalues
n
of the relation x lR have lesser areasx R . As a result, the relation x lR in unlined
irrigation canalsat the samevalue of Froude number, turns out to be by 30-40% less
than this relation is in the cylindrical part of river pools. Accordingly, the values of the
similarity invariant A4 in stableirrigation canalsmainly are grouped in the upper half of
the stability domain (20) that is in the interval of 0.9<M< 1.05. This thesisis illustrated
by graphs of equation (15) based on measurementscarried out in stable canals in
Middle Asia (Fig. 5) and in stablecanalsin India and Pakistan(Fig. 6). As it is seenon
the graphs,the agreementof the quasiuniform flow equationto measurementsis good.
The series of investigators, beginning with Kennedy (1894-95) and Lacey ( 1930)
attemptedto reveal, by empirical means,the relation between cross-sectiondimensions
of a stablecanal and the dischargeof water (velocity of the flow). The slope of the free
surface, the diameter of bottom particles, the roughness coefficient were used as
additional parameters.The obtained results inevitably have local significance.The sum
of theseresults is termed the “regime theory”. Fig. 5 and Fig. 6 show that for canalsin
sandone may managewith one relation - the equationof the quasiuniformflow.

, 1.05

‘m‘i /’
/ /

*d-
/ /
/ /
/
/ A/’
/m A,
1 -
A /’
A&

Fig. 5 Canals: 1 - Palvat; 2 - Gazavat; 3 - Ka&ar Saka; 4 - Shavat Nol: 5 - Shavat No2
(qfter Zamarin, 1951)

200
 M

1.05

A 3

Fig. 6 Canals: I - Lower Chenab; 2 - Lower Jelam; 3 - Lower Bari Doab; 4 - Upper Gang
(after Smith, 1970)

The motion of water in laboratory flumes

The motion of water in a laboratory flume, on the bottom of which the sand is laid and
in which the constant dischargeis maintained,in the course of time becomessteady: the
profile of the free surfacebecomesparallel with the smoothedprofile of the bottom, the
mean height of bottom waves - dunes and ripples - is stabilised.As the conditions of the
uniform motion, constant width of the channel and small size of bottom particles are
satisfied,such state of the flow must satisfy the equation of quasiuniform flow (15) with
A4 = co&. Experimentsshow that things are quite so.
In the treatment of laboratory data, the hydraulic radius R = w lx and the length
of wetted perimeter x were used instead of the mean depth H = w lB and the width
of flow Bin rivers .Three circles of experiments, namely, carried out by Laursen
(1958), by Guy, Simons and Richardson ( 1967) and by Vanoni and Hwong Li San
(1967) were analysed. Total number of experimentsin these three circles is equal to
386. From this sum 131 experimentswere selected,as belonged to the same ranges of
leading parameters, as one can find in rivers with sand bottom:
Fr = c < 0.2 , h, / H > 0.05. The points of these measurementsare plotted in Fig. 7.
gH
The mean value of the similarity invariant A4 is equal to 0.91kO.12. It must be
mentioned that experiment by Guy et. al. disclosed the difTerencebetween values of
A4 on rippled and duned bottoms. Ripples gave the mean value of A4 equal to 0.98 and
dunes- equal to 0.83.
Additional data on the quasiuniformflow in fhtmes may be found in the work by
Garg, Agrawal, Sir@. These authors studied the motion of particles with diameter

201
R. m

A 7
. 8

0.1 -
A 9
0 IO
l 11
+ 12

1 I I

0 0.1 0.2
o.3 $4 7m

Fig. 7 Laursen (1958) : 1 - d,,=O.l mm:

Guy et. al. (1967): 2 - ds,=0.19 mm: 3 - 0.27 mm: 4 - 0.28 mm: 5 - 0.45 mm; 6 - 0.93 mm:
7 - 0.32 mm; 8 - 0.33 mm (uniform sand); 9 - 0.33 mm (nonunifOrm sand): 10 - 0.47 mm
(with suspended clay); 11 - 0.54 mm;
Vanoni and Hwang Li San (1967): 12 - 0.137<ds,<0.206 mm

ranged from 0.29 to 0.53 mm. The mean value of A4 in 52 experimentswas obtained
equalto 0.96. Configuration of the bottom was not recorded.

Conclusion
The presentwork may be consideredas the extensionof the theory of kinematic waves
- the theory initiated by Lighthill and Whitham (1955). Physical objects of this theory
are rivers flowing in plains. Small values of local accelerationsmake rating curves
Q = Q(Z), in these rivers, single valued. The present work shows however that the
minutenessof local accelerationsdoes not give foundation to neglect them completely.
Conserving local accelerationsin the equation of motion, but comparing the sum of
acting forces to zero, one comes to the differential equation of quasiuniform motion of
water and as the result obtains the relation of mean velocity to cross-sectionarea of the
flow. This relation contains one empirical parameter - the invariant A4 of local
similarity transformations. Empirical values of A4 fluctuate over all domain of sandy
channels(rivers and unlined canals)with the mean squareerror f 15% The equation of
quasiuniform flow may be used for interpolation and extrapolation of rating curves (up
to the level of banks top). It permits also to draw the rating curve without any
measurements.It permits finally to avoid rough errors at the choice of cross-section
dimensionsof an unlined canal, that is, it plays the role of the theoretically grounded
regime equation. In the domain of gravel and pebble bottom sediments, parameter
A4 slowly decreaseswith the decreaseof the mobility factor U / ,@ .

202
Bibliography
Garg, S. P. ; A. K. Agrawal and P. R Sir@ (1971) ‘Bed load transportationin alluvial
channels’.J. Hydr. Div. Proc. ASCE, Vol. 97, N5, p. 653-664.
Grishanin,K. V. (1979) Oynumic of alluvialflows, 2nd edition, 3 11 p. (in Russian).
Guy, I. P. ; D. B. Simons and E. V. Richardson(1967) ‘Summary of alluvial channel
data from flume experiments’.Geol. Survey. Prof. Paper 462-l.
Kennedy, R. G. (1894-95) ‘The prevention of silting in irrigation canals’. Proc. Inst.
Civ. Engrs. Vol. 69, p. 281-290.
Lacey, G. (1930) ‘Stable channelsin alluvium’. Proc. Inst. Civ. Engrs., Vol. 229, p.
259-292.
Laursen,E. M. (1959) ‘The total sedimentload of streams’.J. Hydr. Div. Proc ASCE,
Vol. 84, p. l-36.
Lighthill M. J. and G. B. Whitham (1955) ‘On kinematic waves’. Proc. Roy. Sot. L.,
Vol. A229, N 1178, p. 281-316.
Los, M. T. (1973) ‘Uwagi o podobinstwie hydraulicznym korit rzecznych’.
GospodarkaWodna, N1(308), p. 8-l 1.
Smith, K. V. H. (1970) ‘Similarity in unlined irrigation canal systems’.J. Hydr. Div.
Proc. ASCE, Vol. 96, N 1, p. 13-28.
Vanoni, V. A. and Li San Hwang (1967) ‘Relation between bed forms and friction in
streams’.J. Hydr. Div. Proc. ASCE, Vol. 93, N 3, p. 121-144.
Zamarin, E. A. (195 1) Sediment transportation capacity and permissible velocities of
. flow in canals, 2nd edition, 83 p. (in Russian).

Main symbols
B width of a stream
d median diameterof particles
i” function
Fy=c Froude number
gH
g gravity acceleration
H mean depth of flow
hnm maximum depth
I slope of water Surface
I/ friction slope
k coefficient
L. length of a river reach
1 longitudinal co-ordinate
A4 constantof a quasiuniformflow
N arbitrary function
n roughnesscoefficient
Q
R
water discharge
hydraulic radius
t time
u mean velocity
water level

203
a angleof bank slope
P density of water
Ps density of solids
x length of wetted perimeter
co cross-sectionarea

204
On the fractal sinuosity of rivers
V.I. Nikora, D.M. Hicks, G.M. Smart
National Institute of Water and Atmospheric Research(NINA)
Christchurch,New Zealand

Abstract
On the basis of some simple considerationsand relationshipsfor river bends, the fractal
dimension d of the single-channelpattern is determined for the range of scaleswhich
correspondto meandering(d = 1.2- 1.3), The influence of alternation of meanderingand
non-meanderingchannel sectionsupon the estimation of d by Richardson’smethod is
shown. Fractal properties of multi-thread (braided) channel patterns are investigated.
Three degreesof freedom of the channel pattern, which the river can use for regulating
the energ losses,are distinguished:(1) channel meandering or braiding, (2) alternation
of meandering (braiding) and non-meandering (non-braiding) segments,and (3) valley
wandering. It is shown that the total sinuosity of both single-thread and multi-thread
channelscan be describedby the samegeneralrelationshipwhich takes into account their
fractal properties.

1. Introduction
Various approachesare used to quantitatively describe the planform of river channels
(Kondratiev et al., 1982; Richards, 1982). Most existing indices were developed for
special casesof channel pattern (meanders, for instance) and so they do not have a
universal character. At the same time, it is often useful to have some universal indices
which can be applied for both single-threadand multi-thread channels.An exampleis the
total sinuosity which was first defined by Hong and Davies (1979) as length of channel
segments : channel belt length. Richards (1982) provided a similar definition whereby
total sinuosity = total active channel length / valley length Robertson-Rintoul and
Richards (1993) successfullyused the latter as a quantitative characteristic of channel
geometry and establishedquite a close connection between total sinuosity, the stream
power index and bottom particle size, considering meandering and braided rivers
together on the same graph. An alternative way to describe single-thread and multi-
thread channelsusing the samequantitative indices is the fractal approach.
Mandelbrot (1977) first investigated the fractal structure of single-thread river
channels, To evaluate their fiactal dimension d he used Hack’s (1957) relation
0.6
L, CCA rewritten as L1ld
r oc Ao,5 (L, is the river length, A is the catchment area).
This approachwas the basisof severallater works dealing with fractal geometry of rivers
(Tarboton et al., 1988; Hjelmfelt, 1988; Rosso et al., 1991; Rodriguez-Iturbe et al,

205
1992; Liu, 1992). Evidently, the relationship ~!‘d CCAo.5 holds only if the river pattern is
self-similaron all scalesup to that of the river’s overall length (Nikora et al., 1993). On
the other hand, Nikora (1988, 1991), Snow (1989), Nikora et al. (1993), Beauvaisand
Montgomery (1994) showed that the real channel pattern of a single-thread river is
charcterizedby self-similarityonly on small scaleswhich correspondto the scalesof river
meanders.For the self-similarregion the valuesof the dimension d , given in the above-
mentionedworks, fall within the range of 1.0-l. 33, On larger scalesthe channelpattern
is generallyself-affine,with Hurst’s exponent H = 0.5 (Nikora et al., 1993).
Unlike single-thread river channels, braided (or multi-thread) channels have
undergone considerablyless investigationfrom the fractal point of view (Nikora, 1988,
1991; Nikora et al., 1994, 1995).
In this paper we try to show a connection between the fractal dimension and
traditional geometricalparametersof meanderingrivers, we look at fractal properties of
braided rivers, and we determinethe main componentsof river sinuosity.

2. Single-thread channels
2.1 Meandering and self-similarity

Numerous theoretical and field investigationsof channelmeanderingshow that the shape

of river bends and the laws of their developmentare almost universal and practically
independentof the river scale (Richards, 1982; Kondratiev et al., 1982; Ichim et al.,
1989). Moreover, this universality extendsto other meander-likeprocesseslike ingrown
meanders and oceanic and atmospheric currents (Leopold and Wolman, 1960;
Zamyshljaevand Snishchenko,1986; Mienert et al., 1993). Of many facts concerning
channelmeandering,we have selectedonly the following relationshipswhich reflect the
universality(similarity) of their shape
A
- = 5.0 (2-l)
w

s
- = 1.4- 1.6
A

andthe relationof the bendstepA (Fig. 2.1) with the waterdischargeQ

AScQP v-3)

Here w denotesthe flow width, s is the bend length (Fig. 2. l), and the exponent fl
dependson the definition of Q (for instance, p = 0.5 when Q is the bankfull discharge-
Richards, 1982; Zamyshljaevand Snishchenko,1986).
For the channel pattern to be self-similar it is necessaryto have not only
geometricalsimilarity of various size bends,which is expressedby Eqs. (2-l) and (2-2),
but also a definite spectrum of bends. We believe that in natural conditions such a
spectrum exists and is due to the following circumstances.Firstly, as a result of natural
fluctuations of the water dischargeand in conformity with Eq. (2-3), bends of various

206
sizesare formed, each of them correspondingto a definite range of the water discharge
(Sinnock and Rao, 1984; Ichim et al., 1989). Thus, we have a quasi-continuous,
overlapping bend spectrum in the range [ A(Q,,,) - iI 1, where Qmin and Q-
denote the characteristicminimum and maximum water discharges.Secondly, due to a
reduction in streamflowsduring the last climatic epoch, the bend spectrumin the channel
pattern may be extendedup to the A(Qp) scale,where Qp denotesthe water discharge
of the palaeo-river. In other words, modern river bends may be superimposedon the
macrobendsformed by the palaeo-stream(Ichim et al., 1989). Thirdly, even at constant
water discharge the bend spectrum is not a delta-function but comprises a band of
wavelengths. This conclusion stems in part from research in which the meandering
process is considered from positions of hydrodynamic stability (Zamyshljaev and
Snishchenko,1986), but it also relatesto random componentsdue to the heterogeneous
structure of the valley bottom. In this connection it should be emphasizedthat the
numerical ratios in Eqs. (2-l) and (2-2) characterizethe most probable relationships
between the parametersof bends found in nature. These ratios can vary due to both
random factors and the stageof bend development(Kondratiev et al., 1982).

Fig. 2.1 Main parameters of a river bend ccording to Kondratiev et al. (1982)

The above considerationsexplain the existenceof a channel pattern self-similarity

region on a qualitative level. We now try to determinethe fractal dimension d for this
region. As a basis,we take the well-known Richardson relationship for a fractal curve
(Mandelbrot, 1977)
I-dLd
$=1;1 * (z-4)

where LV is the total length of a polygonal chain of line segments,each of length 17,that
approximatethe curve ( 7 can be thought of as the length of a measuringstick used to
estimatethe length of the curve), and L, is the straight-line distancebetween the two
end-points of the curve. Choosing the biggest bend step A(Qmx) to serve as L,, the
streamwidth W(Qmax) to serveas q, and the bend length S(Qmax) to serve as LV, we
rewrite Eq. (2-4) as

s zz W ’-d;ld e-5)

FromEq. (2-5) it follows

207
 s A d-l
y=(w) Q-6)

which connectsEqs. (2-l) and (2-2) and also makesit possibleto find d on the basisof
Eqs. (2-l) and (2-2). Although these considerationsare approximate,the estimateof d
so obtained (1.2- 1.3) is in good agreementwith values dR < 1.3 found previously by
Richardson’s method (Nikora, 1988, 1991; Snow, 1989; Nikora et al., 1993; and
Beauvaisand Montgomery, 1994).
The empirical fractal dimension dR, determined by Richardson’s method, may
dependnot only on variability of the ratios (2-l) - (2-3), but also on the channel pattern
intermittence.If an alternation of non-meanderingand meanderingsectionsis observed
in the plan view of a river, then on Richardson’sgraph InLV = f(nr 7) we will obtain a
(dR - 1) slope which. is lessthan (d - 1) . Figure 2.2 illustratesthis effect. The change of
dR due to changesin the proportion of non-meanderingsegmentsis shown in Figure
2.3. In this figure, dRo is dR at L,, / L, = 0, L,, denotes the length of non-
meanderingsegments,and L, is the total reach length on a straight line.

E
Y

-I
r
1 .o

0.8 !:.I
0.0 0.2 0.4 0.0 0.8 1.0
0.1 1.0 L I L
q, km "Ill t

Fig. 2.2 Influence of the channelpattern Fig. 2.3 Connection betweendR/dRO

intermittence on Richardson ‘sfractal and L,,,JLt , Botna River,
dimension.Botna River, Moldova Moldova

Considering the above results, channel sinuosity conditioned by meandering is

determinedby the relationship

--3c= v-7)
A llX3X

where sw denotesthe low-flow channellength around the largestbend measuredby the

4=Wdn ‘measuringstick’, R.max is the step of the largest bend, and wmin is the low-
flow channelwidth.

208
2.2 Sinuosity of channelson large scales
As alreadymentioned,on scalesexceedingR,, the channelpattern is characterizedby
self-affine behavior with Hurst’s exponent H = 0.5. As a first approximation this pattern
can be consideredas an analog of a Brownian motion trace (Feder, 1988) assumingthat
the transversalfluctuationsof the channelare purely random. One can seethat when the
length of the channelis measuredwith the q a ;1,, ‘stick’its sinuosityis determinedby
the following relationship

Lv wnax =-4nax
-= V-8)
Lo CAz, Ax
where Lo denotesthe distancebetween the source and mouth of the river on a straight
line, 2, is the averageprojection of ;1,, on line Lo, L, is the channel length, and
&,, is the ‘stick’ by which L, is measured.We believe that &, I /2, should be
mainly determinedby the regional drainagebasin slope. A decreasein this slope should
lead to an increase of A,, / il, Also, we consider that L, / Lo, as a first
approximation,can be identified as the river valley sinuositycoefficient K,

2.3 Total sinuosity

The above featuresof channelpattern structure suggestthat in order to regulate energy
losses a single-threadriver uses three degreesof freedom in its channel pattern: (1)
channelmeandering,(2) alternationof meanderingsegmentswith non-meanderingones,
and (3) random valley wandering. In view of these considerations,the total sinuosity of
the channel K, can be expressedas

K, =kCy --sw 4nax a y(-hnax d-l &a,

(2-9)
W )n
L() 4nax a* min X

where L, denotes the river length measuredwith the 11~wmin ‘stick’ and y is a
coefficient characterizingthe alternationof meanderingand non-meanderingsections( y
I 1 ). Assuming R,, of W. , where w. is the river valley width (Nikora, 1988 and
1991), and further assuming K,, E R,, / jl, , we can pass from .Eq. (2-9) to the
following relationship

K, = k a y(-
wO )d-lK (2-10)
v
LO Wmin

Obviously, when flooding occurs small scale meandersare covered by water,

channelwidth w increases,and, as it follows from (2-lo), the total sinuositydecreases.
The explicit determination of the three components of total sinuosity (y ,
SW l&u,, and A,, / IX ) would require an elaboratealgorithm. However, fi-om maps

209
one can easily determine the sinuosity K, of the valley and the channel sinuosity Km
within the valley. Thus as a first approximation, taking into account Eqs. (2-9) and (2-
10), we can write

sW
K, =K,K,, K, cc-4lUX Km OcYp
ax ’ a WX

We have studiedthe connectionsof K, , K,, , and Km with river slope I, (F ),

r

valley slope I, ( $&indd rainagebasin slope 1d (E ) for 44 rivers of the Dniester-

LO
Pruth interfluve (A is the differencein elevation betweenthe river source and mouth).
For theserivers, basin area ranged from 104 to 7760 km2, I, was 0.0007-0.007, I,, was
0.0008-0.009,and I, was 0.0012-0.012. The following scaling relationshipshave been
establishedby leastsquaresmethods:

(2-12)

z 0,g31,-o.059 2; 0.791,-".068 = 0.951;"'039 (2-13)

Kv

Km w 0.981;0'025 e 0.931;"'034 (2-14)

The correlation coefficientsof linearizedforms of relationships(2-12)-(2-14) are

relatively low (range 0.4-0.5) due to the influence of factors other than slope on the
formation of river sinuosity.Nevertheless,the above relationshipsare indicative of(i) the
commensurabilityof the contributions of the three componentsof total channelsinuosity
and (ii) the existenceof a weak inverse non-linear connection between the sinuosity of
rivers and their slopes.Evidently, relationships(2-12)-(2-14) have a regional character,
and they may have another form for regions with different slopes, catchment areas,
geology, climate, and so on.

3. Multi-thread or braided channels

3.1 Fractal properties

Even a simple visual analysisof braided river topography revealsa developedhierarchy

of islands and anabranchescovering a wide range of scales and pointing to quite a
probablefractality of their structure. To investigatethis quantitatively,we have used two
geometrical images of braided channels: (i) the coastline of islands (CI) and (ii) the
braided network of channel anabranchaxial lines (CAAL). We have investigatedthese
from maps and aerial photographs of 14 multi-thread, close to rectilinear,
morphologically homogeneousreachesof New Zealand rivers (Table 3.1). According to
the classification of Kondratiev et al. (1982), all of these reaches are classedas the
“channelbraiding”type. Also, in order to investigatethe influence of the degreeflooding
on the fiactal properties of a braided channel, we have used Mosley’s (1982) aerial

210

-
photographs of one of these rivers, the Ohau River, taken over a range of water
discharge.The main resultsof our analysis,previously summarizedin Nikora et al. (1994
and 1995), follow.
Table 3.1 Fractal dimensionsof somebraidedNew Zealandrivers

River reach 4 ds IV wo. (W L (W

Wairau-1 1.49 1.51 0.00574 0.86 31.45
Wairau-2 1.51 1.52 0.00911 0.44 19.90
Wilberforce-1 1.57 1.59 0.01042 1.30 8.00
Wilberforce-2 1.69 1.62 0.00952 1.67 10.00
Waitaki- 1 1.68 1.70 0.00308 1.53 11.00
Waitaki-2 1.59 1.61 0.00357 1.52 9.05
Waitaki-3 1.58 1.57 0.00377 1.30 22.95
Ftangitata-1 1.56 1.53 0.00276 1.15 9.90
Rangitata-2 1.69 1.62 0.00364 1.48 10.50
GodIey-1 1.63 1.62 0.01029 2.15 17.15
Godley-2 1.60 1.60 0.00465 1.87 9.80
Rakaia 1.65 1.63 0.00421 2.02 27.48
Waimakariri 1.54 1.53 0.00483 1.58 15.00
Ohau 1.70 1.71 0.00650 0.44 1.65
Averaged dy1.61*0.07 ds=1.60~0.06

Working first with the island coastline images,linear relationshipsbetween island

length ( LI ) and width ( wr ), of the form wI = aLI + C, were found for all of the
investigatedreaches.The coefficients a and c ranged from 0.18 to 0.24 and from 3.9 to
49 m, respectively,while the squaresof the correlation coefficientsranged from 0.62 to
0.85. The non-zero c valuesare indicative of the measurementerror. The linearity of the
relation n: = f(~~)) testifies to the self-similarity of the braided channel patterns. The
same conclusion also follows from analysisof the PV~and Lo distributions, which are
both of a power type and whose scalingexponentsproved to be very similar.
The establishmentof self-similarity provides grounds for using the box-counting
method for analyzingthe fractal propertiesof thesebraided channels.The box-counting
method seeksto define a relationshipof the form

-d
Nar (3-l)

where N is the number of cells within a rectangular net containing the investigated
object, r is the cell size, and d is the fractal dimension of the object, Comparison of
braided channel fractal properties with results for meandering rivers (Nikora, 1988,
1991; Nikora et al., 1993) requiresthat Eqn. (3-l) be generalizedto
Nr
- = f(L) (3-z)
Lv Wo

where L, is the length of the investigatedreach and w. is the total bed width of the
braided channel. The Nr /L, ratio is analogousto channel sinuosity in the CAAL case
and to island coastline sinuosity in the CI case. The graphs of Nr /L, vs. r / W. for
CAAL and CI are given in Figures 3.1a and 3.1b, respectively.In all casesthere is a well

211
defined scaling, at least in the range 0.07 JQ - 1.5 w. . No substantialrelationshipsthat
link the individual dimensionsdS (for CAAL) and dI (for CI) with w. or valley bottom
slope I, have been revealed.Thus, as a first approximation,the variations shown by dS
and dr may be regarded as random and their average values ( ds =1.6wO.O6 and
dr =I .61*0.07, Table 3.1) may be assumedrepresentativefor braidedreaches.

1.0
r/W0

Fig. 3. I Fractal scaling for C&IL (a) and CI (b), New Zealand braided rivers

Our analysisof the Ohau River channelpatternsat various water dischargesshowedthat

their fractal properties were independentof water discharge (Figures 3.2a and 3.2b).
Physically,this can be explainedby the compensatingeffects of the appearanceof new
anabrancheson large islands and the flooding of small islands as the water discharge
increases.

r---l
Wakr d*oharge,
3
m I8
10 :

+ 28.5
58.7
3 .
+

z .

Fig. 3.2 Graphs Nr IL,= f (r/ WJ for the Ohau River, (a) CAAL and (b) Cl

An additional comment follows from our analysisof the distributions of water discharge,
width, depth and average velocity in individual anabranchesfrom several other New
Zealand braided reaches(using data from Hicks, 1979; Mosley, 1983; and Glova and
Duncan, 1985). Power-type distributions were revealed for all of these parameters,
indicating that braided channelsshow no characteristicscaleswithin a wide range for
both morphologic and hydraulic parameters.This leads us to supposethat braiding in
rivers is a result of spatialself-organizedcriticality (Bak, Tang, and Wiesenfeld, 1988).

212
3.2 Total sinuosity
The graphsshown in Figure 3.1 and the notion of internal lint and external leXt fractality
scales (Nikora, 1988 and 1991) suggest the following simple model to describe the
tiactal propertiesof braided channels:

_L _ ‘ext d-l
- (I.p (3-3)
LV int

where L is the total length of all anabranchesin the CAAL caseand the total length of
island coastlinesin the CI case. As a first approximation, we assume lint ocwmin and
1ext a, w. (where wmin is the width either of the smallestchannelor the smallestisland).
Then, by analogywith Eqs. (2-9) and (2-lo), we can write the following relationshipfor
the total sinuosity K, of braided channels

L
Kr=-cc--ccY(- L Lv wO )d-lK
V (3-4)
LO Lv Lo “min

Here y (I 1) is a coefficient characterizingthe alternation of braiding and non-braiding

reaches.Thus, as in the caseof single-threadrivers, the total sinuosity of braided rivers
also contains three components:a braiding component, (w. / wmin Jo-’; an alternation
component, y , and a valley wandering component, K, . From this we hypothesizethat
along its path, a multi-thread river can adopt various combinationsof these components
in order to regulateits energylosses.

4. Conclusion
On the basis of simple considerationsand some relationshipsfor river bends, we have
obtained the fractal self-similarity dimension d for single-threadmeandering channels
and have shown the influence of channel pattern intermittence upon estimatesof the
Richardson dimension dR. In addition, we have obtained scaling relationshipswhich
connectthe sinuosityof single-threadrivers with their slopes.
We have shown that multi-thread channel patterns also demonstrateself-similar
fi-actal behavior at scalesless than Ml river width. The total sinuosity of both single-
thread and multi-thread channelscan be describedby the samegeneral relationship(Eqs.
2-10 and 3-4).
We have also distinguishedthree degrees of freedom for the channel pattern
which a river can use for regulating energy losses.The first of these is meandering(or
braiding), the second is alternate meandering (braiding) and non-meandering (non-
braiding), and the third is valley wandering, The first two degreesof freedom occur
within the scaleof river valley width, while the third occurs at much larger scales.
Comparisonof fiactal dimension and sinuosity data for single-threadand multi-
thread channels (Table 4.1) suggeststhat the fractal approach can be used as an
alternativequantitativetechniquefor river channeldescription.

213
Table 1.1 Fractal dimensionsand sinuosity coefficients for river channels

Channel type d K
single-thread rectilinear 1.0 1.0
single-thread,meandering 1.1-1.3 1.5-3.0
multi-thread or braided 1.5-1.7 5.0-15.0

Acknowledgments

The authors are deeply grateful to V.B. Sapozhnikov, T.R.H. Davies, M.P. Mosley, and
M.J. Duncan for discussionsand information, as well as to N.V. Kushnir, N.A. Arnaut,
and A.N. Sukhodolov for their assistancein topographic map analysis.

5. Bibliography
Bak, P.; C. Tang and K Wiesenfeld (1988) ‘Self-organized criticality’. Physical Review
A, Vol. 38, No. 1, p. 364-374.
Beauvais, A. and D.R. Montgomery (1994) ‘Planform scaling properties of rivers and
channel networks’. Supplementto EOS, AGU, November 1, p. 303.
Feder, J. (1988) Fractals, Plenum, New York.
Glova, G. J. and and M. J. Duncan (1985) ‘Potential effects of reduced flows on fish
habitats in a large braided river, NZ’. Trans. of the American Fisheries Sot., Vol.
114, p. 165-181.
Hack, J.T. (1957) ‘Studies of longitudinal stream profiles in Virginia and Maryland’.
U.S. Geol. Surv. Prof. Pap., 294-B.
Hicks, D.M. (1979) ‘The behavior of the Rakaia River at low flow: A case of a study of
a braided river’ Report No. WS 89, Ministry of Works and Development,
Christchurch, New Zealand, 47~.
Hjelmfelt, J.A.T. (1988) ‘Fractals and the river-length catchment-area ratio’. Water
Resour. Bull., Vol. 24, No. 2, p. 455-459.
Ichim, I.; D.Batuca, M. Radoane and D. Duma (1989) Morphology and Lfynamics of
river channels (in Romanian), Tehnica, Bucharest.
Kondratiev, N.E.; I.V.Popov and B.F. Snishchenko (1982) Foundations of
hydromorphological theory of river bed process (in Russian), Hydrometeoizdat,
Leningrad.
Le Ba Hong and T. R. Davies (1979) ‘A study of stream braiding’. Geological Sot. of
America Bull,, Vol. 82, p, 1251-1266.
Leopold, L.B. and M.G.Wolman (1960) ‘River meanders’. Geol. Sot. Am. Bull., Vol.
71, p, 769-794.
Liu, T. (1992) ‘Fractal structure and properties of stream networks’. Water Resour. Res.,
Vol. 28, No. 11, p. 2981-2989.
Mandelbrot, BB (1977) Fractals; Form, chance and dimension, W.H.Freeman, San
Francisco.
Mienert, J.; N.HKenyon, J.Thiede and F.J. Hollender (1993) ‘Polar continental margins:
Studiesof East Greenland’. EOS, Transactions,AGU, Vol. 74, No. 20.
Mosley, M.P. (1982) ‘Analysis of the effect of changing discharge on channel
morphology and instream use in a braided river, Ohau River, New Zealand’. Water
Resour. Res., Vol. 18, No. 4, p. 800-8 12.

214
Mosley, M. P. (1983) ‘Response of braided rivers to changing discharge’, J. of
Hydrology (NZ), Vol. 22, No. 1, p. 18-67.
Nikora, V.I. (1988) Fractalproperties of some hydrological objects (in Russian),Acad.
of Sci. of Moldova, Kishinev, 44~.
Nikora, V.I. (1991) ‘Fractal structuresof river plan forms’, Water Resour.Res., Vol. 27,
No. 6, p, 1327-1333.
Nikora, V.I.; V.B. Sapozhnikovand D. A.Noever (1993) ‘Fractal geometry of individual
river channelsand its computer simulation’, Water Resour. Res., Vol. 29, No. 10,
p. 3561-3568.
Nikora V.I.; D. M. Hicks, G. M. Smart and D. A. Noever (1984) ‘Fractal model of
braided channels’(in Russian). In: Dynamics and thermics of rivers, reservoirs,
internal and marginal seas. Russia’sAcad. of Sci., Moscow, p, 122-125.
Nikora V.I., Hicks D.M., Smart G.M., and Noever D.A (1995) ‘Some fractal properties
of braided rivers’. In: Book qf Abstracts of the 2nd Int. Symp. “Fractals and
Dynamic Systems in Geoscience”, J. W. Goethe Universitat, Frankfurt am Main,
Germany,p, 52-53,
Richards, K. (1982) Rivers: form and process in alluvial channels, Methuen, New
York.
Robertson-Rintoul, M. S.E. and K.S Richards (1993) ‘Braided channel pattern and
palaehydrologyusing an index of total sinuosity’. In: Braided Rivers, Publishedby
the Geological Society,London, p, 113-l 18.
Rodriguez-Iturbe, I.; E.J. Ijjasz-Vasquez,R.L. Bras and D.G. Tarboton (1992) ‘Power
law distributionsof dischargemassand energy in river basins’,Water Resour.Res.,
Vol. 28, No. 4, p. 1089-1093.
Rosso,R.; B. Bacchi and P. La Barbera (1991) ‘Fractal relation of mainstreamlength to
catchmentarea in river networks’, Water Resour.Res., Vol. 27, No. 3, p. 381-387.
Sinnock, S. and A.R. Rao (1984) ‘A heuristic method for measurement and
characterizationof river meander wavelength’, Water Resour. Res., Vol. 20, No.
10, p, 1443-1452.
Snow, R. S. (1989) ‘Fractal sinuosity of stream channels’,Pageoph.,Vol. 131, No. l/2,
p. 99-109.
Tarboton, D.G., R.L. Bras, and IRodriguez-Iturbe (1988) ‘The fractal nature of river
networks’, Water Resour.Res., Vol. 24, No. 8, p, 1317-l 322.
Zamyshljaev, V.I., and B.F. Snishchenko (1986) Meandering of stream channels:
Review (in Russian),VNIIGNI-MTSD, Obninsk, Russia.

215
 Channel processes and their role in river
ecosystems
by R.S. Chalov and A.M. Alabyan

1. Introduction

Contemporary ecologization of the Earth Science arise the ecological approach in

investigation of channel processes. These processes include the totality of phenomena
which are induced by an interaction between moving water and channel material. In
this paper we try to fix the place of channel processes in river ecosystem and to analyse
controls of channel development.
Biogenic processes Linked with the Life and activity of aquatic and subaquatic
plants and animals are traditionally considered as an essence of the river ecosystem.
However these processes develop on the background of fluvial system, formed as a
result of complicated interaction between moving water and underlying rocks. River
channel pattern concentrates and reflects all features of channel processes and can be
considered as an indicator of river state. From the viewpoint of channel morpholog the
river can be plain or mountain, sandy or gravely, meandering or branching, abounding
of shallows and bars or deep and calm. These features are responsible for development
of either type of vegetable and animal societies in the tiver and surrounding area.
As any natural system the system ‘
flow-channel’ has definite capacity to
compensate external impact including an influence of human activity. Two essential
types of the influence can be determined:
(i) Direct influence - changing of river morphoto& and rate of natural processes:
construction of bridges and water off-takes, dredging, sand mining, artificial
cutoEs.
(ii) Indirect influence - changing of main factors controlling channel development: flow
regime, sediment catchment and transport, etc. Consequences of indirect influence
can be very significant, for example, in river sections downstream large reservoirs.
The main goal of ecosystem analysis of river channel pattern is to determine the
natural reserve of the channel allowing to resist technical influence which can induce
irreversible changes in river morphology and regime. Research of threshold or
transitional values for channel pattern and their controls seems to be most important in
this direction.

216
2. River ecosystem

2.1 Structure
The river ecosystem is a combination of channel and floodplain ecological subsystems,
in great extent influencing to each other (Fig. 2.1). This interaction can be displayed
both directly during floods and trough the permanent process of bank erosion and
accretion causing a horizontal channel shifting and forming of yang floodplain.

RIVER ECOSYSTEM

system: t_
’

Channel + Floodplain
ecosystem L ecosystem

system-forming: Biogenic ---

L
processes.. processes -:,

indicators:

factors:

/ ^J
phisical and chemical
I I
i

Fig. 2.1 The structure of river ecosystem

217
 The main system-forming processes are biogenic processes connected with plant
and animal activity, and channel processes creating an environment for development of
biota. The river landscape can be considered as an indicator of the river state and its
various components reflects one or another appearance of system-forming processes.
The fluvial relief and channel morphology are stipulated mainly by channel processes
but a role of soil, vegetation and other biogenic components sometimes can be rather
significant.
According to their origin and behaviour the natural factors influencing river
ecosystem can be subdivided into three main groups:
(i) Hydrological and climatic: solar radiation; temperature and humidity; water
regime; sediment regime; thermic and ice regime of the river; underground water
regime. These factors determine both channel and biogenic processes.
(ii) Geological and geomorphological: lithology and mineral composition of the basin;
size of alluvial particles; rate of material supply from valley boards. This factors are
also important for both groups of system-forming processes.
(iii) Physical and chemical: chemical and isotope composition of water, air and soil.
This factors do not play visible role in channel development, but can be of vital
importance for biotic components of river ecosystem.

2.2 Human impact and channel processes

Channel processes can indicate actual or possible disturbances in river ecosystem and
appearance of ecological tension. According to an extent of intervention manmade
influence on river system (both direct and indirect) can have various sequences:
(9 immediate or quick restoration of the initial natural conditions (episodic dredging
in sand- bed rivers);
(ii) conservation of the river as a natural system but changing of channel pattern and
floodplain landscape;
(iii) aggradation and disappearance of the river.
The latter extreme case of negative manmade phenomena is common for smalI
rivers in steppe zone of Russia died as a result of excessive ploughing in their basins
and channel fulfilling by sediments.
For large and middle rivers main alterations are associated mainly with great
hydrotechnic construction and alluvium mining from channel. Floodplain ecosystems
are particularly vulnerable for changes in water level regime. Large reservoirs intercept
sediments and induce bed erosion in downstream river sections. Active mining causes
appearance of extensive artificial pools leading to redistribution of sediment transport
along the river and consequent channel incision. Range of such lowering of water level
can exceed 2-3 m. As a result of this process a diapason of water discharges passed by
channel without flooding increases, and in turn significant alteration in floodplain soil
and vegetation can take place. Such phenomena were registered in the Tom, the middle
Ob, the Chulym, the Katun, the Irtysh and other large and middle rivers in Russia. This
changes usually are accompanied by changes in channel pattern causing further
alteration of flow dynamics and channel ecosystem transformation.

218
3. Channel processes and channel pattern

3.1 Terminology and classification

The channel processes can be subdivided into two main classes: processes in
mountain rivers and processes in plain tivers. Sharp difference in cinematic properties of
calm (Fr< 1) and rugged (Fr> 1) flow is responsible for different mechanism of flow-bed
interaction in a river. Plain rivers are characterised by calm flow forming channel bars
and dunes with gradual upper slope and steep downstream side. Calm flow of plain river
bypasses an obstacle by fluent trajectory with backwater zone before it. Blustery flow of
mountain river cause forming of alluvial piles similar to antidunes and flow splashes
before obstacles. Resulting system of standing waves on the surface of mountain river
reflects very specific behaviour of the flow.
From the viewpoint of limiting conditions of horizontal channel shifting rivers
can be divided into incised rivers and ones with wide floodplain (wide-floodplain rivers).
In. the development of incised channel vertical deformations play the general role,
morphological features of incised river are determinate mainly by properties of
geological structure and history of the valley. As to rivers with wide floodplain, in this
case the main factor of channel development is the flow, its power and structure.
According to morphodynamic features channels of both classes can be
traditionally determined as straight, sinuous or branched. However, it must be noticed
that straight, sinuous and branching outlines can display in different structural levels
of fluvial relief, three main of them are valley bottom, flood channel, and low water
channel (Fig. 3.1). Concerning channel configuration in macroscale on the background
of valley bottom rivers can be defined as either relatively straight and single-thread, or
forming macromeanders (macrobends). Channel dividing on this level performs
anabranching.
Every anabranch as well as every segment of macromeander may be classified on
three main floodchannel patterns: straight, meandering and split (exactly split but not
braided because the latter term traditionally used for definition of island dividing
(Ferguson, 1984)). The term braided channel is usually used relatively to straight
channels with medial bars, meandering channels with braid bars are known as
wundering. Straight floodchannels with alternated bars are named by Carson ( 1984)
meandetial. Anastomosis is interpreted by Knighton and Nanson (1993) as branchings
on the all levels. So, every known channel pattern may be described as a combination of
three main configurations on three main levels. Of course, like every natural system
there is no any distinct boundaries between neither outlines nor levels, and a lot of
transitional forms may exist.

219
 structural plan outline limiting
level conditions
straight sinuous branched

valley
-----*m?LLLszi
bottom
single-thread macmmeander anabranchlng wide floodplaln

flood / a e
channel
straight meandenng Spilt adapted channel

-0 ‘-4,: s y--Tyq
low water

i
channel
nffle-pool sequense

Typijication of channel patterns

alternate bars medial bars incised channel
!
Fig. 3.1

The influence of limiting factor also may be considered relatively to different

structural levels as sufficient conditions for an appearance of consequent channel form.
In wide floodplains any form up to macromeander can develop. Adapted channel forms
when valley width is approximately equal to the width of meandering belt. Incised rivers
usually have narrow valleys where only bars and undeveloped alluvial forms may.

3.2 Reasons of patterns diversity

Mathematical models of different kind offers similar role of the main controls to channel
forming process. The most popular simulation methods includes investigation of
equations for flow and sediment transport in straight channel with low-amplitude
periodic perturbations of its geometry or hydraulic conditions (Parker, 1976; Fredsoe,
1978). According to these models flow in straight channels is not stable, deep narrow
channels are more unstable to a meander-like perturbations and wide shallow channels
- to braid-like perturbations.
Another path in simulation lies in analysis of flow and sediment transport
equations in complex with some variation hydrodynamic principle (Chang, 1979;
Bettess and White, 1983). It was conclided that for distinct water discharge the value of
equilibrium slope can be calculated. If valley slope is equal this value the channel is
straight and single-thread, if the valley slope is more the river can accommodate this
discrepancy by meandering, and if the valley slope is too much the braiding develops to
decrease the water discharge (the less the discharge the more the equilibrium gradient).
A number of laboratory experiments (Andreev and Yaroslavtsev, 1958; Schumm and
Khan, 1972; Bettess and White, 1983) have shown an agreement with such scheme. So,
the excessive flow power arises meandering and then branching.
In natural rivers the excessive stream power can be spent on the lateral or bed
erosion depending on the geological structure of the valley. If the bed erosion
dominates, incised channel is being formed. Wide floodplain is a result of lateral erosion
prevailing, in this case bank washing causes a loss of flow stability and either
meandering develops or several dynamic axes form and branching appears.
One of the main method of natural channel pattern investigation is discharge-
gradient chart (@-diagram) analysis proposed by Lane (1957) and Leopold and Wolman

220
(1957). The physical essence of this chart is that the higher and further right a river
plots the higher stream power it has (because high value both of discharge and slope
cause heightened stream power). Practically all researches of such kind have shown
that the meandering channels plot lower than multiple ones. At the same time the
discriminant function of meandering and branched channels alters in vast range. One of
the main reasons of this lies in differences in the operational definitions both of the
slope and discharge and channel patterns. Romashin (1968), Ferguson (1984) and
Carson (1984) were right when they contended that exactly the valley slope (but not the
channel gradient) and flood-frequency definition of water discharge (not bankfull or
mean annual) must be used to secure an independence of analysed factors.
The idea of magnitude-frequency analyses of water discharge spectrum for
effective discharge calculation was investigated by Makkaveev (1955), and Woiman and
Miller (1960). They postulated that in alluvial rivers the discharge which transports the
greatest sediment in the long-term plan is the effective (dominant, channel-forming)
discharge. Practically the value of effective discharge can be determined using the
special diagram (effective discharge curve) there values of product of sediment discharge
and frequency of occurrence are plotted against corresponding water discharge values.
The effective discharge has maximal value of ‘sediment-frequency product’ and
corresponds to the maximum of the curve. According to Makkaveev rivers of the Russian
Plain usually have one or several effective values, one of which is close to the mean
annual flood. This fact corroborates the possibility to use mean (or median) annual flood
flow as the first approximation of the effective discharge when constructing Qs-
diagrams.

3.3 Transitions
Carson (1984) and Ferguson (1984) interpreted gently inclined discharge-slope
discriminant lines in diagrams of Leopold and Wolman (1957) by operating with specific
stream power (power per unit of flow length and per unit width) pgQs/w measured in
W/m2. According to them the meandering-braiding threshold corresponds to specific
flow power of 30-50 W/m2.
Romashin (1968) constructed a Qs-diagram using valley slope and median flood
discharge for free developing channels (Fig. 3.2). His results coincided with former in
general, but an incline of discriminant line was more than twice stepper, Analysing this
diagram Antropovskiy (1972) suggested to consider stream power per unit of flow length
N=pgQs, measured in W/m.
After re-examination of these data in terms of modern typification of channel
patterns a field of the diagram can be divided into three parts:
N area of meandering N<4000 W/m;
(ii) area of both meandering and branching 4000<N< 15000 W/m;
(iii) area of branching N> 15000 W/m.
The same steep slope of discriminant lines was found by Alabyan (1992) on the
diagram considering different bar types in branched sandy channel of the Ob river (Fig.
3.3). A tendency is the same: medial bars plot in upper part of diagram, alternate bars
locate lelow them.

