J. Bio. & Env. Sci.

2011

Journal of Biodiversity and Environmental Sciences (JBES)

ISSN: 2220-6663 (Print) 2222-3045 (Online)
Vol. 1, No. 4, p. 35-42, 2011
http://www.innspub.net
RESEARCH PAPER OPEN ACCESS

Analysis of modelling water quality in lakes and reservoirs

Paul C Njoku

School of Engineering and Engineering Technology, Department of Environmental Technology

Federal University of Technology, Owerri and Archna Swati Njoku

Received: 20 May 2011

Revised: 18 July 2011
Accepted: 19 July 2011

Key words: Analysis modeling water quality lakes reservoirs, temperature runoff.

Abstract
Water quality in lakes is analyzed in this study. It deals with modeling of long residence times in water bodies, natural
or man-made. The key driving force of Lake Dynamics is usually temperature. Its vertical distribution defines whether
a lake is stratified or not. The result shows that runoff process is involved the biochemical oxygen demand
concentration of the lake water is reflected mainly by the subsurface runoff component, suspended solid (ss)
concentration is affected mainly by the surface runoff component.

Corresponding Author: Paul C Njoku  [email protected]

35 | Njoku
 J. Bio. & Env. Sci. 2011

Introduction EUTRO, the ecological module of WASP, with a two-

(Cho and Seo, 2007a). assumed that the runoff dimensional finite element hydrodynamic model,
pollution loads, at the time of raining from the and then applied it to a lagoon in Venice. In this
adjacent resort and farmland located at the study, we have applied the WASP model to the water
watershed,affect the lake water quality significantly. quality control of Yeongrang Lake, which is now
Recently, a number of studies were conducted with deteriorated. By using the rainfall-runoff curves, as
respect to the characteristics of suchlike pollution well as the rainfall-runoff pollutant load curves.
loads by rainfall runoff (Bertrand-Krajewski et al.,
1998; Bedient et al., 1978). Since the first flushing by Methods and material
the initial rainfall is considerable and takes a large The structure and characteristics of WASP7 model
portion in the total runoff of pollution loads, The WASP model is widely used in interpreting and
especially in urban areas, research regarding this has predicting the water quality response to the pollution
been actively conducted (Lee and Bang, 2000; Lee et of water resources by natural phenomena or human
al., 2003; Novotny, 2003). Based on the storm water activities. The WASP is a dynamic model for a water
runoff data from the Yeongrang Lake watershed, Cho body including sediment, allowing one-, two-, and
and Seo (2007b) prepared dimensionless three-dimensional calculations, and composed of an
eutrophication module that enable calculation of the
M(V) curves indicating the distribution of pollutant 13 variables related to water quality, a toxicant
mass vs. volume in storm water discharges and module, a mercury module, and a heat module, as
carried out the regression analysis on the shown. The model can be used in connection with the
dimensionless M(V) curves with the power function pollution load data calculated by external models,
and second and third degree polynomial equation such as SWMM, HSPF, and NPSM, using the pre-
curves, in order to develop a tool that allows processor, or it can organize the input data in
quantitative evaluation of first flushing. They have connection with hydrodynamic models, such as
also deducted rainfall-runoff curves and rainfall- EFDC and DYNHYD. In addition, input data, such as
runoff pollutant load curves by investigating the various constants, parameters, segments, loads, and
rainfall runoff from Jangcheon, which provides most boundary conditions, can be easily organized. It also
of the runoff input, as well as the storm sewers that deals with the change of advection, dispersion, point
have a large amount of rainfall runoff, and by source, and diffuses pollution loads depending on
estimating the runoff and the runoff pollution loads time. In the postprocessor, the calculated water
of each rainfall. The WASP model (Ambrose, 1987; quality data can be displayed with the measured
James et al., 1995; Kim et al., 2004; Lung et al., data, so that they can be easily calibrated.
1993; Lung and Larson, 1995; Tufford and McKellar,
1999; US EPA, 2006; Wool et al., 2001), developed The relationships between the runoff components
by the U.S. EPA, is widely used domestically and and water quality changes are finally concluded. The
internationally as the model to calculate the water study shows the Biochemical Oxygen Demand (COD)
quality and phytoplankton in a lake. Recently, Zhang concentration of the lake water is affected mainly by
et al. (2008) have developed the LM2-Toxic model, the sub-surface runoff component, and on the
based on the WASP model, to calculate PCB contrary, Suspended Solid (SS) concentration is
concentration, and they have applied it to Lake affected mainly by surface runoff component.
Michigan. Canu et al. (2004) have combined

36 | Njoku
 J. Bio. & Env. Sci. 2011

Runoff simulation Before repletion of storage: P-E=W2-W1 (1a)