221
 meandering channels
‘incomplete meandering’ (cut meanders)
split channels
anabranching

Q, m3/s
Fig. 3.2 Q-s diagram of Romashm (1968)

(T, alternate bars

0 alternate bars and medial bars

‘3 \ 0 medial bars

0.02 1 : ,
100 200 500 1000 2000 4000
Q, m3/s
Fig. 3.3 Q-s diagram for bars

Comparing of transitional values for different structural levels shows a relation

between flow power and channel morphology (Table 1).

TABLE 1 Flow power and channel morphology

N, W/m Channel outline Bar type

> 15 000 branching
4 000 - 15000 meandering and branching medial bars
3000 - 4 000
1 000 - 3 000 meandering medial and alternate bars
< 1000 alternate bars and riffle-pool sequences

222
3.4 Effective discharge curves
Above mentioned effective discharge curve (EDC) is an integral characteristic of
hydrological and sediment regime of the river and reflects main features of channel
pattern. EDC may be used for channel pattern analysis in complex with transitional Q
values.
Every flow that transports sediment affects channel form. Product of sediment
discharge and its frequency may be considered as an index of channel forming rate.
Constructing of EDC allows to transmit from solitary discharge values (mean, extreme,
bankfull, etc.) to continuous discharge spectrum.
To compare EDC features with typical and transitional Q values several
horizontal lines can be drawn:
(iI the bankfu21 line corresponds to the bankfull discharge (solid line on Fig. 3.3);
(ii) the meandering-branching transition line corresponds to N=4000 W/m (long
dotted line);
(iii) the side bar - midbar transition line corresponds to N=lOOO W/m (dotted line).

a b C

cl, m3t2 Q, m3ts

_---
----~
6000 3, 30000 1500

4000 j 20000 1000

,\

RP
10000

0
2 RP
500

0 1
RP

Fig. 3.3 Effective discharge curves: a - the upper Ob, b - the lower Ob, c - the Chulym.

Correlation of EDC elements and this three lines reflects the channel features. It
can be illustrated on following examples.
The upper Ob River has complicated split channel, midbars prevail in branches.
Almost entire EDC lies below the bankfull line (Fig. 3,3a), so forming activity
concentrates below floodplain tilling level. Both high-water and low-water part of EDS is
located above the side bar - midbar threshold line, so midchannel bars may be formed
and maintained during every hydrological season. High-water part of EDC is situated
above the meandering-branching threshold line, so stream power during a flood is
enough to form splits.
The Ob River in lower reach is characterised by anabranching; branches are
meandering with alternating bars. The upper maximum of EDC is well encountered and
located above the meandering-branching threshold line which in tern plots above the
bankfull line (Fig. 3.3b). So the power suitable to cause channel dividing occurs only
when the floodplain is filled. Similar EDC may be constructed for any anabranch for its
further examination.
The Chulym River (a tributary of the Middle Ob) has meandering channel with
alternating side bars and individual midbars. The meandering-branching threshold line
is situated above EDCJFig. 3.3c), so the channel is not branched. Midbars may be
formed only during a flood without maintaining throughout a low water.

223
4. Conclusion

Any alteration in river basin induced both by natural courses and by human activity
reflects in channel regime. Channel transformation may happen revolutionary or
gradual reformation occur depending on channel stability and on extent of flow
changing.
Forecasting of revolutionary transformation is based on equilibrium pattern and
threshold (transition) investigation. Every main pattern outlines (straight, meandering,
branched) can be distinguished on every main structural level of fluvial relief (valley
bottom, flood channel, low-water channel). Composition of straight, meandering and
branching features on various levels defines the channel pattern as a whole. Natural
data analyses in a form of gradient-discharge charts constructed for channel forms of
different order output a number of Q-s transition criteria, which reflect stream power
per unit flow length.
Using such criteria the main problem is to chose representative discharge value.
Effective discharge curves allow to transmit from solitary discharge value (mean,
extreme, bankfull) to continuous discharge spectrum. Morover EDC can reflect changes
of flow and sediment regimes. Complex EDC and Qs-plots analysis allows to forecast
channel transformation.

Acknowledgements

The investigation was supported by The Russian Foundation for Fundamental Research
(grants No. 95-05 14006 and 95-07- 1905 1).

224
Main symbols

d flow depth, m;

9 gravity acceleration, m/s2;

Fr=&gd Froud number;

N=mQs stream power per unit of flow length, W/m;

P frequency (of water and sediment discharge);

Q water discharge, m3/s;

R sediment discharge, kg/s;

s channel slope;

SO
valley slope;

u flow velocity, m/s;

W channel width, m;

P density of water, kg/m3.

Bibliography

ALABYAN, A. M. (1992) ‘Sediment and sediment transport along plain river’ (in Russian). Vestnik
Moskovskogo Universiteta, seria 5 Geografiya, No. 6, p. 22-28.
ANDREEV, 0. V. and I. A. YAROSLAVTSEV (1958) ‘Modelling of channel deformations (basics)’ (in
Russian). In: N. E. Kondrat’ev (ed.), Channelprocesses. Academy of Science Publ., Moscow, p.
173-181. I
ANTROPOVSKIY, V. I. (1972) ‘Criterial relations of channel process types’ (in Russian). Proc. State
Hydrological Inst., Vol. 190, p. 5-18.
BETTESS, R. and W. R. WHITE (1983) ‘Meandering and braiding of alluvial channels’. Proc. Inst. Civ.
Eng., Vol. 75, p. 525-538.
CARSON, M. A. (1984) ‘The meandering-braided river threshold: a reappraisal’. J. Hydrol., Vol. 73,
p. 315-334.
CHALOV, R. S. (1983) ‘Factors of channel processes and hierarchy of channel forms’ (in Russian).
Geomorfologiya, No. 2, p. 16-26.
CHANG, H. (1979) ‘Minimum stream energy and channel patterns’. J. Hydrol., Vol. 41, p. 303-327.
FERGUSON, R. I. (1984) ‘The threshold between meandering and braiding’. Channels and Channel
Control Structures, Vol. 6, p. 15-29.
FREDSOE, I. (1978) ‘Meandering and braiding of rivers’. J. Fluid Mech., Vol. 84, p. 609-624.
KNIGHTON, A. D. and G. C. NANSON (1993) ‘Anastomosis and continuum of channel pattern’, Earth
Surface Processes and Landforms, Vol. 18, p. 613-625.
LANE, E. A. (1957) ‘A study of the slope of channels formed by natural streams flowing in erodible
material’. U.S. Army Eng. Corps, Missouri River Div., Sediment Series, Report No. V.F.891,
Vol. 9, p. 106.
LEOPOLD, L. B. and M. G. WOLMAN (1957) ‘River channel patterns: braided, meandering and
straight’. U.S. Geol. Survey Prof. Paper, Vol. 282-B, 85 p.
MAKKAVEEV, N. I. (1955) A river channel and an erosion in its basin (in Russian). Academy of
Science Publ., Moscow, 343 p.
PARKER G., (1976) ‘On the case and characteristic scales of meandering and braiding in rivers’. J.
Fluid Mech., Vol. 76, p. 457-480.
ROMASHIN, V. V. (1968) ‘Types of channel processes in relation with determining factors’ (in
Russian). Proc. State Hydrological Inst., Vol. 155, p. 56-63.
SCHUMM, S. A. and H. R. KHAN (1972) ‘Experimental study of channel patterns’. Geol. Sot. Amer.
Bull., Vol. 83, p. 1755-1770.
WOLMAN, M. G. and J. P. MILLER (1960) ‘Magnitude and frequency of forces in geomorphic
processes’. J. Geol., Vol. 68, p. 54-74.

226
 Impact of gravel excavation
on channel processes in the Laba River,
North Caucasus
A. B. Shvidchenkoand Z. D. Kopaliani
State Hydrological Institute, St. Petersburg,Russia

Introduction
Excavation of alluvium from river channelsand floodplains for various economic needs
is one of the anthropogenicfactors producing an unfavourable effect on the river. At
present, intensive excavationsare made from rivers in many industrially developed and
developingcounties.
The Laba River in the North Caucasus is one of these rivers, where the
Mostovskoy crushing and grading plant has excavatedas much as 500,000-700,000 m3
of alluvium per year since 1991.
Brief information on the investigationsmade by the State Hydrological Institute
during 1992-1994 in the Laba River reach with gravel excavationis given in the present
paper. Theseinvestigationswere aimed at the study of channeland hydraulic regimes of
the Laba River, the regime of sedimenttransport under natural and project conditions,
an evaluation of the results of various types of gravel excavation out of the river bed, as
well as a search for scientifically validated and economically and technologically
applicable and ecologically responsible decisions for excavation in the Laba River
channel.

General
The Laba River is a left-bank tributary of the Kuban River (North Caucasus,Russia). It
is 351 km long, its drainage area is 12,500 km’. The study Piedmont river reach is
located near Mostovskoy settlement, 150 km far from the river head, and 13 km down
the site, where the river leaves the mountain terrain (Fig. 1). The Laba River
hydrological regime is characterizedby prolonged spring-summersnowmelt flooding,
rainfall floods all the year round, and low water levels during the winter months
(Fig. 2). According to observations at the Kaladzhinskaya gauging station 10 km
upstreamfrom the study reach, the mean long-term maximum flow of the snowmelt
flood is 411 m3/s, while the maximum recorded water dischargeis 901 m3/s. Mean
water level rise during the snowmelt flood is 1.4 m above the winter minimum water

227

__-~ - --___-_. - .._..----__-- ..-- -

Fig. 1 Aerial \Cew of the Laba River between Kaladzhinskaya and A fosto\skoy at minor flood of 432 m?‘s on May 17, 1987
 12345678910111212345678910111212345678910

Fig. 2 Hydrograph of the Laba River at Kaladzhinskaya for 1991-1993

I, II, III, IPJ, y VI - time periods for which bedload yield was measured

level, the maximum rise being 2.9 m . In general the maximum flow is observedduring
a few hours, in rare caseswithin 2-4 days. In some years the maximum flow of the
rainfall floods exceedsthe maximum flow of the snowmelt floods. Mean long-term
annualwater dischargeis 87.1 m3/s.
The Laba River in the study reach has a braided pattern, which is typical of
Piedmont river reaches.The total channel width is 300-500 m , with mean slope of
0.0048. Mean flow depth during floods is l-l.5 m , and flow velocity is up to 4.5 m /s.

Sediment yield
A peculiarity of rivers of mountain and Piedmontzone is a more distinct subdivisionof
sedimentsinto suspendedload and bedload,if comparedwith plain rivers (Kopaliani and
Romashin,1970). In fact, suspendedsedimentsdo not contributeto the channelprocess
of mountain rivers. The transportation of suspendedsediments is continuous at
velocities close to the flow velocities (which is its qualitative peculiarity), while the
quantity and the granulometic composition depend not only on the hydraulic flow
characteristicsbut also on changeableconditions of weathering and slope erosion over
the drainagebasin. According to measurementsat the Kaladzhinskayagauging station,
the annual suspendedload yield variesfrom 150,000t up to’l,SOO,OOO t and it is equal
to 638,000 t on average.Mean annual suspendedsedimentsconcentrationin water is
220 g/m3,the maximum concentrationwas 17,000 g/m3. Mean observedgranulometric
compositionof the suspendedsedimentsis given in Fig. 3.
Unlike suspendedsediments,the bedload is lessmovable and is transportedfrom
the source of bedload delivery to the river mouth over many years, and is subject to
substantialsorting and abrasion.The bedload motion is governedby the local hydraulic
characteristicsof the flow, it is very non-uniform and unsteadyand is mainly observed
during floods. The amount of bedload is an important characteristicfor estimating
admissiblevolumes,regime and techniquefor alluvium dredgingout of the river channel
and for estimatingthe river capacityto recover natural channeland hydraulic regimes,
Sincebedload yield in the Laba River is not observed,one of the main problem of field,
laboratory and theoretical studiesis to estimatethe bedloadyield and regime of bedload
motion.

229
 1 10
Grain diameter, mm

Fig. 3 Granulometric composition of sediments in the Laba River

1 - suspended sediments at Kaladzhinskaya
2 - bed material at Kaladzhinskaya
3 - channel alluvium at Mostovskoy
4 - surface layer at Mostovskoy
5 - bed material in the movable bed model and flume (recalculated for the prototype)

Field investigations
Field investigationsin the Laba River at the reach of the alluvium excavationswere
made during different phasesof flow regime including spring-summersnowmelt flood,
the period of rainfall floods and the winter low-water period. During field work on the
river reach of about 5 km long, water levels, water surface slopes,flow velocities and
depths were measured; a topographical survey of the study reach was made. To
determinethe granulometric composition of bed material, photographswere taken of
different parts of the channelduring the low-water period (26 photos altogether).As the
analysisof the photographsshows, the surfacelayer of the river bed is composedof
gravel and pebble with median diameterbetween 28 and 107 m m (58 m m on average).
Under the armour layer lo-15 cm thick a m ixture of gravel (75-80%) and sand (20-
25%) occurs. Besides, the results of sieving analysis were used for the material
extractedout of the channel;this sievingwas made at the SedimentLaboratory of the
Mostovskoy crushingand grading plant (13 1 samples40 kg eachwere taken during the
period 1991-1993). According to the data of the Mostovskoy plant, the median
diameter of the channel alluvium varies from 19 to 41 m m and averages26 m m .
Proceedingfrom the bed material samplesdata of the Kaladzhinskayagauging station,
the median diameterof bed materialvariesfrom 10 to 84 m m and is 34 m m on average.
The granulometriccompositionof bed material at Mostovskoyis in generalquite similar
to that at the Kaladzhinskayagaugingstation (Fig. 3).
The granulometric composition of the surface layer varies over a wide range,
which is explainedby the variability of conditions of bed material transportationby the
water flow in various parts of the channel during different time periods. The channel
alluvium under the armour layer is subjectto lesservariation.
Bedload yields for different time intervalswere determinedfrom the filling of the
pit excavatedin the Laba River channel since 1991. In all the volume of bedload

230
retained in the pit was estimatedfor six time intervals covering different phasesof the
flow regime (Fig. 2).
On the basisof field data the following equationwas derived:

QI,= 2.3 1O-9Q’.” (1)

where Qb is bulk volumetric bedload discharge(m3/s) and Q is water discharge(m”/s).
This equation, if computed from the actual hydrograph for appropriate time periods,
resultsin bedloadyields the most closeto the measuredones(Table 1).
Table 1 Comparisonof measuredand calculatedbedloadyields

No. Period Mean daily flow (m3/s) Bedloadyield (thou m”)

minimum average maximum measured calculated
I 1 Jan - 14 Jul 1991 37.8 128 448 130-170 142
II 24 Jan - 7 Jun 1992 96.0 153 274 70-80 76
III lApr-7Jun1992 120 194 274 60-70 63
IV 8 Jun- 19 Jun 1992 224 246 270 20.5 21
V 20 Jun - 7 Nov 1992 59.8 165 493 130-140 142
VI 15Apr-24Apr 1993 138 263 574 45 44

Annual bedload yield of the Laba River computed from mean daily water
dischargesusing Eq.(l) for the period of 52 years dependson the character of the
hydrograph and varies from 31,000 m3 (60,000 t) to 319,000 m3 (622,000 t) with the
averagebeing 132,000m3 (257,000 t). As the computationsshow, the bedload yield in
the study reach of the Laba River is equal to 15-49% (27% on average) of the total
sedimentyield.
In all the calculationscritical water dischargefor bedloadmovement was assumed
to be equal to 30 m3/s.This value was obtainedfrom experimentson the hydraulic non-
distortedmovablebed model of the study river reach .
During 1991-1993 1,400,OOOm3 of alluvium was extracted from the river
channel.The final pit was 1,900 m long, 300-80 m wide and about 3-4 m deep, and
causeda water level fall of 2-2.5 m immediately in the reach of excavationsand l- 1.5
m at a distanceof up to 1,100 m upstreamof the pit. A water level fall downstreamof
the pit was not observed.

Laboratory tests
The advantagesof the hydraulic modelling of the channelprocessesand transport of
sedimentsin the rivers of mountain and Piedmont zone when it is possibleto simulate
the sediments in the scale of the water flow depth are well-known (Kopaliani,
Tvalavadzeand Noselidze, 1985). I. V. Egiazarov, a prominent Soviet scientistwrote in
1935 (Egiazarov, 1935): “On the basisof data analysisfrom field measurementsand on
the basisof laboratory tests during 8 years in the models of mountain rivers, the author
is convincedthat the bedload dischargein the study reach can be determinedby a model
in a laboratory more accuratelythan by direct measurementsin the field; at any rate this
is true until a revolution is made in the measuring methods applied”. In fact, the
revolution in the measuringmethodshas yet not beenmade, and the words of Egiazarov

231
are true even now. Therefore, together with field studies, the channel and hydraulic
regimes of the Laba River, characteristicsof sedimenttransportationand the impact of
the project pit were studied in the laboratory conditions in a space non-distorted
movable bed model of the Laba River and simultaneouslyin a hydraulic flume, 100 m
long.
The model was made in the scale of 1:50 on an outdoor experimental plot of
100x20m, at the Hydraulic Laboratory of the State Hydrological Institute (Fig. 4). The
methodologypresentedin Kopaliani, Tvalavadzeand Noselidze(1985) was used for the
computationof the model. The model was 84 m long and about 10 m wide. The banks
were fixed, in accordancewith the channel survey made in November 1992 for the
study reach of the Laba River. The granulometric composition of bed material in the
model was selectedsimilar to that in the river (Fig. 3). Mean size of sandparticlesin the
model was 0.69 mm (34 mm, if recomputed for the prototype). Suspendedsediments
were not simulated.
Table 2 shows scale coefficients computed proceeding from the condition of
meetingthe Froude number similarity in the model and in the prototype.

Table 2 Scalecoefficientsfor recalculationfrom model to prototype

Characteristic Scalecoeffkient
Linear dimensions(length, width, depth) 50
Size of particles 50
Slopeof bed and of water surface 1
Flow velocity 7.07
Water and bedloaddischarge 17,678
Water and bedloadyield 125,000
Time of hydraulic and morphological 7.07
processes

The model was verified on water surfaceslopes,flow velocity, sedimentyield and

the character of the channel process. First, a flat bed was reproduced in the model.
Then, the bed relief, the channel,arms, micro- and meso-forms were produced by the
flowing water. Moreover, the character of the channel process in the movable bed
model was similar to that of the prototype, i.e. the channel in the model was wide,
multi-armed, and unstable horizontally and vertically. During all the experiments
recirculation of sedimentson the model was provided.
One of the main tasks of the model experimentswas to study the regime of
bedload transport and to establishits relation with the flow. Experiments were made
both with steady and unsteadyflow. Results of some experimentsare shown in Fig. 5
and Fig. 6. The experiments showed that bedload motion in the model was greatly
changeabletemporally and spatially, dependingon unpredictablecasualchangesin flow
and channel structure which is typical of branched rivers in Piedmont zone.
Relationshipsbetweenbedload dischargeand water dischargeobtainedfor the prototype
and for the movable bed model (recalculatedfor the prototype) are shown in Fig. 7.
The scatter of the model data is rather considerable.This means that instantaneous
bedload dischargeis determinedby the local hydraulic flow characteristicswhich have
not stablerelation with water dischargebecauseof the significantchannelinstability. But
for long time periods (e.g. year) bedload yield can be approximately evaluatedusing
middle curve which is well fitted by the rating curve basedon the field data.

232
 0 12 3 4 5 6 7 8 9 10 11
Time, hrs

Fig. 5 Pulsation of bedload discharge in the movable bed model at steady water discharge
of 30 Us (530 m3/s if recalculated for the prototype)
Water discharge

/ Bedload discharge

301234
$ 40-

iii
s 30-
20-

IO- : :.:
:’
;,:I,
02
0 10 1 0123012345

Time, hrs

Fig. 6 Results of experiments with unsteadyJow in the movable bed model

After calibration tests a project pit 1,800 m long, 250 m wide and 4 m deep was
developed.To evaluatethe degreeof the pit impact on the hydraulics and deformations
of the channel bed, a mean long-term snowmelt flood was simulated in the model.
During this experiment sedimentswere deliveredto the model head dependingon the
water discharge,in accordancewith the graph in Fig. 7. The pit functioned in non-
flooded regime and causedthe channelerosion at the depth of 2-2.5 m (if recalculated
for the prototype) in the zone of the active bedloadtransportation,about 1 km upstream
of the pit and at the depth of l-2 m at the distance.of 1 km downstreamof the pit. The
water level fall was 1.5-2.5 m upstreamof the pit and 0.5-l m downstreamof the pit.
Bed erosion and deepeningof the channelimmediatelydownstreamof the pit resultedin
the armouring of the channelbed which provided the stability of the longitudinal profile
of the bed and water surfacein this reach.

234
 I.”

0.8
0.6

0.006

500 600 700 800900

W a ter discharge, m3/s

Fig. - Relationships b e t w e e n bulk volumetric b e d l o a d discharge a n d water discharge

(recalculatedfor the prototype)

S imultaneously e x p e r i m e n tsin th e 1 0 0 - mhydraulicflu m e with th e s a m e b e d

m a terials h o w e dth e a d v a n ta g oe f s h o tpits (shorterth a n 2 0 0 m ), fu n c tio n i n gin flo o d e d
r e g i m ea n dtrappingall b e d l o a da n d s o m ep o r tio n o f s u s p e n d esde d i m e n ts.
R e l a tionship
b e tweenb e d l o a ddischargea n dw a ter dischargeb a s e do n th e flu m e e x p e r i m e n ts is also
s h o w nin Fig. 7 . This relationshipcorrespondsto th e m a x i m u m transportingcapacity
o f th e flo w in caseo f th e straightc h a n n e l .In P i e d m o n trivers such situation m a y b e
o b s e r v e dduring h i g h flo o d s w h e n streamis indivisiblea n d a l m o s tstraight.During
th e low-waterperiodfluvial streamis extrem e lyb r a i d e da n dwinding u n d e r conditio n s
with th e lim ite d s e d i m e ndischarge
t b e c a u s eo f th e a r m o u r e ffect. This is w h y for
low w a ter discharges th e flu m e ratin g curveoverestim a tes b e d l o a ddischargein b r a i d e d
rivers.

A s s e s s m e not f b e d l o a dd ischargeb y calculatin gm e th o d s

In g e n e r a lth, e r eis a g r e a tn u m b e ro f m e th o d sfor th e c o m p u ta tio no f b e d l o a ddischarge
available.T h e s em e th o d sa r e b a s e do n differentphysicalc o n c e p ts;th e y a r e usuallyo f
semi-empirical n a tu r e a n d th e y a r e b a s e do n laboratorya n d field d a ta o b ta i n e dfor
varioushydraulicconditio n sby differentm e a s u r i n gm e th o d s . N u m e r o u scomparisons
o f results o f c o m p u ta tio nby different fo r m u l a s with direct m e a s u r e m e n ts (e.g .
S n ishchenko a n d S h v i d c h e n k o1, 9 9 2 ) s h o w th a t th e calculatedd a ta differ from th e
o b s e r v e d a tain d o z e n sa n de v e nh u n d r e do f tim e s .
T h e availabilityo f d a ta o f field m e a s u r e m e notsf b e d l o a dyields a n d appropriate
hydrauliccharacteristics o f th e flo w m a k e sit possibleto estim a teth e applicabilityo f th e
existingcalculatingm e th o d sfor th e estim a tio no f b e d l o a ddischarge in th e L a b aRiver.

235

.-
Table 3 Results of computationsof bedload dischargein the Laba River by different formulas

Date 9June1992 18June1992 31May 1993*

Q (m3M 195 269 382
Qbm(m3W 0.010 0.026 0.071
QC Er Qbc Er Qbc Er
Author of the formula (m3/s)
. , (m3/s)
. , (m3/s) Sourceof the formula
1. Goncharov(1954) 0.127 12.7 0.257 9.9 0.295 4.2 Karaushev(1977)
2. Egiazarov (1949) 0.183 18.3 0.416 16.0 - - Karaushev (1977)
3. Karasseff (1975) 0.184 18.7 1.129 43.4 - - Karaushev (1977)
4. Kopaliani (1983) 0.190 19.0 0.318 12.2 0.432 6.1 Recommendations...(1983)
5. Kromer ( 1984) 0.005 0.5 0.040 1.5 0.012 0.2 Kromer (1984)
6. Kroshkin (1968) 0.202 20.2 - - - - Kroshkin, Kalinichenko
(1968)
7. Levi (1948) 0.082 8.2 0.175 6.7 0.274 3.9 Karaushev(1977)
8. Nikitin (1951)** 0.018 1.8 0.053 2.0 0.141 2.0 Nikitin (195 1)
9. Noselidze(1992) 0.134 13.4 0.331 12.7 0.355 5.0 unpublished
10. Romanovsky(1977) 0.490 49.0 0.756 29.1 1.997 28.1 Karaushev(1977)
11. Romashin (1990)** 0.017 1.7 0.039 1.5 0.079 1.1 Romashin (1990)
12. Rossinsky(1968) 0.195 19.5 0.412 15.8 0.418 5.9 Karaushev(1977)
13. Rukhadze-1(1977) 0.128 12.8 0.267 10.3 0.929 13.1 Rukhadze( 1977)
14. Rukhadze-2(1977) 0.012 1.2 0.017 0.7 - - Rukhadze(1977)
15. Talmaza (1963) 0.100 10.0 0.192 7.4 0.360 5.1 Talmaza (1963)
16. Timirova-Klassen (1970)** 0.009 0.9 0.029 1.1 0.076 1.1 Timiriva, Klassen (1970)
17. Umarov (1967) 2.880 288 4.313 166 - - Umarov (1967)
18. Shamov(1949) 0.050 5.0 0.092 3.5 0.111 1.6 Karaushev (1977)
19. Ashida et al (1977) 0.492 49.2 0.868 33.4 - - Takahashi (1987)
20. Bathurst et al (1987) 0.167 16.7 0.243 9.3 - - Bathurst, Graf, Cao (1987)
21. Bettess(1984) 0.989 98.9 0.158 6.1 - - Bathurst, Graf, Cao (1987)
22. Chang (1980) 0.545 54.5 1.049 40.3 - - Chang (1980)
23. Einstein ( 1942) 0.334 33.4 0.706 27.2 - - Grishanin (1972)
24. Einstein (1950) 0.331 33.1 0.489 18.8 - - Grishanin (1972)
25. Einstein-Brown (1950) 0.276 27.6 0.674 25.9 - - Sediment (1971)
26. Graf (1987) 0.266 26.6 0.545 21.0 - - Graf, Suszka(1987)
27. Meyer-Peter- Muller (1948) 0.531 53.1 1.000 38.5 - - Sediment (1971)
28. Milhous (1973) 0.023 2.3 0.040 1.5 - - Bathurst, Graf, Cao (1987)
29. Parker (1978) 0.448 44.8 1.007 38.7 - - Parker ( 1978)
30. Parker et al (1982) 0.140 14.0 0.497 19.1 - - Parker et al (1982)
3 1. Rottner (1959) 0.033 3.3 0.090 3.5 0.170 2.4 Rottner ( 1959)
32. Schoklitsch (1943) 0.134 13.4 0.197 7.6 - - Bathurst, Graf, Cao (1987)
33. Smart-Jaeggi(1983) 0.224 22.4 0.361 13.9 - - Smart (1984)
34. Takahashi (1987) 0.714 71.4 1.256 48.3 - - Takahashi (1987)
35. Thompson(1985) 0.078 7.8 0.111 4.3 - - Bathurst, Graf, Cao (1987)
* no data on the water surfaceslope
** regional formulas

Table 3 gives the results of bedload dischargecomputations at water discharge

of 195 m’/s, 269 m3/s and 382 m3/s using 35 formulas specially developed for the
computation of coarse material transport. The first 18 formulas were developed in the
former Soviet Union, while the other 17 formulas were developed in other countries. In
Table 3 the terms Qbrnand Qbcindicate measured and computed bedload discharge
respectively; Er = QbC/ Qbm.As evident from Table 3, a good coincidence with field
data show the formulas of Timirova-Klassen, Rukhadze, Romashin, Nikitin and

236
Milhous. Satisfactory results are provided also by the use of Kromer, Rottner and
Shamovformulas.

Conclusion
On the basisof the multi-purposeinvestigationsthe following conclusionscan be drawn.
The plannedannualexcavationof the channelalluvium out of the Laba River is 4
times higher than the mean annualbedload yield. The excavationof this amount of the
alluvium out of long pits would greatly disturb the natural channel and hydraulic
regimesin the river, both in the pit reach and at the sites downstreamand upstreamthe
pit. This impact would be less significant if excavation would be done out of short
flooded pits (shorter than 200 m), when the whole amount of the bedload is trapped by
the pit together with some portion of the suspendedsediments(coarse fraction). As
computationsshow, the amount of the trapped suspendedsedimentswould be no less
than the amount of the bedload. Proceedingfrom the above, the excavation of channel
alluvium during 10 years or so out of short pits in the amounts annually trapped in the
pits, would not lead to irreversible disturbancesin the channelprocessesand sediment
transportation.During the seconddecade,when excavationsare over, the Laba River is
capableto recover its regime closeto natural.
To solve tasks similar to the describedabove, the main problem is to determine
the exact amount of annualbedload yield. The existing assessments of the applicability
of Bifferent bedload dischargeformulas from the availablepublications seem to be too
optimistic to fulfil this task. It is quite reasonablefor the moment to simplify the
problem and to make an attempt to develop methods for the computation of bedload
transport for different categoriesof rivers individually, i.e. for large, mid-size and small
rivers, for plain, mountain and Piedmont rivers, for the rivers with different types of
channelprocesses,or for particular rivers. As to the rivers in mountain and Piedmont
zone, we shouldnot neglectthe advantageof the laboratory method too.

Bibliography
Bathurst, J. C.; W. Graf and H. H. Cao (1987) ‘Bed load dischargeequationsfor steep
mountain rivers’. In: Thome et al (ed.), Sediment transport in gravel-bed rivers,
J. Wiley & SonsLtd, p.453-490.
Chang,H. H. (1980) ‘Geometry of gravel streams’.Journal of the Hydraulics Division,
Vol. 106, No. 9.
Egiazarov, I. V. (1935) ‘Exchangeof experiencein hydrotechnical tests’(in Russian).
Gidrotekhnicheskoestroitelstvo,No. 6, p. l-7.
Graf, W. and L. Suszka (1987) ‘Sediment transport in steep channels’.Journal of
Hydroscienceand Hydraulic Engineering,Japan Sot. Civ. Engs., Vol. 5, No. 1,
July, p. 1l-26.
Grishanin, K. V. (1972) Theory of the channel process (in Russian), Moscow,
Transport, 216 p.
Karaushev,A.V. (1977) Theory and method for sediment computations in rivers (in
Russian),Leningrad, Gidrometeoizdat,272 p.

237
Kopaliani, Z. D. and V. V. Romashin (1970) ‘Channeldynamics of mountain rivers’.
Soviet Hydrology, 5, p. 441-452.
Kopaliani, Z. D.; 0. A. Tvalavadzeand D. V. Noselidze(1985) ‘Methods and results of
researchof channel deformations in a non-distorted model of the Khara-Mm-in
river’(in Russian).Trudy GGI, Vol. 301, p. 68-80.
Kromer, R. K. (1984) ‘Analysis of formulas for the computation of bedload discharge’
(in Russian).Meteorologiya i gidrologiya, No. 2, p. 82-88.
Kroshkin, A. N. and G. V. Kalinichenko (1968) ‘On the problem of bedloadtransport in
mountain rivers’(in Russian).Voprosy vodnogo khoziaystva,Fnmze, Vol. 2, p.
69-72.
Nikitin, Ya. A. (195 1) ‘Determinationof bedload dischargein the rivers of Central Asia’
(in Russian).Gidrotekhnikai melioratsia,No. 10.
Parker, G. (1978) ‘Self-formed straight rivers with equilibrium banks and mobile bed.
Part 2. The gravel river’. Journal of Fluid Mechanics,Vol. 89, Part 1.
Parker, G.; P. C. Klingeman and D. G. McLean (1982) ‘Bedload and size distribution in
paved gravel-bed stream’. Journal of the Hydraulics Division, Vol. 108, No. 4,
p. 544-571.
Recommendations on taking into account of river-bed deformations for water projects
in the rivers of the Baikal-Amur Railway construction zone (1983) Leningrad,
Gidrometeoizdat,7 1 p. (in Russian).
Romashin,V. V. (1990) ‘On the assessmentof river sedimentdelivery to the beachesof
So&i (in Russian). In: Voprosy sovershenstvovania metodov beregovoi
zashchity, Moscow, p.20-30.
Rottner, J. (1959) ‘A formula for bed-load transportation’.La Houille Blanche, No. 3,
May-June, p. 301-307.
Rukhadze,N. V. (1977) ‘Statisticalrelations between bedload transport and hydraulic
componentsof the mountain river channel’(in Russian).Trudy ZakNIGMI, Vol.
48 (54), p. 29-39.
Sedimenttransportationmechanics:H. Sedimentdischargeformulas ( 1971). Journal of
the Hydraulic Division, Vol. 97, No. 4, p, 523-567.
Smart, G. M. (1984) ‘Sediment transport formula for steep channels’.Journal of
Hydraulic Engineering,Vol. 110, No. 3.
Snishchenko, B. F. and A. B. Shvidchenko (1992) ‘Structural and tmstructural
approachesin the evaluation of sedimentdischarge’(in Russian).Preliminary set
of papers for the XVIth Conference of Danube countries on hydrological
forecasting and hydrological principles for water management(Kolheim, 18-22
May 1992) Koblenz, Germany,p. 571-577.
Takahashi, T. (1987) ‘High velocity flow in steep erodible channels’.In: Topics in
jluvial hydraulics, Lausanne.
Tahnaza, V. F. (1963) ‘On the problem of the transporting capacity of rivers in
mountain and Piedmont zone’(in Russian).Izvestia AN Kirg. SSR, Vol. V, issue
3, Frunze, p. 27-5 1.
Timirova, R. V. and Z. A. Klassen (1970) ‘On the problem of bedload discharge
determination in mountain and Piedmont river reaches’(in Russian). Trudy
SANIIRI, Vol. 124, p. 197-205.
Umarov, A. Yu. (1967) ‘Study of bedload motion in mountain rivers’(in Russian).
Trudy koordinatsionnikhsoveshchaniypo gidrotekhnike,Vol. 36, p. 226-236.

238
Changes of channel morphology of the
Rioni river, Central Caucasus

ProfO.G.Khmaladze, Tbilisi University, Asst. prof 0. D. Shautidze, Kutaisi

Hydrometeorologic Centre.

The present article contains the results of field, investigations of the channel processes
of the Rioni river carried out during 1972-1982 at its three sections with different types
of channels.
In particular, studies of morphology and dynamic characteristics of river bed,
suspended sediments and bedload, turbulent flow characteristics and the specific field
studies for various engineering problems, connected with the channel processes in the
Rioni river were performed in the course of these investigations.
A great number of engineering structures are in operation in the Rioni channel
and floodplain, such as: Rioni, Gumati and Vartsikhe power-plant dams, several water
intakes, pipe-line crossings, numerous bridges, gravel excavations plants, etc. The a
lowland reach of river is leveed for more than 80 km long. Near the town of Poti the
river is straightened by a new by-pass channel. A special regulating structure upstream
of Poti provides the optimal distribution of water and sediments between the old and
the new channels to avoid flooding of the town.
Urban services of Kutaisi and Poti, their water supply and sanitary conditions,
protection of lowlands from floodings, operation of the Poti sea port, and the
conditions of sea beaches greatly depend on the Rioni river behaviour, its hydrological
and morphological regimes,
Rioni, the second largest river in the Caucasus, originates at the height of 2620 m
above m.s.1. and discharges into the Black Sea near Poti (Fig. 1). The river is 327 km
long, catchment area 13400 sq.km, average basin elevation - 1984 m.
Along Rioni river, the territory of high and middle mountain regions of the
Caucasian ridge with its offshoots, are followed by Piedmont and Kolkhida lowland.
Consequently, a great variety of channel types and their interchange,occur from rapids
and waterfalls in the upper reaches to meandering sections in the downstream.

239
 Fig.1. Location of reaches of stationary research of the fluviomorphological process. 1 - Rioni
railway station; 2 - Basbi village; 3 - Sakochagidze village.

The Rioni river recharge is mixed comprising snowmelt, rain glacial and
underground water, where snowmelt and rainfall water prevail. Spring-summer flooding
period is prolonged. The highest water levels are observed during the coincidence of
snowmelt periods and heavy rains. Fluctuations of the water level at this time make up
to 3-4 m, sometimes up to 6-7 m.
Mean annual water discharge in the lower stream is 406 m3/sec, the highest
observed discharge equals -4800 m3/sec. The suspended sediment yield from the Rioni
river to the Black Sea is 7 mlntons on average. Average annual concentration of
sediments is 5 10 g/m”.
The mountain reaches of the Rioni river were episodically studied by a field
research team from the Kolkhida Centre of Hydrometeorology and Environmental
Monitoring (Kolkhidhydromet). Besides, three river reaches with different types of the
channel process were selected in the Piedmont and plain regions, they were: Rioni
railway station (145 km far from the river mouth) with middle and side bars (reach No.
l), Bashi village (115 km far from the river mouth) with multi-armed channel (reach
No.2) and the meandering river reach 85 km long (reach No.3) in the lower part of the
river.
The programme of annual field investigations included topographic surveys,
hydrometric measurements, sampling and analysis of bottom deposits and flow velocity
measurements. Besides, using standard methods and equipment geometric and dynamic
characteristics of the sand dune bottom relief were studied by the echo sounder, using
methods and equipment developed by the Odessa Hydrometeorological Institute.
Investigation of reach No.1 showed that the movement of bottom material
consisting of medium-sized pebbles, initiated at the water discharge of 330m3/s and
coverd certain channel areas. At the water discharge ranging within 330-600 m3/s, local
changes of the channel forms were are observed, during which they were separated or
moved to the river bank. At the water discharge exceeding 600 m3/s the total
displacement of channel mesoforms, middle and side bars were observed. The rate of

240
displacement of the channel forms in this reach during the flood in 1978 (at water
discharge 668m3/s) was (according to our measurement) 90-I 00 m at the end of the
middle bar, 120-140 m at the right-bank side bar and 90-200 m at the left-bank middle
and side bars.
Calculations of the displacement of the mentioned mesoforms by the data of the
flood in 1978, according to the GGI formula (Kopaliani, 1983)

CA = 950V%Fr3 mlday,
showed a close coincidence with our data. At the water discharge of 830 m3/s more
than 500 m displacement of mesoforms was observed. During the IO-year period of
investigations, 1,5 cycles of reversible changes were observed in the dynamics of side
bars, in the process of which the area of the right bank side bar was occupied by a new
one which was further shifted lengthwise.
Horizontal channel changes are not characteristic of this river reach. Vertical
channel changes on various reaches attained 2.0 m at aggradation sites and. 3.0 m at
scour sites.
Mean size of bottom material is in a good correlation with the maximum water
discharge, increasing from 50 mm at 490 m3/s to 90 mm at 830 m3/s of water discharge.

Fig.2. Indices of meander by free meanderirg. il - the step of meander, S - the length of meander, CXI-
inlet angle, a2 - outlet angle.

Analysis of channel process in the meandering reach of 85 km long in the Rioni

river from aerial photographs allowed to evaluate the type and rate of evolution of the
expected horizontal channel changes separately, before and after levees construction
(Shautidze, 1978). Concentration of flood flow within the banks at this section of the
channel leads to the increase of the conveyance capacity of the flow, flattening of river
bend shapes and transformation of a free meandering process into a restricted
meandering, when majority of bends slide down the stream at the rate exceeding 1.5-2.0
times of the previous rate of concave banks scour, instead of the further winding.

241
Acceleration of the rate of meander evolution provided strightening of bends at the
initial section and in the lower reach of river removal into a new channel.
Quantitative relations between morphometric indices of banks on the meandering
section of the Rioni were obtained according to the methodology described in
(Hendelman, 1982) with:
- relative length of bends S/h , and turning angle a0
S=0.045a290 +l. .
a
- relative amplitude of bends Y,/h, and turning angle a0
Y
JL = 0.16aL3':
a
- relative curvative radius of bends r/ h , and their turning angle cxo
r= 0.045aL9+a,;
a

Investigations allowed to describe the regime of horizontal and vertical channel

changes on a certain bend of meandering section when certain protective measures were
made, i.e.: filling the concave bends with large-size stones; to estimate the efficiency of
the above steps, and to obtain information on sand dune motion, as well as kinematic
structure and flow turbulence characteristics,
Comparison of annual topographic plans of the Rioni bend at Sakochagidze
village (1972-1983) revealed a nonuniformity of the concave banks displacement both
in time and space. In some years the displacement of a concave bank was observed only
at some cross sections (1972-1977) and sometimes it covered the whole length of the
concave bank (1977-1980, 1982- 1983). Nonsynchronism of the maximum water
discharge and maximum rate of banks displacement was also observed.
Direction of vertical channel changes in the studied section during this period was
successively alternating: degradation was replaced by aggradation (and vice versa) at l-
2 year intervals proving the reversibility of process of redeposition of bed load.
Absolute values and direction of the vertical channel changes on river bottom on this
section correspond to the water regime. Years with excessive water discharge (1977-
1978) were characterized by maximum absolute values of the bottom scour (-0.6 m)
and by the exceedance of the absolute values of these channel changes above
aggradation (4.0 m). Years with low water (1972-1973) had the smallest values of the
vertical channel changes (fl.6 m).
The combined analysis of the vertical and horizontal channel changes at
Sakochagidze village explain and prove the efficiency of measures arranged to slow
down the erosion of concave banks and bends in the meandering part of the Rioni river.
Long-term channel investigations at Sakochagidze show that the accumulation of
stones in the meander pools caused not scour but aggradation of the river bottom,
which led to the scour of the opposite beach area in 1976-1977 instead of its
aggradation. In that period the scour of the river bottom in the pools began anew. The
river bank protective measures in 1977-1978 resulted once again in the partial
accumulation of stones at the concave bank, and slowing down the rate of the bank
displacement by 3m as compared with lo-16 m observed at the same time in the riffle
section, where normally the bank ‘displacement should not be observed.

242
 Geometric and dynamic characteristics of microforms - sand dunes - were
obtained by the bottom sounding of the above mentioned section of the Rioni river. The
rate of dunes movement, derived according to (Kondratiev, Popov and Snishchenko,
1982) was close to measurements made at Sakochagidze. The total annual yield of bed
load, calculated with the account of dunes movement, was equal to 868000 tons per
year, i.e. close to calculation results of the “Kavgiprotrans” Institute.
Numerous turbulence and geometric characteristics over the channel microforms
were obtained during field measurements in cooperation with the Department of
hydrology of the Odessa Hydrometeorolagical Institute using their methods and
equipment (Grinvald, 1974):
- turbulence intensity and its distribution over the open flow area and along the
dunes;
- standardized time correlation functions;
- type of velocity distribution and flow eddies longitudinal size relation to the
flow depth;
- spectral turbulence characteristics and distribution of the dissipation rate of the
flow turbulence energy along sand dunes.
The following research results have been obtained:
- the logarithmic formula for velocity distribution gives the best agreement with
field measurements;
- standardized time correlation functions for the sand dune bottomed flow
represent periodically subsiding curves, cutting the zero line depthwise.
Negative correlation is the result of prevalence of low frequency fluctuations in
the spectrum. Microforms influence the type of distribution by depth of the
time correlational function;
- river section with sand dunes on the bottom has the minimal turbulence
intensity equal to 0.045 near the water surface, which tends to increase up to
0.279 when approaching the bottom. Dune roughness of the channel increases
the turbulence;
- eddies generated at the river bottom covered by sand dunes have prevailing
longitudinal size of the depths order, which grows up to (8-10)H by
approaching the flow axis and decreases to a certain extent in the surface layer.
Eddies have the form of a largely extended ellipse;
- in the bottom layer at the level of dunes the longitudinal size of eddies is
comparable with the dunes length and flow depth;
- dissipation of kinetic energy of turbulence at any level above the sand dunes is
significantly increased as compared with the dissipation values in rectilinear
sections of lowland rivers above the flat bottom;
- validity of Zhelezniakov’s formula for “C” and Snishchenko’s formula taking
into account the dependence of hydraulic friction coefficient on the sand dunes
steepness is proved by the measurements taken in the Rioni river;
- friction coefficient in the Rioni river bed covered by sand dunes is about 60%
of the total coefficient of hydraulic resistance, whereas the effect of roughness
due to sand particles size does not exceed 8%.

243
References
Hendelman, M.M. (1982). Hydromorphological mechanisms of the river channel
meandering and ways of their engineering application (Irtysh river case-study).
Candidate’s dissertation, Leningrad, 253 p. (in Russian).
Grinvald, D.I. (1974). Turbulence of channel flows. Gydrometeoisdat, Leningrad, 167
p. (in Russian)
Kondratiev, N.I., IV.Popov and B.F.Snishchenko (1982). Principles of
hydromorphologic channel processes. Gidrometeoizdat, Leningrad, 27 1 p. (in
Russian).
Kopaliani, Z.D. (1983). ‘Approximate method of calculation of the river channel
mesoforms’. Trudy GGI, vol. 288, p.9-15 (in Russian).
Shautidze, O.D. (1978). ‘Some results of investigations of the Rioni channel processes.
Collection of works on hydrology’, No. 14, Gidrometeoizdat, Leningrad, p. 103-
126 (in Russian).

244
Airborne data for river-bed deformations
study
D. V. Snishchenko
State Hydrological Institute, StPetersburg, Russia

Air photographs, together with cartographics material, offer wide opportunities for a
formulation and further development of a hydromorphological theory of the channel
process(Kondratiev et al., 1982; Popov, 1960). Methodology for using airborne survey
data and satellite images for river-bed deformations study has been described in a
number of publications(Prokachevaet al., 1982; Anon., 1983, 1984; SnishchenkoB. F.
& SnishchenkoD. V., 1986, 1991a, 1991b). Extensive investigations of the channel
process were made and new data were obtained for the rivers of Trans-Baikal, East
Siberia and Altai regions on the basis of satellite images and surveys from aircrafts,
cartographica data and land measurements (Prokacheva et al., Anon., 1983;
SnishchenkoD. V. & SnishchenkoB. F., 1991a). Space images have been used to
compile a large-scale schematic map for the types of river-bed deformations in the
rivers of West Siberia and Altai regions, and a small-scaleschematic map has been
compiled for almost the whole territory of the former USSR. This work has not yet
been completed for the reason independentof the author. During the map compilation,
all river reachesfrom head to mouth were studied, types of channel processeswere
identified and appropriate singswere made in the schematicmap, just as it wasdone for
the rivers of BAR zone (Baikal - Amur Railway) (Anon,, 1983, 1984). The available
images made it possible to get data on channel width, river bend height, width of the
meandering belt and width of the floodplain of meandering rivers. Previous ratios
(SnishchenkoD. V. & SnishchenkoB, F., 1991a, 1991b) were used to get information
about the changes in the dominant discharge (mean long-term discharge among
maximum discharges).These data are essentialfor all regions, becausein majority of
mid-sizesrivers water dischargesare measuredat the outlets only.
Useful material has been obtained for the rivers of the Yamal Peninsula. The
analysisof remote-sensingdata shows that most of the rivers there are free meandering
rivers, and each successivebend exceedsthe previous one (due to increasingdominant
discharges,though without any evident lateral inflow). In the floodplains of some
meandering rivers it is possible to see marks of bends produced by other water
discharges.These bend marks were used to reconstruct water dischargescontributed to
a formation of channels,floodplains and even valleyshundreds and thousandsyears ago
in many dozens of rivers in different regions of Russia (Snishchenko D. V. &
Snishchenko B. F., 1991a, 1991b). Satellite images were used to discover reasons
which had causedchangesin those river reaches.
At the presenttime forest cut (or forest fires) in river basinsis the main reason of
river change. Forest cut leadsto runoff transformation during a season,most often due

245
to higher maximum discharges in spring. Top cover destruction (often top soil
including) led to greater sedimentstransport to floodplains and to river channels,
During the analysis of air photographs and satellite images of the former USSR
territory, all possible stageshave been discovered for the changesoccurred in rivers
under the effect of intensiveanthropogenicload (partial forest cut, or forest fires in river
basinsup to complete land ploughing from watershed-divideup to the banks of small
and midsized rivers). Forest cut and land cultivation in Europe began many centuries
ago, while in Siberia (permafrost zone, in particular) this processis going on and it is
shown in photographsmade in different time. Man’s influence on river basin, as evident
from space images,may be shown for the Barguzin River (Lake Baikal basin), After
slopes deforestation and ploughing, a great amount of sedimentsdischargedto the
floodplain and to the channel. The transported sedimentswidened the meandering
channel and made it straight. This transformation of meanderingchannelsinto straight
ones is observedin many regions, e. g. in tributaries of the.Volga (Vetluga and Viatka
rivers), in the tributariesof the Ob river (Chumyshand Aley tributaries), etc.
These figures make it possible to see changes occurred in rivers caused by
different water projects (dams, weirs, pits, etc.). In Kazakhstan and partly in Altai
region, the bends in many rivers downstream the structures decreased in size,
sometimes up to a complete disappearenceof the channel and river runoff. This
decreaseof the bend size is typical of the rivers in the deserts of the Central Asia
(Syrdaryariver)
Air photographs and satellite images make it possible to discover zones (river
reaches)with the marks of disastrousevents (earthquakes,cracks in the Earth’s crust),
resulting in the changeof river channels,floodplains and even valleys. In the reachesof
intensivecontemporaryuplift of the Earth’s crust, the river bends appearto be cut into
solid rocks; from the size of these bends it is possible to recover dominant water
dischargesduring different geologicalepochs,
Methods for channel processanalysisinclude a comparison (superimpositionof
different surveys)of airborne surveys,satelliteimagesand topographic maps of different
scales,and made in different periods of time. Air photographsmake it possibleto apply
old topographic maps in which many regions were shown quite schematically,Air
photographsshow the actual situation of any river reach or channelmacroform.
Rather long reachesof large rivers (such as the Amur or Ob and the tributaries of
theserivers) were comparedas well. These reacheswere many hundredsof kilometres
long. Data on horizontal displacementof channel in some particular reach for definite
type of the channelprocesswere obtained,as well as on the changesin the longitudinal
river profile. The laws of cyclic channel evolution in plane and along the longitudinal
profile due to cyclic transportationof huge massesof sedimentsalong the channelhave
been discovered for such large rivers and Amur and Ob. The cycle of river-bed
deformation in the Ob river reach (upstreamKamen-na-Obi) is 120 years. This cycle is
divided into four stages,about ( 30 years each). This material, too detailed for the
analysis,was obtained due to using detailed topographic surveysof the Amur and Ob
rivers made at the beginning of the XXth century. These data of about 90 years old
provided the detailsand proved the reliability of the analysis.It was also stimulatedby a
wide use of various-scalesatellite images, beginning from large-scale airborne 1 -p
surveys(1: 2000 scale),and satelliteimages(1: 200 scale)up to small-scaleTV-surveys
(1: 2,500,OOO scale).

246
 Air photographsand satellite images, in particular, offer unlimited opportunities
for analysing all processeswhich may occur on the surface of river basin. This
conclusion was made by the author as a result of studying various surveysdifferent in
scale, skill and quality for may years during a comparative analysisof the processes
observedin river basins,in valleys, in floodplains and in channelsall over the territory
of the former USSR.
Thus, air photographsand satelliteimagesprovide the following kinds of work to
be done:
- analysisof geomorphologicalsituation in small and large river basinsaffectedby
man’simpact (forest cut, forest fires, land plough, etc.);
- analysisof geomorphology of valleys, floodplains and channelsof the rivers of
all orders;
- establishmentof the type of the channel process (river-bed deformations
scheme) with a simultaneousanalysisof changesin rivers under the influence of an
intensiveanthropogenicload;
- recovery of the historical evolution and transformation of rivers with various
types of the channelprocesses;
- recovery of mean long-term maximum water dischargesduring 2 - 3 millennia,
for meanderingrivers in particular;
- analysisof changes(in plane) in channelmacroformsand mesoformsall over the
study river, as well as of cyclic transport of large massesof sedimentsdownstreamthe
river;
- multi-purpose wide-scale estimatesof the state of ecological systems in the
watershed(erosion network, river network, water bodies);
- correct planning and implementationof land surveysof river channelsand basin
surface.
Let us briefly describe the methodology to classify the types of the channel
process.
A quantitative classificationof B. F. Snishchenko(Fig. 1) has been acceptedas
the initial or basic scientific classificationof the types of channel process(or schematic
river channel changes).According to this classification,the types of river channelsare
arranged in a definite order, beginning from free meandering channels up to multi-
armed channelswith numerous side bars in them. This gradual distribution of river
channelsis explainedby the increasedamounts of transported sedimentsfrom winding
channelsup to straight channels,with a redistribution of hydraulic resistancesfrom
floodplains (meanderingchannels)to straight channels(with side bars) (Kondratiev et.
al., 1982)
Classification of river channels in Siberia and in Altai shows that there are
transient types of the channel processfrom one type of channelsto another one in the
classificationof B.F. Snishchenko;besides,there are types of channelsappropriate to
specific conditions of the channel process manifestation (swamp rivers, rivers under
aufeisconditions, rivers on mountain slopes,etc.)(Fig.Z). The order of types of channel
processdistribution, however, is in a good agreementwith the laws establishedby B. F.
Snishchenkoi. e., they are arrangedin the order of the increasingsediments,increasing
channelresistanceand decreasingfloodplain resistances.

247
 free meandering
A=36.60

incomplete meandering
A=1465

Floodplain multi-armed
channel
QT A=7.93

restricted meandering
A=!593

side-bar type
A=2.60

middle-bar type of
multi-armed channels
A=l.97

Fig. 1 C’lms[ficntion ofthe chmnel processes (according to B.F.Snishchenko).

BO - width of valley bottom,

B - channel width:

IO - volley bottom slope:

I - Mnter swfaceslope.

248
 1 - Free menntierinp; 2 - incomplete v~eontlering; 3 -,floorlp/nin mrtlfi-rvwetl channel; 4 - resfricferl menndering; 5 - side-hnr type; 6 - rnulfi-owned chnnnel; 7 -
motmlnin turrlfi-onrred chnnnel; 8 - rrufies mlrlfi-nnrrerl chnnnel. 9 - overlai7tl trrirlti-nrttred cl7nnnel; IO - Induced (orogtwphic) ttrenndering; 11 - runoJ,j,j~~orn
.swotrlp.s.
REFERENCES
Anon., (1983). Recommendationson the account of river-bed deformations for
different projects in rivers in the zone of Baikal - Amur railway construction.
Leningrad, Gidrometeoizdat,68 pp (in Russian).
Anon., (1984). Recommendationson the use of airborne and space information to
study channelprocess,Leningrad, Gidrometeoizdat,68 pp (in Russian).
Kondratiev N. E., Popov I. V., Snishchenko B. F. (1982). Principles of
hydromorphologicaltheory of channelprocess.Leningrad, Gidrometeoizdat,272
pp (in Russian).
ProkachevaV. G., SnishchenkoD. V., UsachevV. F. (1982). Remote-sensingmethods
for hydrological studies of the BAR zone. Leningrad, Gidrometeoizdat, 224 pp
(in Russian).
SnishchenkoB. F., SnishchenkoD. V. (1986). Establishmentof the types of river
channels and components of their morphology from the data of airborne and
satellitesurveys.“Meteorologiya i gidrologiya”, No. 11, p, 117 - 119 (in Russian).
SnishchenkoD. V., SnishchenkoB. F. (1991a). Airborne and satellite information as a
source of data on long-term changesin water availability in rivers. In: “Materialy
nauchnoi konferentsii po problemam vodnykh resursov Dalnevostochnogo
ekonomicheskogoraiona i Zabaikalia”St.Petersburg,Gidrometeoizdat,p. 187 -
195 (in Russian).
Snishchenko D. V., Snishchenko B. F. (1991b). Assessment of present
hydromorphologicalstate of small rivers from the satellitesurveysinformation. In:
“Sovremennoyesostoianiemalykh rek SSSRi puti ikh ispolzovania,sokhraneniai
vasstanovlenia”,Issue2, Leningrad, Gidrometeoizdat,p. 85 - 95 (in Russian).

250
4. RIVER CHANNEL DESIGN CONCEPTS

AND APPLICATIONS
Theory and practice of river channel
processes
B.F. Snishchenko& Z.D.Kopaliani,
State Hydrological Institute, St. Petersburg,Russia

The general idea of this short report makes it hardly possible to describe the above
problems in a detailed and systematicway. Therefore, it is inevitable that the nature of
the present report would be quite fragmentary. The fragments, however, would cover
most sign&cant aspectsof the problems title meant under the given.
About 3,000,OOOrivers flow on the territory of the former Soviet Union (FSU).
Among them, 2,959,284 rivers are small up to 100 km long, 3,844 rivers are mid-size
up to 500 km long, and 260 rivers are large, out of which 63 rivers are longer than
1,000 km. About 2,000,OOOrivers flow over the territory of Russia.These rivers flow in
different geomorphologicaland physiographicconditions.
River basinsin Russiahave already suffered, suffer or will be suffered from man’s
activity. Besides, different water projects are planned, are constructed or under
operation in river channelsand floodplains every year.
The processes of aggradation and degradation in a river basin include the
following three components:water erosion, produced by water yield of the eroded soil
from the basin to the valley bottom and channelprocessesobservedin the rivers of all
kinds. Degradation and aggradationprocessesare included into the water-erosion cycle
on the basin combining the processesof water and eroded products formation and
motion in all the links of the hydrographic network of the basin or all over the
hydrographic system of the basin. Under the conditions of natural variability of
physiographicfeatures,the systemis in a dynamic equilibrium, except its marginal links.
The system state is determined by the main forming factors, and the parametersof the
system - by stable relations between the characteristicsof the determining factors and
degradation-aggradationprocesses(Snishchenko, B., 1988; Snishchenko, B., 1989;
Snishchenko,B. 1992).
It is assumed,that at present, during the epoch of Holocene, (the last lO,OOO-
17,000 years) the river systemsare in a dynamic eguilibrium, i.e. they are characterized
by stable parameters, on average. During typical climate periods of the Holocene,
however, which cover rather long time intervals (2,000 or 3,000 years), changes in
water availability caused significant though slow variations in river size. It can be
assumedthat during those periods the river systemswere in equilibrium,
A new factor of water-erosion processes,i.e. anthropogenic factor, appeared
severalthousand years ago in Asia and during the last millenium in Europe (since Xth
century A.D.) (Starhel and Hotinskiy, 1985)The last severalcenturiesare characherized
by a peculiar “competition” between physiographic and anthropogenic factors, where
the anthropogenicfactors become more and more intensive and important. As a result,

251
the water-erosion cycle on the basin becomes not so regular; relations between its
characteristicsand natural characteristicsof the determining factors become more
random. The problem of predicting any parameterof the system,on the one hand, goes
to the sphere of probabilistic relations, and on the other hand, the solution of this
problem dependson the stability of the systemin general.
Thus, two factors should be consideredwhich determinewater-erosionprocesses
over the basin,i.e. physiographicclimate factor and the anthropogenicfactor.
In order to predict the eventsin the links of the whole degradation-aggradation
process(sedimentyield and channelprocessesin particular), it is necessaryto see quite
clearly the state of the hydromorphology in the systemand the trends of its evolution.
Let us briefly describethesetwo factors.
Phisiographic-climatefactor. Changesin the air temperatureand humidity during
the Holocene period in Siberia and Central Asia have been used as characteristicsof the
climate factor taken from (Snishchenko,D. and Snishchenko,B., 199la). In general,
thesecharacteristicsdetermineriver runoff. It appeared,that rivers with larger channels
correspondedto the periods of higher river runoff, if compared,with the presenttime,
when river runoff is lower.
Consideringthe Ob River basin, the largest river basin in Siberia, it is possibleto
see,that mean long-term maximum water dischargesfrom the Ob River tributaries were
severaltimes less 3,400 or 4,000 years ago, if comparedwith 1,000 or 2,000 years ago.
The river dimensions during this intermediate period were reduced greatly
(Snishchenko, D. and Snishchenko,B., 199la). At present, the increase of water
availability caused the increase of channel size. A direct relation between water
dischargein a river and its width is well known in hydrology. Theserelations are shown
in detailsin (Snishchenko,D. and Snishchenko,B., 1991a).
It should be noted, that climate changesoften do not causeany changein the type
of the channelprocessin the river, but in the size changeonly, at leastin lowland rivers.
The types of the channelprocesscorrespondto typical schematicchangesin the
river channel. Classifications of the types of channel processesin lowland rivers,
developedat the StateHydrological Institute in Russiaare given in (Kondratiev, Popov,
and Snishchenko,B., 1982; Snishchenko,B and Kopaliani, 1994). In publications
(Snishchenko, D and Snishchenko, B, 1991a, 1991b; Snishchenko, D 1994) a
transformation of three left-bank tributaries of the Ob River where water availability
was suddenlyreduced is investigatedin a great details.For instance,water dischargein
the Burla River reduced ten times during the period of 3,500 years, if compared with
the ancient period. The size of the present Burla River channel is much less, and the
ancient channel is now used as its floodplain, without any changein the type of the
channelprocess.A similar situation is observedin another tributary - in the Kuhmda
River. In the floodplain of the Chumysh River it is possible to see large separated
ancient river bends and minor river bends, separatedtoo, which correspondedto the
intermediateperiod and to small water discharge.At present,the water availability and
type of the channelprocessin this river are different.
Among physiographicfactors it is necessaryto distinguishtectonic movementsof
the Earth’s crust, resulting in valley capture, blocking the channels, etc. during
geological breaks and earth-quakes.The time of physiographic-climateoccurrence of
the events is rather long, it is compatible with the time of geological epochs
(Snishchenko,D., 1994).

252
 The nature of processesis quite different in case of the anthopogenicimpact on
river basins.According to the results,they are also compatiblewith the eventsoccurred
during geologicalepoch,but the duration of theseprocessesmay be by severalorders, if
comparedwith the natural processes.
At present there are no large river basins on the Earth which are not subject to
man’simpact. These are river basinsof the Danube,Rhine, Mississippi,Niger, Yangtze,
Hwang Ho, Amour and the mentioned Ob, etc. The maximum anthropogenicload to
the Danube is quite evident, becausethis river flows across the territories of eight
developedcountriesof Europe (Snishchenko,B., 1988).
The major types of the anthropogenicimpact on the transformation of river
channelsand floodplains are well known and can be reducedto the following:
- land and water use practices;
- forest cut and forest fire;
- basinplough;
- constructionof reservoirs;
- excavationof sandand gravel from river channeland floodplains;
- open mining in river basins,in channelsand floodplains;
- urbanisation:urban constructionwith the use of floodplain areas;
- construction of communication lines (roads, bridges, pipelines, power
transmissionlines, etc.);
- irrigation;
- different hydraulic structuresin channelsand on floodplains (water diversion
sites,sitesfor waste discharges,etc.);
- navigation and timber floating;
- flood controls, etc.;
- disasters resulted from man’s activity (forest fires, landslides, mudflows,
inundations,etc.).
Let us briefly describesome peculiaritiesof hydrographicnetwork transformation
of all river basins.As it is evidentfrom the geomorphologicalpublications(Dedkov and
Mozzerin, 1984), in particular, severalperiods have been identified with various man’s
impact on fluvial processes,which correspondto the epochsof man’s cultivation of the
terrain. The first and the secondepochsare usually related to the beginning of the last
millennium and to XII-XVI centuriesand they are characterisedby the land use with
previous forest cut and burn, and by more intensive farming accosiatedwith partial
forest cut and land plough. The third epoch is extendedup to the presenttime and it is
relatedto an almost completeforest cut and intensiveplough.
We assume,that the.forth epoch began at the beginningof the XXth century and
it is characterisedwith a radical technogenictransformationof the river network.
There is no integral quantitative hydromorphological criterion of these epochs.
Therefore, they are characterisedby different factors of different components of the
degradationand aggradationprocesses.
The intensity of soil scour causedby basin plough is indirectly characterisedby
sedimentationof eroded products on the floodplains, by a stratified soil composition
(according to different epochsof man’s activity). Silt depositionin the floodplains over
the RussianPlain is up to 2.5 m deep and its age is up to several centuries. The silt
depositionrate in the zonesof intensivefarming during the third epoch exceeds1 cm/yr.

253
 A comparativeassessmentof the specific sedimentyield dependingon the rate of
the anthropogenic cultivation of the landscapeis an objective characteristicsof the
erosionrate.
On the plains (Dedkov and Mozzerin, 1994) mean specific sedimentyield equals
70 t/km2 per year; in poorly disturbed basins it equals 13 t/km2 per year and in
intensivelyplough basinsit attains 130 t/km2 per year.
The amount of soil washedof the slopesaccordingto investigatorsgreatly depend
on the steepnessof the slopes and these values cannot serve as an objective
characteristic of erosion rate because they hardly follow the standards of the
measurementsmethod and factors which determine these values. The erosion rate
varying from 1 to 124 t/ha are observedin the European territory of Russia both in
poorly eroded and severelyerodedregions.
Small rivers in the river basin, i.e. watercoursesof lower orders (according to
R.Horton} in the river system serve as most evident indicator of man’s activity.
Degradationof small rivers is causedby man’simpact over the basin and in the channels
of theserivers. The first reasontypical of the early stageof man’sactivity is observedat
presenttoo. River control is typical of the last, fourth stageof man’sactivity.
As to extreme values of specific sedimentyield in the world, the famous Hwang
Ho in China, with its specific sedimentyield of 1400t/km2 per year is just the fifth river.
The highestvalue of 2580 t/km2 per year is observedin the Purari River in Hew Guinea
(Gwynne Pover, 1988).
Ploughing up of the river basins,including floodplains, has led to silting of small
river channelsin the south of the European Russia. Construction of ponds (tens of
thousandsponds) during 1950-1965 in the small rives stopped any water flow and
sedimenttransport there (Lapshenkov, 1991). Survey of rivers along the Baikal-Amur
railway road (Snishchenko, D. and Snishchenko,B., 1991a; Snishchenko,D. and
Snishchenko,B., 199lb; Snishchenko,D., 1994) showed a crucial role of forest fires
and forest cut in the permafrost zone for an increaseof the sedimentyield from the
slopesof theseriver basins.For example,when the forest was burnt on the slopesof the
Muya river valley, an intensive erosion of the slopesbegan and the scoured soils were
deposited on the floodplain; aggradation of the floodplain began, the amount of
sedimentin the river was great and the type of the channelprocess in the river was
changed.After forest f&-esand forest cuts gullies and “dry streams”were formed in the
Muya basin.In the Charariver the forest fires changedthe level marks of the floodplain.
The hydrographic network of the rivers of low orders is subject to constant
irreversiblechangesgetting specific features of the evolution. Silting of channelsmakes
the channel profiles more gentle, reduction of the river, according to the data of
M.Ya.Prytkova, are closely connectedwith the rate of basinploughing. Silting of rivers,
river training, construction of ponds and land plough in the upper reacheslead to a
reduced density of the river network and to a less number of watercourses. For
example,according to the data of V.M.Shirokov, the length of small rivers in Belorus
was reduced by 30%, meanwhile, the network of canalswas increasedand it exceeds
the network of small rivers in 4-7 times.
Intensive whash out of soil from the basins leads to changesin the sediment
budget in river basins. In general (according to N.I.Makkaveev), the balance of
sedimentsis a function of degradation-aggradationprocessesat different links of the
hydrographic network. In case of an intensive man’s activity in drainage areas and in
river channels,the sediment balance components in the river basin are subject to

254
permanent changes,it mainly concernsthe overland and channel components of the
flow. A ratio betweensuspendedsedimentsand bedloadis subjectto changestoo. Both,
sediment deficit and sediment excessdue to man’s impact on the basin and in river
channelslead to continuous and often fundamental changesin the lower link of the
chain of the single degradation-aggradation processover the basin, i.e. fluvial progress
in the channel and in the floodplain. In facts change in the sediment yield, as the
determining factor of channel formation, is not the only reason of this change. The
man’s activity in the basin primarily leads to changesin flow characteristicsand to the
conditionsrestricting a free evolution of channels.Thesefactors cannot be describedin
detailswithin the limits of the presentpaper. It should be only noted that an assessment
of the degradation and aggradation processeson the river basin would require an
analysisof all the determining factors of each of the three componentsof the cycle, i.e.
soil erosion,sedimentyield and channelprocesses.
At presentthe problems of river channelchangesdue to man’simpact and due to
changesin the first two cycle components are investigated quite inadequately for
practicalpurposes. The reasonis that the analysisof this problem requiresnew research
methods,basin-wide approachinsteadof the routine local approachwhen the restricted
short river reach in close vicinity to the future project is being considered,is given by
air-borne photographs and by satellite images (Snishchenko, D.and Snishchenko,
B.,1986).Air photographs and satellite images applied for the analysis of the channel
processesin the Ob River basin within the Altai region made it possibleto establishvery
rapidly (during 3 years) fundamentalnew lows in the channelprocessesof this large
river basin,which may be discoveredby other methodsfor a longer period or they may
not be evennoted at all (Snishchenko,B., 1992; Snishchenko,D., 1994). The analysisof
satelliteimagesand air photographsmade it possibleto ascertaina degradationand even
completedisappearanceof the certain links of the hydrographicnetwork, a reduction of
river length, changesin river width and changesin the type of channel process. The
prepared schematic map of the channel process in the basin, together with the
discovered trends of channel evolution, made it possible to outline a substantiated
forecast for a future developmentof the hydrographicnetwork in the region.
In general, an intensiveman’s activity on the watershedsaffecting basic channel-
forming factors (runoff characteristics,sediment yield, restricting conditions) would
sooner later lead to a changein the type of the channelprocess. Fig. 1 shows a new
channelof the Viliuy river (a tributary of the Lena river), developedinto the side-bar
type in a rectilinear channel,insteadof a free-meanderingtype. A changein the channel-
processtype occurred due to a great amount of sedimentsdischargedto the river from
basin slopes denudated by intensive forest fires (Snishchenko, D.and Snishchenko,
B., 1991b).
Similar transformation occurred 3,000 or 4,000 years ago in the Amur river basin,
when the right-bank part of the Amur river basinwas cultivated in China. Forest cut and
land plough ware expandedfrom south to north. At present,a destructiveforest cut on
the left-bank area of the Amur basin is ever growing which would result in a further
changeof the hydrological and channelregimesin the Amur river (Snishchenko,Dand
Snishchenko,BJ991b).
A transition of channelsfrom meanderingto rectilinear due to intensive sediment
yield from the basin slopesis typical not only of the Amur and Ob river basins. This
processis also observedall over the Europeanterritory of Russia,even in the northern

255
Fig. 1 Changed type oJchanne1 process of the Viluy River

256
rivers of Vetluga, Viatka and Vychegda. This process is observersin the rivers of
Bashkiria and in all left-bank tributaries of the Volga river.
Forest cut and basin ploughing causedchangesin the redistribution of maximum
water dischargesduring a year. In general, greater maximum water dischargescaused
wider channelsand river bends. For example,the ratios of widths in ancient channels
(2500-1000 years ago), intermediate channels (1000-2000 years ago) and in
contemporarychannels(200-50 years ago) in the Ob river basin are as follows: 3.5 : 1 :
2 (Snishchenko,D and Snishchenko,B., 1991a; Snishchenko,D. and Snishchenko,B.,
1991b).
A number of disastrousfloods in spring of 1994 in the rivers of Russia may be
regarded as a result of wrong man’s activity on the watersheds,in floodplains and in
river channels,i.e. the result of a suddendecreaseof the channelconveyancecapacity
due to sedimentationof the channels,due to different water projects in the floodplains
when the structuresare made at the very water edge,due to floodplain plough, etc.
Irreversible processesof river degradation are observed also in the permafrost
zone in the polar areas of European and Asian Russia, in the zone of intensive
developmentand extraction of preciousmetals.
A contemporary river network is affected by a peculiar “creeping hydrographic
gangrene”resulting in a successivedegradationof watercoursesof all the orders in the
river network, from minor orders to major ones. It should be kept in mind, however,
that runoff formation and sedimentformation are by 50% (on average)observedin the
small watercourses, and degradation of these small streams is very important to
maintain mid-size and large rivers. It should be also emphasized,that many small rivers
in Russia, and not only in Russia, are observedin the air photographsonly in spring,
when water surplusin the fields forms dark marks of the former meanderingchannels,
presently filled with eroded soil and sometimes ploughed (Snishchenko, D.and
Snishchenko,B., 1991b).
It is high time to introduce appropriate changesinto geographic manuals and
atlasesin each country of the world to the data on rivers, e.g., instead of river No. we
have to put “the former river” and insteadof its usual length of, say, 1000 km, we have
to put 900 km. For example,the river training causedby great sedimentyield made the
Ob river reach of about 500 km by 100 km shorter during the period of 1800-1986
(Snishchenko,B., 1992).
The very fact of changesin the river network is not only of a purely scientific
interest.Millions of hydraulic structureshave been made in rivers, from large dams for
power generationup to power transmissionlines, all over the movable and varying part
of our planet. In 1975 the damagein the USSR from the negativeeffect of river channel
changeson thesestructuresmade at least 4 billion USD (Snishchenko,B., 1994).
The hydraulic structuresthemselves,made in the rivers, affect the regime of these
rivers.
Most significant disturbances in the natural channel regime and sediment
transportationin rivers are observedas a result of reservoirsconstructionin rivers.
At present,more then 30,000 large reservoirshave been constructedin the world.
More then 4,000 such reservoirshave beenmade in the rivers of the FSU (Reservoirsof
world, 1979). The rates of reservoirs construction since the beginning of the XXth
century (when there were only 41 reservoirsin the world) have beenintensifiedgreatly.
During the last 30 years the number of reservoirsincreased4 times, and the total
volume of the reservoirs increased 10 times (in Asia it increased in 90 times). At

257
present,from 300 up to 500 new reservoirs are being constructedin the world every
year; in future it is projected to control 2/3 of the world rivers FSIJ (Reservoirs of
world, 1979).
Excavation of sandand gravel out of the channelsand floodplains for construction
purposesis one of the most intensivekinds of the technogenicimpacts (Snishchenko,B.
and Meserlians,1994).
Before 1950s’excavationof sand and gravel out of river channelsand floodplains
for construction needs was not a damage for the environment and for the ecology
becauseit was not intensive and it was of a local nature, if compared with other
anthropogenicimpact on the channelevolution.
At the end of 1950s however, ecologically and technically invalidated excavation
of sand and gravel from the rivers in many countries of the world caused the river
network degradationand becamedisastrousfor the environment.
During the last 20 or 30 years the removal of send and gravel from the river
channelsfor constructionneedsis ever growing in Russiaand in the CIS. The Ministry
of the River Fleet of Russia excavated 160,000,000 m3 of sand from the pits in
navigablerivers up to 171,000,000m3 of sand. Sandwas removed quite intensively out
of the rivers of Lena, Ob, Irtysh, Tom, Belaya, Ufa, Oka and Kuban. Unprecedented
(by rates and volume) amounts of sand and gravel were excavated from the river
channelsin the zone of the Baikal-Amur railway construction.
Excavationsfrom the pits in many rivers of Russiaand CIS made low-water levels
in the rivers by 1,5-3,5 m lower. The results of excavationsand lower water levels in
rivers are well known. They createproblems for waterwaysnavigation and operation of
ports, for water diversions, they destroy bridges, pipelines across the rivers, power
transmissionlines, communicationcablesas well as dykes and other structuresalong the
river banks. Gravel excavation from rivers often cause disappearanceof sea-shore
beachesin the vicinity of river mouths.
Long river reacheswhere sand and gravel are excavatedare inthtencedby these
pits; the stability and stationarity of relation between water discharge and levels are
disturbed; depressioncurves of the ground water are lowered which leads to plants
suppressionin the floodplain, to intensivelandslides,to reduced spawningsites, deficit
of sedimentsand reducedbeachareason the seacoast.
Similar situationsmay be taken from numerouspublicationsin different countries
of the world, in Austria, Germany, Hungary, Bulgaria, Italy, France, India, USA and
Canada.
According to the assessments of the Japanesescientists,the intensive excavation
of sand and gravel fi-om the rivers of Japan simultaneouslywith the construction of
power generation plants became a national disaster. According to the data of the
Ministry of Construction,the gravel excavationduring 1966 an 1970was equal to 374
and 477 mln tons, respectively. As a result the rates of the overall degradation
(irreversible scour) of the channelsin some rivers of Japanwere equal to 10 cm/year.
Multi-armed channelswere transformedinto meanderingchannelswith deepeningof the
channel depth; the stability of the hydraulic structures as well as ecosystemswere
disturbedin the zone subjectto channeltransformations(Human Influences..., 1989).
It is a hardly possibleto mention other kinds of water projects in rivers within the
limits of this paper.
To solve appliedproblems of the channelprocessesfor any water project in rivers,
a system of predicting the channel processes has been developed at the State

258
 Hydrological Institute; this system made it possible to combine all variable water
projects in rivers, all kinds of river channel changes,all kinds of predictions, kinds of
information required for the predictions and to solve practical problems on water
projects validation in a simple way (Snishchenko,B., 1994).
A combination of natural processesand man’simpact on the water projects often
causehardly solubleproblemsfor scientists,designersand decision-makers.
Let us take a casestudy.
On the request of the Government of Argentina, the scientistshorn the State
Hydrological Institute were invited in summer 1993 to make an international expertise
for the problem of the Pilcomayo river, a tributary of the Paraguayriver, which in turn
dischargesto the Paranariver.
The Pilcomayo river flows from The Andes in Bolivia and runs across this
country over 600 km; then the river crosses the frontier between Argentina and
Paraguayand flows as the boundary between these countries over 900 km long. The
river transports a great amount of fine sands (125,000,OOOt/year) - the maximum
observed amount was 288,000,OOOtons. Distribution of runoff and sediment yield
during a year is extremely uneven. The maximum observedwater dischargewas 5,000
m3 Is, mean among maxima was 1843 m3 /s, mean annualdischargewas 2 11 m3 is, and
minimum water dischargewas 9.5 m3/s.
The river channelis meandering.During 1930sthe river flow was broken in the
middle reachesand the transit water flow was stopped.The river was separatedinto two
individual streams. Since that time, the upper reachesof the river are being silted by
sedimentsevery year during the flood and during the flood recessionup to the edge
over the distanceof 3.5 km (1990)up to 45 km (1984). The silting of the Pilcomayo
channel produced a number of socio-economic, research, technical, ecological and
frontier problems.
The Pilcomayo valley and floodplain is a habitat for Indian tribes since the time
immemorial. Thesetribes consumefish, animalsand plants in the floodplain. They drink
water out of the river. The Indians move their housesfor the third time upstream the
river, as far as the river dry channelis extendedupstream. Since 1940 the river moved
upstreamby 400 km.
The secondproblem is to provide equal water distribution betweenArgentina and
Paraguay,during the dry seasonin particular (June-November).
The major research and technical problem is to discover the reason and
mechanismof this phenomenonand to decide whether it is possibleto stop the process
of channelsedimentationand in what way.
The Pilcomayo river channelis raised, if comparedwith the surroundingterrain.
Here numerousnatural diversion canalsoccur from the channelto the floodplain which
are effective during the high-water period.
Two experimentalcanalsconstructedby the specialistsof Argentina and Paraguay
during 1976-1983 on this principle completely, stopped the dry channel extension
upstream.
The study of this problem makes it possibleto conclude that even in case of the
appropriatepolicy of the three countriesconcernedin the Pilcomayo river and in caseof
the availablefunds, the major problem would bein the scientific and technical aspects
matter.
Fig. 2 gives the superposedcross-sectionsof the Pilcomayo river channel at La
Paz gauging site 40 km upstream the dry boundary of the channel. Hydrometric

259

-- ---
 6 1 h(m) 6
.-.-.-.-.-.- .-.-.-. 1
3
5-
MI MD

4- .-.E.-.-.- .-.-. 2
o-o-~-o-o--o-~70-
5
3-

Qm;. Data
2-
1 889 12.02.90
2 12 22.09.90
l- 3 1881 15.01.91
4 16 06.09.91
5 609 30.11.91
O- 6 2201 18.01.92

-l-

-2-..,,...,.,,, , , ,,,, ,
20 40 60 80 100 120 140 160 180 200 m

Fig. 2. Superposed cross sections, Pilcomyo River, 1990-1992.

E a

IMeyer-P$te/ 1

Fig. 3. Predicted scour andjll downstream porn Chapeton reservoir, river Parana. 2.5year after dam
closure (Prendes. 1984).

260
observationsat La Paz are made since 1960; water dischargesare measuredevery day.
Here cross-sectionprofiles for 1990-1992were compared,theseprofiles were obtained
at the station. Sedimentsconcentration during flood was 30 kg/m3. As evident from
Fig. 2, the water depth at the cross-sectionarea 2 was within 0.5-l .2 m during the low-
water period on 22 August, 1990. The rise of the water level produces not only the
widening of the channel but a simultaneoustotal scour into depth by 3.5 m (cross-
section area 3). Then, as far as water dischargebecomes smaller, up to 16 m3/s, the
channel is tilled with sediments again (cross-section area 4) and the channel size
approachesthe size of the channel during the low water period for the previous year.
The subsequentflood (cross-sectionareas 5 and 6) restores the channel within the
boundariesof the channelduring the flood at the previous year, etc. The quasi-dynamic
equilibrium of the channelis expressedin the annual “respiration”of the cross-section
area within a definite range of elevation marks, during which there is a clear relation
between water discharge and maximum channel depth, and this relation is never
disturbed.
Here we approach the most important theoretical and practical problem of the
channel process on the quantitative assessmentof discharge and yield of channel
forming sediments.This problem is more than 200 years old. According to a general
assumptionit is one of the deadlock problems of sedimentationand theory of fluvial
processes.In our opinion, the reasonis methodologicaland the way out of this situation
shouldbe made in this methodologicaldirection.
At present, there are more than 150 design formulas to compute bedload yield;
about a half of these formulas were developedin the FSU. Most of these formulas are
not known abroad, as it follows from the referencesto the papers.Besides,some dozens
of works have been published during the recent time in Russia and abroad where
computation results obtained from different formulas and/or from intercomparison
between the laboratory measurementsand field data are given. In general, the basic
conclusionin such works is that the author’sformula is the best or amongthe best.
Therefore the generalprogressin this matter is an imperative sincethe availability
of variety of formulas at presentservsas a factor with disorient engineersand designers
who is to choose“the best formula out of the other very good ones”.
Fig. 3 gives one of such casestudies.Prendes(1984)used severalformulas applied
by different scientists (Raynov, Pechinov and Kopaliani, 1986) for a mathematical
model of predicting bottom scour in the Paranariver downstreamthe Chepeton dam.
This case study is interestingfor us by the fact that the formula of Mayer-Peter more
frequently used abroad and the formula of Shamov (often applied in Russia) are used.
As evidentfrom the figure, theseformulas provided the extrememaximum scoursin the
downstream of the dam: the use of Shamov formula gave 1 m, and the formula of
Mayer-Peter - 8.5 m. Commentsare not neededhere.
Indeed, what is the problem? It appears,that when formulas are compared, very
often different formulas are appliedwithout a detailedconsideration,whether they were
developedfor bedloadyield, or the total sedimentyield; for lowland or mountain rivers,
and even mudflow are not distinguished, etc. Moreover, the term “bedload” is
understoodby different authorsin different ways.
Computationsmade by the scientistsfrom the State Hydrological Institute, for a
determinationof the forms of sedimentstransport bedloadmotion, saltation, suspension
using different methodsprovided the resultswith deviationsin dozensof times.

261
 We can put a question: if there is no a definite co-ordinated terminology and
methodology even for a determinationof the forms of sedimenttransportation,how is it
possible to get compatible computation results, to say nothing of compatible data
observed and computed if different object of measurement and techniques of
measurementsare quite individual in each case?Besides,as evident Iiom the Pilcomayo
casestudy, it is not valid to solve the problem of bedload yield separatelyIi-om regime
of river channelchanges.
It is quite evident that currently it is more promising to develop design methods
separately for a certain category of rivers, type of channel process, length of the
watercourse, peculiarities of the granulometric composition and hydrological river
regime.
Another fundamentalproblem, the solution of which also reached a deadlock, is
the assessmentof hydraulic resistancesin the natural watercoursesand canals.Without a
solution of this problem the progress in the improvement of methods for the
computationof the channeland floolplain conveyancecapacity is impossible,as well as
in modification of methods for extrapolationof water dischargecurves, methodologies
for hydraulic modelling, technical validation of channel reconstruction, methods for
canaldesignand many other appliedhydraulic problems.
A future progress in this field would require expansion of conventional
approaches, application of new theoretical and methodological ways and a new
empiricalbasis.
The results on the laboratory study of the conveyance capacity and hydraulic
resistancesin meandering river channels are given below; these results have been
obtained at the Laboratory of Channel Hydraulics at the Department of Channel
Processesat the StateHydrological Institute (Kopaliani and Hendehnan,1989).
The laboratory research method is used in the following two basic directions
within the framework of methodology of the hydromorphological concept of the
channelprocesses,i.e. for the study of generallaws of the channelprocessesas a natural
phenomenonand for a solution of particular appliedproblems.
In the first casehydraulic flumes are applied and different laboratory installations
are used together with the scale models of some type of the channel processes.The
secondproblem is solved on hydraulic (fixed and eroded) models of river channelsand
floodplains.
Numerous studiesof any type have been made at the State Hydrological Institute.
Thesestudiesare characterisedby a great interest to a discretemorphological elements
of the channeltopography, sedimenttransport and water flow structures.
Studiesof the first kind may be illustratedby the tests on the conveyancecapacity
of the channel and on the hydraulic resistancesin the channelsof meandering rivers
using schematic laboratory installation. The effect of pools and shoals to hydraulic
resistanceswas investigated,in particular.
A meanderingreach of a real river 112 km long with well-developed 14 river
bends was simulated on the model in the horizontal scale of 1 : 800 within fixed
boundariesof the river banks (Snishchenko,B. and Kopaliani, 1994). The model bed
with the slope of 0.00 1 was covered by sand with mean particles of 0.33 mm in
diameter. First, water dischargeswere delivered to the model which produced no
motion of the bedload. After each delivery a position of the water surface was lixed
along the whole model reach. At the channel-forming water discharge of 47 l/s the
position of the water surface was fixed twice, i.e. at the beginning of the test and 36

262
60

100 120 m
0 20 40 60 80

Fig. 4. Longitudinal profile of the bottom and water surface m the schematic modeelof the meandering
channel reach

1 - initial profile ofapat bottom;

2 - water surface profile at the channel-forming water discharge in case missing and available shoals
and pook;

3 - profile of the water surface (Q4.6 1is)with available pools and shoals;

I - ldem. without pools and shoals:

5 - longitudinal profile of the bottom in case of the channel-forming water discharge along the right
(5) and left (6) banks with available shoals and pools.
hours later when pools and shoals were completely formed on the model. Then, the
tests were repeated, but the delivered water discharges were less than the channel-
forming discharge (less than 47 l/s), with available pools and shoals, however, resulted
from the channel-forming water discharges.
The results of the above experiments are shown in Fig. 4 with the comparison of
longitudinal proties of the water surface: initial prolYe and the profile fixed at the
channel-forming water discharge, respectively at the absence or availability of pools and
shoals, as well as in case of small (not channel-forming) water discharge of 6.6 l/s at the
absence and availability of pools and shoals.
Since water surface curves at the channel-forming water discharge with missing
and available pools and shoals coincide, it appears that pools and shoals do not
contribute to hydraulic resistances and do not affect the conveyance capacity of the
channel. They are probably produced as the mechanism for sediments transportation
only.
The second conclusion is as follows. In case of water discharges less than channel-
forming ones, pools and shoals are the forms of resistance to the water flow motion,
because they reduce the conveyance capacity of the channel. As evident from Fig. 4,
one and the same water discharge of 6.6 l/s with available pools and shoals (relict forms
left by previous channel-forming water discharge) is transported through the channel at
higher water surface marks, if compared with the case of missing pools and shoals.
In accordance with the methodology of the hydromorphological concept of the
channel process, the use of the laboratory method is reasonable in combination with
other methods (analysis of cartographic and air-borne materials, field studies, theoretical
methods and computations).
Proceeding from the peculiarities of the specified problem, research object and
nature of the basic information, the laboratory and other methods are applied in
different combinations. The effect of laboratory method is different in the solution of
the specified problem, the applied model is diEerent too (schematic accurate or
approximated model, rigid or eroded model, nondistorted or distorted hydraulic
models).
For instance in case of the channels with coarse sediments it is possible to
reproduce the most accurate scale model of the study event.