The Xinanjiang runoff model is used in this study to After repletion of storage: P-E-R=WM-W1 (1b)
simulate the runoff process. This model has been Where, W1, W2-the areal mean tension water storage
successfully and widely applied in humid and semi- at beginning and end of specific time interval. R-the
humid areas in China since its development in the total runoff generation in specific time interval, and
1970s (Zhao et al., 1992). It is a semi-distributed R=RS+RG.
runoff model for use in humid and semi-humid Therefore, the total runoff generation can be
regions. The whole catchment is divided into several separated as two parts:
sub-catchments according Thiessen method. The RS=P-E-Fc; RG=Fc (2)
runoff generation in each sub-catchment is Strictly speaking, equation (1) and (2) are only
simulated, and routed to the catchment outlet by suitable to a specific point in the catchment. Because
using of the Muskingum method. the spatial and temporal distribution of tension water
storage is not possibly uniform in the whole sub-
catchment area. The Xinanjiang model uses a
parabolic curve to represent the distribution
variation of tension water storage, which showed in
equation (3) and Fig. 3:

Fig. 1. Conceptual model of heat budget of a Lake.

Runoff generation model

The main feature of the Xinanjiang model is the
concept of runoff generation on repletion of storage,
which means that runoff is not generated until the
Fig. 2. The concept of runoff genera.
soil moisture content of the aeration zone reaches
maximum capacity, and thereafter runoff equals the
rainfall excess without further loss. This concept of
runoff generation can be expressed in Fig. 2 (the
shaded box represents a soil column).

In Fig. 2, P, E and Fc are precipitation (mm/day),

evapotranspiration (mm/day) and final infiltration
rate (mm/day) in specific time interval, respectively.
W is areal mean tension water storage (mm) and WM
is the capacity of W. RS and RG is surface runoff and
Fig. 3. Distribution of tensionwater storage capacity.
ground water contribution respectively. According
the assumption of this concept, by using the water
balance calculation, we have:

37 | Njoku
 J. Bio. & Env. Sci. 2011

Where, f-the partial pervious area of the catchment direct runoff and includes two parts: the “real”
whose tension water capacity is less than or equal to surface flow and interflow. In the three runoff
the value of the ordinate W΄m. The tension water components’ Xinanjiang model, the total runoff
capacity at a point W΄m varies from 0 to its generation is separated into three parts: RS-surface
maximum W΄mm. F is total catchment area and B is runoff, RI-a contribution to interflow and RG-the
exponent of the tension water capacity curve, which contribution to the groundwater by introducing
represents the non-uniform of tension water storage another 4 parameters instead of final infiltration rate
distribution (Figure 3). Therefore, the areal mean Fc. They are: S-free water storage (mm), SM-areal
tension water capacity WM can be expressed by mean free water storage capacity (mm), KI-
integration: coefficient related to interflow component and KG-
coefficient related to groundwater component. The
water component equation can be expressed as
(4) follows:
And the ordinate which corresponding to W is: RG=KG×S×FR (8a)
RI=KI×S×FR (8b)

RS=0, when P-E+S SM (8c)

(5) RS=(P-E+S-SM) ×FR, when P-E+S>SM (8d)
The runoff generation can be expressed by following R=RS+RI+RG (8e)
equations: Where, FR represents the area where runoff
when P-E 0, R=0 (no runoff generation); generates.
when P-E>0, R>0 (runoff generates):
Evapotranspiration model
In the Xinanjiang model, a three-layer model is used

when P-E+A (6a) to estimate the evapotranspiration depending on the

pan evaporation. The soil column is divided into
, when P-E+A< (6b) three layers called upper, lower and deepest layer,
The result of above generation equation is total which having areal mean tension water storage WU,
runoff generation, which including surface runoff WL and WD respectively (WU+WL+WD=W), and
and sub-surface runoff. Because the variation of tension water storage capacity WUM, WLM and
runoff generation area is considered, the runoff WDM respectively (WM=WUM+WLM+WDM). The
component separation equation (2) must be changed evapotranspiration firstly begins from the upper
to equation (7): layer, and after the WU is exhausted,
When P-E>Fc: RG=Fc×(f/F)=Fc×(R/(P-E)) (7a) evapotranspiration starts from lower layer. The
RS=(P-E-Fc) ×(f/F)=R-RG (7b) actual evapotranspiration in the upper and lower
When P-E Fc: RS=0, RG=R (7c) layer EU, EL can be expressed as follows:
EU=K×EM, when WU>EM (9a)
The equation (6) and (7) are the so-called the EU=WU, when WU EM (9b)
Xinanjiang model with two runoff components.
EL=(K×EM-EU) ×WL/WLM when EL C×EM (9c)
Actually, the surface runoff RS separated from
equation (2) and (7) usually includes most of EL=C×EM, when EL<C×EM and WL C×EM (9d)