Conclusions
Summarising the above, it is possible to state that the problem of radical transformations
metamorphosis of the river channel and floodplains caused by man’s impact becomes of
the utmost importance during the technological epoch.
Since there are no general theory and generally accepted methodology and
verified methods for a solution of this problem, it is necessary to develop them at
present.
It should be noted that the approach proposed by S. Shumm (Schumm, 1971) is
most successful and timely, where an attempt is made to develop a methodology for an
assessment of trends for basic transformations metamorphosis of rivers depending on
the nature of changes in the governing factors of channel formation, i.e. characteristics
of water discharge and sediment yield irrespective of reasons of their changes (climate
or anthropogenic impact). However this is not a method for quantitative estimates, but

264
assessmentof trends only. Unfortunately, Shumm’s approach is restricted by
meanderingrivers only and his estimatesdo not contain the time factor. Besides, it
seemsquite reasonableto assumein such approachesbased on the laws of hydraulic
geometry and the ideas of the regime theory, the chamiel size (width, depth, steps of
river bends) would increasesat a suddenincreaseof water discharge,but the opposite
situation is not always possible,i.e. the river bend would not be reduced at a sudden
decreaseof water discharge(especiallyin rivers with coarse alluvium). Low-powered
flow will not be able to do this work.
Thus, assumingthe priority of the problem on the recovery of the natural channel
regime of rivers or, which is more realistic, maintenance of river channels and
floodplains under the controlled technogenicstable state regulated by the society, and
assumingthat the progress in this matter is most effective in case of an intensive
international cooperation, the State Hydrological Institute proposed via the National
Committee of Russia for the IHP for the Vth IHP phase an implementation of the
project on the methods for estimation, computation and prediction of river channel
changes,and sedimentyield characteristicsin the conditions of intensive anthropogenic
impact to river basins, to river channelsand floodplains. The implementation of this
project would make a clear global situation of this event end of the scalesof the event, it
would generalisethe experience(both, positive and negative ones) of the practice from
many countriesand it would provide a developmentof coordinatedprinciplesfor some
general strategy and tactics for interrelations between the modem society and river
systems.
The implementation of this project would require much efforts and it would
probably stimulate a development of numerous specific problems of morphology,
theory and methods for a computation of bedload transportation in river systems,
hydraulic geometry and resistancesus well as the problem of interaction between river
systemsand water projects.

Bibliography
Dedkov, A.P. and Mozzerin, V.I.(1984) ‘Erosion and Sediment yield on the Earth’.
Kazan University Press,p. 264 {in Russan).
Dedkov, A.P. and Mozzerin, V.I.( 1994) ‘The sistem of erosion and sedimentload on
plains of temperatezone’. In: Proc. of the Int. Symp. “East-West,North-South
Encounter on the State-of-art in River Engineering Methods and Design
Philosophies”.St. Petersburg, ~012, pp. 24 - 3 1.
Gwynne Pover. ‘Siltation is threat to whole Word’s StorageDams’. June 1988. World
Water. vol. 11 N 5.
Human Influenceson Hydrological Behaviour: an International Literature Survey. IHP
3 Project G. 1 TechnicalDocumentsin Hydrology UNESCO, Paris, 1989,p. 195.
Kondratiev, N.E., POPOV, I.V.and SNISHCHENKO, B.F. (1982) ‘Principles of
hydromorphological theory of channelprocess’. Leningrad, Gidrometeoizdat,p.
272 (in Russian).
Kopaliam, Z.D.and Hendehnan, M.M. (1989) “Cannel Process and Hydraulic
ResistanseProblemsof up-date Hydrology.” Leningrad, Gidrometeoizdat,p. 288-
304 (in Russian)

265
LapshenkovV. S. (199 1) ‘Puti uluchsheniasostoianiei ispolsovaniamalykh i srednykh
ravninmkh rek’. In: “Sovremennoye sostoianie malykh rek SSSR i puti ikh
ispolzovania, sokhraneniavasstanovlenia”,Issue 2, Leningrad, Gidrometeoizdat,
pp. 16 - 23. (in Russian)
Raynov, S., Pechinov, D.and Kopaliani, Z. (1986) ‘River Responce to Hydraulic
Structures’.Technical Document in Hydrolody. UNESCO, Paris, p. 114.
Reservoirsof world. Moscow, “Nauka”, 1979 (in Russian).
Schumm,S. (1971) ‘Fluvial Geomer Jology: Channel Adjustment and River Metamorphosis’.
River Mechanics, edited by H.W.Shen. Chapter 5. Snishchenko,B.F. (1988) ‘Puti
kompleksnogo prognozirovania erozii, stoka nanosov i ruslovykh protsessovna
rechnom vodosbore’. In: Papers of the 14-th Conf of Danube countries on
hydrological forecasts.Kiev, pp. 349 - 356. (in Russian)
Snishchenko,B.F. (1989) ‘Methodologies of monitoring water erosion during various
phases of the hydrological cycle. Abridged Proc of the Int. Symp. on Water
Erosion, UNESCO, Paris, pp. 711- 85.
Snishchenko,B.F. (1992) ‘Multipurpose studies of erosion and sedimentationin the
Upper Ob Basin’. Prot. of the Int. Symp on Erosion and Sediment Transport
Monitoring Progammes in River Basins. Oslo, IAHS-NVE, Publication N2 10,
pp. 405 - 411.
Snishchenko,B.F. (1994) ‘Economic efficiency of hydrological validation of water
projects in river channels and floodplains’. In: Abstr. of paper of conferenceon
the economicbenefits of meteorologicaland hydrological services.WMO/TD - N
630, Geneva,pp. 25-27.
Snishchenko,B.F. (1994) ‘Methodolody for the Inventory of Channel Processesfor
Water Projects’. In: Proc. of the Int. Symp. “East-West,North-South Encounter
on the State-of-art in River EngineeringMethods and Design Philosophies”. St.
Petersburg,vol 1, pp. 102 - 107.
Snishchenko, B.F.and KOPA~IANI, Z.D.( 1994) ‘Hydromorphological consept of
Channel Process and its Applications’. In: Proc of the Int. Symp. “East-West,
North-South Encounter on the State-of-art in River Engineering Methods and
Design Philosophies”.St. Petersburg,~012, pp. 9 - 23.
Snishchenko,B.F.and Meserlians, G.G.( 1994) ‘Hydraulic Structure of the Flow and
SedimentTransport in River Pits’. In: Proc. of the Int. Symp. “East-West,North-
South Encounter on the State-of-art in River EngineeringMethods and Design
Philosophies”.St. Petersburg,vol 1, pp. 334 - 345.
Snishchenko,D.V.( 1994) ‘Airborne data form river-bed deformationsstudy’. In: Proc.
of the Int. Symp. “East-West,North-South Encounter on the State-of-art in River
EngineeringMethods and Design Philosophies”.St. Petersburg,vol 1, pp. 127 -
132.
Snishchenko,D.V.and Snishchenko,B.F. (1986) ‘Establishmentof the types of river
channelsand components of their morpholody from the data of airborne and
satelhte surveys’. “Meteorologiya i gidrologiya”, N 11, pp.. 117 - 119. (in
Russan).
Snishchenko,D.V.and Snishchenko,B.F.( 1991a) ‘Airborne and satelliteinformation as
a source of data on long-term changesin water availability in rivers’. In: St.
Petersburg,Gidrometeoizdat, pp. 187 - 195.
Snishchenko, D.V.and Snishchenko, B.F.(1991b) ‘Assessment of present
hydromorphologica1 state of sma11 rivers from the satellitesurveysinformation’.

266
 In: “Sovremennoye sostoianie malykh rek SSSR i puti ikh ispolzovania,
sokhraneniai vasstanovlenia”Issue 2 Leningrad, Gidrometeoizdat,pp. 85 95 (in
Russian).
Starhel, H and Hotinskiy, N.(1985) ‘Prirodnie antropogennie rubeji galogena na
territotii centra Evropeiskoy chasteUSSR i Polshi’. IZV AN USSR, Ser. geogr.,
N 5, pp. 27 - 39. (in Russian).

267
Methodology for the inventory of channel
processes for water projects
B. F. Snishchenko State Hydrological Institute, StPetersburg, Russia

The discussedmethodology has been developed at the State Hydrological Institute

(SHI); it is describedin monographs,articles and in standarddocumentsand it is widely
used for different projects in Russia (Snishchenko, 1980; Kondratiev et al., 1982;
Anon., 1981, 1983, 1985).
This methodology envisagesa definite content and sequenceof the procedure for
the inventory of channel processesfor any water project in a river. It is based on the
hydromorphological theory of the channel processes,developedat the SHI (Kondratiev
et al., 1982;). The concept of this theory is presentedin the paper of B. Snishchenkoand
Z.Kopaliani, included in these proceedingsof this symposium.This theory describesall
the eventsin the channel from the viewpoint of the systemapproach widely applied in
natural sciences.A theoretical validation of this methodology, i. e. a systemtheory of
the channel processes,makes it possibleto create the structure of this methodology on
the systemapproachprinciples as well. In fact, the methodology should be basedon the
systemof interrelationsbetween hydraulic engineeringstructuresand channel processes
characteristics.
Forecast of any hydrological regime component is an important aspect for the
validation of any water project, A forecast of the channel processesis a more specific
notion than the inventory of channel processes.But in any case a determination of
channelevolution should bear a systemcharacter.
Information on the objects to be analysed is an essential component of the
methodology. It is quite natural, that the use of information on the inventory of channel
processesshould be also basedon systemapproach.
Thus, four interrelated components (subsystems)of researchand applied activity
of a hydrologist are subdivided into a system of methods for channel processes
inventory for different water projects:
I. Subsystemfor inventory of hydraulic engineeringstructures.
II. Subsystemsfor inventory (prediction) of channelprocesses.
III. Subsystemsfor inventory of predicted componentsof channelprocesses.
IV. Subsystemsfor information inventory.
The first subsystem. There are about -,3 000,000 rivers in Russia,with dozens of
millions structures in these rivers. To construct the 1st subsystem, all hydraulic
structureswere classifiedaccording to their effect on the characteristicsof factors which
determinethe channel processes(river runoff, sedimentdischarge,conditions restricting
a free channel evolution) (Table 1). Regular combinations of factors characteristics
create definite types and parameters of channel forms (microforms, mesoforms,
macroforms, megaforms)- seethe above paper of Snishchenkoand Kopaliani. ’

268
 Table 1

ANTROPOGEMC ACTIVITY IN RIVER BASINS

AGROTECHNICAL MEASURES ON HYDRAULIC STRUCTURES AND MEASURES

RIVER BASINS IN RIVERS

Category I

Dams

Power Transmission
Bridges
Communication
Lines, across Rivers

Water Diversion from a

- River in Case of Water
Transfer

Dykes, Fills for roads

Numerous Alluvium Pits

in the Channel and in the
Floodplain
i
Bridge Supports

According to this principle, all hydraulic structuresare classifiedinto active and

passive. The construction of active structures causes changes in the factors which
determinechannelprocesses,while passivestructuresdo not produce this effect.

269
 The active structures are sibdivided into two categories, The construction of
structuresof categoryI may lead to unidirectional changein any of the three factors and
to retransformation of the channel processesat each structural level (microforms,
mesoforms, macroforms or megaforms). This effect may be also produced by
agrotechnicalpracticesin the basin (Table 1).
Changesin channel processesat the level of megaforms and macroforms can
affect the stability of the structuresof this category.
The construction of structuresof category II leads to the local changesin some
characteristicsof the determining factors, This change, however, does not affect a
fundamentalmegaformsand macroformsredistribution, but concernschannelevolution
at the level of meso- and microforms. The stability of such structuresalways dependson
the mega- and macroformsredistribution, and sometimeson the dynamicsof meso- and
microforms, if channelproceesesinventory is inadequite.
Passive structures, size, location and operational life do not affect the flow
hydraulics, regime of sedimentstransportation and river-bed deformationsover a long
distance.In case of intensive operation of passivestructures (e. g., water diversions)
their effect may be similar to that causedby active structures.Passivestructures are
subjectto the influence of channelforms at each structural level and a reliable operation
of these structures requires the inventory (forecasting) of channel deformations (by
changingthe structuredesign,and location, river training, etc.).
The second subsystem. First, let us determinethe term “forecast”. The forecast
of the channel process means time-spacecomputation of any change in the channel
morphology basedon the knowledge of the laws of channel evolution under particular
physiographic features of a river/canal. Table 2 shows basic features and types of
channelforecasts,which may be used for up-date application.Table 2 is general,but its
nature is that of principle.

Table 2. Classificationof river-bedforecast.

Featuresof Forecasts Types of forecasts

GeneralPurpose Scientific and Cognitive
Engineering
Nature-protective
WatercourseGenesis Forecastsfor Rivers
Forecastsfor Canals
Type of Watercourse(River) Forecastsfor Plain Rivers
Forecastsfor Mointain Rivers
Forecastsfor Rivers in Specific Conditions(permafrostzone,Karst zone)
Scheme for Interrelations Forecastsin Caseof Hydraulic Structuresof CategoryI
between Hydraulic Structures Forecastsin Caseof Hydraulic Structuresof CategoryII
and River-BedDeformations Forecastsin Caseof PassiveHydraulic Structures
Forecastson Urban&d River Reaches
ForecastsReriods Forecastsof Stagesof ChannelForms Evolution
Forecastsof Cyclesof ChannelForms Evolution
ForecastsMethods Hydromorphologic
Hydraulic-Morphometric
Modeling

It follows from Table 2 that the channelforecast in some particular reach should
be selectedfor particular water project and it should be studied for particular water
project only, i. e, with the account of all kinds of forecasts. The time factor of the
forecastshould be noted. From the viewpoint of the channelprocesses,time division of

270
forecastsrequiresthe forecastsubdividionnot into different time intervals in its absolute
expression(as acceptedfor hydrological forecasts)but into the period of a complete
genetic cycle of the channel form evolution (e. g., river bend) or of its portion (some
particular stageof the river bend evolution).
The third subsystem. This subsystemreflects the basic features of the channel
processwhich should be primarily taken into account for predicting this process.It is
convenient to express these features through the predicted components and
characteristicsof the channel process (Table 3). The major feature of the channel
process,i. e. its discreteness,reflects the channelforms separationinto six types in this
subsystem(block I, Table 3). The types of forecastsgiven in Table 2, should be applied
to appropriatechannelforms, noted in block I. Separationof block II (Deformations) is
explainedby the fact that river channelscan be deformated in different ways: due to
unidirectional, irreversible deformations; due to sign-variablereversible deformations;
and due to joint effect of the two deformation kinds. Block III in Table 3 takes into
account interrelationsbetween discrete channel forms: i. e. effect of any form on the
other form may occur at the same level or at different structural levels of the channel
process.

Table 3.

BASIC PREDICTED ELEMENTS AND CHARACTERISTICS OF

RIVER-BED DEFORMATIONS

I CHANNEL FORMS ~ TI DEFORMATIONS 1 III INTERRELATIONS 1

’ , A 1 ;;=EN;NNEL ~
I

-L - 1 1

6 7

al
3
2 ( -0
z
.-iii
8 2
s
z 0”
-

The fourth subsystem.Determinationof the optimum amount of information for

the inventory of the channel processis a very important section in the methodology.
Unfortunately, federal hydrological institutions do not get standard information on
channel and floodplain deformations. Therefore, the required information should be
collectedfor eachparticular project. This imposesspecialeconomic and methodological
requirementsfor such information. The main requirement is its correspondenceto the
above three subsystems.When the type of the hydraulic structure is determined,when

271
the relation betweenthis structure and the determining factors of the channelprocessis
estimated,when the type of the forecast is selectedand forecastedcharacteristicsare
determined,it is possibleto establishthe scope and volume of information and methods
for its collection and processing.
The briefly presentedmethodology for the inventory of channel processesfor
different water project in rivers of the former USSR made it possibleto solve applied
problemsfor individual hydraulic structuresand for categoriesof similar structures.This
methodology also provided to formulate a correct problem for a developmentof river-
bed deformations theory and for a development of particular types of forecasts and
computations. Nevertheless, a number of applied and theoretical aspects of this
methodologyrequire improvementand long-term efforts,
According to this methodology assessmentof the channel processwas made at
the State Hydrological Institute for the construction and operatian of all projects given
in Table 1. Let us take a case study. It cancernsa hydromorphological validation for .
repairing oil and gaspipelinesacrossthe Volga river. The river bottom on the reach of
these pipelineswas eroded during the spring snowmelt flood in 1991; the pipes were
exposedand one of the pipes, i.e. the gas pipeline, was destroyedby w’k- mv,Q the
water flow and exploded. Appropriate ripraps of stone bankets were made on the
exposedreaches of the oil pipelinesup to 6 m high (Fiq. 1). It was decidedto replacethe
broken duplicatepipe by a new one at greater depth.
In accordancewith the methodology, the following schemewas accepted and
introduced into practice.
1. Determination of the Structure Type
The pipeline was determinedas a passivestructure, but the protective banketson
the pipeswere determinedas active structuresof categaryII (Table 1).
2. Assessmentof the River Effect.
Three types of discretechannelmorphological formations affect the pipelines,i.e.
micro-forms meso-forms and macro-forms; the pipes are placed in the zone of
resersible(sign-variable) and irreversible (uni-directed, downstream the power plant)
channelchanges,as well as in the zone of interrelationsof the above three types of the
channelforms (Table 3).
The stone banket causeschangesin the local characteristisof water flow and the
channel,which may disturb the pipeline stability.
3. Selection of Types of Channel Forecasts
In accordancewith iteme 1 and 2 and Table 2, the forecastshould be hydraulic, it
should be made for natural lowland rivers and for the rivers with runoff control, it
should take into account the peculiaritiesof passivestructuresand of the structuresof
categoryII; the predictedperiod should take into accountthe ratio of the designtime of
pipeline operation to the time of the three types of channel forms evolution noted in
item 1. The solution of the problams requires the application of the hydraulic-
morphologicalmethod of the forecast.
4. Selection of Information
The amount and details of information required for a forecast should follow the
requirementspresentin items 1,2 and 3, above.
As a result of the above schemerealization, and with the account of economic
capacity of Customers,the State Hydrological Institute (WI) suggestthe following
order of the hydraulic engineering de cistions to improve the reliability of pipelines
under operation.

272
 Firstly, it was proposedto fill the erodedpits with a specialground and reducethe
height of stone bankets above the oil pipelines (Fig. 1). These works were supervised
continuosly by the field tram from the SHI before the spring snowmelt flood. Control
measurementsof channel depths the flood proved a high efficiency of the proposed
measures.
As long-term measuresto be taken, it was proposed to re-install the broken
pipeline to the depth in such a way the above three types of the channel forms would
not touch the upper wall of the pipeline operation (30 years) (Fig.2). At presentposition
of the pipelines the channel changes occur around the pipelines often causing the
pipeline uncovering (Fig.3).
A method of horizontal directional drilling under the Volga channelis considered
as another variant of long-term engineeringmeasure;in this casethe pipeline would be
placedinto the drilled hole. The SHI also contributesto a Hydrological validatin of this
project.

Bibliography
SnishchenkoB. F. (1980). Problem of forecasting channel processand ways to solve
this problem. Meteorologiya i gidrologiya, No. 11, p. 71 - 78 (in Russian).
Anon., (198 1). Recommendationson the location and design of sites for waste water
dischargeto rivers. Moscow, Stroiizdat, 224 pp (in Russian).
Anon. (1983). Recommendationson the inventory of river-bed deformationsfor water
projects in the zone of Baikal - Amur railway construction. Leningrad,
Gidrometeoizdat,72 pp (in Russian).
Anon. (1985). Inventory of deformationsin river channelsand in the shoresof lakes in
the zone of underwater pipelines (oil and gas pipelines). VSN 163 - 83.
Minneftegazstroy.Leningrad, Gidrometeoizdat,144 pp (in Russian).
Kondratiev N. E., Popov I. V., Snishchenko B. F. (1982). Principles for
hydromorphologicaltheory of channelprocess.Leningrad, Gidrometeoizdat,272
pp (in Russian).

273
 m
-10,

Volga ~ , :
i

-40' I '
50 100 150 200 250 300 350 400 450 500 550 m
29-30.04.94 03.05.95
21.05.94 12,13.05.95
22.08,06.09.94 - - - - 16.05.95
---- 17-18.11.94 - - - - 2-6.07.95

Fig. I Combined longitudinnl profile of the river bottom on the reach of oil and gas pipelines
across the Volga

274
 -30

3
ul -40

-50
100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 m
1985 15.03.93 -.- ._.------ 06.09.94 - - - 21-25.08.94
09.91 --- 10.05.94 - - - 17-18.11.94 - - - 06.05.95
.-. --.--- 26-30.".92 -- 28-30.01.93 - - - - 21.05.94 - - - - 16.0595

IGg.2 Combined cross-section of the I,hlgcl river nt the site qfmnin oil pipeline crossing
 m

left bank water level -10 m

-10 --.+-.---

---l--- >
-20

--
-30

-40
w 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 m
4
cn
1985 23.06.93 - - - 21.05.94 - - - 17-18.11.94
12.07.91 10-11.05.94 .-- -_- -.- 21-25.08.94 - - - - 06.05.95
-.- 26-30.11.92 - - - 16.05.94 - - - 06.09.94 - - - - 16.05.95
---- 2-6.07.95
Changes in sediment transport and river
engineering methods: case study of the river
Drau in Austria
by H.M. Habersack and H.P. Nacbtnebel

1. Introduction

1.1 Human Impacts

The impacts of humans on river systemshave considerablyincreasedduring the last century,

so that today only few rivers can be termed as “natural rivers”. Human activities associated
with navigation, hydroelectric power generation,water supply, waste managementand urban
development have had a major impact on many river systems,primarily due to alteration of
regional or local hydrological conditions (Bellamy et al., 1992).
The consequenceof anthropogenic actions and associatedmanagementobjectives are
typically manifest through initial loss of habitats and population diversity (Imhof et al., 1990,
1991), followed by larger-scale, accelerated, and sometimes longterm geomorphological
adjustmentsof the river system.

1.2 Sediment transport and river engineering methods

Since aggradation and degradation problems due to anthropogenic measuresoccur, sediment

transport takes a more central role in river engineering methods and design philosophies..
Proper sediment management has become a key issue in the development of the basins of
sediment laden rivers, and also of rivers with moderatesedimentcontent (Yuquian, 1992).

1.3 Aim of the paper

The aim of the paper is to analyse the developments of sediment transport and river
engineering methods, as reflected in the casestudy of a large Austrian gravel-bed river.
Changes of sediment transport are often caused by engineering measures,that were
implemented in the course of the last century. The results of those changes have led to
degradation problems in many Alpine rivers (Jaggi, 1992; Hunziger, 1991; Nachtnebel and
Habersack, 1993).
Owing to the developments in sediment transport research, the limits of “classical
engineering methods” have to be discussed,in order to determine whether new approaches
should replace the old ones.

277
1. Introduction

1.1 Human Impacts

The impacts of humans on river systems have considerably increased during the last century,
such that today only few rivers can be termed as “natural rivers”. Human activities associated
with navigation, hydroelectric power generation, water supply, waste management and urban
development have had a major impact on many river systems, primarily due to alteration of
regional or local hydrological conditions (Bellamy et al., 1992).
The consequenceof anthropogenic actions and associated management objectives are
typically manifest through initial loss of habitats and population diversity (hnhof et al., 1990,
1991), followed by larger-scale, accelerated, and perhaps longerterm geomorphological
adjustmentsof the river system.

1.2 Sediment transport and river engineering methods

Sediment transport is taking a more central rule in river engineering methods and design
philosophies since aggradation and degradation problems due to anthropogeneticmeasuresoccur.
Proper sediment managementhasbecome a key issue in the developmentof the basins of sediment
laden rivers and also of rivers with moderate sediment content ( Yuquian, 1992).

1.3 Aim of the paper

The aim of this paper is to analyse the developmentsof sedimenttransport and river engineering
methods, regarding a casestudy of a large Austrian gravel-bed river.
Changes of sediment transport are often caused by engineering measures, that were
realised during the last century. The result of those changeshave lead to degradation problems in
many Alpine rivers (Jaggi, 1992, Hunziger, 1991, Nachtnebel & Habersack, 1993).
Due to the developments in sediment transport research the limits of “classical
engineeringmethods”have to be discussedto determine whether new approachesshould replace
the old ones.

278
2. Study reach

2.1 Location of the study reach

‘0 l), which enters Austria in Eastern Tyrol (coming
The study reach is part of the river Drau (Ft,.
from Italy) and flows to Slowenia. There are two dominant geological units in the catchment.
which are segeratedby the Drau :
1) central Alps in the North (granite, gneissj; and
2) lower limestone Alps.
Concerning sediment transport and especially bed load supply the Central Alps in the
Norm (including the nighest mountain in Austria and some glaciered areas) supply the most
resistant sediment. Thus. they are most important for the study reach of the Drau. Therefore, in
comparison to the other tributaries, the river Isel draining a large area in the central Alpine region
has the greatestinfluence on the sediment regime.

Cet~trul Alps
Isel
Lienz. Oh*aubwz Sachsenburg
:
1 .. .., DRAU .,.:

: gauging stat& study reach

Eastern Tyrol ,’
Carinthia
‘.. _
, Limestone
_,,_ .I’. .’ Alps
,_ .. . .. .’”

ITALfY

studyreach

Fig. 2.1 Study rruch at the river Drau in Austria.

279
2.2 Hydrological, hydraulic conditions and settlements
The hydrological data for the Drau in the study reach are given in Table 2.1.

Table 2.1 Hydrological data for 3 selectedsectionsof the Drau

gauging stations Lien2 Oberdrauburg Sachsenburg
location (km) 0 15 57
catchmentarea (km2) 1.876,2 2.112 2.561,4
slope (m m- l) 0,0025 0,002 0,0015
channelwidth (m) 30 30-40 40 - 50
averageflow (my/s) 55 65 76
HQm HQ I nn (m3/s) 632.7; 760,8 706.4; 854,7 840,9; 1029,4

The distribution of the seasonaldischargeis dominatedby the contribution from glaciers,

with minimum in winter and a maximum in June/July. Discharges above 300 rn3 s-’ cause
flooding, covering the whole valley, which is densily populated in several reaches.The villages
are mostly situated on the fans of the tributaries, but more and more spread in the valley itself.
The economicaldevelopmentof the Drau valley hasbeen closely related to river engineeringflood
control measures.which started at the end of the 19th century.

2.3 Reasonsfor the initiation of the project

Adverse ecological developmentssuch as lack of dynamic gravel bars, decreasingground water
levels and reduction of “Aue” areas as well as interruptions of the river continuum necessitated
the inception of an interdisciplinary river managementproject which started in 1991. Within this
project river managementstrategiesfor the next centuries should be suggestedfor a length of 60
km, beginning at Lienz in EasternTyrol and ending at Sachsenburgin Carinthia.

280
3. Data base

3.1 Elements of the data base

In order to predict future river developmentsand to plan measuresfor a relatively long river reach
of 60 km a sufficiently detailed and large data basis is requested.The Drau project has allowed
the analysis of historical developments of sediment transport and river engineering methods.
Surveys in shorter time intervals and more detailed measurementswould be desireable.
The data baseincludes the following elements:
Cl maps beginning at 1820
0 annual low water measurements
0 annually repeatedcross sectional measurementssince 1991
C! data (projects, events ..) from water authorities of the Drau and its tributaries
Cl data concerning gravel mining; and
0 sediment size distribution of subsurfacematerial

3:2 Analysis methods

Maps from 1824 and 1880, that show the situation of the river reach before regulation measures
were realised, were analysed. Annual low water measurementsbegan in the late nineties of the
last century, these were undertaken since 1931 with exeption of World War II and 1965-1967,
when large floods occurred, every year in the Carinthian part of the study reach. Low water
(November till February) measurementsof the water level at 200 m steps along the whole reach
show the temporal developmentof the river bed.
These results can be compared to annually repeated measurementsof 14 cross sections,
undertaken since 1990 and regression analysis of low water level developmentsat the gauging
stations and river bed developmentsat bridge piers. Gravel mining data and results of sediment
analysisare compared with low water measurementsto identify the different influences.

281
4. Results

4.1 Analysis of historical maps

The historical maps show that the Drau previous to 1890 in long reaches in Austria was
characterised by a large sediment supply from the Alpine sources. This resulted in a braided,
aggrading channel system. Hence, one hundred years ago engineers were required to find
solutions that would reduce aggradation and, particularly, the risk of flooding. At that time the
reduction of channel width and bank protection building were regarded as the acceptablesolution
to aggradation.
As a consequencefrom 1890 to 1930 the river was channelizedwith a constant, uniform
channel width of 30 - 50 m and a reduction of the previous braided area of 34 % (reduction was
calculated from an overlay of an old map and planform informations from projects). As bed
aggradation ceasedand flooding was reduced the engineeringmethods seemedto be successful.
After 1930 and especially after 1965/66, where high floods occurred, further regulations and
bank protection measures,but also torrent control structures in 19 small, sediment transporting
tributaries were erected. These measures, the realisation of water power plants in the catchment
and gravel mining affected sediment transport and consequentlyriver morphology.

4.2 Analysis of annual low water level measurements

Analysis of annual low water level measurementsfor the period 1931- 1991 shows that the river
bed degraded at a mean rate of 1,O cm/year. Maximum degradation was 1,53 m (Fig. 2). If there
would be a general degradation of I,0 cm/ year for the whole reach, the suggestionsfor further
engineeringmethods would be not difftcult. However, the high temporal and spatial variability of
the river bed developmentwith interactions between sediment supply, different geological bedload
material, gravel mining asks for local specific measures.Volumetric comparisonsbetweengravel
mining and variations of the bed level showed that 60 % of the changeswere caused by gravel
mining. In principle, it can be concluded, that gravel mining should be prohibited. At few
locations a surplus in sediment deposits after major floods would be available for gravel mining.
In Fig. 2 an obvious split of the Drau river in the Carinthian section into 5 separate,
significantly different river reachescan be observed, in terms of the averagedegradation rate of
the whole reach (- 0,6 m) compared to average values of short reaches.Reaches 1, 3, 5 have
lower mean degradation rates (-0,32 m, -0.35 m, -0,28 m), whereas reaches 2 and 4 are
characterisedby higher mean values (-0,95 m, -0,87 m). The engineeringmethods applied at the
Drau hitherto did not regard these spatial and temporal variabilities, this is why the whole reach
was designed as a uniform channel. In order to stop the high degradation rates in the reaches2
and 4, other engineeringmethods will have to be applied (e.g. increasing the width again) than in
reaches 1, 3, 5, where at some sections the increase of width might cause further aggradation,
that cannot be tolerated, becauseof the necessityto keep flood control.

282
4.3 Cross section measurements
Comparison of repeatedcross section measurementsfor the years 1990 and 1991 demonstrates,
that 66.9 % of the whole river reach degradedand 33,1 % aggraded.This could be interpreted as
reduction of the degradationtendencyresulting from me low water level measurements,where 96
% degraded,only 3.6 % aggraded and 0.4 % maintained in equilibrium. But the tune period of
one year is too short in order to get representativeresults and 14 cross sectionsare not enough to
concludefor the whole river section.

L
F -0.2-
c
z
-&a .‘-
5
5
3
g -0.6--
.-c
g -0.8-
20
.i‘z -
3 -l-

-1.2-

-1.4- reach 5

-1.6 i I I I I I I I I I 1
0 10 20 30 40 SO
5 15 25 35 45
Drau river (km)

- comparison 1991-31 m-mm-*

averagevalue c

Fig.4.1 Spatial variability of bed level developments at the river Drau due to sediment
transport combined with degradation and aggradation

As a consequencelow water level and cross section measurementshave to be performed

as part of a future monitoring program to obtain data about the further river bed development.
Regression analysis of low water level developmentsat the three gauging stations correspond
with the results of the low water level measurementsat steps of 200 m as well as the changesof
bed level observednear bridge piers. Hence the whole analysisis basedon high accuracy.

283
4.4 Results of mathematical calculations
Besidesthese measurementsmathematical calculations and the application of simulation models
allow to analyse the present situation and to predict developmentsof the river bed. For me river
Drau the bedload transport rate was calculatedwith the formula of E. Meyer-Peter and R. Miiller
(1949). In Eastern Tyrol there exist leveeson both sides of the Drau, which allow a flood control
for HQlOO, whereas in Carinthia already fltxxLs of HQl to HQ5 inundate the whole valley. Of
course, these different engineering methods affect the sediment transport rates (Fig. 3). In
Carinthia the transport rate up to 330 m3/s is higher than that in EasternTyrol, which is due to a
greater catchment area and discharge. The transport rate is then rapidly stabilized at 600 m3/s
becauseof the large floodplain areas in Carinthia. For me transport rate in Eastern Tyrol this
point is reachedat a dischargeof 900 m3/s, when leveesare alreadyflooded.
The result of this calculation shows, that it is very important for the design of flood
control structures to consider the effect on the sedimenttransport capacity of the river. As the
river Isel is the most important tributary for sedimentsupply, wrong river engineeringmethodsat
the Drau in Eastern Tyrol could lead to a lack of sedimentsin Carinthia. where problems with
degradation already occur (Fig.2). The higher sediment transport rate in Tyrol for discharge
exceeding330 m3/s means,that bedload derived from the Isel should be transported to Carinthia
to be sufficient for stabilisation. Below 330 m3/s bedload from me smaller tributaries in
Carinthia will be also required for a reduction of the degradationrate.
Grain size analysis of the subsurface (volumetric samples) has demonstrated that in
reacheswith large degradation the mean diameter is less man that of reachesin equilibrium or
with slight degradation. For some reasons this is due to small tributaries transporting in some
reachescoarsermaterial to the Drau, which has createdan armour layer. In reach 2 the dm of the
subsurfacematerial varied between 22 mm and 26 mm, in reach 3 between32 mm and 33 mm, in
reach 4 cit. 20 mm and in reach 5 ca. 33 mm. The application of the formulas of Gessler (1965)
and GiJnter (1971) allowed to demonstratethat in the whole reach of the river Drau exists a
tendencyfor a strong armouring layer.

250

400 500 600

discharge (m3 s-l)

The existanceof dynamic bars, which are very important for biological aspects,is limited
to very short sectionsof the river Drau. Mostly there are just point bars at bends and no alternate
bars at straight reaches,which was also verified through the application of theoretical criteria
(Jaggi, 1983).

284
5. Summary and conclusions

Low water level measurementsat the River Drau in Austria showed for the period 1931-1991 a
degradation rate of 1.0 cm/year with great spatial and temporal variablility (sections with
different degradation values and sediment size distribution), mainly caused by engineering
measures.Sixty percent of the volumetric changeswere caused by gravel mining. Variability in
degradation and aggradation processes might be the reaseon why gravel mining cannot be
completely halted, especially in relation to flood control and river bed developments.Various
flood control structures have been realised, affecting sediment transport in different ways.
Instability of bank protection measures and ecological problems in connection with lack of
dynamic gravel bars lead to the following conclusionsincluding a different design approach:

1) there exists a great temporal and spatial variability in sediment transport rate and
related degradation and agradation;
2) sediment transport is a key variable for engineering methods in Alpine rivers;
3) only interdisciplinary teamscan achievesound solutions which satisfy economical and
ecological needs,especially when long river reachesare considered;
4) not one schematicproject for the whole reach but a variety of smaller and larger
measureswith different methods is necessary;and
5) a step by step realisation as well as accompanyingmonitoring programs, including the
whole catchment,have to be taken into considerationin order to improve the
hydrological and ecological functions of a fluvial system.

5.1 Acknowledgements
We acknowledgethe Ministery of Agriculture for financial support of the whole interdisciplinary
project. Further we thank Jonathan B. Laronne, Ben Gurion University of the Negev in Israel for
scientific discussionsand cooperation.

285
Bibliography

BELLAMY, J., BEEBE, J.T., SAUNDERSON, H. C., IMHOF, J. (1992) ‘River morphology,
sedimentsand fish habitats’. Erosion and Sediment Transport Monitoring Programs in River
Basins. IAHS Publ. No. 210.
GESSLER, J. (1965) ‘Der Geschiebetriebbeginnbei Mischungen untersucht an namrlichen
Abpfhisterungserscheinungenin KanMen’. Mitt. der Versuchsanstalt fiir Wasser und
Erdbau, ETH Ziirich, No. 69.
GUNTER, A. (1971) ‘Die kritische mittlere Sohlenschubspannungbei Geschiebemischungen
unter Beriicksichtigung der Deckschichtbildung und der turbulenzbedingten
Sohlenschubspannungsschwankungen’ Mitt.
. Nr. 3 der VAW der ETH Zurich.
HUNZIGER, R.( 1991): ‘FluBmorphologie. Modelle in der Geomorphologie - Beispiele aus der
Schweiz’. Fribourg.
IMHOF, J.G.A., PLANCK, F.M., JOHNSON, F.M., HALYK (1990) ‘Watershed urbanization
and managing stream habitat for fish’. 56th N. Am. Wildlife and Natural Res. Conf., 269-
285.
IMHOF. J.G.A., REIGER, H.A., PLANCK, R. J., SCI-IRIMPF, A. (1991) ‘Urbanization and
stream habitats for fish- a synoptic review and perspective’. Proc. 1st World Fisheries Conf.
Athens.
JAGGI, M. (1983) ‘Alternierende Kiesbanke’. Mittleiung VAW Ziirich 62.
JAGGI, M. (1992) ‘Sedimenthaushaltund Stabilitat on FluBbauten’. Mitteilung VAW Ztirich
119.
MEYER-PETER, E., MULLER, R. (1949) ‘Eine Formel zur Berechnungdes Geschiebetriebs’.
Mitteilung aus der Versuchsanstaltfi.ir Wasserbauund Erdbau an der ETH-Zurich Nr. 16.
NACHTNEBEL, H.P., HABERSACK, H. (1993) ‘Erhebung von gewbsermorphologischen
Daten - Aufwand im Verhahnis zur Aussage’. Stand der Technik im Landschaftswasserbau.
14. Seminar Landschaftswasserbauan der TechnischenUniversitat Wien.
YUQIAN, L. (1992): ‘The design and operation of sedimenttransport measurementprogrammes
in river basins: the Chinese experience’. Erosion and Sediment Transport Monitoring
Programs in River Basins. IAHS Publ. No. 210.

286
Improvement of navigable width in river
bendsby periodic dredging;Casestudy of
the river Waal, The Netherlands
Marco Taal
RIJKSWATERSTAAT
Institute for Inland Water Managementand Waste Water Treatment
(RIZA), P.O.Box 9070, 6800 ED Arnhem,The Netherlands

Hermjan Bameveld
DELFT HYDRAULICS,
P.O. Box 152, 8300 AD Emmeloord, The Netherlands

Ton Swanenberg
RIJKSWATERSTAAT, Directorate East Netherlands
P.O.Box 9070,680O ED Arnhem, The Netherlands

Abstract

Several bends in the River Waal in the Netherlands form bottle-necks in this vital shipping route
because of insufficient navigable width. Dredging of sand from the shallow inner bends and
dumping of the dredged material in the opposite outer bend appears to be an attractive measure
to increase the navigable width.
Initial calculations with the two-dimensional morphological model SEDREDGE were carried out
so as to study the feasibility of the intended dredging measure. Based on the results of these
calculations a prototype dredging experiment was carried out. In two dredging periods 300,ODD
m3 of sediment was dredged. An intensive monitoring campaign was carried out so as to
evaluate the experiment. In addition, analytical and computational tools were applied in the
evaluation.
Initial calculations, the prototype experiment and its evaluation are discussed in this paper.

Introduction

The river Waal is the main branch of the river Rhine in the Netherlandsand a
very important chain in the international transport between the port of
Rotterdam and Germany. Eachyear approximately 170,000 vesselspassthe
Dutch-German border, carrying 150400,000 tons of goods.

For the river Waal the Dutch governmenthas set the highestrequirements
with regard to the quality of transport and bottlenecksshould be tackled with

287
priority. Upstream of the city of Nijmegen, four bends are located which are a
bottleneck from a nautical point of view (seeFigure 1).
The navigable width in these narrow bends does not meet the safety require-
ments, especially not in the dry season. This problem launched the study
‘Bend improvementsriver Waal’. The objective of this study is to guaranteea
minimum navigable width of 170 meters and a depth of 2.50 meter at an
Agreed Low Discharge Level (OLR). The corresponding discharge has a
probability of exceedanceof about 95% and mainly occurs during the low
water seasont?om April to October.

A few mitigation measures, restrictied to the main channel, have been

considered. One of the measuresis dredging the point bar periodically on a
yearly basis. This measurerequires yearly maintenancedredging becausethe
lowered point bar will recover to its original equilibrium cross-profile.
Conform Government policy, the dredged material is dumped in the deep
pools of the opposite outer bend, in between the groynes. This policy pre-
scribes that all dredged sedimentsshould be retained in the main channel due
to the autonomous morphological development of the river Waal, which
shows a continous lowering of the river bed.

Based on promising results with a 2-D morphological model it was decidedto

start a prototype experiment in two bends near the village of Hulhuizen. The
aim of the prototype experiment is to clarify the working of this measureand
to take away uncertainties in both implementation and realisation. An evalu-
ation of the experiment has been carried out by comparing the measured
morphological development of the river bed with simulations performed with
both an analytical recovery relationship and a numerical morphological model.

Initial morphological computations

The two-dimensional morphological model SEDREDGE has been applied to

simulate the recovery of the cross-sectionalprofiles in river bends of the
River Waal in the Netherlands. SEDREDGE has been developed by DELFT
HYDRAULICS. In the software package the river is modelled by two parallel
branches. The water movement in the model is based on a quasi-steady
approach. Exchange of water and sediment between the two branches is
calculated taking into account the secondary current and the influence of

288
 ,,,’
,:’
,;’
,,,. ._,.. I’
,;’
,:

NETHERLANDS ,,,’ ,.,’ GERMANY

~,,’
,,,’
Figure 1: The Upper Waal
transversebed slopes. A straight line through the calculated bed levels in the
two branches appearedto give a fair approach of the cross-sectionsin the
rivers. SEDREDGE enables the simulation of dredging activities and non-
erodible layers.

Simulations were carried out with SEDREDGE to investigate the impact of

dredging activities on the river morphology and navigable width. A repre-
sentative hydrograph for the River Waal was applied as a boundary condition.
The results of the simulations showedthat dredging material in the inner bend
and dumping of the dredgedmaterial in the opposite outer bend:
could be an attractive alternative for improving the navigable width;
should not be concentratedin one or two locations along the point
bar, but should be equally spreadover the total bend;
- should be carried out during periods of low flow;
does not affect the morphology in the downstreamriver reach;
does only slightly increase the hydraulic roughness in the reach
causedby flatter transversalbed slopes.

Based on the initial results it was decided to start a prototype experiment in

two bends of the River Waal near the village of Hulhuizen. The upstream
bend is located between chainage 867.5 and 869.0 [km]. The downstream
bend is located betweenchainage869.0 and 871.5 [km] (see also Figure 3)

Dredging activities and measurements

Since the start of the prototype experiment in january 1992 two series of
dredging activities were carried out. The first period lasted t?om January to
March 1992 and the second from Septemberto December 1992. The total
amount of dredged sediment in the two bends was 106,000 and 194,000 m3
respectively.

Before, during and after the dredging activities soundings have been carried
out on a monthly basis. The measuredcross-sectionswere equally spacedwith
a distance of 25 meter. The soundingswere presentedas sounding charts with
contour lines and separate cross-sections.Plots of cross-sections, covering
several successivesoundings, gave a clear indication of the recovery of the
point bar.

290
10

*- . . . . . . (_.............,.._..... _ .
DLR~129
.’
.*’
if! I
6 _...... cp.. _.... ;. .

!ii
DLR (-2.50)
:.. ., I
....
4_ _ .......... :I .-I., .... -:: *.,r ::::.+:: . .
“a^,

\.-
k...
‘L-l
‘\ -.
-7
\ \
2- . .................... . ..:..._ .

o- . ........ ......... _.
,i,

-2
G 0 li0 ; D 240 3t 1 3 0

Figure 2: Soundings before and after dredging of the point bar

WIDTH
WI
As in the 2-D morphological model the river is modelled by two parallel
branches,the measurementswere also translated into average bed levels for
the left and right branch, covering the width between the river axis and the
river bank minus 30 meter.

Based on the soundingsthe following conclusionscould be drawn:

In the upstream river bend the point bar recovers much slower
compared to the downstream bend. This is probably due to the
influence of the bifurcation just upstream, which introduces a differ-
ent distribution of the flow comparedto normal developed river bend
flow.
A part of the bend forms a bottleneck as the cross-sectionalprofile
shows an additional elevation of the point bar on top of the more or
less sinusoidal shape.This reducesthe available navigation width.

. In the downstream river bend a quick recovery of the dredgedpoint

bar occurred betweenNovember 1992 and February 1993. This quick
recovery was caused by periods with high discharges and sediment
transport resulting in reducedtime scalesfor morphological processes.
The bottleneck in this bend shows a steep transversal slope resulting
in a reducednavigable width.

Model simulations

Simulations with an analytical recovery relation and the two-dimensional

morphological model SEDREDGE were carried out to study the applicability
of these tools for predictive purposes. Such tools would be very useful for
planning optimum and thus economical dredging activities.

For deriving the analytical recovery relation the following exponential relation
for the water depth development near the bank after dredging activities was
assumed:
(1)
H(t) = H(m) +lJ-I(O)-H(-)]e-’

292
 In which:
H(t) : water depth near the banks at time t;
H(O) : water depth at time t=O;
H(a) : equilibrium water depth near the bank;
T : time scale.

The time scale T appears to be the governing parameter in this relation.

Crosato (1990) analyzed the linearized time-dependent differential equation
for the bed developmentnear the bank, which reads:

Asa2H+@J-3)
as2 w w+--+--
2Tlx+h at=F(c)
s,&at
S,kw as iJH H h,*s a?H hoAs aH (2)

in which F(c) is a function basedon curvature terms and local parameters:

F(c) = ;hOB&(b-I)&$ +[Af(Q -(2-u)-- (b-l) k X Af@‘&) (3)

2 Awlz+T- w
where:
H : water depth near the bank;
L : adaptationlength of flow
C2h
& = -2 (4)
2g
adaptation length of bed-topographydeformations
As = fh0[$2f(0,J (5)
0
b exponent in the sedimenttransport formula;
for the Engelund & Hansenformula b=5;
ho averagewater depth;
so
k,
averagesedimenttransport (including pores) per unit

I- width;
wave number in transversaldirection = x/B;
curvature term

(J calibration coefficient (for secondaryflow convection,

293
-- -
 about 2~64);
A : weighing coefficient of spiral flow intensity;
l? : curvature of the stream-linesdue to the non-uniformity of
the flow field:
2 = -Ld” (7)
% as
The behaviour of the system is determined by the homogeneous part of
equation (2) which has the following solution:
(8)
H = fi,[sin(Ks)-efsin(Ks+~t)]+~[cos(Ks)-e’cos(Ks+~t)]

Based on the boundary condition H(x,O)=Othe following expression for the

time-scale T can be derived:

(9)

Simulation results

The prototype experiment has been simulated with the analytical recovery
relation and SEDREDGE. Not only the calculatedchangesin morphology, but
also the calculated navigable width was compared to measured values. For
calculating the navigable width the cross-sectionalprofile in the analytical
recovery relation was assumedto follow a sine-function. In SEDREDGE a
straight line through the calculatedbed level in the two parallel brancheswas
applied.

The schematization of the cross-sectionsin SEDREDGE and a comparison

between SEDREDGE results and measuredbed levels are shown in Figure 3.

294
 chonnei 1 channei 2

1 1.0
O.OlO
I,
11
I
12
I
13
I
14
I
15

- OISTANCE WI

LEFT BRANCH (CALCULATE01

----e-m LEFT SAANCH IMEASURE

RIGHT BRlNcH ICALCLA*lEOl

------. RIGHT BRANCH lMEASUREOI

Figure 3: Model simulations

Conclusions

From the soundings it appeared,among other things, that between km

867.725 and 868.075 the point bar was dredgedtoo deep and to near to the
river banks. Due to this a trench was dredgedwhich did not contribute to an
enlargementof the navigablewidth.

295
Conform to the expectations, it appearedfrom the soundings and the model
simulations that the largest morphological recovery occurs during high
discharges.During low dischargesthe sedimenttransport and the morphologi-
cal changesare small.

The development of the point bars in both up- and downstream bend can be
simulated by the analytical recovery relation and SEDREDGE to a reasonable
extend. However, the expected yearly amount of sediment transport should
therefore be increased considerably with a factor 2.6 and 2.0 respectively to
calibrate the models.

At several locations the calculated navigable width does agree less with the
measuredwidths and the widths as stated by the river supervisors.During and
after the first dredging period, at the beginning of the low water season,the
calculatednavigable width gave a pessimistic view, while during and after the
second dredging period, at the beginning of the high water season, the
calculatednavigable widths showed almost no changesin time and were 10 to
20 meter larger than the measuredwidths.

It is assumedthat the recovery of the point bars is influenced by processes

that are not simulated in detail by the used models. One can think of addi-
tional sediment supply from the dump locations for dredged sediment in the
deep pools of the outer bends between the groynes, collapse of banks, or
three-dimensional processes that have been neglected or linearized in the
mathematical equations. In particular the quick recovery of the downstream
point bar between November 1992 and February 1993 supports this assump-
tion. The calculated recovery of the upstreampoint bar appearsto be strongly
influenced by the way the upstreamboundary conditions, at the bifurcation of
the Rhine into Pannerdenschchannel and Waal, were modelled.

Studying the cross-sectional profiles in both up- and downstream bend

showed that especially in the bottlenecks of these bends the shape of these
profiles differ from the ones adopted by the analytical recovery relation and
SEDREDGE.

Based on the present evaluation it can not be concluded yet that the recovery
of the point bars, and consequentlythe decreaseof the navigable width, can
be simulated and predicted properly with the analytical recovery relation or

296
SEDREDGE. Calibration of the models with new soundings,especially during
high discharges,is therefore important.

It is advised to dredge in particular during the beginning of the low water

seasondue to the slow recovery of the point bar and the benefit of a suffi-
cient navigable width during the lowest discharges,that mainly occur at the
end of the low water season.
Dredging activities should be concentratedat the bottlenecks. It is expected
that in that casea maximum navigable width of 150 meter (at OLR-2.5m) can
be obtained. Enlarging this width would introduce a considerableincrease in
the volumes to be dredged and of the stretch over which dredging would be
necessary.

Reference

111 Rijkswaterstaat, Institute for Inland Water Management and Waste Water Treatment
(RIZA), 1994;
Evaluation dredging experiment Hulhuizen (report 94.059).

PI Rijkswaterstaat,Directorate Gelderland, 1993;

Bend improvements river Waal, Fase 1.

[31 DELFT HYDRAULICS,1993;

Evaluation dredging experiment Hulzuizen (report 41699)

[41 DELFT HYDRAULICS,1991;

Dredging upper Waai, SEDREDGE calculations (report Q1363)

I51 Crosato, A., 1990;

Simulation of meandering river processes,Communications on Hydraulic and Geotech-
nical Engineering, No. 90-3, Delft Technical Universiteit, Faculty of Civil Engineering.

I61 Taal, M.C., 1989;

Time-dependent near-bank beddeformation in meandering rivers, Delhi University of
Technology, Faculty of Civil Engineering.

297
Program for computation of channel
deformations downstream from dams: case
study of the Votkinsk hydropower plant on
the Kama river
A. B. Veksler, V, M. Donenberg, Y. L Manuilov, R. S. Frid B. E. VedeneevAll-
RussianResearchInstitute of Hydraulic Engineering (VNIIG), StPetersburg, Russia

Mathematical program

General
A system of equations for uni-dimensional flow with suspendedsedimentsis used as
the basis of method to compute channel deformations in the lower pool of the power
plant unit, due to liquid discharge control and accumulation of sediments in the
reservoir. In case of low concentrations of suspendedsediments,usually observed in
channelflows, this systemis reducedto equationsof Saint Venant for a flow of mixture
supplied with the sediment balance equation. The system is closed made by an
additional equation of relations between the rate of bottom deformation, eroding
capacity of the flow, sedimentsdischargeand conveyancecapacity of the flow.
If we assumethat the channel flow is slowly variable, and if we replace the real
hydrograph by a stepwise hydrograph with a permanent water discharge within each
step, the system of equations for unsteady water flow in erodible channel may be
reduced to a system of equations of quasi-steady(within each hydrograph step) non-
uniform flow and equation of channel deformation. This system may be presentedby
finite differencesand it can be solved by separatingdifferent processesfrom each other:
(1) water flow, and (2) channel deformations.
If the schemeof identification of indices for design channel reachesand division
of reaches over length is introduced (Fig. l), the finite-difference system of basic
equationsmay be presentedas follows:

(1)
(2)

298
 M = PskQw Qs24
At
-
(3)
Pgrk &I’ B’

Fig. 1. E+stem of specijication of indices for design reaches (a) and division of reaches Into steps over
the length (3).

The following symbolsconnectedwith the generalschematizationof the channel

and its transformationare used in the systemof equations(l)-(3) and in the schemeof
indicesgiven in Fig. 1.
Indicesk = I, 2, . . . . K indicategauge-lineswhere the initial channelgeometryis
specified(bottom marks Z, , and width B, of the channel,assumedto be rectangular
with fixed banks).Among gauge-lines1, 2, .. ., K there are such gauge-lines,for which
initial rating curvesfor dischargeQ and water level Y are specified.Thesegauge-lines
are separatedfrom each other by Y reaches(Fig. 1) where r is arbitrary, and includes
gauge-linesI and K by all means.

299
 During computationsthe reach lengths AX, = X,_, - X, are divided into steps
AX = X, - X, , where sliding indices 1 and 2 indicate inlet and outlet of the designstep
hx long, respectively.
Division of reachesover the lenght (discretization of the channel) can be made
manually or a special subprogramcan be applied, using the principle of approximate
equality of bottom areasin designreaches,

At At, At,
The duration of hydrograph steps (j = 1,2,. , J) is also divided into stepsby
time = t, - t,-,2 where indices {- 1 and r indicate initial and final time moments of
design interval computed according to a special subprogram from the condition
that deformation of the bottom does not exceedthe specifiedportion of the flow depth
in any reach.
For averagedvalues and for differencesthe sliding indices 1, 2 and < - 1 are
omitted.
The following symbolsare also applied in equations(1) - (3): v - velocity of the
flow; h - flow depth and Q,- dischargeof sedimentsat the gauge-linewith appropriate
index psk and pgrk- densitiesof particlesof the bottom ground and in newly deposited
ground in k-th reach; i, - slope of flow energy losses for friction over the reach
determinedfrom uniform flow equation with the use of Chezy-Manning formula, hrr-
local lossesexplainedby changein the channelwidth along the flow.
The systemof equations(1) - (3) is solved with specified/initialconditions:
z(o,x) = Z,,Y(o,X,Q) = r,,
and boundary conditions:
QiV) = Q,7 QsW = Qso>
Q<cL) = QWO+ 2 Q,,,
n=l
W> L) = UQW
whereindex n - is the number of concentratedinflow, ZZ- number of inflows.

Consideration of various factors

Before the channeldeformationis computed,roughnesscoefficientsare identified in the
reachesbetween gauge-linesk and k + r, and initial curves of free surface at the fixed
water discharge are plotted. This problem is solved with the help of the method of
exceedencesduring determinationof free surfacecurvesand by the method of iterations
during roughnesscoefficientsdetermination.
The algorithm of computation and program RUSLO, realized by this algorithm,
take into accountthe following factors:
1. concentratedinflows;
3Y. channel harnessingby coarsefractions using method of Ts. E. Mirtskhulava
and A. V. Magomedova(Anon., 1987);
3. pit excavationsin the channels;
4. local resistancesdue to changesin the channel width along the flow using
dependencespresentedby formalized generalizationof well-known proposals
and recommendations(Anon., 1988; Grishanin, 1990);

300
 5. rate of flow saturationwith sediments,
The account of the first three factors is somehow described in our previous
publications,including (Veksler, at al., 1988). Therefore, we shall briefly explain some
problems of the account of local lossesof energy flow of gradual saturation of flow
with sediments.
According to the data from manuals and classicaltextbooks on river and open
canal hydraulics,the local lossesare most significant in they are due to flow widening in
plane. If the channel bends, lossesof turbulent flows (Reynolds’number Re> 3 1500)
with a relative radius of the bend R/B > 3 are very low at any h/B value (Anon., 1988).
In rivers, Re >>3 1500 and R/B >3 in general,which gives us the right to neglect losses
at the channelbend.
Changesin the altitudinal configuration of the channelbottom (alternateddunes,
side bars, shoals and pools) may also explain local losses of flow energy, but they
should be probably regarded as lossesover the river length during water flow round
rough obstaclesproduced by channel mesoformsand these lossesshould be integrally
taken into accountby roughnesscoefficient,
Indicating local lossesfor flow contraction and widening in plane through some
generalizedcoefficient 6.

and transforming equation(1) as follows

r, -y; = AY=(l+<)

we obtain rather simple and suitable algorithm for the account of local losses
where coefficient < is a formalized compilation of well-known proposals and
recommendations,generalizedin (Anon., 1988; Grishanin, 1990). We only took the
liberty to approximatethe expressionsand to fix strict limits for g computation in case
Of
02 -01 >o:
B2 -4
‘.55 if p< - 90”

-0.55 p+ 7.5” if - 90”s p< - 7.5”

82.5”
0 if - 7.5”s p< 0
(+ -o.gP~z -wl if oqk40”
40” co2+o,

if 4o01 p< 60”

if

where angle l3 = 2 arctg B2 -4

2/w

301
 It should be emphasizedthat when we speakabout flow contraction in plane, we
mean acceleratedwater flow, and when we speak about flow widening, we mean
deceleratedflow. Thesevalueswould coincide if w, - w, and B2 - Bl would be of the
samesign i.e. 02B2 -CL)1
_ By >o, If (32 - @i CO,we shall assumethat local lossesdue to flow
B2 -4
width change are missing (<= 0). The latter assumptionis most rough in the whole
schemeof local lossesaccount,but we have to be reconciledwith it, becausewe do not
know any recommendationfor that case.
Computationsmade for wide plain rivers show that the valuesof 6 in majority of
cases are within 0.0 1 t 0.05, being 0.16 in some cases only. For mountain and
Piedmont river reacheswith alternated narrow canyons and wide hollows (e. g., the
Vitim river discharge to Muyskaya hollow and its successivecontraction at the
Paramskyrapid) the < valuesfar exceedthose in the caseof plain rivers.
Introduced conventionallimits and approximationsused in (6) may be modified as
far as new data are collected and the conceptsare changed.Moreover, the structure of
the local lossesaccount may be kept. The account of a gradual saturationof water flow
with sedimentsconsistsin transformationof equation (3) as follows:

(8)

where 8 = - S = sQ - rate of flow saturationwith sediments;

St, P,
St7 - extreme concentration,correspondingto the transported (conveyance)capacity
dZ
of flow P, , if the rate of channeldeformation - 1spresentedas a sum, as follows:
at
(9)

where: f(0) is somegravimetric function, taking the valuesoff0 = I andf(l) = 0.

The extremevaluesof g are as follows:

-dZ =-.-- ps P (10)

dt e=o Pgr e

where: P,- eroding capacity of the flow, determined in accordance with

recommendationsin (Anon,, 1981; Anon,, 1988):
(11)
if Q, = P, is assumed for a complete saturation of flow with sediments
3s .
(6 = 1),equation (11) is usually applied, in which it is also assumedthat term -.-A- IS
dt
low in value if comparedwith other equationterms.
Substitutingequations(9), (10) and (11) into equation(8) we shall obtain:

302
 Equation (12) can be solved relative to 8, if a kind of fknction f( t9) is specified.
After derivation of 8, it is possible to find the value of $ using equation (8).
Moreover, equation (8) would automaticallysatisfy the extreme casesof (10) and (11)
which is regulated by expression [1- 0-f(B)] in equation (12), equating the second
term of the right part of the equation with zero at 8 = 0 and 8 = 1. It should be noted
that if 8 = 1, the whole equation (12) is transformed into identical zero, degenerating
equation(8) into (11).
Determination of the kind of function 4 0) requiresfurther investigations.Hence
we shall acceptthat
f(e) = (1- 0)” (13)
which is a generalizationof the hypothesisof G. R. Foster and L. D. Meyer (1972),
who assumedk=l. On the basisof previous studiesit would be more correct to accept
k=5 t9.
Equation (12) is solved relative to 0 in the finite differencesat the left boundary
condition of
wo = e,(t) (14)
and at initial condition of
e(x,o)= e,(x) . (15)
Conditions (14) and (15) are not homogeneous and should be accepted
dependingon the ratio of flow velocity V to non-eroding velocity V,, for the ground of
the bed:

1 if v I v,
e,(x) = o if
i v > v,

303
 Al

Reading,print and control of input

data;
information processing

Discretization of the shematized

channel
[ over its length
Dl .
Identification of roughnesscoefficients;
printing of roughnesscoefficientsand of
8
initial free water surfaces
.
Are 011the hydrograph steps
looked through?

Step selectionby time

r
4 L

Gl I Computation of sedimentdischarge,
volumesof channeldeformation,bottom
marks after deformationwith the account
of pit excavationand flows
Hl I

1 Printing of computation rezults

I1

73 output

Fig. 2. Flow diagram for RUSLO program.

304
Program characteristics

Program RUSLO consists of utility-envelope in PASCAL, main program and nine

subprogramswritten in FORTRAN-77.
IBM PC is required for program operation or a compatible computer of any
model with operation systemcompatiblewith MS-DOS versionsnot below 3.30.
Flow diagram of the program is shown in Fig. 2. The program has following
technical and economic factors. The maximum memory capacity required for solving
problems equals 5 12 kbyte. Time for one problem solution dependson the PC model
and on the rate of designarea discretizationand variesfrom 5 to 120 min.
The main functional restriction of the program is connectedwith the degree of
specifiedarea discretization.The number of gaugelinesis limited by 999; the number of
computedyears- by 999; the number of inflows - by 10.
The utility-envelope (input-output both in interactive mode and in file, input data
control, output file scan and computation results print) makes general monitoring
functions.
The utility-envelope menu contains the following items: input, update,
computation, scan,print, output.
For a better evidenceof input and output data “GRAPHICA” program has been
developedwhich makesit possibleto get input data and computation resultsas graphs.
The program is written in PASCAL. The plotting of graphs required a modification of
the main program RUSLO to provide the output information required for plotting
graphs.
Information output in the graphic way facilitatesand acceleratesthe interpretation
of the obtainedresults.
Program-RUSLO and its modification at the stage of its developmentwere used
for retrospectivecomputationsand prediction of channeldeformation downstreamfrom
the Novosibirsk HPP on the Ob river, Dnestrovsk projects on the Dnestr river,
Mokskaya HPP on She Vitim river and Uotkinsk HPP on the Kama river.

Case study of the Votkinsk hydropower plant on the Kama river

Introduction
The Votkinsk power generationplantain the Kama river was put in to operation from
the end of 1961 up to 1966. During the plant operation the river channel deformations
occurred and water level in the lower pool of the dam fell, which induced deviations
from the project rules of the power plant unit operation. According to the project, the
discharge of at least 900 m3/s should be delivered through the spillways to provide

305
navigation on the Kama river. During last years it was necessaryto dischargeat least
1200 m3/sof water to the lower pool to maintain navigation, Thus, the project schedule
of runoff control by the Votkinsk Reservoir was broken and the power-plant
contribution to peak power generationwas reduced,
The fall of water level in the lower pool caused great deformations of the left-
bank slope in the diversion canal from the hydropower plant (HPP), i. e. settle and
damage of mounting slabs, as well as destruction or concrete walls of the HPP
structures,non-resistantto frost and located in the zone of variable water levels (piers,
divide wall, battering walls of abutments).A necessityof their repair was emphasizedby
the Commission of the Ministry of Energetics in July, 1989. The Commission also
recommendedto considera problem of admissiblevolumes of inert materialsexcavation
in the lower pool of the Votkinsk HPP.

Variation of satge-dischargerating curves

Investigationswere made during 1990 - 1992 at VNIIG; during which the processes
causing channel deformations and water level fall in the lower pool of the Votkinsk
HPP were studied.The integral estimationof theseprocesseshas been derived from the
comparisonof stage- dischargerating curves at characteristicgaugelinesof the lower
pool; these curves have been plotted on the basis of data processingfor the period of
HPP operation. To plot these curves, the appropriate data on the steadyregime period
were selectedfrom daily reports of the HPP accordingto methodologyfrom (Veksler &
Donenberg, 1983) and processed.The plotted curves for the gauge-line of the HPP
(Fig. 3) give some idea on the stagefall variations. The water stagefor about 30 years
of HPP operation is constantlyfalling,

Fig. 3. Stage-discharge rating curves at the T’otkinsk power phnt gauge-line (observed data)

306
despite any usual channel deformations observed in the lower pools with similar
grounds and hydrological conditions (e. g., in the lower pool of the Gorky HPP on the
Volga river (Veksler& Donenberg, 1983)). By 1990 - 1991 the water stage fall at the
gauge-line of the HPP and at “Olkhovka”station, located downstreamthe joint of the
lower inlet canal of the sluice and Kama river channel,was as follows:

At gauge-lineof power-plant At gauge-lineof 1

if comparedwith if comparedwith
projectcurvein 1961
19.56
At minimum prospect water
dischargeof 890 - 900 m3/s to
providenavigation 1.10 0.90 0.82
At meanlong-termwater discharge
of 1710m3/sin the Kama river at
the site of HPP 0.97 0.80 0.74
At maximum water dischargeof
7000 m3/sthroughHPP 0.90 0.60 0.55 1

Displacement of sedimentation waves

A changein the volume of dredging made by river crafl agencieson the shoals in the
lower pool to provide standarddepthsis another factor of channeldeformationsduring
the HPP operation. This factor is to an certain extent subjective,becausethe volume of
transit dredging made on some shoal dependsboth on the individual approach of an
engineer designing a cutoff on the basis of data on shoal survey and on different
production reasons.Nevertheless,data on the amount of transit dredging given in
annual reports of “Water Ways of the Kama Basin” Production Association (PA),
makes it possible to present a situation on the displacementof zones of sediments
accumulationin the lower pool of the Votkinsk HPP Unfortunately, the data analysis
given below on the dredging was made for 22-year period only, from the start of HPP
operation in 1961 up to 1983, becauseafter 1984 only total data are given on the work
madein some pool, i. e. in the navigablelong reach.
To estimatethe volume of dredging, method was applied previously develop d for
the evaluation or the Ob channel deformations in the lower pool of the Novosibirsk
HPP ( Veksler & Donenberg, 1983): the lower pool was separatedinto reaches8 - 12
km long each.This separationwas madeto maintain uniformity of channelforms and of
the channelprocessestype within every reach. Then annualvolumes of transit dredging
on shoals within every reach were summed up, and graphs of annual dredging
variations were plotted for 1961 - 1983 for these reaches.Years (or periods) were
separatedin the graphs correspondingto peak dredging. It should be noted that at this
approacha regular outlined increaseof standardnavigabledepthsfrom 240 cm in 1961
up to 335 cm in 1970 did not affect any peak observedin someparticular year.
Peak years (or periods) separatedin these graphs were plotted as dashed or
shadedareas (evident peaks are given by solid lines, other peaks - by broken lines),
years are laid off as abscissa,and distancefrom HPP along the navigation channel- as

307
ordinate (Fig. 4). The drawing formed by these dashedlines (areas)made it possibleto
outline the lines connectingthem. These lines are also given in Fig. 4, They determine
variations of the displacementof the silted zone downstream. The first sedimentation
wave (upper line) leavesthe secondline behind by 3 - 6 years and it is characterizedby
lower peaks (Fig. 4) if comparedwith the upper line (except area No. 4). The second
wave determinesthe displacementof the absolutepeak of dredging (except some rare
cases).It is possibleto assumethat first wave is the result of the Kama river channel
closure and displacementof bedload composedby the products of local bottom erosion
formed behind the side pitching of the spillway dam during the first years of HPP
operation. This dune was gradually eroded and moved downstreamby the floodwaterof
subsequent years.

Fig. 4. Graph of displacement of zones of maximum sedimentation in the lover poll of the Votkinsk
power plant unit.

The secondwave of sedimentation,by our opinion, reflects the major processof

channeldeformation occured in the lower pool of the power plant unit and it precedes
the zone of predominant scour distribution downstream, explained by the solid
dischargesilted in the reservoir (Veksler & Donenberg, 1983). This wave originates
from the dischargeof first rather high prolonged floods with maximum dischargesof
about 8000 m3/s through the spilways in 1965 - 1966 which happenedfirst since the
time of HPP operation.Next, the third wave 4 - 5 years later, testified the continuation
of this processand was probably causedby a relatively high flood of 1968. In Fig. 4 the
origin of the fourth wave is quite evident, causedby the flood in - 1979, i. e. the highest
flood during the HPP operation, with its maximum discharge of 9630 m3/s
Unfortunately, inadequatedata on dredging after 1983 make it impossibleto procede
the analysis.Ideas expressedabout the sedimentationwaves origin moved in the lower
pool of the Votkinsk HPP explain this processsomehow.
As evident from the graph analysis, if the flood water is discharged through
spillwaysto the lower pool of the Votkinsk HPP and the intensity of this flood exceeds
the other previous maximum floods, then new waves of sedimentsmay originate and
move through the spillways again The site of thesewaves origin would be as far from
the HPP as the distancerequired for a stablereach of the scour zone by the time of this
flood passthrough the dam.
At present,the stablereach of the scour zone is propagatingover the length of 13
- 15 km far from the site of the HPP (according to the estimatesof the personnelof the

308
Analysis of channel surveys made in the lower pool of the Votkinsk HPP by
reconnaissanceteams of “Water Ways of the Kama Basin”PA as well as computations
of channeldeformationsshow that besidesthe channeldeformationsdue runoff control
and lesssedimentsconcentrationin the water flow, excavationof sandand gravel out of
the pits in the Kama river channel since 1975 is of a great significancefor the Kama
river-bed deformations. According to the data of the Kama steamshipcompany, the
amount of sand and gravel excavatedduring 1975-1991from the reach of HPP site to
Sarapul, 68 km long, was as much as 12.5 min m3, i. e. about 6 % of the channel
capacity at the quaranteednavigation depth. The total increaseof the channel volume
(emsion and excavation)in this reach for the sameperiod was equal to 34.2 min m3 or
about 18% of the channelcapacityat the navigation depth.

Computation of channel deformation

To compute channel deformationsand to evaluatethe effect of various factors on this
processevolution a schematizationof the Kama river channelwas made in the reach of
68 km long from the Votkinsk HPP up to Sarapul. This channel schematizationwas
based on channel surveys made in 1961 by the Kama Basin Office of Water Ways.
Those surveyswere suppliedby Lenhydroproject surveysfor 1963 - 1964 for the 2-km
reach adjacent to the HPP. Channel schematization was made according to
recommendationsin (Anon,, 1981). The following information was applied as basedata
for the retrospectivecomputationscovering 30-year period of HPP operation:
- schematichydrographsof discharges,according to (Anon., 198l), through spillways
of the dam during 1961- 1990 (from the data on HPP operation);
- volumes of annual sand and gravel excavationfrom the channel pits during 1975 -
1991 (from the reports of Kama steamshipcompany);
- characteristicsof alluvium deposits in the Kama channel (from the reports of
Lenhydroproject, suppliedby the reports of the river craft agency).
Design characteristicsof the bottom ground were accepted on the basis of
averaged data available: ground density pRr= 1500 kg/m”, density of particles ps
=2650 kg/m3, mean size of particles d =1,38 mm; content of particles less or equal to
meanPd = 0.81 in the bulk of bedload; size of particlesdgs= 8 mm, with of which are
availablein the bedload,by weight.

309
 03

64

63

62

bislancc fmm ?GP,kniT --

Fig. 5. (Yomputation of the Kama channel deformation in the dam reach of the lower pool of the
J btkinsk power plant

(a) retrospective computation for 1962-1990; (b) prediction for 1990-2006 I-20 - numbers of design
gauge-lines; 1 - original bd and free water surface in 1961; 2 - design bed and free water surface in
1990 without of pit excavation in the channel; 3 - idem,with the account of pit excavation; 4 -
understurned schematic bed in 1990; 5 - original bed and free water surface in 1990; 6 - prediction of
the bottom position and of free water surface, if dregding is stopped; 7 - idem, if pit excavation is
continued

The computationresult were comparedwith the observeddata on 1971, 1981 and

1990 appropriate changes were introduced to the channel schematization, thus
approximating computed and observer data. Channel schematization accepted for
retrospectivecomputation was used predicting a future evolution of the Kama channel
for the nearest15 years,i.e. before 2006.
Fig.5 presents the results of computation for the 25 km reach adjacent to the dam

310
 72

II

’ I I lwn

67 ti

66 ,
t. ---be 70

65 69

68

67

I I I I I I lbb

Fig. 6 Stagedischarge rating curves e = .f (H) at the 1btkinskpower plant gnuuge-line (a) and nt
“Olkhovkn ” gauge-line (b) porn th e results of retrospective computation andforecast before ,7006.

1 -project curve; 2 - curve of 1961; 3 - curve of 1990; 4 - predicied curve for 2006, ifthe excmwtion
is stopped; 5 - idem, if excavation is continued.

Conclusion
The perfomed computations make it possible to judge about the Kama channel
evolution in the lower pool of the Votkinsk power plant unit:
(1) The total deformationsof the river channelin the reach from HPP to Sarapul
is causedby erasion due to Votklnsk Reservoir construction (its contribution is 60%)
and by pit excavation (34%). During the period of pit excavation (1978 - 1991) the
portion of erosionwas 52%. Pit excavationscontribute 48% of the total deformationsin
this reach. Additional water level fall due to excavationsat the HPP gauge-linewas 39
cm, and 18 cm - at the “Olkhovka”gauge-line.
(3) If pits excavationsproceed in the same amount in the Kama channel, an
additional water level fall (if compared with its present level) should be expected,thus
by 2006 it would be 38 cm lower at the HPP gauge-line, and 36 cm - at “Olkhovka”

311
gauge-line.If excavationsare intensifiedin the Kama channel,the water level fall would
be more significant. If no excavationsare made, the additional water level fall would be
24 cm at the HPP gauge-line,and 23 cm - at “Olkhovka” gauge-line(Fig. 6).
(3) Predicted deformations of the channel would lead to even geater water
releasesfor navigation purposes - up to 1600 m3/s by 2006 if excavations are
continued; and up to 1500 m3/s - if excavations are ceased after 1992. Thus, the
contribution of the Votkinsk HPP to peak generation of electric energy would be
reducedmore Intensively.

Bibliography
Anon (1981) ‘Recommendationson the computation of channel deformations in the
lower pools of power plant units’, II 95 - 8 1, VNIIG, Leningrad (in Russian).
Anon ( 1987) Determination of admitted non-eroding waler flow velocities for various
grounds .for canals projects. Textbook to SNIP 2.08.03 - 85 Meliorativnye
sistemyi sooruzhenia,Tbilisi, Gruzgiprovodkhoz, 116 p. (in Russian).
Anon (1988) Hydraulic computations for spillwuys: Manual. - Moscow,
Energoatomizdat,624 p. ( in Russian).
Foster, G. R and L.D.Meyer (1972)A closed-form soil erosion equation for upland
areas.. Sedimentation, Edited by H. W. Shen, Colorado, Fort Collins, Ch. 12, p.
19
Grishanin,K.V. (1990) Principles of afynamics of channelflows. Moscow, “Transport”,
320 p (In Russian).
Veksler, A. B. and V. M. Donenberg (1983)Channel deformatione in the lower pools
of large hydro-electric power plants. Moscow, “Energoatomizdat”,(in Russian).
Viksler, A.B.; V.M.Donenberg and VlManuilov (1988) ‘Application of numerical
methodsI& grcdiction of channeldeformationsin the lower pools of power plant
units’. Proc. of the Vth All-Union Hydrological Congress, vol. 10, “Channel
processesand sedimentation”,book 1. Leningrad, Gidrometeoizdat,p. 390 - 396
(in Russian).

312
 Design of bank revetment based on
reliability concept

Prof. Dr. SC. Miodrag Bozinovic Institute for Water Management, Faculty of
Agriculture, University of Novi Sad, Yugoslavia

Abstract
A reliability calculation is proposed in which the bank revetment is treated as a technical
system composed of three elements: the base, the surface and the filter beneath the
surface, while the causes of system failure are incidental in character. The position is
taken that the system fails if any of the elements of the system ceases to function, and
thus the system is defined as a system of elements connected in order (in series). It is
supposed that the intensity of failure of individual elements of the system leads to a
decrease in the calculated reliability of the system, but the evaluation is made that the
results obtained can be accepted in the phase of ranking the projected variants of the
technical plan of the bank revetment. A graphic method is presented and recommended
for the quick definition of approximate reliability values of a sloped bank revetment
exposed to the effects of waves caused by wind.

Introduction

Bank revetments are relatively simple structures which are usually built in order to deter
the erosion of concave shores on the banks of alluvial rivers. ln practice, the most
common are so-called “sloped bank revetments” which are composed of the following
construction elements: base (foundation-support), permeable surface of the bank slope
and the sub-surface filter beneath the surface (Fig. 1).

313

_- ----__ .-____
Fig. 1. The cross-section and basic elements of the sloped bank revetnlent: 1 - base, 2 - swface. 3 -
sub-sut-face,filter.

The base is most often built of crush stone in the shape of a trapezoidal prism and
serves to receive the weight of the surface and to deter river bed erosion. The surface of
the bank slope is most often made of crushed stone, but also of concrete blocks,
concrete panels, gabions and other materials. The surface serves to fix the bank slope,
to counterbalance the effects of the load and the effects which directly tend to deform
and destabilize the bank slope. The sub-surface filter beneath the surface is made of
gravel-like and sandy materials or of porous geo-textiles. The basic function of the sub-
surface is to control the process of the washing away of materials from the bank
through the filter itself, the surface, and the base: it stops the process of the internal
erosion of non-compact materials. Bank revetments are exposed to numerous and
specific dynamic and static loads and influences which are variable in space and time,
and many of them have a stochastic character (the static and dynamic effects of water
flow, the effects of waves caused by wind and watercraft, the effects of seepage from
the area around the bank, etc.). The character of such loads and effects arises from the
fact that bank revetments are located in the “triangle” of solid, liquid and gaseous states
of material (embankment soil, water, air) in which all three phases change their
properties in a different way. These changes are especially significant in alluvial rivers
with an inconsistent water regime and an intense channel deformation. It should also be
stressed that bank revetments (as well as other constructions in the river channel) also
cause additional changes in the river channel, which interact with the bank revetment.
The analysis of the stability and reliability of the bank revetment becomes
complex because of the number, type and character of the loads and other factors which
should be included in the analysis. Apart from that, complete information is not
available in practice for the defining of those loads and factors, and thus dilemmas occur
in the planning of revetments. As pointed out by S.Bruk ( 1988) either:
a) simplify the problem so that only the most important phenomena are included
in the analysis, reducing the complexity of the calculation model, but this risks
compromising the adequacy of the system simulation and its performance;
or:

314
 b) create a complex model which simulates, in principle, the performance of the
system much more precisely, but this increases the risk of having inadequate (low
quality) input which can reduce the predicting of capability of the model.
It is estimated that, for planning sloped bank revetments in alluvial rivers,
solution “a” is more acceptable. In doing so, there should be a tendency to accept
adapted - flexible and fixable systems and elements, along with the evaluation of
reliability, or risk, according to the recommendations of V.Yevjevich ( 1977).

Application of the concept of reliability in the planning of river

Bank revetments
Using the suppositions of Section 1 the possibility and effectiveness of reliability
analysis for planning sloped revetments on alluvial rivers will be presented. It should be
emphasized that this analysis is not based on statistical evaluation of the reliability of
already built revetments for very simple reasons: it is practically impossible to form a
representative statistical series of empirical data about failures, that is, about the
reliability of identical revetments which functioned under the same conditions. This is a
problem common to most structures which are not produced in large series (Augusti,
Baratta and Casciati, 1988) and in those cases varying approximate calculations and
analyses are made in calculating the reliability parameters in the planning phase.
Different - varying construction (system) solutions are used as starting points. the
physics of failure for the various parts and the structure as a whole are analyzed, and
then an approximate calculation for the quantitative evaluation of reliability parameters
is done for individual variants in order to determine the most suitable solution. An
illustration will now be given of such a concept which could be applied in the starting
phase of planning of a sloped revetment of crushed stone (Bozinovic, 199 1).

The structural plan and method of calculation for sloped

revetments of crushed stone
A sloped revetment is a system of three elements; the base, surface and filter (Fig. l), in
which it is assumed that the elements are connected in series. For such a system, and
under the condition that individual elements are independent and complete, the intensity
of system failure will be:
a=a, +a, +a, (1)
and the reliability of the system R is:
R= R&-R, (2)
where RI, R2, and Rj are the reliabilities of individual elements of the system.
The presented plan of the system and the assumption of the independence of
failure for individual elements of the system leads to a certain decrease in the calculated
reliability of the system. However, it is evaluated that the obtained results cannot
significantly influence the comparison of individual observed variants in the beginning
phase of planning the revetment.

315
 The reliability of the elements which are exposed to the work load s is defined
from the condition that the probability of work (functioning) without failure by the
elements exists if the work load s has not exceeded the resistance of the elements Y, that
is, that Y ’ s. From the mentioned condition, the calculation for defining reliability is
extracted, if the distributions of s and r are known, i.e. ,f(J;, and ,f(r) (Vujanovic, N.,
1990).

Fig 2: Defining reliabili@ when the work load .s exceeu’sthe resistance r ofthe elements

In Fig. 2 one can see the probability that the load si exists in the interval d as
represented by surface Al that is:

From the same illustration (Fig. 2) one can see the probability that the resistance
of the elements Y is greater than the work load s expressed by surface A? that is:

I’(r > s) = j,f (s)ir = A3 (4)

-5
The probability of work without failure is equal to the production of the two
obtained probabilities, that is:

and the general equation for the reliability of the elements has the form:

R = IdR = if(s) 05 (6)

u)
Since there is not exact mathematical solution for all types of distributions of
work load and resistance, an approximate graphic method has been developed for
defining the reliability according to the following approach (Vujanovic, 1990).
Let:
<*I
(7)
and:

316
 H = jf(s)ds (8)
s

which means that:

dH = f(s)ds (9)

and then equation 6 obtains the form:

R=jGdH (10)

The borders of the integrals are changed because the quantity under the integral is
also changed. While the resistance boundaries were from -XI to +co , and for the work
load from s to +a, the new boundaries for quantities G or H are from 0 to 1.
Therefore, the surface under the function G = f(H) represents the reliability value (Fig.
3)

Fig. 3: The graphic definition of reliability

The cited graphic method allows the definition of reliability for any kind of
distribution of work load s and resistance r. An illustration of the application of this
method, when the empirical integral functions of the work load F(s) and resistance F(r)
are given, is shown in Fig. 4. The procedure is contained in the following:
For known integra!s of the empirical distributions F(s) and F(r), the helping
functions are defined: H(s)=1 - F($ and G(s)=l-F(s=r). Then the diagram G-f(H) is

drawn, and the integral R = [C(H)dH, that is, the surface between line H=O, G=H

and G==f(N) is planimetered. This surface presents the approximate reliability value R

Example of calculating the reliability of revetments of crushed

stone exposed to the effects of wind waves
The basic elements for reliability culculating (Fig. 1) are the following:
- height of the revetment, H = 5 m;

317
 - slope of the surface, m = ctg a= 2;
surface and base are of blocks of stone, ok= 2.6 t/m3 ;
- sub-surface - filter is of unwoven polyester geo-textile, g = 300 grim’;
height of waves cause by wind, h = I m;
- wave length, h = 10 h;
- friction coefficient between the surface and sub-surface, f0 = 0.4
- friction coefficient between base and channel base,f, = 0.3

The choice of materials and dimensions of the revetment are given in (Bozinovic,
1991) while the reliability calculation is done for the following work loads s (active
effects):
for the surface 5’1- effects of waves caused by wind,
for the base Sz - effects of the horizontal force of the surface,
for the sub-surface S3 - effects of the seepageflow expressed in the speed of
filtration of sub-surface waters from the area around the bank behind the
revetment

S H(s) G(s)
0 1.00 1.oo
0.5 0.96 1.00
1.0 0.83 1.00
1.5 0.65 0.99
2.0 0.44 0.97
2.5 0.20 0.92
3.0 0.08 0.73
3.5 0.02 0.47
4.0 0 0.15
l\\\\\\Q 4.5 0 0.03
h’.\\\\\ 5.0 0 0

I
r
R = J G(H)dH

;=o. 92

Fig. 4: The graphic method for defining reliability R when the emplncal integral functions are grven
of the distribution of work load F(s) and the resistance of the elements (load cnt.) F(r)

As critical loads Y (resistances) for individual elements of the individual elements,

the following are adopted:

318
 for surface r the mass of the stone blocks of equivalent effects of the effects
of waves according to the relation from (SNIP 2.06.04.~82., 1982)
-
increased by the safety coefficient: ?f, = g = 1.11

for the base r the friction force between the foundation of the base and the
-
soil, increased by the safety coefficient: ?(7*=F= 1.1,
2

-
for the sub-surface r the value of filtration speed through the geo-textile,
-
increased by the safety coefficient: q3 = F = 1.5.
3

The integral distribution functions for all actions and critical loads (resistances)
are shown in Fig. 5.

Fig. 5. Reliability calculation for a slope revetment of stone blocks and,filter of unwoven textile

For the given work loads and resistances, the calculated reliability related to the
state of the stone blocks in the surface exposed to waves, the state of the base’s slide

319
under the effects of the gravitational load of the surface, and the state of the sub-surface
filter exposed to colmatage from the area around the bank.
The reliability calculation of individual elements of the revetment is done
according to graphic methods based on equations 6 and 10, and the reliability
calculation for the entire revetment (system) is done according to equation 2. The
results of the calculations are shown in Fig, 5.
As seen in Fig. 5, the following reliability values are obtained for individual
elements of the revetment:
R,= 0.76, RI = 0.78, R3 = 0.91,
and for the revetment as a whole:
R = R,.R2.Rj = 0.54
If an increase in the planned reliability of the whole revetment were required, it
would be achieved by altering the character of the distribution of resistance (decreasing
0,) and/or increasing the safety coefficient for individual effects. For example, if the
integral curves of the resistance distribution obtained the form shown by the dotted
lines in Fig. 5, the reliability of individual elements would increase to the following
values:
R,‘ = 0.86, Rz‘ = 0.94, R3‘ ==0.996,
and the reliability of the entire revetment would be:
R =R,‘.RZ‘.Rj‘ = 0.80
In the given example, only one combination of the possible failures of individual
elements of the revetment is treated, from which one reliability value for the entire
revetment resulted. In order to obtain a more complete picture of the planned reliability
of a defined (single) type of revetment, it is also necessary to calculate the other
possible combinations of loads and effects which could cause failure in the integral parts
of the revetment. For example, it would be possible to calculate reliability for the
surface with regard to the effects of ice flow torsion (pulling blocks from the surface),
ice flows or river craft which hit the revetment, etc.; the reliability of the base could be
calculated with regard to the effects of the pulling force of the water flow or of the local
erosion of the channel, while the reliability of the contra-filter could be calculated with
regard to outflow or contact erosion.

Conclusions
1 .Although the basic suppositions and calculation methods include significant
simplification and approximation, the application of the concept of reliability
can contribute to the improvement of planning revetments on alluvial rivers.
2.The projected reliability, as an important characteristic of the structure, depends
on a series of factors which are deterministic and stochastic in character, such
as: the number of construction elements, contact between the elements,
characteristics of relevant load and construction resistance, etc.
3.From the standpoint of revetment reliability, intended for the stabilization of the
river bank in “field” conditions, it is most suitable that the entire revetment be
made up of one continual, flexible and reparable carpet (“sandwich”) which
would cover the entire contour of the erodible bank, and which would carry
out the function of the base, surface and contra-filter.

320
 4.The evaluation of planned reliability depends a great deal on the knowledge
about the distribution of resistances and loads which are expected in the
process of constructing, using and maintaining the structure; thus, research of
this type must be a mandatory part of the reliability analysis of any SUU~
structure.

References
Bruk, S. (1988). Design Data for River IManagement. International Conference on
Fluvial Hydraulics ‘88, Budapest, Hungary.
Yevjevich, V. (1977). Risk and Uncertainty m Design of Hydraulic Structures.
International Symposium on Stochastic Hydraulics, Lund, Sweden.
Augusti, G., Baratta, A. and Casciati, F. (1988). Probabilistic Methods in Structural
Engineering. Chapman and Hall, London, New York (translation from English to
Russian), Strojizdat, Moscow.
Vujanovic, N. (1990). Reliability Theory of the Technical Analyses. Vojnoizdavacki i
novinski centar, (in Serbian), Belgrade, Yugoslavia.
Bozinovic, M. (1991). Stability and Reliabiliry Analyses on Revetments in Alluvial
River Beds. (in Serbian), Dissertation, Poljoporivredni fakultet, Novi Sad,
Yugoslavia.
SNiP 2.06.04.-82. (1982). Loads and InJuences on the Hydraulic Structures. (in
Russian), Moskva.