interflow part. Strictly speaking, it should be called as

38 | Njoku
 J. Bio. & Env. Sci. 2011

Where, K-coefficient to adjust evaporation; EM-pan Where, T-air temperature in ℃; esat-in hPa.
evaporation; C-minimum evapotranspiration
efficiency, which is a constant in a specific basin area. Optimization of parameters and runoff simulation
When the tension water storage in lower layer WL is The main parameters of runoff generation model,
decreased to a value of C×EM, the model assumes evapotranspiration model and runoff separation
that evapotranspiration in the deepest layer (ED) equation can be summarized as following:
begins at following rate: Runoff generation: WM, B
EL=WL, when WL<C×EM (9e) Evapotranspiration: K, WUM, WLM, C
ED=C×(K×EM-EU)-EL when WL<C×EM (9f) Runoff Separation: SM, KG, KI

Estimation of potential evapotranspiration K is a coefficient related to pan evaporation. Among

The Xinanjiang model includes two inputs: the above parameters, the simulation result is
measured areal mean precipitation and measured particularly sensitive to this parameter which
pan evaporation on the sub-area. Because there is no controls the water balance of whole catchment. This
pan evaporation data set in Japan, in this study the parameter should be determined by field experiment.
potential evapotranspiration data is used instead of Due to lack of pan evaporation data, this study firstly
pan evaporation data. Usually it is difficult to tries to find suitable K value for the Kamafusa Lake
estimate the potential evapotranspiration, because catchment by optimization. The parameter SM, KG
evapotranspiration is dependent upon some and KI are also more sensitive to output and
parameters, which is difficult to estimate their value. optimized according to the recommended value for
This study uses Hamon equation to estimate humid area. For other parameters like WUM, WLM,
potential evapotranspiration: B and C, the values recommended for humid area in
literatures are adopted.
(10) The two indexes are used to evaluate the simulation
Where, Ep-daily average potential evapotranspiration results of each year:
(mm/day); D0-available sunshine duration
(12hrs/day); pt-saturated absolute humidity (g/m3).
Actually the saturated absolute humidity is the (13)
density of water vapor (kg/m3), i.e., mass of water
vapor in unit volume of air under saturation vapor
pressure. It can be calculated by using following (14)
equation: Where, WBD- annual water balance difference (%);
NTD-Nash index, which means the value is closer to

(11) 1, the better result is simulated; Qobs-observed runoff

Where, esat-saturation vapor pressure (hPa); T-air flow rate (m3/s); Qcalc-calculated runoff flow rate

temperature (K); pt-in g/m3.

(m3/s) and - daily average calculated runoff
Then saturation vapor pressure can be approximately
flow rate (m3/s).
calculated by Tetens equation:

The mathematical modeling of lakes is based on the

(12)
following equations:

39 | Njoku
 J. Bio. & Env. Sci. 2011

Conservation of mass At- £• - Qjout^ - Qj Outflow

Conservation of momentum ╧Qg Groundwater
Transport of contaminants
Chemical and biological process kinetics Where, Vj-= volume of the jth layer
conservation of heat Qjin Qjout = horizontal surface inflow and outflow
(abstraction) from the ith layers
The first two equations are the foundations of all Qj = outflow vertically from the jth layer
hydrodynamic problems, while the third is used in Qj-1, Qj+1 = inflow vertically to the jth layer from the
the water environment and other environments (air, j - 1 and j + 1 layers
soil, and biota). The fifth, the conservation of heat,
has not been used so far in this text and while it may Model application
be applicable to the air environment, it is usually not Several of the above terms require information on
of relevance to rivers. the heat budget of the lake as depicted in Fig. 2.
Assuming that P,E and Q are included in the vertical
Usually lakes require to be modelled when there are fluxes {Q,. i and g,+i) then the mass conservation for
problems with water quality and most often occur in a typical layer of thickness Az is
both shallow and deep lakes when they are stratified. əVj = Qjin ╧ (Qj-1 + Qj+1)z –Qjout ət
Stratification is when there is a significant where, Qjin' Qjout = horinzontal flow advection
temperature difference between the epilimnion Qj-1, Qj+1' = vertical (low (into or out of the element)
(upper waters) and the hypolimnion (lower waters).
The narrow band between both with a steep Results
temperature gradient is the thennocline. The conservation of heat equation stored in a
Ifthedensiometric Froude number) is < 0.01, and horizontal control volume layer is described by Orloh
then the water body stratified. The physical processes (1981). Solution of the mass balance F.2 requires
involved in the heat budget of a lake are depicted in information on the heat energy balance. The
Fig.1. end result of modelling of reservoir is the production
of vertical profiles of temperature to determine the
Model setup lake status for different inputs of solar heat and
One technique to model stratified lakes is to divide freshwater flows.
the lake into a number of horizontal layers as shown
in Fig. 2. The more simplistic model will have three Detailed descriptions of the above and solution
'layers': the epilimnion, the thermo cline and the techniques are to be found in several articles (Oriob
hypolimnion. More capable models will have several 1981; James, 1993; Fischer et al., 1979). A computer
layers. On each layer, a mass balance is computed, program DYRESM detailed by (Fischer et al., 1979)
allowing for the transfer of flow and constituents and Imbergcr (University of West Australia) is
from one layer (vertically) to another. commercially available to solve lake hydrodynamic
and lake water quality problems.
Model performance
For instance, the mass flow balance of the jth Conclusion
(horizontal) layer is These model are used predict temperature profile in
Avj = Qjin = (Qj -1 + Qj +) = P inflow stratified lakes and are ideal for lake water quality

40 | Njoku
 J. Bio. & Env. Sci. 2011

management in identifying incipient seasonal model(LM2-Toxic) for the Lake Michigan Mass
eutrophication. Complex models may extend to two Balance Project, Ecological Modelling 219, 92-106.
and three dimension and/or incorporate reaction
chemistry and ecological models. Recent work by James RT, Bierman VJ. 1995. A preliminary
lake models, DYRESM. modeling analysis of water quality in Lake
Okeechobee, Florida: calibration results, Water
References Research 29(12), 2755-2766.
Ambrose RB.,1987. Modeling volatile organics in
the Delaware Estuary, Journal of Environmental Kim D, Wang Q, Sorial GA, Dionysiou DD,
Engineering 113(4), 703-721. Timberlake D. 2004. A model approach for
evaluating effects of remedial actions on mercury
Bedient PB, Harned DA, Characklis WG. speciation and transport in a lake system, Science of
1978. Stormwater analysis and prediction in Total Environment 327, 1-15.
Houston, J. Environ. Eng. Div. ASCE 104(EE6),
1087-1100. Lee JH, Bang KW. 2000. Characterization of
urban stormwater runoff, Water Research 34(6),
Bertrand-Krajewski JL, Chebbo G, Saget A, 1773-1780.
1998. Distribution of pollutant mass vs volume in
stormwater discharge and the first flush BALWOIS. 2010. Ohrid, Republic of Macedonia -
phenomenon, Water Research 32(8), 2341-2356. 25, 29 May.

Canu DM, Solidoro C, Umgiesser G, 2004. Lee JH, Yu MJ, Bang KW, Choe JS. 2003.
Erratum to modelling the response of the Lagoon of Evaluation of the methods for first flush analysis in
Venice ecosystem to variations in physical forcings, urban watersheds, Water Science and Technology
Ecological Modelling 175, 197-216. 48(10), 167-176.

Cho JH, Seo HJ. 2007a. Parameter optimization Lung WS, Martin JL, McCutcheon SC. 1993.
of SWMM for runoff quantity and quality calculation Eutrophication and mixing analysis of embayments
in a eutrophic lake watershed using a genetic in Prince William Sound, Alaska., Journal of
algorithm. Water Science and Technology: Water Environmental Engineering 119(5), 811-824.
Supply 7(5-6), 35-41.
Lung WS, Larson CE. 1995. Water quality
Cho JH, Seo HJ. 2007b. Lake Water Quality modeling of the upper Mississipi River and Lake
Modelling Considering Rainfall-Runoff Pollution Pepin., Journal of Environmental Engineering
Loads, journal of Environmental Impact Assessment 121(10), 691-699.
16(5), 341-350.
Novotny V. 2003. Water quality: diffuse pollution
Zhang X, Rygwelski KR, Rossmann R, Pauer and watershed management. John Wiley & Sons,
JJ, Kreis RG. 2008. Model construct and Inc., New York, USA.
calibration of an integrated water quality

41 | Njoku
 J. Bio. & Env. Sci. 2011

Tufford DL, McKellar HN. 1999. Spatial and Oriod GT. 1981. models for stratified
temporal hydrodynamics and water quality modeling impoundments in models. for water quality Mgt. all
analysis of a large reservoir on the South carolina Biswas (ed) Mc Graw Iri New York.
(USA) coastal plan, Ecological Modelling 114, 137-
173. Casmit, John G, Schladraw G. 1993. Vertical
Distribution of particles in stratified lackes ASCEL.
US EPA. 2006. Watershed and Water Quality Of Environment Engr.
Modeling Technical Support CenterWater Quality
Analysis Simulation(WASP) Version 7.1. James A. 1993. an introduction to water quality
modeling John Wiley, New York.
Wool TA, Ambrose RB, Martin JL, Comer EA.
2001. Water Quality Analysis Simulation(WASP) Fischer HB. 1979. Mixing in Inland coastal
Version 6.0 Draft: User's Manual, US EPA. coaters. Academic Press New York.
Zhao RJ. 1992. The Xinanjiang model applied in
China, J. Hydrol. 135, 371-381.

42 | Njoku