L
321
5. MODELING STRATEGIES AND APPROACHES
River modelling

G. Di Silvio
U riiversity of Padua, Italy

1. MODELLING AS A FIGURATIVE ART’

It has been observed already that a model is never conceived, developed and
implemented in the same neat, logical way in which it is later presented in a scientific
paper or a technical report. No model, in other words, is born beautifully shaped and
provided with all i 7~ armour, shield and weapons like Minerva from Jupiter’s head.
Modelling is rather the product of fantasy and labour, where the typical working style is
the trial-and-error procedure (inductive method) more than a linear sequence of
operations stemming from a set of postulates (deductive method).
Probably for this reason it has also been observed that modelling resembles more
“art” than “science” (although both the inductive and the deductive methods are amply
used by science as well as by art). In fact, the resemblance between modelling and art
makes more sense if one thinks of art (and especially of painting) as a way of
representing the real world.
By this metaphor it follows that there is not one model of a river, but rather a
large number of models portraying different aspects of the river itself. And even the
same aspect of the river may be seen with different eyes, depending upon the
impression which is to be conveyed to the observer: like the same view of the Seine
river has been reproduced, year after year, by tens of painters interested to send
completely different “artistic messages”.
The specific “artistic message” is in fact the driving force that motivates every
valuable work of art and that, consequently, dictates how the reality should be rendered
in the painting. In the case of modelling the motivation is (or should be) the specific
“problem to be solved”: only if this motivation is clear, the modeler will be able to
transfer into his model all the necessary details of the real world. Too many times, by
contrast, a model is built without any motivation, just for the sake of playing with a
“nice” methodology or under the academic pressure of publishing. These models will
only reveal themselves as useless and misleading representations of the reality, as the
amateurish or academic paintings produced in analogous circumstances.

322
2. THE CREATIVE ITINERARY OF A MODELER.

In Fig. 1 is traced, roughly schematized, the laborious itinerary of a modeler along his
work.
As said in the previous section, a meaningful model can only be originated by a
scientific or a technical problem to be solved. In the field of engineering, technical
problems typically come out from a project, that is from an organized set of practical
goals to be achieved through appropriate means (a number of goals and means in the
field of river engineering will be seen in Section 6).

IL-- C.----l1’ iit

1 --‘r--- ---l-------e----J
---- t; ‘I
/ II/I
Ii
Lk ,‘I

Fig. I - The creative itinerary of a modeler.

323
 When a problem is singled out, its solution can be achieved through an adequate
model that can simulate the system behaviour. The simplest way of solving a technical
problem is by applying the engineer’s educated guess (which is, after all, a rather
reliable “statistical model” based on a well-stocked data base). Engineer’s educated
guess is indeed the tool by which many important problems are solved (that is, many
important choices are made) in a project.
If a numerical model is deemed necessary to solve a problem, the model should be
first conceived, then implemented and finally applied: through all the three phases, the
modeler is to be prepared to go back on his steps, as indicated by the broken line shown
in the figure, whenever the need arise for corrections or improvements.

3. CONCEPTUALIZATION: FROM THE PROBLEM TO THE EQUATIONS.

This is the most delicate and creative phase of the modeler’s itinerary (Fig. 2). Very
often the problem immediately dictates which processes are to be reproduced by the
model. In some cases, however, the processes are different according to the approach
the modeler decides to follow (Bruk, 1994).

Fig. 2: The conwptudizution phnse: from the problem to the equations.

324
 If the problem can be tackled only from the hydraulic point of view, the processes
to be reproduced by the made1 are those described by the water-flow equations: a typical
purely hydraulic problem is deciding how high should be a levee in order to contain the
design flood discharge. Sometimes the same problem may require the reproduction of
other physical processes: in the previous example, this occurs when the waterlevel in
the river depends not only on the water discharge but also on the excessive bed
aggradation which may take place during the flood (Fig. 3). To simulate the possible
bottom change during the flood (morphological approach), sediment motion is to be
properly reproduced besides waterflow.
30600
I-
m I.1

al bl
II 3----'
t n-- *A - --‘ri- ----e
-0 10 20 30 40 10 20 30 40 ,~ 50
m

1.!0,+

v 198.02 m s.l.m.m

Fig, 3. Excessive aggradation in some rivers of Northern Italy during

an exceptional flood in 1953.

326
In other cases the approach is water-quality oriented and the model should include
transport and dispersion processes in the river, together with some kinetics of reactive
tracers (e.g. BOD).
Even more demanding is the ecological approach, which requires simulating the
dynamics of animal and vegetal populations (cenosis) living in or around the river.
The most complex approach to solve a problem is seeing it from the management
(or cybernetic) point of view. In this way (Bruk, 1994) the river system does not
comprise only water, sediments, chemical and biological components, but man himself
as engineer and manager. Thus the river model should in principle be able to simulate
not only physical and biological processes but also the interaction between nature and
man with his social and economical demands.
Not all the problems, of course, request the most complex approach. On the
contrary, in choosing the processes to be simulated, complexity is to be limited as much
as possible, since the more complex is the approach, the more empirical is inevitably
the model WILL ian actually be implemented. Indeed, as the system grows bigger and
bigger and the waterflow becomes just one component of the system, the possibility of
applying the basic equations of theoretical mechanics moves further and further away.
Another circumstance which is fundamental in the conceptualization phase of
modelling, is the time- and space-scale of the process to be simulated.
From the morphological point of view, the space scale is defined by the fluvial
_form one is interested in. Such fluvial forms are often classified as (1) microforms (e.g.
ripples, dunes, antidunes, namely bottom’s undulations with a vertical size smaller than
the depth); (2) mesoforms (e.g. banks, pools, meanders, with a vertical size comparable
to the river depth and a horizontal size larger than the river width) and (3) megaforms
(e.g. the structure of the river network, with a size comparable to the river length).
Systematic study of fluvial forms has been, up to recent times, almost entirely
confined in the province of geographers and geologists (or, better, geomorphologists).
However, the commitment of hydraulic engineers in this field has rapidly increased,
also because of the practical interest of these forms for several engineering problems.
For example, several “local” quantities (waterdepth, waves velocity, sediment
transport) are basically controlled by flow resistance, which depends in its turn on both
the grainsize and the size of the bedforms (microforms). Now, bedforms seem to be the
result of various kinds of instability, involving sediment transport and waterflow, which
control their formation, evolution and disappearance. In very simple situations (uniform
flow, uniform grainsize) it is even possible to predict the shape and size of bedforms,
corresponding to the most stable configuration for the given hydraulic conditions; but
the analysis becomes extremely complex when the flow is unsteady, the material is not
uniform and the geometrical boundary conditions are complicated. In this case the only
wav to cope with this problem is resorting to approximate empirical formulations.

327
 The situation is similar as far as mesoforms are concerned: for example, the
occurrence of a braided stream or of a meandering stream depends upon whether a
certain planimetric configuration is stable for the given flow conditions and bottom
material. Up to a point, it is possible to predict theoretically the stable mesoforms, but
in most cases an empirical approach is still the most convenient for having an idea of
the planimetric configuration of the river.
Also at the largest space-scale (tluvial megaforms) some attempts have been
made to explain, by a theoretical analysis, the morphological properties of the
hydrographic networks that have been empirically recognized since a long time.
Finally, the spatial scale relevant to the problem, is obviously important not only
from the morphological but also from the management point of view, since the
economical and social parameters involved in the problem depend substantially upon
the extension of the territory under the river influence.

4. RELEVANCE OF TIME- AND SPACE RESOLUTION.

Besides the size of the fluvial structure, there is another aspect, indicated in Fig. 2, that
the model should take into consideration. This is the detail, in time and space, that is
expected from the simulation, namely the required degree-of-resolution of the model
results.
Very often, for example, one is not interested to know in detail how the flow
velocity is distributed over a river cross-section during the passage of a flood wave, but
only to know how the water discharge and the water level vary in time and space along
the river. In this case, the only independent variables that are really important are the
time and the longitudinal distance on the river axis: all this implies that the vertical and
the transversal dimension can be somehow removed from the model. It does not mean,
of course, that these dimensions do not exist anymore, but rather that they can be
treated in a simplified way: e.g. by assuming a horizontal waterlevel over the cross-
section and a prescribed profile (similarity hypothesis) for the velocity distribution.
This hypothesis permits to perform an averaging operation on the velocity and to
consider the horizontal waterlevel elevation and the cross-section averaged velocity (or,
equivalently, the water discharge) as the only dependent variables of our problem.
Correspondingly, the only independent variables remain time and longitudinal distance.
Averaging (or integrating) in space and time is in fact a very common way for
simplifying complicated equations. The notion is quite familiar to the waterflow
modeler: in Fig. 4 is schematized the sequel of operations which provides, starting from
the most general Navier-Stokes equations, the 3-D Reynolds equations and,
successively, the 2-D and the O-D waterflow models.

328
 Averaging of non-linear equations, as it is known, generates residual comnonents
which may be either neglected (if it is permissible) or conveniently expressed as a
function of the averaged quantities: these temrs are indicated on the left column of Fig.
4. Other terms which should again be expressed as a function of the averaged
quantities, are the boundary conditions at the intepration limits: these are listed on the
right column of the same figure.
By the same token, also the most general three-dimensional and instantaneous
transport- and diffusion equations of a non-reactive tracer, can be integrated in time a
space following a similar sequence (Fig. 5). In the case of a non-reactive tracer a time-
averaging operation can also be performed over a relatively long period of time (e.g.
one year), in such a way that hydrological short-time oscillations (dry-and-wet seasons,
floods, tides) are removed from the equations as a sort of “macroturbulence”.

RESIDUALCOMPONENTS CONDITIONS
AT THE
INTEGRATION
LIMITS

AVERAGING ON THE DEPTH

t
I

2-DIM MODEL
Uniform-flow
formulae

,-------I AVERAGhVG
OVERTHECROSS-SECTION I-------,
I
I I I

A VEFMGINGOVER THE BASIN

I

Non-uniformvelocity Stage-dischargecurve
distribution over the basin

Fig. 4 - Water-flow models (with constant density) yielded by subsequent space-

and time- averaging c!f’Navier-Stokes equations.

329
 It should be noted that sediments and several pollutants can be just considered as
non-reactive tracers. The possible sequels indicated in Fig. 5 lead again to the 3-D, 2-D,
1-D and O-D models, which may perform either with a fine time-resolution (time steps
of minutes, for the simulation of single events) or with a coarse time-resolution (time
steps of years, for long-term simulations). Particularly important, in this regard, are

RESIDUAL COMPONENTS CONDITIONS AT THE

INTEGRATION LIMITS

2-DIM UNSTEADYMODEL flux of tracer through surface

I I
Flux of tracer
(Cross-sectionaveraged)

AVERAGINGON THE
I
Tidal or seasonal oscillation

AVERAGINGOVER THE BASlN

I I

Fig. 5 - Transport models yielded by subsequeru space- and time- averaging of

the tracer balance equations.

330
the so-called long-term morphological models for the simulation of riverine systems at
the historical scale (decades or centuries) and even at the geological scale (millenia).
A last observation should be made about the importance of the residual terms
when the tracer is constituted by sediments: while these terms are generally negligible
in rivers with relatively coarse transport, they become more important for fine
sediments and when the sediments are markedly non-uniform.

5. IMPLEMENTATION AND APPLICATION.

Let us now go back to the Fig. 1 that describes the creative itinerary of the modeler.
After the conceptualization phase, that is after having formulated the equations which
describe the relevant processes, the modeler should move to the other two phases, by no
means less important than the first one.
The model is usually constituted by a system of partial differential equations
and/or ordinary differential equations and/or algebraic equations which must be solved
with some kind of numerical technique. Model implementation comprises the
numerical solution (which would possibly require the expertise of an applied
mathematician) and the formulation of the computer code (which is preferably to be
carried on with the collaboration of an informatics expert). During the implementation
phase, moreover, the model is subject to a systematic investigation (sensitivity analysis)
about the relative importance of the “calibration parameters”, that is of the parameters
which cannot be directly and precisely evaluated by a direct inspection. For this part of
the implementation phase, the collaboration of the “problem” expert (hydraulic
engineer, potamologist, sanitary- or chemical engineer, ecologist, economist, etc.) with
the “method” expert (numerical techniques and informatics) is fundamental.
The present trend in modelling is to give particular attention to the
implementation phase. The old computer-programme, somehow put together by an
individual solely interested in the solution of his particular problem and totally
incomprehensible by any other person, tends to disappear. The prepartion of a computer
code usually needs such a large investment, that this cost should be shared by a large
number of users.
For this very reason, also the last phase of the itinerary in Fig. 1 (model
application) is extremely important. It is more and more frequent, in fact, that the user
of the model is not the modeler himself; this means that, before letting his model go out
in the world, the modeler should go repeatedly through the last phase,. with great care
and patience, validating the model for a large number of different situations and frankly
reporting the results of these tests in the user manual.

331
6. PROBLEMS IN RIVER ENGINEERING AND MANAGEMENT.

In Fig. 6 a few examples are given of goals and means in river engineering, together
with the problems which can correspondingly arise.
The goals are listed, more or less, according to the traditional priorities of an
European hydraulic engineer: safety first, then navigation, water utilization,
hydropower etc. Mass-movement control in the more and more developed mountain
areas, is subsequently mentioned, together with recreation and the recent must in
several countries: re-naturaiization.
As far as the means are concerned, they are briefly mentioned in correspondence
to each specific goal, even if in some cases they may serve multiple purposes.
The problems that can arise are in fact of various types and may result as a
consequence of different interventions. Their classification in the graph of Fig. 6 is
rather made according to the different approach which should be followed to solve the
problem itself: purely hydraulic approach, hydraulic and sedimentological approach,
water-quality approach, chemical and biological approach, ecological approach. Each
approach, as said in section 3, requires a different type of model.

Fig. 6 - A few examples of goals and means in river engineering, together with
the problems which can correspondingly arise.

332
 It is apparent that a complete examination of all the problems and a comparative
discussion of all the models is not possible here. Instead, for sake of example, the
attention has been focused on some problems arising from the construction and
management of reservoirs for water storage. In the following sections these problems
will be examined and the available (or desirable) models to be used will be discussed.

7. GENERAL DEGRADATION DOWNSTREAM OF A RESERVOIR.

This problem has often been perceived as one of the most serious posed by the
construction of a reservoir. In fact, the almost complete interception of the solid
transport (especially of the coarser fractions) by the dam will substantially deplete the
natural input of material, with consequent degradation of the river downstream.
The application of conventional morphological models, e.g. the HEC model used
and distributed by the U.S. Corps of Engineers, provides wrong results (exaggerated
degradation) inasmuch uniform sediment-size is considered in the sediment balance
equations. Indeed, if the model does not reproduce the progressive coarsening of the
bottom composition (armouring effect), the erosion rate results to be much larger than
in reality. A too large degradation, for instance, was predicted by assuming uniform
size material in the river Rhone between Geneva and Lyon, where unnecessary
remedial measures were taken before the construction of the barrage (Bouvard, 1992).
Different attempts have been made to incorporate the armouring process in the
uniform sediment-size evolution model (as HEC). It should be understood, however,
that most methods are not physically based as they inevitably violate the mass-
conservation equation of each single size fraction.
On the other hand, an evolution model is not necessary if one is not interested in
the space- and time variations of the river far downstream of the dam but contents
himself with the prediction of the final equilibrium configuration not far from the
reservoir. This is in fact what is more important for the safety assessment of the dam
and of the structures nearby. The equilibrium composition of the bottom (“static”
armoured bed) which will eventually be reached downstream of the dam, together with
the corresponding final degradation, can be predicted by some experimental formula
(e.g. Gessler, 1971, or similar) which has been proved reasonably reliable.
An evolution model, by contrast, is needed for evaluating the progressive
reduction of solid discharge, coarsening of bottom composition and bed degradation all
over the hydrographic network down to the lowland stretches of the river and
eventually to the sea. These phenomena are important for their delayed consequences
also far downstream of the dam, where they may be exalted by other causes: reduction
of sediment input due to reforestation and torrent training in mountain watersheds;
quarrying of river sand and gravel; presence of other barrages; river canalization for

334
navigation purposes, etc. For a correct simulation of these processes (which, according
to the size of the river, have adaptation times of decades, centuries or more) a new
conceptual model is needed, capable to simulate grainsize sorting while respecting the
mass-conservation principle of each size.

7.1. The mixing-layer model.

Such a model is obtained by writing the differential balance-equationb of each grainsize

class not only for the stream proper, but also for the bottom layer (mixing layer) where
the sediment particles are provisionally stored but can still exchange with the stream.
The concept of mixing layer, first introduced by Hirano (1973, has been later resumed
by a number of authors (among others, Armanini and Di Silvio, 1982 and 1988; Bettes
and White, 1982; Ribberink, 1987; Parker and Andrews, 1987; Holly and Rahuel, 1990;
etc.).
Models based on the mixing-layer concept give reason for the bottom armouring
process as well as for its final (“static”) configuration as expressed by the Gessler’s
equation.
A mechanism similar to, but less convincing than the mixing layer, has been
proposed by Borah et al. (1982) to put into account the exchange between bottom and
stream.
The concept of mixing layer has been further extended in order to explain the
vertical sorting which sometimes appears in gravel bed mountain stream: namely, the
so-called “pavement” or “dynamic” armouring, which may occur also when coarse
particles are moving and therefore is basically different from Gessler’s “static”
armouring. The extension has been made either by subdividing the “mixing layer” in
sublayers (Di Silvio, 1991) or by considering a continuously variable composition of
the bottom in the vertical direction (Armanini, 1994). It should be noted however that
only the original mixing layer concept has been tested already and extensively applied
with success to practical prdblems.
An important aspect somehow related to the structure and dynamics of the
“mixing layer” is the so-called “hiding and exposure” effect: namely, the decrease of the
(large) intrinsic mobility of the finer particles, protected by the larger ones, combined to
the increase of the (small) intrinsic mobility of the coarser particles, more prominent
than the smaller ones. Although the hiding and exposure effect is usually understood as
a modification of the horizontal entrainment capacity of the flow, it seems more
reasonable, in fact, that it represents a sort of vertical protection exerted by the coarser
particles that provisionally cover the finer ones within the mixing layer.
This effect is put into account by some empirical correction coefficients to be
introduced in the original transport formulae when applied to a non-uniform size

335
mixture. A number of corrections have been proposed (Ashida and Michiue, 1973; Day,
1980; Proffitt and Sutherland, 1983; Misri et al., 1984, etc.).
The experimental correction coefficients proposed by each author should be
applied in principle to the specific transport formula they have been tested for. For
example, the coefficients mentioned before are to be used with, respectively, the
formula of Meyer-Peter and Miiller, Ackers and White, Ackers and White again, Paintal
and Misri.

7.2. The transport formula.

It is always to be remembered that the sediment transport formula introduced in the

model is a crucial factor for any morphological computation. As each transport formula
has been obtained from flume experiments performed in particular conditions, one
should be careful when the formula is extrapolated outside its experimental range. The
reliability of a transport formula, moreover, is substantially related to the reliability of
the roughness predictor (Chezy or Manning coefficient): as the roughness depends on
both grainsize and bedform size, empirical predictors are available for these quantities
(see, for example, the one by Engelung and Hansen, 1967, and the more recent one by
van Rijn, 1984). Unfortunately they are not fully satisfactory because rising,
development and disappearance of bedforms (ripples, dunes, antidunes, also called
fluvial microforms) is a quite complicated phenomenon not yet very well understood.
Bedforms being the result of a flow/transport instability, various mechanisms have
indeed been proposed to explain their (stable) configuration, namely .the size and shape
of the bedforms corresponding to a given flow and a given granulometric bottom
composition. Although very interesting from the phenomenological point of view,
however, similar analysis are rather complicated and not yet developed for practical
applications. Even less known is the phenomenon of bedforms when the sediment is not
uniform (apparently there is a tendence to reduce bottom undulations) and when the
flow is not stationary.
Therefore, if a theoretical transport formula is to be implemented in a
morphological model, one should expect errors up to one or two orders of magnitude in
the estimate of the solid transport. Of course things are much less dramatic if the
formula has been calibrated on the same watercourse (or on a similar one). In this case
it is probably better resorting to a “compact” synthetic formula valid for very specific
conditions, rather than a “complex” analytical one pretending to have a much larger
general validity.

336
 8. GENERAL AGGRADATION UPSTREAM OF THE RESERVOIR.

The rapid deposition of coarse material taking place when the sediment-carrying water
reaches the reservoir, gives origin to a typical delta which progressively increases its
size as the deposition goes on. While the delta front very slowly propagates
downstream within the reservoir, the aggradation pr’.c’1 Y nrogressively propagates
upstream along the river.
Assuming a uniform sediment composition appears to be less crucial for the
aggradation process upstream of the reservoir, than for the degradation process
downstream (see previous section). For this reason a one-dimensional, one-grainsize
morphological model (like HEC) is generally sufficient to describe both the increase of
the delta height and the propagation of the delta profile. By schematizing the river as a

ii ZOj-
after2Y35

4000 5000
Distance from the dam
T Mdlnla

1 1 I
4 '30 '75 '101
1 Mnrslon

4000 5000
Distance from the dam

Fig. 7 - General aggradation upstream of a reservoir obtained by analytical solution.

337
prismatic channel and introducing a few other simplifications, it is also possible to give
a diffusive form to the sediment continuity equation. An analytical solution of this
equation (similar to the heat-diffusion or Fourier equation) can be found, which in
many instances gives very good results when compared to the measured bottom
variations (Fig. 7).

The material deposed along the river (i.e. above the maximum storage level) does
not contribute to the reservoir’s siltation. It should be remembered that while the river
aggradation is certainly beneficial as far as the reservoir’s capacity-loss is concerned, it
may be as well dangerous since, in the long run, it increases the flooding risk for the
riverine areas upstream of the reservoir. Thus, the amount of the progressive deposition
along the river and the propagation celerity of the aggradation profile should be
checked on the model.

9. EROSION AND DEPOSITION PROCESSES WITHIN THE RESERVOIR.

As it was said in the introduction, the processes taking place in the deepest part of the
reservoir (strongly dominated by the vertical velocity component and by the vertical
density gradient) will not be treated in this paper, devoted to the “river” processes.
However, due to the strong oscillations of the water level, a large part of the storage
area is alternatively subject to typical “reservoir” and “river” processes.
For example, the progressive advancement of the delta within the reservoir,
mentioned in the previous section, is markedly controlled by this alternance: if the
waterlevel oscillation during the hydrological cycle is very ample, the delta front, year
after year, tends to move downstream comparatively faster and with a smaller height
than if the waterlevel is steady. It should noted that, consequently, also the aggradation
along the river proper tends to move upstream with a flatter profile.

9.1. Flushing.

Another example of typical “river” process taking place within the storage area is
sediment flushing by emptying the reservoir. During the flushing, coarse particles are
transported not only near the delta but also in the very deepest part of the reservoir; on
the other hand, when the waterlevel is high, only deposition of very fine particles takes
place in this area. In order to have a comprehensive reproduction of the flushing
mechanism, it is necessary resorting again to a not-uniform grainsize model. Moreover,
since very fine particles are transported by the flushing stream in the suspension mode,
a non-eauilibrium formulation is recommended for the solid transport.

338
 In fact, while the transport of coarse material (bedload) exclusively depends on
the local hydrodynamic conditions and on the local bottom composition, the transport
of fine particles (suspended load) also depends on the conditions upstream. To take into
account this circumstance, the one-dimensional sediment balance equation in the stream
is written

JTi - 21
3X
1
(bi Tci
- Ti) (1)

where Ti represents the actual sediment transport of the i-th grainsize class; bi is the
percentage of the i-th grainsize class in the bottom composition; 4” is the adaptation
length of the i-th grainsize class; T,i is the transport capacity of the i-th class, i.e. the
transport in equilibrium conditions with a uniform bottom composition (bi = 1). The
value of T,i is provided by any uniform-flow transport formula (Meyer-Peter and
Miiller, Engelung and Hansen, Ackers and White, van Rijn, etc.), properly corrected
because of the hiding-and-exposure effect; the adaptation length L: can be obtained
either experimentally or by an asymptotic solution of the two-dimensional suspended
transport equations (Di Silvio and Armanini, 1981; Galappatti, 198.5, etc.); an
evaluation of L* is given by the following approximate fomlula:

(2)

where U is the horizontal flow velocity, h the waterdepth, a the bottom roughness, U*
the fiction velocity, wi the particle falling velocity (including flocculation).
Note that for fine particles L* tends to become equal to the particle’s falling
distance Ly = e; this means that the adaptation length of silt and (flocculating) clay

may be even larger than the reservoir length. By contrast, for coarse particles Ly tends
to be zero and Ti to biTci; in this last case the erosion rate from the bottom becomes
abiT,i
as in the usual models where one assumes equilibrium transport (instantaneous
3X

adaptation).
The evolution of the bottom during the reservoir flushing is extremely
complicated because of the irregular fomlation, displacement and disappearance of the
“flushing channels”; yet, it can be reasonably simulated by the one-dimensional model
described above. In fact, as the bottom of a flushing channel is degrading its banks
collapse and the channel moves; on the average, however, the cross-section of the
channel tends to maintain an equilibrium configuration depending on the waterflow and
on the bottom material. In other words, when a bank collapses the channel tends to
move laterally in such a way that the channel cross-section remains more or less

339
unchanged. If the bottom material is homogeneous, one may assume that the channels
will wander over the entire reservoir, eroding progressively its bottom, while keeping
constant the total width of the flow. This width, according to the “regime theory”, can
be assumed proportional to Q m, being the exponent m about 050.6; these values are
apparently supported by the so-called “extremal approach” (Chang, 1988), which
introduces an extra equation based on the minimization or maximization of some
significant hydraulic or sedimentological quantities.
Due to 1:: emnirical foundation (the very concept of the “extrem,: :?proach” has
been often questioned; e.g. see de Vries, 1993), this procedure is necessarily rather
approximate. Indeed, the behaviour of flushing channels is a very complex mechanism
somehow related to the development of fluvial mesoform (bars and meanders). Like
fluvial microforms (ripples, dunes, antidunes) also mesoforms are the product of
flow/transport instabilities which have been studied in a few special cases; but, again,
the development of these theories has not reached the point to be easily applied to all
practical cases.

9.2. Long-term sedimentation.

When the reservoir is filled-up by water, its purely hydrodynamics behaviour should, in
principle, always be simulated by a two-dimensional (laterally averaged) or even by a
three-dimensional model. However, two- and three-dimensional models are not always
necessary. For instance, when density currents are created by the flow through the
bottom outlets, an acceptable schematization is given by a strongly two-layers flow-
pattern in which the sediment transport is confined in the lower and heavier layer.
With other types of discharge from the reservoir (spillway, intake, or no flow at
all), the one-dimensional model described in the previous section is still be applied if
the purpose is to evaluate the long-term process of reservoir sedimentation and its
trapping capacity. The model, in fact, can simultaneously take care of the coarser
particles (gravel and sand) that stop on the delta, as well as of the finer particles (silt
and clay) which in part settle down, selectively, along the deeper part of the reservoir
and in part remain in suspension and are eventually discharged through the spillway
and/or the intake.
The sedimentation rate depends very much on the position of the waterlevel and
on the corresponding liquid (solid) discharge from the river. Due to their continous
variation during the year, a detailed time-history of these data is necessary for a correct
simulation of the filling process. On the other hand, apart from the difficulties to obtain
these data, an hour-by-hour simulation over a period of several decades would require a
huge amount of computer time. A more affordable procedure is to average the
equations and the boundary conditions or, alternatively, to recognize a number of

340
“equivalent operational years” by a statistical analysis. As mentioned in Section 4,
however, the averaging operations or statistical techniques should account for the
“residual terms” due to non-linear character of the transport equations.

10. SEDIMENT DISPOSAL DOWNSTREAM OF THE RESERVOIR.

Disposal of material downstream of a reservoir may give place to various problems.

The first one is related to the ctuantitv of sediments which lllL~y uepobe in the river
during the flushing(or siphoning) operation. If the sediment concentration is very high,
the aggradation may be large enough to cause flooding and inundations in particularly
vulnerable places (e.g. where the river slope locally decreases) downstream of the
reservoir.
Since the sediment concentration depends on the flushing (or siphoning)
techniques, a trade-off should be found between economy (the larger is the
concentration, the more economical is the sediment disposal) and safety. For this
purpose, the one-dimensional, non-uniform grainsize model discussed in the previous
section should be applied to the river reach downstream to find out an acceptable policy
for discharging the sediments from the reservoir. This is in fact generally possible, on
the condition that the sediment disposal is regularly made every two or three years, or
even less whenever necessary.
The second type of problem created by sediment disposal is the water quality in
the river downstream of the reservoir. The water quality here obviously depends on the
quality of the sediments within the reservoir and, therefore, on the quality of the
sediments (often of agricultural origin) washed out from the watershed. Again, the
water quality downstream the reservoir will be negatively affected if the sediment
disposal has a very high concentration anti produces an excessiv; ?~~~+tion of polluted
sediment. Also in this case, the already mentioned one-dimensional model may be
utilized to evaluate the fate of sediments downstream of the dam, with special attention
to the finest fractions as pollutant carriers.
Besides the poor quality of the water, possibly caused by agricultural non-point
pollution, a further problem may be caused by the turbidity itself. Although adult fishes
can apparently resist (for a limited period of time) to a reasonable increase of turbidity,
their eggs in the spawning places along the river seem to be particularly vulnerable to
the infiltration of fines in the river bed, especially for very delicate and rare species
present in some torrents. Also this problem may of course be tackled by the same
model, in a strict collaboration with fish specialists.
In regard to sediment disposal, it is to be mentioned that, due to a scarce
knowledge of the processes mentioned above, in some countries any flushing or
siphoning operation from reservoirs has formally 5,:en forbidden. The sediments should

341
be dredged and mechanically transported far away; but since the regulations prevent the
opening of the bottom outlet, the water level in the reservoir cannot be lowered
anymore below the intake elevation.
It is the opinic:. of the writer that an interdiscip!inan/ analysis of each particular
problem created by sediment disposal, carefully supported by model simulations, would
lead in many cases to a more practical solution both from the economical and
ecological point of view.

11. MODELS FOR UPLAND RESERVOIRS: CONCLUSIONS

For most of the problems connected to the river processes upstream of, downstream of
or within an upland reservoir, a one-dimensional model is capable to give satisfactory
results, provided that the model can account for different grain-size classes ranging
from gravel to silt and clay. For the particles transported in suspension within the
reservoir, moreover, the model should also consider their adaptation (deposition)
length. Since for very fine sediments the deposition length is much longer than the
reservoir itself, it explains why the trapping effect of a reservoir is less than one.
It is to be stressed that failing of accurate predictions in mountain river
computations does not depend so much on the model inadequacy but rather on a poor
knowledge of boundary conditions (especially of the sediment input).
One of the best ways of assessing the annual sediment input in the reservoirs is
probably by computing it from the hydrological, geometrical and sedimentological
characteristics of the inflowing river. A reliable transport formula, needed for this
purpose, should be calibrated against sediment measurements in the river itself or in
rivers with similar characteristics.

12. REFERENCES

- Ackers, P. and White, W.R. (1973), “Sediment transport; new approach and
analysis”, J. Hydraulic Div., ASCE, Vol. 99, HY 11.
- Armanini, A. (1994), “Non uniform sediment transport: dynamics of the active
layer”, Dept. of Civil and Environmental Eng., University of Trent, Italy.
- Armanini, A. and Di Silvio, G. (1982), “Sudden morphological modifications along
a mountain river simulated by a mathematical model”, 3rd Congress of the Pacific
Regional Division of IAHR, Bandung, Indonesia.
- Armanini, A. and Di Silvio, G. (IS@), A one-dimensional model for the transport of
a sediment mixture in non-equilibrium conditions”, J. of Hydraulic Res., IAHR, Vol.
26. n. 3 and discussion.

342
- Ashida, K. and Michiue, M. (1973), “Studies on bedload transport rate in alluvial
streams”, Trans. ASCE, Vol. 4.
- Bettes, R. and White, W.R. (1981), Proc. Institution of Civ. Engs 71, Part 2, pp. 879-
892.
- Borah, D.K.; Alonso, C.V. and Prasad, S.N. (1982), J. Hydraulic Engineering ASCE,
108(12), pp. 1486-1500.
- Bouvard, M. (1992), Mobile Barrages and Intakes on Sediment Transnorting Rivers,
IAHR Monograph Series, Balkema.
- Bruk, S. (1994), “River response to natural and man-made change”, IHP-IV, Project
H-l-2, UNESCO (in preparation).
- Bruk, S., rapporteur, (1985), Methods of comDutinP sedimentation in lakes and
reservoir, UNESCO, Paris.
- Chang, H.H. (1988), Fluvial Hvdraulic Processes in River Engineering, John Wiley
& Sons.
- Day, T.J. (1980), “A study of the transport of graded sediments”, HRS Wallingford
Report, n. IT190, April 1980.
- de Vries, M (1993), Use of models for river problems, Studied and rep. in
hydrology, 5 1, UNESCO.
- Di Silvio, G. (1986): “Validity and limitations of different transport models with
particular reference to sediment transport”, Workshop on Agricultural Nonpoint
Source Pollution, Development in Environmental Modelling, n. 10, Elsevier Scienze
Publ.
- Di Silvio, G. (1991), “Sediment exchange between stream and layer model”, Proc.
Int. Grain Sorting Seminar, Ascona, Switzerland, 21-26 October 1991, Vaw,
Technischen Hochschule Zirich, Mitt. 117.
- Di Silvio, G. (1994), “Mobile bed river modelling”, UNESCO, IHP-IV, Project H-l-
2, Int. Workshop on Erosion, River Bed Deformation and Sediment Transport in
River Basins as Related to Natural and Manmade Changes, St. Petersburg, 16-21
May 1994.
- Di Silvio, G. and Armanini, A. (1981), “Influence of the upstream boundary
conditions in the erosion-deposition processes in open channels”, Paper n. 22, Subj
A(d), XIX Congress IAHR, New Delhi, India.
- Engelung, F. and Hansen, E. (1967), A monorrraph on sedimei;t transport in alluvial
streams, Teknick Forlag, Copenhagen, 62 p.
- Galappatti, R. and Vreugdenhil, C.B. (1985), “A depth-integrated model for
suspended sediment transport”, J. of Hydraulic Res., Vol. 23, n. 4, p. 359-375.
- Gessler, J (1970). “Critical shear stress for sediment mixtures”, Proc. XIII Congress
IAHR, Paris.
- Graf, W.H. (1971), Hvdraulics of Sediment Transport, McGraw Hill.

343
- Hirano, M. (1971), “River bed degradation with armoring”, Trans. of JSCE, Vol. 3,
part 2.
- Holly; F.M. and Rahuel, J.L. (1990), “New numerical/physical framework for
mobile-bed modelling”, Part I, Numerical and physical principles, J. Hydraulic Res.,
IAHR, Vol. 28, n. 4 Part II, Test applications, J. Hydraulic Res., IAHR, Vol. 28, n. 5.
- Misri, R.L., Garde, R.J. and Ranga Raijn, K.G. (1984), “Bedload transport of coaarse
nonuniform sediment”, J. of Hydraulic Eng., Vol. 110, HY3.
- Parker, G. and Andrews, E.D. (1987), Water Resources Research, 21(9), pp. 1361-
1373.
- Pianese, D. and Rossi, F. (1989), “Morphological changes and grain-sorting in
mountain gravel-bed streams”, Proc. Int. Workshop on Fluvial Hydraulics of
Mountain Regions, Trento, Italy, pp. C 15-C36.
- Proffit, U.T. and Sutherland, A.J. (1983), “Transport of non-uniform sediment”, J.
Hydraulic Res., IAHR, Vol. 21, n. 1.
- Raudkiwi, A.J. (1990), Loose boundary hvdraulics (3rd Ed.), Pergamon Press,
Oxford.
- Ribberink, J. (1987), “Mathematical modelling of one-dimensional morphological
changes in rivers with non-uniform sediments”, Ph.D. Thesis, Delft University of
Technology, the Netherlands.
- van Rijn, L.C. (1984), “Mathematical modelling of morphological processes in case
of suspended sediment transport”, Delft Hydraulics Communication n. 382.
- van Rijn, L.C. (1984), “Sediment transport”, Part I: Bed load transport, Part II:
Suspended load transport, Part III: Bedforms and alluvial roughness, J. Hydraulic
Eng., ASCE, Vol. 110.

344
 On methods for the prediction of river
planform changes

by Erik Mosselman
Delft Hydraulics, P.O. Box 152,830O AD Emmeloord, The Netherlands

Abstract

Theoretical and empirical approaches to the prediction of river planform changes are
discussed. Theoretical models based on the mechanics of flow, sediment transport and bank
erosion have been developed in the past fifteen years, but empirical methods cannot be done
without. The empirical methods, however, should go beyond the traditional regime theory.
The use of extremal hypotheses is strongly discouraged.

Acknoweledgements

Most of this paper were written duriong my employment at Delhi University of Technology.
The study on the Brahmaputra-Jamuna River was carried out at Delft Hydraulics for FPCO
(Flood Plan Coordination Organization, Bangladesh), with funding from KfW (Kreditanstalt
fur Wiederaufbau, Germany) and CFD (Caisse Fraqaise de DCveloppement, France).

345

--
1. On methods for the prediction of river
planform changes

1.1 Introduction
Land use planning in alluvial river valleys and the choice of locations for bridges and
hydraulic structures require predictions of future river planform changes. The usual
mathematical models for river morphology are based on the conservation of mass and
momentum of small volumes of water and sediment, whereas predictions of river bank
migration commonly span long periods of time. There is a considerable gap between the
scales of the phenomena and the processesactually modelled. Therefore, it is important to
reflect upon the possibilities and limitations of this approach, and to see whether other
methods could be more powerful.

1.2 Justification and limitations of a mechanistic approach

The general method of scientific inquiry is Aristotle’s procedure of induction and deduction,
in which first principles or fundamental laws are induced from observations and then tested
by comparing statementsdeduced from them with other observations. Fundamental laws are
usually on a lower level of complexity than the observed phenomena. The fundamental laws
of classical mechanics, which includes river mechanics, have reached a high cognitive status.
They are descriptions for underlying meChdrbS on a small, but superatomic scale.
Formally, they are still empirical, but their greater generality and more inclusive explanatory
power justifies that they are called ‘theoretical’ rather than ‘empirical’. As they appear to be
the soundest basis of knowledge, deduction must be a progression from smaller towards
larger scales. Ikeda & al (1981), Thorne & Osman (1988), Johannesson& Parker (1989),
Crosato (1990), Mosselman (1992) and Darby (1994) follow this approach in their models
for river planform changes. However, deducing predictions of phenomena on larger scales
by formal integration over time and space is not a direct integration, but a series of
integration steps to ever higher levels in a cascadeof scales. Due to the nonlinearity of the
system, each step requires the introduction of empirical closure relationships. This implies
that even ‘theoretical’ models become to a large extent empirical, and hence inaccurate.
Additional problems arE:the possibility ot Chauk t‘-haviour, the interaction at larger scales
with the dynamic systems of climate, geological activity ancl :?getation, and the necessity of
including exceptional events that can be disregarded in predictions for shorter time spans.

346
8

1.3 The need of empirical approaches

Because of the limitations in deduction from underlying mechanismson a small scale, also
deduction from relationships on a higher level of aggregation deserves attention, i.e.
relationships formulated on a larger scale for integrated quantities. These relationships must
be induced directly from observed large-scale river phenomena. In fact, the deduction from
fundamental laws and the induction from observations are complementary. Both are
necessaryand should be used to cross-fertalize one another, as in the ‘mixed approach’of de
Vriend’s (1991) methodology of process analysis for coastal morphology.
Interestingly, deduction and induction of relations for river meander migration have
produced similar meander simulation models. Parker (1983) argues that the kinematic
meander model induced by Howard (Howard, 1983; Howard & Knutson, 1984) is equivalent
to the bend equation deduced by Ikeda & al (1981). A difference is that Howard’s kinematic
meander model contains an additional factor based on Hickin & Nanson’s (1984) nonlinear
relationship between migration and channel curvature. It is tempting to speculate that this
additional factor corresponds to the deviations between the first-order model of Ikeda & al
(1981) and the more complete second-order models of Johannesson& Parker (1989) and
Crosato (1990).
Empirical approachesshould not be limited to the establishmentof some rules of thumb
for equilibrium states, such as in the traditional regime theory, but should also be applied to
transient process-responselinkages on a high level of aggregation, such as in Howard’s
kinematic meander model. A recent example is the development of a model for planform
changes of the braided Brahmaputra-JamunaRiver in Bangladesh O(laassen & al, 1993).
Here the phenomenaare decomposedinto a number of aggregatedmechanismsfor individual
channels within the braided system, which are studied and modelled separately. These
aggregated mechanisms are width adjustment, channel migration, island formation, channel
creation and channel abandonment. Individual channels gain importance or decline
continuously due to changesin the supply of water and sediment from upstream. This supply
depends on the distributions of water and sediment at upstream bifurcations, which are
partially determined by the geometry of the bifurcations. Mosselman & al (1995) find a
correlation between channel abandonmentand detlection angle (i.e. angle between channel of
interest and channel upstream, at the bifurcation), but no correlation between width
adjustment and deflection angle, cf Figures 1 and 2. Apparently channels are abandoned
through shallowing rather than narrowing. Also no correlation between width adjustment and
other geometrical parameters is found. Hence no process-response linkage for width
adjustment can be induced, but Figure 2 shows that a probabilistic approach is possible in
which the annual variations in width are modelled as a random process with average zero
and standard deviation equal to 0.46 times the width. Klaassen & al (1993) explain that
thresholds in the morphological behaviour dictate that the probabilistic elements cannot be
dealt with by a parameterization of stochastic distributions, but must be included through
monte car10simulation.

347
 9

percentase
100%

80%

60%

I
30-39 40-49 50-59 >60
deflection angle (degrees)

Fig. 1 Relation between probability of channel abandonment and dejlection angle

200 n

a
m
r 100 0

-100 I I I I I
0 20 40 60 80 100 1 !O
deflection angle (degrees)

Fig. 2 Relation between width adjustment and deflection angle

1.4 The misuse t” qime theory

The regime theory predicts equi!ib-i*~~lI C!WS of width, depth and streamwise slope as a
function of discharge and, in later versloils, GUIi,xnt and bank properties. Lacey (1930) and
others induced the regime equations from observations in alluvial irrigation canals in the
Punjab in present-day Pakistan and India. These car.als are uniform straight channels with a
constant discharge, a sand bed with dunes, steep cohesive banks, small amounts of sediment

348
10
transport and Froude numbers below 0:05 (Blench, 1969). The regime equations are
statistical best-fit relations and large deviations from them are observed. For rivers in
regime, these deviations are usually explained from (1) not having included all the relevant
variables into the equations, and from (2) dealing with a dynamic equilibrium in which the
geometry is in a continuous state of delayed adaptationto a fluctuating discharge.
Those who only give these two explanations assumetacitly that the equilibrium hydraulic
geometry of an attuvial channel is a unique hmction of water discharge, sediment supply and
bed and bank material. This hypothesis, however, is a misconception. The equilibrium width
of a river, and hence the other parameters of its hydraulic geometry, can take any value
from a certain range between an upper and a lower limit. The hydraulic geometry actually
established depends on the initial and boundary conditions which govern the preceding
transition towards the equilibrium. This multivaluedness of equilibria is receiving renewed
attention due to the recent adoption of concepts of nonlinear dynamics in geomorphology
(e.g. Phillips, 1992), but is already stressed by Blench (1969) who writes ‘rf the attack
exceedssome limit then the sides will erode till the attack drops to what they can withstand.
If the uttuck fulls short of a lower limit then materials from the total loud can deposit on the
sides, and the channel breudth will decrease. If the attack is between these limits nothing
will happen to the sides’. The multivaluedness implies that the averages represented by the
regime equations have no special dynamical significance. The ASCE Manual on
Sedimentation Engineering (Vanoni, 1975) rightly prescribes that ‘one should study Blench’s
rather thorough exposition of the “regime’.’philosophy before undertaking use of the regime
method‘.

1.5 The fallacy of extremal hypotheses

Extremal hypothesesare another type of formulations on a high level of aggregation. They
state that a river adjusts its geometry in order to minimize or maximize a certain quantity,
such as energy expenditure, stream power, or sediment efficiency. They do not have the
generality of fundamental laws of physics, but they appear to be mathematically equivalent
to equations deduced from fundamental laws for special simplified cases. For instance, Yang
& Song (1979) show that the velocity distribution which satisfies a iinearized momentum
equation without inertia terms, is the one that minimizes the total rate of energy dissipation.
The limited validity is substantiated by Griftiths (1984) and Lamberti (1988, 1992), who
demonstrate that, generally, conclusions from extremal hypotheses are not compatible with
observations.
.\nother objection against extremal hypothesesis the teleological form in which they are
presented. This form carries a connotation of anthropomorphism, as if rivers are living
organisms with conscious goals. A few centuries ago, Leibniz and Maupertuis even held the
opinion that the possibility to express a relation in physics teleologically by an extremum
principle was evidence of the purposive activity of the Creator. Although teleological
explanations of things and processesare accepted in biology as merely, a way of speaking
and not necessarily based on the premise of a conscious goal, they are not accepted in
mechanics where they do imply the connotation (Nagel, 1961). Extremal hypotheses are
sometimes called ‘modern’, but the adjective ‘old-fashioned’ might be more appropriate
since their teleological character places them closer to Aristotle’s notion of final causes than
to modern concepts in the philosophy of science.
The fundamental equations for the morphology of rivers with mobile banks form a
determinate system which, theoretically, can be solved when the initial and boundary
conditions are known (Mosselman, 1992). Indeterminacy arises, however, when only the
steady state is considered, which complies with the multivaluedness of the equilibrium

349
 11
width. Some researchers believe that the equilibrium should be singlevalued and that hence
an extra law of nature should exist to close the system of equations. An extremal hypothesis
is sometimes proposed to provide this extra closure relationship. But as an extremal
hypothesis appears to be just an alternative formulation for a simplified version of a
fundamental law already used, the only new information it brings in is equivalent to the
conditions under which the simplification of the fundamental law is valid.

1.6 Compatibility of mechanistic width adjustment with regime

width predictors
The observational evidence that river width, B, is proportional to the square root of
discharge, @ ‘, is quite strong, despite the shortcomings of regime theory and
notwithstanding Rhoads’ (1992) suggestion that some researchers introduce bias by an a
priori assumption that the exponent in the width-discharge relation equals 0.5. Mechanistic
approachesshould be able to produce similar relationships. Parker’s (1978) approach based
on lateral momentum exchange and a balance between erosion and deposition of suspended
sediment does not lead to this relation between discharge and width (cf Bettess & al, 1988),
but Darby’s (1994) approach to river widening with a geotechnical bank stability model
does.
A simple and rough analysis of how mechanistic width adjustment can comply with
regime width predictors is given below, following essentially the same lines of thought as
Bettess & al (1988) in their ‘initial approach to channel width’. A straight narrow river with
cohesive banks and a consld.. ,iischarge will widen until it reaches its smallest possible
stable width. The flow velocity along the banks, ubank,is then equal to the critical flow
velocity, u,, below which no bank erosion occurs. This critical flow velocity is a soil
material property and can be assumedto be constant

‘bank = ‘c = constant (1-l)

It is plausible that the ratio of near-bank flow velocity to cross-sectionally averaged flow
velocity, zi, is a function of the width-to-depth ratio, B/h,

O-2)

in such a way that ubankapproachesti if the river is very narrow

(l-3)

and u,, becomes much smaller than z.iwhen the river is so wide that the major portion of
the flow is concentrated in a few deep channelswithin the cross-section

bu,=o
B (l-4)
-..a -
u
ii

A simple functional relationship satisfying these conditions is

350
 12

u-3

Substituting the identity of Eq.(l) and eliminating ti by using the continuity equation,
Q = Bhi, one obtains

B = <Qo3 (l-6)
UC

This is not a proof that B is proportional to @ ‘, but it demonstrates that a mechanistic

approach does not need to be incompatible with regime width predictors. Furthermore, it
shows that the shape of the cross-section, the associatedcross-sectional distribution of flow
velocities (or shear stresses) and the resistance to bank erosion are key elements when
modelling width adjustment.

351

- - --- ..-~-- ..___

References

BFITESS, R., W.R. WHITE & C.E. REEVE (1988) ‘On the width of regime channels’. Int. Conf.
River Regime, 1988, Wallingford.
BLENCH, T. (1969) Mobile-bedfluviology. Univ. of Alberta Press, Edmonton.
CROSATO, A. (1990) Simulation of meandering river processes. Communications on Hydr. and
Geotech. Engrg., No.90-3, Delft Univ. of Technol., ISSN 01696548.
DARBY, S.E. (1994) A physically-based numerical model of river channel widening. Dissertation
Univ. Nottingham.
GRIFFITHS, G.A. (1984) ‘Extremal hypotheses for river regime: an illusion of progress’. Water
Resources Res., AGU, Vol.20, No.1, pp. 113-118.
HICKIN, E.J. & G.C. NANSON (1984) ‘Lateral migration rates of river bends’. J. Hydr. Engrg.,
ASCE, VoI.110, No.11, pp. 1557-1567.
HOWARD, A.D. (1983) ‘Simulation model of meandering’. In: C.M. Elliott (ed.), River Meandering.
Proc. Conf. Rivers 1983, New Orleans, ASCE, 1984, pp.952-963.
HOWARD, A.D. & T.R. KNUTSON (1984) ‘Sufficient conditions for river meandering: a simulation
approach’. Water Resources Res., AGU, Vol.20, No. 11, pp. 1659-1667.
IEDA, S., G. PARKER & K. SAWAI (1981) ‘Bend theory of river meanders, Part 1, Linear
development’. J. Fluid Mech., vol. 112, pp.363-377.
JOHANNESSON,H. & G. PARKER (1989) ‘Linear theory of river meanders’. In: S. Ikeda & G. Parker
(eds.), River meandering. AGU, Water Resources Monograph 12, pp.181-213.
KLAASSEN, G.J., E. MOSSELMAN & H. BROHL (1993) ‘On the prediction of planform changes in
braided sand-bed rivers’. Adv. in Hydro-Sci. and -Engrg., Ed. S.S.Y. Wang, University
MS, pp. 134-146.
LACEY, G. (1930) ‘Stable channels in alluvium’. Minutes Proc. Inst. Civil Engrs., 1929-1930,
vol.229.
LAMBERTI, A. (1988) ‘About extremal hypotheses and river regime’. Int. Conf. River Regime, 1988,
Wallingford.
LAMBJZRTI, A. (1992) ‘Dynamic and variational approaches to the river regime relation’. In: V.P.
Singh & M. Fiorentino (eds.), Entropy and energy dissipation in water resources. Kluwer,
Water Sci. and Technol. Library, Vol.9, pp.507-525.
MOSSELMAN, E. (1992) Mathematical modelling of morphological processes in rivers with erodible
cohesive banks. Communications on Hydr. and Geotech. Engrg., No.92-3, Delft Univ. of
Technol., ISSN 0169-6548.
MOSSELMAN, E., M. HUISINK, E. KOOMEN & A.C. SEYMONSBERGEN (1995) ‘Morphological changes
in a large braided sand-bed river’. In: E.J. Hickin (ed.), River geomorphology. Wiley,
pp.235-247.
NAGEL, E. (1961) 7he structure of science.Second Ed. (1979), Hackett, Indianapolis.
PARKER, G. (1978) ‘Self-formed straight rivers with equilibrium banks and mobile bed. Part 1, The
sand-silt river’. J. Fluid Mech., vol.89, part 1, pp.109-125.
PARKER, G. (1983) ‘Theory of meander bend deformation’. In: C.M. Elliott (ed.), River Meandering.
Proc. Conf. Rivers 1983, New Orleans, ASCE, 1984, pp.722-732.
PHILLIPS, J.D. (1992) ‘The end of equilibrium ?’ Geomorphol., Vol.5, Nos.3-5, pp. 195-201.
RHOADS, B.L. (1992) ‘Statistical models of fluvial systems’. Geomorphol., Vol.5, Nos.3-5, pp.433-
455.

352
14

THORNE, C.R. & A.M. O&AN (1988) ‘Riverbank stability analysis. II: Applications’. J. Hydr.
Engrg., ASCE, VoI.114, No.2, pp.151-172.
VANONI, V.A. (ed., 1975) Sedimentation engineeritzg. ASCE Manuals and Reports on Engrg. Practice,
No.54.
VRIEND, H.J. DE (1991) ‘Mathematical modelling and large-scale coastal behaviour. Part 1: Physical
processes’. J. Hydr. Res., IAHR, VoI.29, No.6, pp.727-740.
YANG, C.T. & C.C.S. SONG (1979) ‘Theory of minimum rate of energy dissipation’. J. Hydr. Div.,
ASCE, Vol. 105, No.HY7, pp.769-784.

353
Prediction of long-term evolution of
lowland river channels
V.A.Bazilevich, V.V. Kozitsky, J.A.Gaiduchenko
The insnitute of Hydromechanicsof the
Academy of Sciencesof Ukraine

Abstract:
Methodology is presentedto predict long-term channel evolution of lowland rivers; this
methodology is based on the calculation of the balance of suspendedchannel-forming
sedimentsand bedload. Recommendationsare made to determine sediment yield, to
separatesuspendedsedimentsinto channel-forming and transit fractions, to take into
accountthe nature of bed evolution (scour, sedimentationand transport of sediments)at
each river reach, to plot a design, mean long-term hydrograph, to take into account
bank erosion, and to calculate annual channel-forming sediment yield variable due to
channelevolution.
General channel changesare calculated on the basis of the computed balance of
channel-forming,sedimentsdown the river. In lowland rivers with sandy bottoms there
are suspended channel-forming sediments and bedload; moreover, the portion of
suspendedsedimentsprevails.
To calculatethe yield of suspendedchannel-formingsedimentsit is recommended
to apply the following equation (Bazilevich, V.A. and V.V.Kozitsky, 1987; Bazilevich,
V.A. and V.V.Kozitsky, 1985):
Qp=A- hP4 b, - *S,,qb,
gw, wP
where: A is coefficient of proportionality which should be established(Bazilevich, V.A.
and V.V.Kozitsky, 1987; Bazilevich, V.A. and V.V.Kozitsky, 1985; Bazilevich, V.A.,
N.K.Gribovskaya and V.V.Kozitsky, 1988)
for each river reach from the results of field measurementsof sedimentyield (for the
Dnieper river at Kiev (Bazilevich, V.A., N.K.Gribovskaya and V.V.Kozitsky, 1988) the
averaged,value of Ap, / g = 3gs2I m4 ; PH is density of sediments; u - mean flow
velocity; g - free fall acceleration; w, wrp - mean weight fall velocities for channel-
forming and transit suspendedsediments;bH - width of the flow transporting channel-
forming sediments; srp - gravimetric sediment concentration in suspended transit
sedimentsacceptedto be constant all over the study reach,just as the value of wrp , (i.e.
independentof the water flow velocity).

354
 Suspendedsedimentsare subdivided into channel-forming and transit fractions
with the account of their granulometriccomposition,accordingto (Bazilevich, V.A. and
V.V.Kozitsky, 1986):
wT= 0.04 i!d+ (2)
where: wr - boundary fall velocity; U* - dynamic flow velocity determined from the
Chezy-Manningequation:
u, = J-gnp I H’l6 (3)
np - roughnesscoefficient of the channel(for the Dnieper river at Kiev the value of np,
derived from the results of water surface slope measurements,was equal to 25 mslm’”
which was proved by the analysis (Bazilevich, V.A., N.K.Gribovskaya and
V.V.Kozitsky, 1988) of the results of suspendedsedimentyield measurements);H -
mean flow depth.
Analysis of data on the yield and granulometric composition of suspended
sedimentsin the Dnieper at Kiev and in the Desnariver at Chernigov (with the account
of time lag between these sites) made witb the account of equation (2) and given in
hydrological yearbooks made it possibleto acceptvalue of srp=243 g/m3 for the study
reach in the Dnieper, at the total water discharge at gauging station to be equal to
Q=19.700 m3/s,st,=17g/m3at Q=6000 and 2000 m3/sand w,,=O.250mm/s (diameter of
transit sedimentsd,=O.O2mm) - at all Q values.
Computation of channel-forming suspendedsediment yield by the considered
methodology should be made with the account of the nature of bottom changes(scour,
sedimentation or transport of sediments) which is determined by a comparison of
channel-formingsediment yields transported over the study reach and imported from
beyond (Bazilevich, V.A. and V.V.Kozitsky, 1985). Proceeding from this, when
computationsare made by equation (1) the fall velocity of sedimentsscoured from the
bottom (WY) should be accept as wp value for each design reach, and then, if it
happensthat sedimentationor transit of sedimentsoccurs on the study reach, the fall
velocity of suspendedchannel-forming sedimentscontributing to the study reach from
the upper reaches.Here methodology is used for a determination, of the size of the
scoured sediments(Bazilevich, V.A. and V.V.Kozitsky, 1986) on the basis of data on
the granulometriccomposition of sedimentsin the river channeland the value of u in the
lower end of the study reach. When, the initial data are restricted by a single curve of
the granulometric composition of sedimentsfor the whole study reach (as it was in the
caseof the Dnieper at Kiev) the determinationof the value of wp may be simplified, if
we apply a graphic dependenceof w,bum=f(u+ preliminarily plotted for that river reach.
To determine the bedload yield (Qb) it is recommendedto use a dependence
obtained for the case of H 2 1 m on the basis of recommendationsin (Snishchenko
B.F., 1980) resultedfrom field and laboratory tests on sand-dunesmotion:
al =~~b~(O.ll/H~‘~ +0.06/H”*) (4)
When long-term channelevolution computed, it is necessaryto take into account
changes in water discharges (and, consequently change in water flow effect on
sedimentstransport and channelchanges)during eachyear of the designtime period. It
is impossibleto predict any alternation of different floods during a long term period.
Therefore, the only design long-term hydrograph of the maximum discharge(say 1%
frequency,which occurs every year but which would neverthelesstakes into accountthe
probability (frequency) of each water dischargefor a long-term period) was used in the

355

-.- ___.
consideredmethodology to describe changesin water discharge during a year. This
hydrograph should be plotted on the basis of the analysisof bydrographsmeasuredat
the study reach during different characteristicyears. Besides,if a power generationplant
is constructedjust downstreamthe study reach, it is necessaryto take into account all
the changesintroduced to river hydrographs by the regime of water control. For
example, the Dnieper river reach at Kiev is in the downstream of the Kiev power
generationplant (operating in the regime of daily control); when that portion of the
designhydrograph was plotted which correspondedto the low-water period arid low
floods (Q I 4,000 m3/s),the working graphsof the power plant were analysedtogether

t,ts,
with the resultsof model tests of flood wave transformationalong the study reach.
The analysisof hydrographsmeasuredin the Dnieper at Kiev made it possibleto

tl,
establishthat each water discharge during a year (with some approximation) had a
certain duration irrespectiveof the fact that the dischargecorrespondedto the flood
peak dischargeduring that year or it was less than peak dischargesduring other years.
Using parameter mean long-term steppedhydrograph,its so-called“shelf’ for each
s-th water dischargein the hydrographmay be determinedas follows:
11s= QOl(Os - $s-1+(,-l))
where 8, - water dischargefrequency, in %; s - number of water dischargein the
design hydrograph, counting from the maximum discharge ( 1%~frequencyfor the
Dnieper at Kiev).
The design hydrograph is used to determine the yearly yield (R,) of chamrel-
forming sedimentsat the outlet of the designreach under the conditions of the original

0
channel,i.e. in the first cycle of calculations(i = 1):
qkl) = gQ&L~6400
where: po - density of channelsediments(for the Dnieper river it
was accepted that p. = 1700 kg/m3); Qz - yield of channel-forming sediments
(suspendedsedimentsand bedload) transported from the outlet of the design reach
down the river at the s-th water dischargeand it is derived Corn equations(1) and (4):
Q:=Q;+Qt
Sediment yield QE for all s-th water discharges (into which the design,
hydrographis separated)should be determinedfrom graph dependencesof Q,” = f(Q)
preliminarily plotted for each gauging site on the basis of discharge Q,” at the gauging
site computed for three typical yields of Q (for the Dnieper at Kiev sedimentyields of
19.700, 6000 and 2000 m”/s correspondingto maximum flood water dischargesof 1,
50 and 100% frequencieshave been selected).
Total river channelchangeon the study reach during a year is determinedby the
yield of annual channel-forming sedimentstransported to the study reach (I?:) and
dischargedat the outlet ( RH ) . When long-term channelevolution is computedwith the
account of subsequentchangesin water flow parametersat dischargesites it is most
reasonableto consideryearly changesin the cross-sectionarea at the outlet, which are
causedby bottom deformations:

Ao, I At = (&H - I?,$1 - 0.05.0.5(~k& I b,&, +RHkbR I bnycp) (5)

356
and by bank erosion:

Aor I At = 0.05RHkb”,I bn&,Kb”, (6)

where: 1 - length of the designreach, i. e. the distancebetween design gauging sites in
the middle of the channel;kbR- the portion of annual sedimentyield correspondingto
possiblebank erosion (determined for each gauging site under the condition that mean
flow velocity at the bank foot exceedsits nonerodiblevalue for soil which makes bee
foot of the bank); on the basis of analysis of epures of plane velocities distribution
(Bazilevich, V.A. and V.A.Tkachenko, 1982) it was establishedthat in undisturbed
slightly winding channelsit is possibleto use velocity u mean over the cross-sectionas a
velocity at the bank foot which is quite sufficient for practical computations; b,., -
width of the flow transporting channel-forrtung sediments, averaged for design,
hydrograph dischargesin case of mean water dischargeQC,,there (for the Dnieper at
Kiev the value of QCp= 6000 m3/s); kb.p - amount of channel-forming sediments
averagedfor the dischargesof designhydrograph (in parts of 1) in the soil which makes
the river banks (determinedfrom the data of granulometriccomposition of the bank soil
or channel deposits and equation (2) for boundary fall diameterstransit and channel-
forming fractions); indices “H” and “I?” correspondto lower and upper sites of the
designrepch.
Formula (6) and the second term in equation (5) here obtained fr”nr field
investigations(Vikulova, L.I., 1973) and from experiments(Bazilevich, 1986) for the
casesof intensive erosion of the both banks in a slightly winding channel. For the
Dnieper at Kiev these formulas were used only for calculating a branch reach in the
floodplain (the Desenkariver) where a man-madechannelmade during the construction
of the Moscow bridge should have a trend to widen up to the width of a dynamically
stable channel.Moreover, it was accepted,that the bottom of that man-made channel
and its banks were composedof the same soil as bottom sedimentsin the ZL:Ter (d =
0.44 mm in mean diameter). The value of kh.+was equal to 0.815 for the granulometric
composition, of those sedimentsat any value of u+,In caseswhen the channelwidth in
slightly winding channelswas formed by the water flow in undisturbed conditions, the
bank erosion may be neglected,and when long-term channelevolution is computed the
value of Ace: /At (6) and the second term in equation (5) should be acceptedto be
equalto zero. This condition was acceptedat the computationof long-term evolution of
the main charu3elof the Dnieper and of its other branchesin the floodplain (moreover,
the horizontal position of the channelthere was fixed by bridges, dykes and houndings).
The duration if the computation cycle t: (in years) according to considered
procedure is determined procep:dingCorn the condition that channel evolution during
the cycle (in the limits of which the flow parameters are taken constant) in each
dischargesite does not exceed 15 % of the cross - section area of the flow 0$ in the
caseof averagedischargein hydrograph:

f’ = O,lfiwH,/ I(A,:: / At)i +(Ao,H / At)ii.

357
 Minimum value of all values of tl!’for all dis-h- 7 sitesis taken as designvalue of
G8.ahge
t ir~+i.Using this value one can f%rdthe changeof cross - section in all dischargesites
during this cycle:
AcoZi

A4
=

= (A
(A

o,HlAt
w;IAt )i t paa -;

)i t pat,.
While calculating the long - time channel evolution the conditions of channel
forming depositsshould be taken into account;theseconditionsare due to the changeof
the channel cross - section (and consequentlythe averageflow velocity) owing to its
deformation. Making allowancefor this, the analysisin its generalform should include
the calculationof Q,”and R,H.
The calculation may be significantly simplified if it is carried out only for one or
two channelcross - sectionsand then the results of this calculationfor severaldischarge
sites of the consideredriver reach are used for revealing a dependencecommon for all
designdischargesites giving a possibility to determineimmediately the annual yield of
channelforming sedimentskf on the basisof data concerningthe changeof flow cross -
section causedby channelevolution in the given design site. For the Dneeper at Kiev,
the obtaineddependenceis as follows:

Besides,the scatteringof designpoints correspondsto the range exponentchangein the

dependence(7): 4-5.
Cross-sectionarea of the channtl of the outlet of the design reach in the case of
averagewater dischargein hydrographin the i-th calculationcycle wz is determinesas

1 1

The main results after the i-th cycle of the calculationare total for all the previous
cucles(including the i-th one) bottom deformations

the bank erosion during the i-th computationcycle

where Hby - the height of the erodedbank,

i-l

z,,- the mark of the flood plain surfacenear the bank edge of the channel(determined
as the averagevalue flong both banks);z, , Hcp- the mark of water surface and average
flow depth in the caseof averagedishargein hydrograph.
Long-term channelevolution is calculateduntil the ultimate bottom deformations
are completedin each dischargesite proceedingfrom the condition that the extreme

358
 annualyield of channel-formingsediments(RnTi) appropriateto the dynamic stability of
the channelshould be establishedin the channelresultedfrom long-term evolution; this
yield shouldbe similar for all gaugingsitesin a one-branchchannelwithout tributaries.
If this channelwas acceptedat the first computationcycle for the inlet, the value of R is
acceptedas the sedimentyield at that inlet. To determinethe extremevalues of the flow
area(in caseof averagedischargein the hydrograph) w$ and bottom deformations
AZ+,the dependencies(7) and (8) may be usedunder the condition that 4: = R,$,

In the channelsectionwhere the banks erosionis not determined,it is necessaryto

take 4’& = bkAi=lj in formula (8) in the case of calculation of bottom and banks
deformation using the value b:,cp = bz,,,, it is necessary to assume that with
deformation devilment the intensive of channel width noticeably decreases;this was
determinedat the definite stageof calculations.
Long term bottom deformation calculatedaccording to recommendedprocedure
and averagedon the channelwidth may also give the insight concerningmaximum long-
term bottom deformations in each dischargesite, if one uses the calculation results of
intensity of bottom deformations in plane jets obtained at the Institute of
Hydromechanicsof the Academy of Sciencesof Ukraine. According to theseresultsthe
maximum bottom deformations in the flow cross-sectionmay be taken as equal to
double value of this deformation averaged over the channel width. Overall bottom
deformationsmay be also supplementedby the local sign changingdeformationscaused
by the movement of sand dunes (Kondratiev, N.E, I.V.Popov and B.F.Snishchenko,
1982; SnishchenkoB.F., 1980).
The results of the calculation, carried out for the Dnieper at Kiev, show that
stability disturbance of the channel by pits long-term deformations by silting and
degradationof the bottom in the reachesdownstream these pits and in the zones of
bridges, pipelines etc. these expected long-term bottom deformations over
constitute 2-7m, design duration of their action untill reaching the limiting values
width
constitutes 30- 150 years, moreover the increase of this duration is observed
downstreamwith in the limits of the design river reach. The notable difference from
bottom deformations< causedby channelpits, was recorded in the river reach without
any pits and the channelmay be taken as dynamically stableeven at the beginningof the
calculation.Long term designbottom deformationsin this casewere insignificant(up to
1 m) and they had sing-changingnature of alteration with a small interval (5-10 years).
for the section of flood plain branch Desenka river the calculation pointed to the
possibleintensivewidening of the new man-madechannel(approximatelyto 90- 115 m).
The procedureto its natural (undisturbed)width.
The methodology has been applied for prediction of long-term evolution of the
following rivers: the Dnieper at Kiev on the reach of numerous pits causedby sand
excavations:the Wpiat at Chernobyl on the reach of sand excavationsand bank dukes;
some small Ukrainian rivers with intensivewater diversions,etc. Computationsmade by
the proposedmethodology on the reach of the south bridge acrossthe Dnieper at Kiev
made if possibleto specify the validated depth for bridge supports and to reduce the
constructioncost.

359

-- -- ___-.- -
References
Bazilevich, V.A. ‘Experimental investigations of banks erosion of channels
composed of gravels and pebbles’. Hydraulics and hydraulic engineering:
Resp.interdepart.scient.-techn.toll., No. 35, p. 60-65. (In Russian)
Bazilevich, V.A. and V.V.Kozitsky (1986) ‘Composition change of suspended
sediments in the process of deformation of channel bottom’. Hydraulics and
hydraulic engineering:Resp.interdepart.scient.-techn.toll., No. 43, p. 41-46. (In
Russian)
Bazilevich, V.A. and V.V.Kozitsky (1987) ‘On the problem of determining discharge
of channel-forming suspended sediments in river flows’. Hydromechanics:
Resp.interdepart.scient.-techn.toll., No. 56, p. 61-63.&r Russian)
Bazilevich, V.A. and V.V.Kozitsky (1985) ‘Calculation procedure of polydimensional
deformations of channelsof lowland rivers’ Hydrophysicalprocesses in rivers
and water storages.-M.: Nauka, p. 115-122.(In Russian)
Bazilevich, V.A. and V.A.Tkachenko (1982) ‘Planned distribution of velocities in
channelswith low curvature’. Hydraulic engineeringand land reclamation.,No.
10, p. 27-29. (In Russian)
Bazilevich, V.A., N.K.Gribovskaya and V.V.Kozitsky (1988) ‘Discharge
determination of suspended sediments in rivers with sandy channels’.
Hydraulics and hydraulic engineering.,No. 46, p. 32-38. (In Russian)
Vikulova, L.I. (1973) ‘The design problems of channelsregimes in channels’
Trans. of the Ins-t Gydroproject., No. 30, p. 126-187.(In Russian)
Kondratiev, N.E, I.V.Popov and B.F. Snishchenko ( 1982) Principles of
hydromorphological theory of channel process. Leningrad, Gidrometeoizdat,
272 p. (In Russian)
SnishchenkoB.F. (1980) ‘The connection of sand dunes height with parameters
of river flow and channel’.Meteorology and hydrology,.- No. 6, p. 84-91. (In
Russian)

360
Hydromorphological aspect of channel
process modelling
A. B. Klaven
State Hydrological Institute, St. Petersburg,Russia

The present practice of channel process and flow structure studies, as well as the
solution of individual economic problems, is based on a wide application of the
laboratory method of hydraulic modelling. This method, based on generalprinciples of
the similarity theory, is formulated by various scientific schools in different way. The
present paper describes quite definite methodological approaches to a laboratory
hydraulic modelling of the channelprocesseson the basisof hydromorphological theory
developedby N. E. Kondratiev and I. V. Popov (Kondratiev, 1959) at the Deportment
of ChannelProcessesof the State Hydrological Institute (SHI) as early as 1950s’.
The term “hydromorphological”both in the theory of the channelprocess and in
the methodology of its modelling means that during the study of the event an equal
emphasisis made on the contributing liquid and solid ingredients. The liquid ingredient
(water flow) is consideredin the terms of hydrology, hydraulics and hydrodynamics,
while the solid ingredient (suspendedsedimentsand bedload) is considered at various
structural levels of the forms of sedimentstransportation. Therefore, if some particular
water body is to be investigated,it is necessaryto consider it within a great variety of
channel forms and it should be recognised that the method of laboratory modelling
appears to be most effective only a combination with other ways of this study:
hydrological and morphological analysesof the available data on the particular water
body, field studies of the water body, and design methods for the estimation of
individual components of the process. Some problems often arise, however, when the
laboratory modelling appears to be practically the only and sufficient way for the
solution of theseproblems.
The hydraulic modelling method has been used at the SHI for more than 30 years
in such variants. Since the beginning of 1960s’more than 50 models of river reaches
have been made at the Channel Laboratory; most of these models were rigid; they
served as an indirect tool to investigatechannelprocesses;thus, they were less perfect,
if compared with erodible models. Attempts were made to create such models in the
laboratories of West-European countries at the very beginning of the XXth century.
Theseattemptswere not a successbecauserifned forms were produced in those models
which created a great hydraulic resistance;thus, the model was neigher hydraulically
nor kinematically similar to the prototype. Those forms produced in the model were
never used as the researchobject, but they preventedmodelling. It should be noted, that
the recent literature and discussionsoften expressedthe viewpoint that bed forms,
similar to those of the prototype, could not be obtained in laboratory (in flumes,
included).

361
 A certain positive experienceis available,however, on the tests in erodible models
of mountain river reaches(Kopalian, Tvalavadze and Noselidze, 1985). This experience
proves the validity of the initial position of the hydromorphological aspect, which
confirms a fundamental possibility of assessment(with the help of the model) of channel
forms produced in the study water body during different hydraulic situations of project
conditions. In caseof a rigid model, this estimationis made by an indirect way, while in
the erodible model - by a direct way, i. e. during the formation of the channelforms just
in the model, the geometric and dynamic characteristics of these forms can me
measuredand then re-computed for the prototype.
Thus, the main difference of the hydromorphological aspect of the channel
processmodelling from other approachesis in the fact that the channel forms produced
in erodible models do not impede the researchbut they are consideredas the goal of the
research. Secondly, the hydromorphological aspect of the channel process modelling,
unlike other approaches,requires a complete (geometrical and dynamic) similarity of
the channelforms in the model and in the prototype. And, f$rally, the third difference of
this approachis in the requirement of the unified systemfor the computation of all scale
factors for liquid and solid phases,basedon the geometric scaleof the model.
These three peculiarities, specifying the essenceof the hydromorphological aspect
of the channel process modelling, may be easily realised if certain conditions are
satisfied,which may be reduced to the following. A similarity of the channelprocess at
particular structural levels can be obtained only in case when the similarity of the
properties of the interrelated liquid and solid media (being determinantsfor the study
structural levels) is provided, and, besides, when a similarity of the mechanism of
interrelation between the contributing media is attained.
If this modelling problem is formulated in this way, it is necessaryto know the
properties of the interrelated media and the mechanismof this interrelation, which, in
fact, makes it possibleto formulate a theoretical basisfor the modelling method.
Let us briefly consider the results obtained from the investigation of the properties
of the channelflow, underlying alluvial bed and flow-bed interaction.
At the study of flow velocity field, the idea of its discretenessformulated by N. E.
Kondratiev (195 1) serves as the initial point. In fact, quite similar approacheswere
formulated on the structure of the turbulent channel flow by M.A.Velikanov (1958)
V.N. Goncharov (1954), K.V.Grishanin (1974), A.V.Karaushev (1969) and
V.M.Makkaveev (1965). These approachesare reduced to the following: a system (or
group) of individual volumes of fluid different in size, is formed in a flow. The largest
of these volumes are usually termed as macrostructural components of turbulence, or
macroturbulence.
As to energy mechanism and the laws and properties of different scales of the
turbulence spectrum, the channel flow is regarded from the viewpoint of Richardson-
Kohnogorov ( 1941).
A seriesof experimentshas been made at the Channel Laboratory of the SHI with
the use of stationary and sliding photographs and shooting particles suspendedIn the
flow. The experimentswere made in flumes with smooth glassbottom (Klaven, 1966)
and with uniform grained roughnessof the bottom with different size of bed particles
(Klaven and Kopaliani, 1973). Then, measurementsof macrorurbulencewere made by
a contact method in a laboratory and in the prototype in case of different roughnessof
the boundary surface (from smooth surface and uniform grained roughnessup to large

362
dunes)(Zaitsev and Klaven, 1983; Zaitsev, 1984). The main results are reduced to the
following.
(1) Dependenceshave been obtainedfor the estimationof the size of large eddies,
(in the function of depth and conditions on the boundary surface (Zaitsev, 1984)
(Klaven, 1966). In particular, the dependencederived for the step of large eddies
(Klaven, 1966) is similar to a formula of K.V.Grishanin (Grishanin, 1979) derived
theoreticallywith the account of dissipationof the turbulenceenergy.
(2) A structure of large eddieshas been discovered(Klaven, 1966). Large eddies
with vertical sizes close to depth appearedto be complicated structures composed of
three or four smaller eddies,while the latter ones consistedof even smaller eddies.This
may be supposedthat all these three groups of eddies are equally important for the
channelrelief formation.
(3) Dependenceshave been obtainedfor the estimationof characteristicvalues of
pulsation velocities in the function of dynamic velocity (Klaven, 1982) first proposed
by B. A. Fidman (Fidman, 1948) and then used by N.K.Nlkitin (Nikitin, 1963) as a
scaleof pulsationvelocities.
(4) Conditions have been formulated for the origin of characteristic dynamic
situationson the bottom Surface(Klaven, 1976), similar to those consideredby N. E.
Kondrariev relative to wind waves (Kondratiev, 1953) and by V. M. Makkaveev
(Makkaveev, 1960), where the mechanismof the turbulent channelflow is identified to
the auto-oscillatingsystem.
(5) A model of the profile of the averagedlongitudinal velocity, similar to that in
(Klaven, 1966) has been considered,including the subchannelflow (Klaven, 1987). A
phaseshift in the pulsations of the channel and subchannelflows has been discovered
(Klaven, 1987) (Klaven and Kokovin, 1987) establishedby J F. Karasseff at the
analysisof the mechanismof the cohesiveground scour (Karasseff, 1975).
The study of the solid medium properties, contributing to the channel process,
was basedon extensivelaboratory and field data and led to the following conclusions:
(a) simple relations have been obtained which connect geometric and dynamic
characteristicsof microforms and mesoformswith averagedhydraulic characteristicsof
the flow (Kondratiev, Popov and Snishchenko,1982), (Kopaliani, 1983);
(b) a formation of stable zones of higher and lower (if compared with the
averaged)hydrodynamic pressurehas been discovered in the f?ltrating layer of bed
particles (KLAVEN and KOKOVIN, 1987). The effect of the particles size on the space
characteristicsof thesezoneshas been established;
(c) the influence of the relative roughnessand grained filtrating medium on the
hydraulic resistance has been established and the automodelity of the hydraulic
resistance,according to the relative roughness,has been discovered(Grishanin, 1974),
(Klaven and Kokovin, 1990);
(d) the importance of the uplift impulse effect has been discoveredfor the removal
the bed particles from the immobility state and for the mixing of the bed particles
(Klaven, 1987), (Klaven and Kokovin, 1987),
(e) conclusionshave been drawn on the determinantrole of macroturbulencein
the microforms genesisand on the effect of the structure of the averagedvelocity fields
and intra-annualliquid flow variability in the genesisand evolution of the channelrelief
mesoforms(Klaven and Kopaliani, 1974), (Kondratiev, Popov and Snish~henko,1982).
Generalisation ,of the research material on the mechanism. channel-flow
interrelationhas led to the following conclusions:

363
 - the mechanism of flow and bed ground interrelation greatly dependson the
propertiesof interrelatedmedia;
- this mechanismhas a remarkable capacity, i. e. a rather narrow stripe, most
greatly effecting the bedload transport under particular conditions of combination of
hydraulic characteristicsof the flow and size of the particles,is always separatedfrom a
wide spectrum of turbulent pulsationsof the channelflow. K. V. Grishanintermed this
property as “selective capacity of the flow-channel system”- (Grishanin, 1979). In fact,
M. A. Mikhalev noted on that occasionthat during the channelprocessmodelling It was
necessaryto provide the similarity of the forms of bedload transport, without which it
was impossibleto expect any similarity of the processas a whole (Mikhalev, 1988);
- a selective capacity of the flow-channel system and the form of the bedload
transport are presentedin a particular way by the organisedinterrelation between the
channeland subchannelflows. This conclusionrequiresa more detailedexplanation.
The analysisof different combinationsof hydraulic characteristicsof the flow and
bed particlesmakes it possibleto separatethree typical casesof channeland subchannel
flows interaction (Klaven, 1987): the first case is characterisedby the fact that fine-
grainedbed particles, and, consequently,too small spacebetweentheseparticlesprevent
the velocity pulsationsand channelflow pressurefrom penetrationinto the subchannel
flow, At >> T, A, + 0 ; the second case -- when coarser bed particles allow velocity
pulsationsand pressureto penetrateinto the subchannelflow, but with a certain delay by
phase,and tend to decreaseby size, 0 < At I T,O < A, < A, ; the third case -when very
coarsebed particles allow a free penetrationof the velocity pulsation and pressureinto
the subchannelflow without any delay by phase and without any decreasein size,
At -+ 0, A, + A, (here AI and AZ are amplitudesof velocity pulsation or pressureof
channel or subchannelflow respectively, At - phase shift, T- period of large-scale
pulsation of the channelflow).
This presentationof the mechanism of channel and subchannelflows shows a
necessityto combine liquid and solid media in such a way when thesemedia interaction
would correspondto one of the above three casesfor channel and subchannelflows
interactionin the prototype.
The results of laboratory experimentsin flumes (Klaven and Kokovin, 1990)
prove quite convincingly that this condition should be satisfied. Otherwise, even if the
large-scalemodel (e. g., 1:2) is provided, horizontal and vertical scalesare equal, and
conditions of Re, > Re, and Fr = idem are satisfied,the channelprocessappearsto
be different in the model not only quantitatively but even quantitatively, if compared
with the prototype, if the prototype bed particles are reduced in the geometric scale of
the model. Moreover, the hydraulic characteristicsof the flow appearto be not similar
to those of the prototype.
Thus, to get a similarity of the channelprocess(when the modelling is made in the
hydromorphologicai aspect), it is necessary to take into account the following
circumstances.
(1) A similarity of the open turbulent flow (if characteristicsare averaged) is
attainedby satisfyingthe major conditionsobtainedfrom the equationsof 0. Reynolds:

Re M > Re K, Fr = idem, Sh = idem. (1)

364
 (2) A similarity of the turbulent flow in pulsating velocities, in the structure of
macrotnrbulenceand in the structure of secondarycurrentsis attainedby satisfying the
conditions obtained by A. V. Karaushev (Karaushev, 1948), from the equations of
motion in the form of V. M. Makkaveev, or suppliedby conditions(1) as follows:
BIH =idem (2)
which means a nondistortedrelief of the model. Here B and H - typical width
and depth of the channel.
(3) A similarity of bedloadis provided by sati@ing the following condition:
H/d=idem (3)
Here: H and d - characteristic depth and diameter of bed particles. The
experienceshows, that satisfying condition (3) automatically satisfiesthe condition for
similarity of bedloadmobility, which is presentedas follows:

WIU=idem,or U,,,lU=idem. (4)

Here: W - fall velocity, U and U,, - mean and non-eroding velocities of the
flow.
As experience shows, the channel process similarity at the structural level of
micro- and mesoformsis attainedwhen conditions(l)-(4) are satisfied.
(4) As mentioned above, however, in some cases the satisfying of all these
conditions does not lead to any similarity either by flow hydraulics or by channel
morphology. Therefore new additional criteria are proposed which are more general
conditions for similarity, if compared with conditions (3) and they reflect the
peculiarities of channel and subchannelflows interaction. These conditions may be
presentedin the following way:

AtlT=idem,and A,lA, =idem. (5)

(5) Satisfying conditions (5) sometimesrequires coarser bed particles applied in

the model, if comparedwith the requirementsof the geometric scale.In such casesit is
necessaryto reduce the density of the model material up to the values at which
conditions(4) are satisfied.It is evident,that at coarserbed particles, conditions (3) are
to be disturbed. This disturbance,however, is quite acceptable(in connection with the
observedautomodelity of-the hydraulic resistanceaccording to relative roughness),if
the following model condition is satisfied:

d,lH, 50.001. (6)

tool of the method becausethey contain two characteristicsof At

It should be noted that at the presenttime, conditions(5) are not yet the working
and A, , which are
not easily determined.The developmentof ways to estimatethese characteristicsis the
first priority problem.
Moreover, it appears, that the opportunities of the erodible models are not
restricted so greatly. For example, mountain river reaches (according to the
investigationsof Z. D. Kopaliani (Kopalian, Tvalavadzeand Noselidze, 1985)) can be

365
simulatedby models quite reliably if conditions (l)-(4) are satisfied. Many reachesof
the plain rivers can be related to the first caseaccordingto the characterof channeland
subchannelflows interaction. In this situation it is probably quite sufllcient to satis@the
conditions of (1) (2) (4) and (6). Moreover, the bed particles in the model by the size
should satisfy the conditions of At >> T and A, + 0. Taking into account possible
difkulties in selectingsubstitutesfor natural bed material, the model may be computed
in accordancewith the fall velocity of the selectedsubstitute.
In conclusion, it should be noted that the development of method for the
assessment At and A, characteristics is not the only problem of the
hydromorphologicalaspectof channelprocessmodelling.
Neverthelessthe problems which are not yet solved, should not prevent the use of
most perfect laboratory research methods (e. g. the channel process modelling by
erodiblemodels) for the investigationof the channelprocesses.

REFERENCES
Vasilchenko, G. V. (1973)‘Investigations of relations between the turbulent
characteristicsof the flow at fhe bottom zone and in the underlying loase
ground’. - In: ‘Dynamika I term&a rek”, Moscow,‘Stroyizdat”.p. 118-126.(in
Russian).
Velikanov, M. A. (1958) Channelprocess (theoretical basis). - Moscow. “Fizmatgiz”.
395 p. (in Russian).
Goncharov, V. N. (1954) ‘Dynamics of channelflows’. Leningrad. Gldrometeoizdat.
452 p. (in Russian).
Grishanin, K. V. (1974) Channel stability of rivers and canals. Leningrad.
Gidrometeoizdat. 144 p. (in Russian).
Grishanin,K. V. (1979) @ namics of channelflows. Leningrad. Gidrometeoizdat. 3 12
p. (in Russian).
Zaitsev, N. I. and KLAVEN, A.B. (1983) ‘Kinematic structure of the Zow above large
sanddunes’.Trudy GGI, vol. 288, p. 86-96. (in Russian).
Zaitsev, N. I. (1984) ‘Large-scaleeddy structure of the turbulenl flow in a rectillinear
flume’. Trudy GGI, vol. 318. p. 3-17.
Karasseff, J. F. (1975) Channel processes in case of water transfers. Leningrad.
Gidrometeotzdat,288 p. (in Russian).
Karaushev,A. V. (1948) ‘Open flows similarity at the turbulent regime’. Trudy GGI,
vol. 4(58), p. 159-174.(in Russian).
Karaushev, A. V. (1960) Problems of natural water flow enamics. Leningrad,
Gidrometeoisdat,392 p. (in Russian).
Klaven, A. B. (1966) ‘Investigation of the turbulent flow structure’. Tr,rdy GGI, vol.
136, p. 65-76. (in Russian).
Klaven, A. B. and KOPALIANI, Z. D. (1973) ‘Laboratory investigationsof the kinematic
structure of the turbulent flow with very rough bed’. Trudy GGI, vol. 209, p.
67-90. (in Russian).
Klaven, A. B. and KOPALIANI, Z. D. (1974) ‘ On the relations between the length of
dunes with longitudinal dimensions of large-scale turbulence components’.
Trudy GGI, vol. 216, p. 19-24. (in Russian).

366
Klaven, A. B. (1976) ‘Some propertiesof large-scalechannelturbulence’. Proceedings
of the IVth AII-Union Hydrological Congress, vol. 10, p. 252-259. (in
Russian).
Klaven, A. B. (1982) ‘Estimation of the characteristics of the channel flows
turbulence’. Trudy GGI, vol. 278, p. 36-43. (in Russian).
Klaven, A. B. (1987) ‘Interaction between channelflow and loose bed particles and its
account for channelmicroforms modelling’. Trudy GGI, vol. 307, p. 71- 81.
(in Russian).
Klaven, A. B. and KOKOVIN, V.N. (1987) ‘Some results of channel flow properties
researchat the flow-bottom interface and in the mass of bedload’. Trudy GGI,
vol. 307, p. 82-91. (in Russian).
Klaven, A. B. and KOKOVIN, V. N. (1990) ‘Experience on the hydraulic modelling of
bottom relief. Trudy GGI, vol. 337, p. 109-123.(in Russian).
Kohuogorov, A. N. (194 1) Local structure of turbulence in non-compressible viscous
fluid at very high Reynolds numbers. Dokl. AN SSSR, vol. 30, No.4. (in
Russian).
Kondratiev, N. E. (195 1) ‘ Conditions for continuity and discreteness relative to
channelflow’. “Meteorologiya i gidrologiya”, No. 12, p. 36-41.
Kondratiev, N. E. (1953) Computation of wind waves and trasnformation of reservoir
shores. Leningrad. Gidrometeoizdat.(in Russian).
Kondratiev, N. E. (1959) ‘Channel process’. Ed. by N. E. Kondratiev. Leningrad.
Gidrometeoizdat, 37 1 p. (in Russian).
Kondratiev, N. E., POPOV I.V., SNISHCHENKO B.F. (1982) Principles of the
hydromorphological theory of the channel process. Leningrad,
Gidrometeoizdet,271 p. (in Russian).
Kopaliam Z. D., (1983) Approximate method for the computation of mesoforms
motion along the river channel. Trudy GGI, vol. 288. (in Russian).
Kopaliani, Z. D., TVALAVADZE O.A., NOSELIDZE D.V. (1985) ‘Methodology and
researchreluts of river-bed deformationson non-distoriedmodel of the Khara-
Mm-in river’. Trudy GGI, vol. 301, p. 68-80. (in Russian).
Makkaveev, V. M. (1960) ‘The effect of autofluctuating processesin turbulence
mechanics’.Proceedingsof the IIlrd AII-Union Hydrological Congress,vol. 5,
p. 5-9. (in Russian).
Makkaveev, V. M. (1965) ‘Channelflows turbulence’. Trudy GGI, vol. 124, p. 40-54.
(in Russian).
Mikhalev, M. A. (1988) ‘On the pmblem of hydraulic events modelling in a rigid and
erodible open channel’. Procedings of the Vth AII-Union Hydrological
Congress, vo1. 10, book 2, Leningrad, Gidrometeoizdat, p. 271-280. (in
Russian).
Nikitin, I. K. (1963) Turbulent channel Jlow and processes at the bottom zone. Kiev,
Izd. AN USSR 142 p. (in Russian).
Fidman, B. A. (1948) ‘On the effect of walls roughnesson the turbulent flow structure.
Izv. AN SSSR,ser. geogr. i geofis., vol. 12, No. 3. p. 255-260. (in Russian).

367
Flume investigations into the influence of river
training structures on sediment transport

by S. Wieprecht and W. Bechteller

Institut fur Wasserwesen, Unversitat der Bundeswehr, Munchen

1. Introduction

In the past, most natural rivers and streams have been regulated in a schematic
monotonous way. In recent years, the attempt has been made to “re-naturalize”rivers which
were straightened in former times. The foreshores are often built on with houses, roads and
rails, or agriculturally cultivated, so that there is no possibility to lead the river back to its
former meanders. Measures like planting trees, structural elements or widening the cross
sections are very important. These “re-naturalization works” have a great influence on the
hydraulic conditions, as well as at the sedimenttransport of the river.

At the Institut fur Wasserwesen,Unversitat der Bundeswehr, Munchen, a hydraulic

model was used to perfoirm fundamental envestigations into the influence of river training
structures.

368
 2. Basic Remarks about Sediment Transport

The interaction between discharge (water depth, flow velocity) and sediment is characteristic for a
river with movable bed. The following parametersdescribe this interaction:
- Resistance of the river bed depending on the bed forms (flat bed, ripples, dunes, antidunes,
scours etc.)
- Sediment transport (begin and quantity of sedimenttransport, bed load, suspendedload)
- Morphology (meanders, sedimentation,erosion, etc.)
There is no generally acceptedexplanation for the occurrence of bed forms. Thus most of
the existing criteria rely on empirical approaches.On the basis of measuredand observed results
relations were developedin order to define areas, where one or the other bed form will occur. As
an example a diagram by Zanke (1982) is shown in Fig. 2.1. Here the relation between grain
diameter (dcJI and mean velocity (v,,,) is described.

--
I I I I ,,
no motion
I I I II I I ,I
0,Ol 0.1 1 dchlmml

Fig. 2.1 Bed forms, Zanke (1982)

Up to now the Manning-Stricklerequation(2-l) is the best-known and most-appliedformula for the

calculation of river roughnessresp. discharge.

V, = k,, Rz3 11’2 (2-l)

V, mean velocity [m/s]

k 5% Manning-Strickler coefftcient [m”3/s]

369
 R hydraulic radius, for wide rivers (b > 30 h) R = h [m]
b width [m]
h water depth [m]
I slope [-]

The essentialassumptionof equation (2-l) is, that the roughnessand the wp*er depth do not depend
on eachother. But in.natural rivers, especially with vegetationon the embankmentsor “obstacles”
such as big stoneson the ground, roughnessis very different for certain parts of the channel. In
such cases, especially for small and varying water depths, the use of the Manning-Strickler
equationcannotbe recommended.Roughness(in particular for small rivers) has to be determined
as exactly as possible. Fig. 2.2 shows the different factors making up the total flow resistanceof a
river.
.
FLOW RESISTANCE
OF A RIVER

,
RESISTANCE OF RESISTANCE OF RESISTANCE OF
EMBANKMENTS RIVER BED VEGETATION

RESlSiANCE OF
GRAIN (flat bed)
/**kii~::..~,
BARS, SCOUR

Fig. 2.2 Flow resistanceof a river

370
3. Model Set-up

The model tests, ordered by the local water managementauthority, were performed becauseof
agglomerationsof sedimentin the Weissachriver, that occurred during a flood period in the lower
sectionof the Weissachriver before it flows into lake Tegemsee. This part had been “renaturalized”
slightly with some stones and groins. In a hydraulic model with movable bed (scale 1:20) the
existing conditions and possible improvements were studied in order to prevent agglomerations.
Furthermore fundamental investigationswere made into the influence of obstructions on sediment
transport and water levels.

3.1 Model Description

The model consistsof two equally long sectionsthat are connectedby a hinge. It is supported by
two I-shaped steel girders, the elevation of which can be adjustedto any slope desired, up to 1%,
for the different investigations. River cross-section is simplified to a trapezoidal model cross-
section. The lateral embankmentsare made of sheet iron. Roughnessis simulated by coarse sands
glued on the surface. A bed material ground layer of 10 cm thicknessavoids the ground plate to be
washed free in erosion ranges and the flume roughnessto be changed. Two metal rods serve as
guide rails for the measuring carriage. Fig. 3.1 shows the model cross-sectionand plan view.
As measuring installations serve an inductive flowmeter for discharge regulation, point
gaugesfor the registrationof water depths and a profile follower for the assessmentof river bottom
changes. The sediment input comes as a sand-water-mixture out of a plexiglass tube having a
vertical slot, which is openedat a constantpredefined speedand thus provides a constant inflow of
material per time unit.

371
 t30+7Ocm ---+-30*
guide rail for the
measuring carriage

sheet iron profile

embankment
inlet gauge 1 slope 1:2 hinge movable bed

- weir

sediment a measuring section -outlet + sand irap with

feed = 16m section weighing device

Fig. 3.1 Cross-sectionand plan view of the model

3.2 Model Tests

In renaturalisationand bed protection works of smaller rivers, whose course had been straightened
in previousyears, often constructionalelementsare used to force meanderingof the flow from one
bank to the other at low discharges.The systematicinvestigationstest the influence of constructional
elementson the sedimenttransport behaviour of rivers. The subsequentlydescribedtests intended
to obtaingeneralconclusionsfor the river renaturalisationworks. These tests were conductedwith
varying slopes, grain sizes, dischargesand constructionalelements.
First testswere donewithout constructionalelements. They served as a basis of comparison
with the following ones. At the beginning tests with constant discharges provided a transport-
discharge relation. They were executedfor three different slopes (S = 0.002, 0.004 and 0.006),
and for three different characteristic grain diameters (d,b. Table 3.1 shows the test program.

372
6

Table 3.1: Test program

d cb 0.65 mm 0.85 mm 2.00 mm

0.002 Q = 10 l/s Q = 10 l/s

20 l/s 20 l/s

0.004 Q = 10 l/s Q = 10 I/s Q = 10 l/s

20 l/s 20 l/s 20 l/s
40 I/s 40 l/s 40 l/s
0.006 Q = 10 l/s Q = 10 I/s
20 l/s 20 I/s
40 l/s 40 l/s

The choiceof model sand was guided by the idea to keep the part of the suspendedmaterial as low
as possible.A geometricalreduction would have contradictednatural conditions. Therefore almost
uniform sandswere selected.
The test program comprisesinvestigationsof three series of different arrangementsof con-
structionalelements,eachof which was to be testedfor three grain sizes, three dischargesand three
slopes.The three constructionalelementsare groins on both sides, alternating groins (length about
20 cm) and ground sills (heightabout 6 cm) with a distanceof 1 m. They are modelled using stones
of 5 to 6 cm diameter. The obstaclesare placed on the fixed solid ground plate, which lies 10 cm
lower (Fig. 3.2).

1 0: without constructions II: Groins, on both sides,

I I distance 1 m I

I I I
III: Groins, alternate. ’ VIII: Ground sills,
distance 1 m distance 1 m

Fig. 3.2 Plan view of different constructional measures(schematic)

The executionof the test is similar with and without arranging construction! elements. Bed profiles
were registered before the test after levelling the flume bed, and after test execution. Comparing
both registrations allows the assessmentof bottom changes.Their differences show where in the
16 m measuring range there are regions with erosion or sedimentation.

373
3.3 Test Results
The comparisonof test seriesresultswas made possible by introducing a number characterizing the
degreeof obstructioncausedby the constructio~l elements. This degree of obstruction is calculated
to be the ratio of obstructed volume to unobstructedvolume (Fig. 3.3). Reference length is one
meter.

Jf lm L
1
lm
7k

A: unobstructed Volume
B: obstructed Volume

obstructed Volume (A)

Obstruction Degree =
unobstructed Volume (B)

Fig. 3.3 Obstructed and unobstructedvolumes

3.3.1 Hydraulic Conditions

For all tests the same tendency, as expected, could be recognized: with an increasing obstruction
degree the water level and the roughnessare increasing, while the FROUDE numbers and the
velocities are decreasing.

374
 8

0
mm
13
-

d ch= 0,65 mm
1,4
d ch= 0,85 mm
12

o,o z5 5,o 7,5 IO,0

degree of obstruction [%]
Fig. 3.4 Relationsbetweenthe degreeof obstructionand the ratio of water depths with and without
constructio~l elements, S = 0.004

The dimensionof the increaseof the water depthis strongly dependenton grain size. Fig. 3.4 shows
the relations between the degree of obstruction and the ratio of water depths with and without
obstructions.The tendencyis the same for all tests. With an increasingobstruction degree the ratio
of water depths is increasing. Comparing the results of the tests having a discharge of 10 l/s the
samesystematicscan always be recognized. The biggest rise of water depth occurs for sandswith
dd, = 2.00 mm, secondin order are sandswith d, = 0.65 mm and third are sandswith dch= 0.85
mm. The reaw; fx ZL order lies in the resistanceof bedforms (Fig. 2.2). The highest bed forms
could be recognizedfor sandswith dc,,= 0.85 mm and the smallestor even no bed forms for the
sand with 2.00 mm. That meansthe biggest changeof roughnessdue to introducing the obstacles
happensin the caseof the coarsematerial. At the same time the roughnesscausedby the bed forms
is at its lowest.
In Fig. 3.5 the Manning-Strickler coefficient for all tests with dch= 0.65 mm are shown.
The k,, - values are dependingon the degreeof obstruction independentof slopesand discharges.
With an increase of the degree of obstruction from 0% to about 3% the k,, - value decreases
considerably (from 72 down to 45 m”3/s). When the degree of obstruction is greater than 3% the
influenceof obstruction increaseis nearly insignificant. The k,, - value decreasesfrom 49 down to
39 m1’3/s.This systematicscould be recognizedin all test series.

375
 9

A 10 I/s

0 20 I/s

x 40 I/s

d ch = 0.65 mm

5:o 7.5 to:0 1215

degree of obstruction [%I

Fig. 3.5 Relation between k,, - value and degree of obstruction (dch= 0.65 mm)

3.3.2 Sedimentological Conditions

For all tests the transport rate is decreasing with an increasing degree of obstruction (Fig. 3.6).

..
..
.
‘q - .
-y -.
-. -.
-.
“= Q=40I/s

. *
- Q = 20 I/s
-- -v------4-- -- -*Q = 10 I/s
8 I I I I / I I 1 / / r I 1 I / I I

w 23 570 7,5 10,o

degree of obstruction [%]
Fig. 3.6 Relations between transport rate and degree of obstruction

The first “step” from the situation without constructional elements to alternating groins has a
relatively big influence on the transport behaviour. The next step to groins on both sides is not as
significant as the last step to the ground sills.

376
10

a) Variant III b) Variant II

Fig 3.7 Transport area for variant III and variant II (schematic)

In Fig 3.7 a) the area of the bed is shown where transport occurs in caseof variant III. In Fig. 3.7
b) the tranport area for variant II is indicated. In case of alternating groins the bed load moves
“meandering”.The transportarea for variant III is almost the same as for variant II. Becauseof the
symmetricobstructionthe cross section is reduced very strongly. Therefore the mean flow velocity
in the compressedflow section is higher and causesan increase of the transport capacity in the
middle of the flume. In caseof sills (VariantVIII), the transport is restrainedover the whole cross-
section.

377
4. Conclusion

Renaturalisationworks have consequences on the hydraulic and morphologic equilibrium of a river,

which can hardly be foreseen. It is therefore necessaryto rely on simulations with physical models
if there exist no experimentsfrom nature to predict possible effects of constructio~1 measures.
There are already calculation proceduresavailable that deal with overgrown cross-sections.
The influence of constructio~l elements, however, still cannot be describedexactly.
In all test seriesthe samesystematicdevelopmentcould be observed. With an increase in the
degree of obstruction the water depth also increases, and velocities, FROUDE numbers and
roughnesscoefficients decrease.Important is the influence of the hedforms on the increase of the
water depth. If there are smaller bedforms (flat bed, ripples, e.g. dc,,= 2.00 mm) the effect of the
river training structures was more important, in case of bigger bed forms like dunes (e.g. d, =
0.85 mm) groins and silk causean increaseof water depth that is nearly unimportant.
First investigations into changesof hydraulic conditions and transport capacity causedby
river training structures have been performed, but are yet to be generalized.Actually additional
studies are executed, that investigate the influence of a widening of the cross-section and
furthermore the combinationof both renatumlisationmeasures,obstructionsand changing the cross-
section.
5. Bibliography

BECHTELER, W., H.-J. VOLLMERS and S. WIEPRECHT (1994) ‘Model Investigations into the
Influence of Renaturalisationon Sediment Transport ’ DYMIII~CSand Geomorphology of
Mountain Rivers, Springer Verlag
DVWK (1980) ‘Hydraulic Modelling’ Bulletin No. 7, Ed. H. Kobus, Verlag Paul Parey
DVWK (1991) ‘Hydraulische Berechnungvon Flieggewassern’ Merkblatt 220/ 199I, Verlag Paul
Parey (in German)
DVWK (1994) ‘Hydraulisch-sedimentologischeBerechnungennaturnah gestalteter FlieBgewHsser’
Mitteilungen Nr. 25 (in German)
VOLLMERS, H.-J. (1989) ‘The State of the Art of Physical Modelling of Sediment Trans-
port ’ Sediment Transport Modelling, Proc. International Symposium, New Orleans
ZANKE, U. (1982) ’ Grundlagender Sedimentbewegung‘, Springer-Verlag, Berlin Heidelberg (in
German)

379

-__
___-- -
A perspective of mobile bed river
modeling
Dr.-Ing. Gunter Glazik D-29462 Wustrow/Germany (formerly FAS Berlin)

Historical review
Nearly 100 years ago, in 1898, Hubert Engels at the Technical University of
Dresden/Germanyfounded the first permanent “river engineering laboratory”, which
became an example to the whole world. He not only investigated numerous objects of
river engineering by means of an experimental method elaborated by himself, but he
also solved many basic problems by systematicinvestigations not yet cleared up at that
time using physical hydraulic models with movable bed - for example the suitable
design of transverse structures (groynes) in rivers with bed load transport (Bollrich,
1987).
Only a few years later Theodor Rehbock establisheda hydraulic laboratory at the
Technical University of Karlsruhe; there he later on carried out test series for
regulation work for the River Rhine by studying the river morphology in movable bed
models using a technique largely also developed by himself. In these tests he already
studied the transport of bed material and the development of bed formation in a longer
river stretch as a function of the time and with changing discharges(Larsen, Bernhart,
Theodor Rehbock, 1987).
Also only some years later, in 1903, there was founded as a governmental
institute the “PreuDischeVersuchsanstaltf?ir Wasser-, Erd- und Schiffbau (VWS)” in
Berlin (Krey, Winkel), whose vigorous developmentwas affected by Krey, which took
over the directorship in 1910. He led the institute with great scientific and practical
success,up to his early death in 1929. Engels and Krey as co-authorsjointly published
scientific papers about both hydraulic model tests (Engels, Krey, 1916) and river
training (Engels, Krey, 191S), the latter in further explanation of the general guidelines
which Engels had already pointed out in “Handbuch des Wasserbaues”(Engels, 1914).
Very soon the department for hydraulics of the VWS had developed from a small
beginning to a great and distinguished researchorganization, of which G. de Thierry
stated in 1926 in the well-known issue “Die WasserbaulaboratorienEuropas”(Thierry,
Natschoss, 1926) that “under all existing laboratories ... this one is the first ... as for
both, the number and the versatility of the experimentshitherto run”. In this issue the
then leading scientists of European hydraulic laboratories reported about whose
equipment and work; and Winkel, one of the nearestco-workers of Krey and then head

380
of the hydraulics department, especially tried already to outline some generalized
scientific drafts about similarity problems of river model tests also with movable bed
(Winkel).
In those years the work of the often so-called Berlin laboratory influenced
considerably the development and methodology of hydraulic research and especially this
of hydraulic modelling in many countries. An example for orders from abroad for river
model tests are these for the Volga-Don-Canal, especially the junction of the canal into
the Don River, in the former Soviet-Union (Fig. l), which a contemporary publication
characterized as the most remarkable task at that time confronted to a hydraulic
laboratory (Korner, 1937). An acknowledgement of these irradiations may be seen in
the fact, that the first congress of the “International Association for Hydraulic Research
(IAHR)” in 1937 took place in the Berlin laboratory (IAHR,l938), at which occasion
Engels became the first Honorary Member of the organization. It is very interesting and
reflects the then phase of river training projects, that this first congress to a great extent
dealt with river engineering problems, especially the question of distorted river scale
models with fixed as well as with movable beds (Rehbock, Meyer-Peter, 1938;
Rohringer, 193 8; Seifert, 193 8).
It must be accentuated, that it was a speciality of the VWS, to develop research
on movable bed river models and on fundamentals of the problem of bed load
movement, among these the up to now well-known and appreciated work by Shields
(Shields, 1936). The successor of Krey as director, Seifert, promoted together with the
then chief engineer Zschiesche the development of such river model tests and in (Liebs,
1942) and (Seifert, 1942) it was reported in detail about the then state-of-the-art of
hydraulic modelling of river processes, especially as it was developed and carried out in
this institute. A very remarkable place took scale model tests for stretches of the Elbe
River with intensive bed load movement. Already in the years 19 18-l 921 Krey himself
had done pioneer work in this direction, which the reviser of the later on publicated
results characterized as follows: “Never before a model test had been prepared and
carried out as carefully as this great work. This can be seen by the time expense of four
years, .._ a time, in which the department for hydraulics was nearly exclusive engaged in
this one task. Preparing and carrying out, however, is not the really permanent value of
the tests; much more important is the mental insight into this difficult matter... At the
time of the experiments, the scale model test with bed load was a new and much
controversial problem, which only could be mastered, if the research engineer in
addition to great practical experience had a deep theoretical ,knowledge and a sensibility
into the natural processes of our rivers; qualities, which not often can be found joint in
one person. After all there must not lack self-criticism, which just in experimentation is
strictly necessary. These qualities, which Krey had to a high degree, enabled him to
influence hydraulic modelling directively and far sighted by his fundamental statements”
(Krey; Korner, 1935). These river model tests with movable bed might have been the
first large-scaled and systematic conceptioned of this type. Similar test series later on
were the already mentioned above by Rehbock for the Rhine River. Together with the
famous large-scale model tests by Engels in the thirties in the open air hydraulic
laboratory Obernach/Walchensee for the Hwangho River (Yellow River) in China
(Scheuerlein, 1980) these river model tests are historical steps in river engineering
methods.
An other remarkable movable bed river model at the Berlin laboratory - like that
already mentioned above for the Don River carried out in the large open air laboratory

381
Fig. 1
Model tests for the Wolga-Don-Canal:
transformation of the movable bed
of the river model after a test run

382
of the dependency in Potsdam/Marquardt - was that of an experimental stretch of about
5 km length of the Elbe River near Magdeburg comprising a bend, transition and
counter-bend and trained by groynes. In cooperation with the waterways administration
it was possible to compare direct and immediate results obtained in prototype (nature)
and model. These investigations were interrupted owing to the World War II. After
1945 as one succeeding institution of the then VWS the “Forschungsanstalt fur
Schiffahrt, Wasser- und Grundbau (FAS)” Berlin continued not only the investigations
at this interesting model but also in general the tradition of movable bed river models
and research about bed load transport, especially in alluvial lowland rivers, as
established and developed at the formerly Berlin laboratory. For it the great experience
of Zschiesche, previously co-worker of Krey and Seifert and then director of the FAS,
was an inestimable support (Zschiesche, 1954).

Basic remarks about movable bed river modelling

A general description of the problems of river scale models with movable bed goes
beyond the frame of this paper. Only some general questions can be broached, which
are on principle significant for the practical valuation of such models.
Fundamental of hydraulic modelling is the mechanics of similitude. But it is
evident, that no scaled down hydraulic model is possible, which in the strong sense
fulfilled the exact conditions of similarity. Proved by the results, many difficulties were
compensated by increased experience, that means especially the interpretation of
extensive realized experiments. A special type of hydraulic models are river models with
movable bed, at which additionally to general hydromechanic similarity the
morphological similarity of the river bed configuration is desired.
On principle each scaled down physical model only can have approximate
similarity to nature or prototype. Foremost just is valid for river models with movable
bed, where beside the not or only limited compatible hydromechanic similarity laws of
Froude and Reynolds are acting the up to now not entire recognized and formulated
conditions of the river-morphological processes and their similarity conditions. River
models with movable bed are consequently on no account similar in the exact sense of
the physical-mathematical formulation. It is the objective of the approximate simulation,
to estimate the necessarily appearing differences from exact similarity both qualitative
and quantitative, to minimize them adequate to the concrete investigated problem by
means of appropriate simulation methods and to interpret the test results experienced
and relevant to practice. In the process in several hydraulic laboratories partly differing
simulation methods or conversion factors for river models with movable bed were
developed and used, modifying theoretical statements of similarity mechanics by new
knowledge of open channel hydraulics and results of the fluvio-morphological analysis,
which substantial were found empirical. Both these branches of knowledge in recent
time developed considerable and in this situation it is no wonder, that in consequence of
the absence of a closed theory of river bed processes and accordingly of river models
too, several different methods of simulation and scale determination were set up.
In recent times from scientists all over the world efforts were made to formularize
sets of criterions and equations for the computation of scales, which should enable the
construction and running of naturelike models with movable beds. Renowned. for
instance, is the rational approach by Einstein and Chien (Einstein, Chien, 1954).

383
 In Germany in recent years the approach by Gehrig was published and repeatedly
quoted (Dorer, 1984; Gehrig, 1967; Gehrig, 1978). The basic idea is, that certain
dimensionless numbers, which are characterizing sediment movement, in prototype as
well as in the model must have the same value and that moreover the similarity law of
Froude is valid. In this way the required scales can be calculated. In this context the
well-known Shields diagram was complemented with data of several lightweight
materials which may be used as model bed material.
But comparisons with movable bed river models actually running in the
laboratories shows more or less deviations from the calculated scales, the reason of
which may be seen both in limited conditions of the laboratory (e. g. capacity of area
and pumps) and in otherwise not fulfilled morphological similarity. An interesting
example is the comparison of the scales of the then model tests by Engels for the
Yellow River with these according to the now used approach mentioned above, done
by Scheuerlein (Scheuerlein, 1980). Whereas H. Engels worked with depth-distortions
of 1:2 and 1: 1.5, the present calculation yielded 1.4 or 1:5. Own experience with a great
number of river models however shows that such great distortions in the case of
movable bed scale models produces no good results and that a distortion as used by
Engels is appropriate.

Similarity conditions, scale effects and methods of compensation

For more than 40 years in the then FAS river model tests with movable bed, on
principle, were carried on and furthered on the basis of concepts developed at the
formerly Berlin laboratory (VWS), which essentially date back to the pioneers of
hydraulic modelling Engels and Krey. They based on the application of the regularities
of the shear stress concept, which by Krey was written in the manner

K= 20: initial movement of single grains

K= 13: initial, though slow, transformation of the bed
K = 8: beginning of intensive bed load movement
From equation (1) follows
H.S
-=-..- 1 -EKn
or (2)
D K H .S
which formulation later on (1942) Bogardi (Bogardi, 1974) introduced as SO-
called “dimensionless bed stability number b”. As a treshold criterium for the beginning
of movement, Bogardi gives the value b = 12,91, which is in correspondence with the
above value K = 13 according to Krey.
The first known relation of this kind as a stability criterion is the Lochtin number
(1897)
r=”
i (3)
s .
and already Engels gave for the purpose of river modelling the conditional
equation

384
 (4)
where capital letters are standing for nature (prototype) and small letters for the
model. It means
H, h - water depth
D, u’- grain diameter (bed load)
S, s - slope
Equation (4) is only a simplified description of relations; later on. Krey et al.
extended it by coefficients as well as mnctional dependencies with reference to the
influence exerted by characters of the sediment, friction effects etc.
Further, the following relations exist:

with 11~
= 3 (shear velocity) (6)
Pw
and Fr+ = f, being the so-called shear stress coefficient or Shields factor.
These,relations quality the formulations by Engels and Krey as preliminary stages
of the well known Shields concept (Shields, 1936). With these relations, starting from
eq. (4) we also fulfil the condition of equal dimensionless numbers in prototype and
model as pointed out in the foregoing chapter. But we don’t demand to observe the law
of Froude unconditional. This law is neglected widely for the benefit of more
morphological similarity (of the fluvio-morphological process), correcting scale effects
by means of compensations (e. g. distortions of depth, slope, grain diameter), which are
not in conformity with the Froude law. The nature-like modelling of the river bed
formation should be proven by comparisons of the topography surveyed in nature
(prototype) and in the model and eventual divergences must be corrected by
adjustments. An essential is, that the bed forms (dunes, ripples) in model and prototype
are the same. This procedure means, the model has to be calibrated and verified with
field data.
River models with movable bed managed in this way, therefore are not similar in
the exact sense but are what is called of an approximate similarity. This is not a rational
but a semi-empirical approach.
Applications of this movable bed river modelling to practice of river engineering
with good results in solving the related problems can be found, for example, in (Blau,
1963; Faulhaber, Kebelmann, Thieler, 1986; Glazik, 1969; Glazik, 1969; Glazik, 198 1;
Glazik, 1984; Glazik, 1986; Glazik, 1987; Liebs, 1942; Zschiesche, 1954). In these
publications are also recommendations for the practical performance of such model
tests, which not all can be quoted here. But here it must be emphasized that the
experience has shown: As prerequisite to nature-like results river models with movable
bed should have certain minimum absolute dimensions! According to that it is
recommended:
- a width of the movable bed in the order of meters (at least about 1...2 m),
- a water depth of at least 5 cm,
- not or only little depth distortion (maximum about 1.5...2).
In most cases we worked with quartz sand mixtures, which allows sorting as in
nature.

385
 In case of restricted area and pump capacity in several laboratories river models
with movable bed are often operated at smaller and, partly, even much smaller
dimensions. The interpretation of such test results calls for much care, experience and,
possibly, even restraint. Qualitative knowledge deduced from them often, however,
already could be valuable, e. g. for the investigation of variants. Final proposals for
important projects should always be checked on large-scale models, eventual in
cooperation of several laboratories.

Reflections in international literature

What about the evaluation of this kind of method for movable bed river modelling in the
frame of international experience as reflected in publications and conferences?
For instance, the IAHR-Symposium in 1986 in Toronto was particularly devoted
to scale effects in modelling sediment transport phenomena and it reflects very well the
todays situation in this field. A keynote speech by H.W.Shen (Shen, 1986) gave a broad
and authoritative review on movable bed river models and he accentuated the distress as
follows: “For movable bed processes, there is not a single governing equation (or a set
of equations) that can be found to determine the variation of alluvial bed resistance,
suspended sediment load, bed load, channel widths, etc. and thus the modelling
procedure involves the selection of different governing parameters for different
problems.. Unfortunately, not all of scale ratios are practical to use and therefore, some
preliminary adjustments are done based on previous experiences”. In his concluding
comments he asked for more communication and exchange of information between
model investigators with a view to testing “the theoretical approaches as introduced by
Einstein, Chien, Yalin, Bogardi and many others”. According to a report by the
chairman of the symposium “it is now past the time when sediment transport studies
should continue to proceed under so many disparate theories and approaches” (Dick,
1987). But here it also should be mentioned, that Yalin as the chairman of the Technical
Committee of the Symposium in his Opening Address gave a summarized account on
various shortcomings inherent in physical models in general and in the models with
movable bed in particular. in which he said: “But if the criteria of similarity (in simple
cases) of sediment transport are known, why the modelling of sediment transport
cannot be carried out, why... all these hesitations and uncertainties? The reason for
ambiguities lies in the design and operation of a conventional physical model, not in the
lack of knowledge of the criteria of similarity” (Yalin, 1986).
Some years before his early death I had (at the occasion of a IAHR-Congress) the
fortune to discuss related problems with H A. Einstein personally. I told him about the
praxis in our laboratory and he accentuated, that the various sets of formulas only can
be means for preliminary calculations when designing movable bed models, while
certain empirical assumptions and/or procedures are indispensable, particularly the
verification and calibration of the model. Therefore already in the then publication by
Einstein and Chien (Einstein, Chien, 1954) it was pointed out, that “because of these
difficulties it is absolutely necessary to verity any such model and its scales. Such a
verification consists in the reproduction of a known prototype development in model by
similar flows. Only if such a verification is possible and successful can the model be
depended upon for the prediction of future developments”. And they continued, that as

386
a common experience the greatest difficulty of most these model tests is the supply of
the necessary data for the design, effect and verification of the model.
We find similar statements by nearly all investigators which practise really model
tests with movable bed. Thus Dorer from the “Bundesanstalt tir Wasserbau (BAW)” in
a summarizing representation about similarity of river models finally stated, that the
computations about scaling especially for models with movable bed only can be so-
called “orientating calculations” which allow to tit the selection of scales into a
physically founded frame (Dorer, 1984).
These statements were fully confirmed by own comprehensive experiences with
movable bed river models as explained above.

Hydraulic river models with movable bed and approximate

similarity - a valuable tool for solving river engineering
problems
Many years experience proved that hydraulic river models with movable bed provide
results and knowledge of river engineering projects, which could hardly be received by
other resources, because in the model the expected morphological changes can be
studied in detail and such a model delivered a visuality both of the streamflow and of
the river process (see figs. 2 and 3) For solving river engineering problems of difficult
nature the higher expenses of a model with movable bed compared with that one with
fixed bed are justified, although it is only of approximate similarity (but what in another
manner in some cases is also true for fixed bed models). If one accepts the validity of
modelling a river process within the framework of some constraints and interpreted the
results within the background of modern hydraulic research, the possibility of misdesign
can greatly be reduced or even eliminated.
In the earlier years of hydraulic modelling there were sometimes reservations
against it, but the success had convinced most of the sceptics. But in some few cases
there are also such tendencies today, especially in river engineering too. Here it shall be
quoted a relevant event already over sixty years ago but from which it can be learned
also today.
Nearly legendary are the remarks by the senior of hydraulics and hydraulic
modelling, H.Engels, that it would be erroneous to suppose, “that the real natural
realization of a building or river training would be the best test... Effectual tests... on
realized river stretches mostly will be unreasonable expensive and will last very
long.. .The often praised test in life-size therefore.. commonly has only few value and in
no case can be confronted to the proper model test as equivalent” (Engels, 1926). In a
paper (Faber, 1927) referring to these comments Faber went so far as to maintain, that
“to experiences, which are in hand, it must be excluded, that the authoritative research
in river engineering can be ceded to the laboratories”. Faber then was one of the first,
which deserved themself about clarifying the relationships between river training and
river morphology on the base of observations in nature, and his efforts and also his
success about the development of river training technique are too important, as to
dismiss the cited remark easy as disproved by practice. He saw “the source of
knowledge.. hided under measurements and observations made at these rivers for more
than hundred years”. Taking into account the statement also by Engels, that “a trial...at

387
Fig. 2
River model with movable bed during test run:
river reach with different types of regulation structures,
upstream revetment changes to groynes
a realized river stretch in connection with model tests for the purpose of checking the
transferability of the results of model tests can be very valuable, both for science and
practice”, then the discussion by Faber in the true sense can not be estimated as a direct
controversy, but as demonstration of the necessity of a complex approach including
observations in nature as well as in model.
In this connection it must be emphasized the modern research about fluvial
processes. Especially in the last decades beside theoretical approaches it was fertilized
by the work about the partly semi-empiric so-called hydro-morphological theory of
river bed processes at the State Hydrological Institute in St. Petersburg and which is
connected with authors as N.E.Kondratiev, I.V.Popov, N.S.Znamenskaya,
B.F.Snishchenko and others (see e. g. (Kondratiev, Popov, Snishchenko, 1982). There
were explored the concepts first suggested by Lochtin. The results of hydro-
morphological analysis contribute to the solution of many problems involved in river
engineering and they should be regarded as a foundation for the most purposeful
performance of hydraulic studies of the river bed processes. The thesis included in it
about the river bed process as self-regulating offered new chances for the further
developing of the similarity theory and thus especially for the performance of movable
bed river models.
Although this paper directly dealt with physical modelling of fluvial processes
here it should be referred to mathematical models too. These models are based on the
mathematical formulation of the processes which take place, both the hydrodynamic and
the morphological process. Such models are all the better the more the interaction of
stream and sediment is investigated and described formulawise. This emphasized the
great importance of the exploration of the fluvial processes for the successful
application of mathematical models too. In the framework of development of computers
and hydroinformatics these models are indispensable means also in river engineering.
But like hydraulic models with movable bed also mathematical models demand
verification by means of prototype data, when using them for solving practical
problems. Their evidence depends to a large extent on the quality and quantity of the
available data. The development of mathematical models cannot fully replace the
conventional hydraulic model tests. Both methods have their specific advantages and
disadvantages. The use of mathematical models e. g. may decrease the extent of the
expensive tests with movable bed models, but certain problems may be reserved for
such models. A great methodically advantage for developing mathematical models is its
verification by the hand of hydraulic models as it is possible above all in hydraulic
research institutes, which investigate both mathematical and river scale models (Eckert,
Glazik, Thieler, 198 1).
Also up to date the following statements by Harleman on the occasion of IAHR’s
first panel discussion on new developments and needs in hydraulics are valid (IAHR,
1979): “What are the necessary conditions for improvement of simulation methods ?
More interaction between simulation and field data collection. More effort on
verification. More attention to intercomparison of simulation methods”.
And thus the circle closed themselves. Also nowadays hydraulic river modelling
with movable bed by means of an approximate similarity and by hand of orientation
calculations in the frame of a physically founded set of equations based on the basic
regularities of bed load transport appears furtheron sound, the more so as it offers
possibilities, to take into consideration certain quantified results from analysis of the
investigated natural river course empirically. This last inevitable is the fact too with the

390
more recent statements or formulations. In comparison with older publications which
are so to speak classic some recently papers provide hardly substantial accession of
knowledge. Several statements differ only in shades. To each hydraulic-modelling
specialist and each river engineering expert too it would be a genuine gain if from time
to time he has the opportunity to study the earlier publications, (The detailed list of
references included a greater number of such papers.) Also H.W.Shen in his speech
mentioned above cited preliminary axiomatic observations by Engels on indispensability
of experimentation to the I?uther development of hydraulics. In this sense this
contribution may conclude with the quotation of a statement by Engels too (Engels,
1926):
“Especially the river engineer shall lead the natural forces for his measures and
master them up to a certain degree in such a manner that the purpose of these measures
is reached in an as economical and complete way as possible. This, however, requires in
first line a knowledge of the natural processes themselves, a knowledge, which only can
be gained by observations. And theory in its real meaning is nothing else than the
impartial and methodical investigation starting from exact observations... In the hand of
a qualified theorist and coincident experienced engineer, model tests are of inestimable
value for science and practice”.
This is exactly true also today for hydraulic river models with movable bed, which
are a valuable tool for solving river engineering problems and consequently are a
significant part in the arsenal of river engineering methods.

References
Blau, E. (1963) ‘Die Entwicklung und der gegenwartige Stand der Versuchstechnik bei
Flubbauversuchen mit beweglicher Sohle’ “Wissenschaftliche Zeitschrift der
Technischen Universitat Dresden”, 12, 6, S. 166 1
Bogardi, J. (1974) Sediment transport in alluvial streams. Akademiai Kiado, Budapest
Bollrich,G.( 1987) ‘Julius Weisbach’s and Hubert Engelscontributions to hydraulics and
hydraulic research development’. In Garbrecht(ed.): Hydraulics and hydraulic
research - A historical review, IAHR 19351985; Balkema, Rotterdam, p. 275
Dick, T.M.(1987) ‘Symposium on scale effects in modelling sediment transport
phenomena (Report on the meeting)‘. “IAHR-Bulletin”, 4( 1987)4, p. 43.
Dorer, H.( 1984) Ahnlichkeit bei jlubbaulichen Modellen. “Mitteilungsblatt der
Bundesanstalt fiu Wasserbau”, Nr. 54, Karlsruhe.
Eckert, P., G. Glazik and H.Thieler, (1981). ‘Contribution to a determination of river
bed variations by means of hydraulic and mathematical models for maintaining the
navigation depth’. XXVth International Navigation Congress, Edinburgh; Section
I, Vol. 3, p. 433.
Einstein, H.A., N. Chien (1954). ‘Similarity of distorted river models with movable
bed’. “Proc. ASCE”, Vol. 80, Separate No. 566
Engels, H. (1914). Handbuch des Wasserbaues, I. Band. Verlag von Wilhelm
Engelmann, Leipzig und Berlin.
Engels, H., H.Krey (1916). ‘Ober den Wet-t von Nodellversuchen’. “Nitteilungen der
Versuchsanstalt fur Wasserbau, Wien”, Heft 3.
Engels, H., H. Krey, (1918) ‘Allgemeine Gesichtspunkte fur FluBregelungen’.
“Zentralblatt der Bauverwaltung”, Jg., S.60.

391
Engels, H.( 1926). ‘Geschichtliche Entwicklung, Wert von Modellversuchen’. In VDI-
Verlag, Berlin (English version: Hydraulic laboratory practice; ASCE, New
York), S.41.
Faber, E.(1927). ‘Der Wert baulicher Versuche in den geschiebefirhrenden Flussen’.
“Die Bautechnik”, Jg., S. 371,
Faulhaber, P., G. Kebelmann and H.Thieler, (1986). ‘Contribution on sediment
transport in open channels and river models’, IAHR-Symposium “Scale Effects in
Modelling Sediment Transport”; Toronto, August 1986; Proc. Supplement, p, 1,
Gehrig, W. (1967). ‘Ober die Frage der naturlichen Nachbildung der Feststoffbewegung
in Nodellen’. “Nitteilungen des Franzius-Instituts der TU Hannover”, Heft 29.
Gehrig, W. (1978). ‘FluBmodelle mit beweglicher Sohle’. In “Wasserbauliches
Versuchswesen”, Deutscher Verband fur Wasserwirtschaft, Mitteilungsheft Nr.4.
Glazik, G. (1969) ‘Notes on consideration of river morphological processes in river
model tests with movable bed’. XIIIth Congress of the IAHR, Kyoto 1969; Proc. .
Vol. 5-l (Seminars), Paper 3-4, p. 207.
Glazik, G. (1969) ‘Practical performance of river scale model tests with movable bed’.
See preceding, Paper 3-5, p. 211.
Glazik, G. (1981) ‘Morphological processes in hydraulic river models with movable bed
and practical performance of scale model tests’. In “Advances in Sediment
Transport”; Polish Academy of Sciences, Conference Jablonna. Nov. 1978.
Wydawnictwo “Ossolineum”, Wroclaw, p. 123,
Glazik, G.( 1984) ‘Influence of river model distortion on hydraulic similarity of
structures arranged at the channel’. Symposium on Scale Effects in Modelling
Hydraulic Structures, Esslingen, Proc. Paper 3.3.
Glazik, G., H.Schinke (1986). ‘Demonstration of scale effects in river models with
movable bed by means of comparisons of prototype and model family’. IAHR-
Symposium, IAHR-Symposium “Scale Effects in Modelling Sediment Transport”;
Toronto, August, p. 16.
Glazik, G. (1987) ‘Hydraulic river models with movable bed and approximate similarity
- a valuable tool for solving river training problems’. Conference “Actual
Problems of HydroEngineering: Model investigations in hydro-engineering”,
Wroclaw; “Scientific Papers of the Institute of Geotechnics of the Technical
University of Wroclaw”, Series Gonferences, No. 23.
IAHR (1979) Panel discussion “New developments and needs in hydraulics” XVIIIth
IAHR-Congress, Cagliari.
IAHR( 1938) Report on the First Meeting (Berlin 1937). Stockholm.
Kondratiev,N.E., IV. Popov and B.F. Snishchenko (1982). Eirndamentals of the
hydromorphological theory qf river processes. Gidrometeoisdat, St. Petersburg
(In Russian)
Korner, B.(1937) ‘Modellversuche fur den Wolga-Don-Kanal, ausgefuhrt in der
Berliner Versuchsanstalt fin- Wasserbau und SchifIbau’. “Mitteilungen der
Hannoverschen Hochschulgemeinschaf”, Heft 19.
Krey, H., R.Winkel (1926) ‘Die Preubische Versuchsanstalt fur Wasserbau und
SchifIbau in Berlin’. VDI-Verlag, Berlin, (English version: Hydraulic laboratory
practice; ASCE, New York). S, 192.
Krey, H., B.Korner (1935) Modellversuche jdr einen Flub mit starker
Geschiebebewegung ohne erkunnbare Bankwsnderung. Verlag von Wilhelm
Ernst und Sohn, Berlin.

392
Larsen, A.P., H.H.Bernhar-t (1987) Theodor Rehbock (18641950). IAHR 1935-1985;
Balkema, Rotterdam, p. 281.
Liebs, W. (1942) ‘Die Nachbildung von Flussen mit beweglicher Sohle im Modell’.
“Mitteilungen der Preubischen Versuchsanstalt fi.rr Wasser-, Erd- und Schiffbau”,
Heft 43, Berlin.
Rehbock, Th., E. Meyer-Peter (1937). ‘Diskussion zu Rohringer und Seifert’. Report
on the First Meeting (Berlin), Stockholm.
Rohringer, A. (1937). ‘Verzerrte Flubmodelle’. Report on the First Meeting (Berlin),
Stockholm.
Scheuerlein, H. (1980) The historical model tests of Engels for the Yellow River
reclamation in 1930-193.5 in perspective of modern research methodLy.
“International Symposium on River Sedimentation”, Beijing.
Seifer-t, R. Qber 193 8) Tiefenverzerrung und Gefalleverstirkung im wasserbaulichen
Modellversuch. Report on the First Meeting (Berlin). Stockholm.
Seifert, R. (1942). ‘Allgemeins Ahnlichkeitsbetrachtungen uber Modelle
geschiebefuhrender Flusse nach praktischen Gesichtspunkten’. “Die Bautechnik”,
20(1942)36/37, S. 327.
Shen, H.W. ‘Movable bed river models’. IAHR-Symposium “Scale Effects in Modelling
Sediment Transport”; Toronto, August 1986; Proc., p. 156
Shields, A.( 1936) ‘Anwendung der Ahnlichkeitsmechanik und der Turbulenzforschung
auf die Geschiebebewegung’. “Mitteilungen der Preubischen Versuchsanstalt fur
Wasser-, Erd- und SchitIbau”, Heft 26, Berlin.
Thierry, G. de, C.Natschoss (1929) ‘Die Wasserbaulaboratorien Europas’ VDI-
Verlag, Berlin 1926 (English version: Hydraulic laboratory practice; ASCE, New
York).
Winkel, R.( 1926) ‘Die Grenzen der Obertragbarkeit der Versuchsergebnisse und die
Modellahnlichkeit bei flubbaulichen Versuchen’ In VDI-Verlag, Berlin (English
version: Hydraulic laboratory practice; ASCE, New York), S. 55.
Yalin, M.S (1986). ‘Some aspects of physical modelling of flows with a free surface’.
IAHR-Symposium “Scale Effects in Modelling Sediment Transport”; Toronto,
August, Proc., p. 1.
Zschiesche, 0. (1954). ‘Ergebnisse von Flubmodellversuchen mit beweglicher Sohle fiu-
zwei Elbestrecken’. “Veroffentlichungen der Forschungsanstalt fur Schiffahr-t,
Wassenmd Grundbau”, Nr. 4, Akademie-Verlag, Berlin.

393