ZevenbergenAndThorne DigitalTerrain EarthSurfaceProcesses1987
ZevenbergenAndThorne DigitalTerrain EarthSurfaceProcesses1987
ZevenbergenAndThorne DigitalTerrain EarthSurfaceProcesses1987
56 (1987)
ABSTRACT
- Land surface topography significantly affects the processes of runoffand erosion. A system which determines slope,
f and curvature in bolh the down-slope and across-slope directions is devclorxd for an altitude matrix. Also, Ihe
upslope drainage area and maximum drainage distance are determincd for every point within the altitude matrix. A
FORTRAN 66 program performs the analyss.
INTRODUCTION
Geomorphologists and engineers have long recognized the importance ot- the study of land surfaces and the
effects of topography on hydrologic and sedimentary processes. Slope, the most widely used topographic
measurement, influences flow rates of water and sediment by controlling the rate of energy expenditure or
stream power available to drive the flow. Aspect defines the slope direction and therefore the direction of flow.
: Knowledge of how aspect varies throughout a catchment provides the information necessary to determine
*What upslope land area contributes to the flow at any point in the catchment. Profile curvature, the rate
of ,change of slope, affects flow acceleration and
and therefore influences aggradation and
'degradation. The curvaturc of the land surface transverse to the slope direction, called planform curvature,
Influences flow convergence and divergence. A FORTRAN 66 program which analyses an altitude matrix for •these
four topographic indices plus upslope drainage area and the distance of the maximum drainage path has
n developed. T his program improves previous analyses because it objectively determines the
upstrcam drainage area for every point within an altitude matrix.
PREVIOUS WORK
isdiscussion centres on topics of particular relevance to this paper. For a more in-depth review, Evans ( 1972)
rovides a thorough discussion on tcrrain analysis.
Rhind (1972) used patches of twelve elevations for fitting quadratics and cubic surfaces to aid in contour pping,
though no topographic parameters were evaluated from these surfaces. Young (1978) and Evans 1979) developed a
method of topographic analysis yielding four topographic indices. Their analysis extends methods presented by
Tobler (1969) and Sharpnack and Akin (1969). The four topographic indices oduced by Evans' system are slope,
aspect, profile curvature and plan curvature. The analysis is based on a
IOS05.oo
1987 by John Wiley & Sons. Ltd.
48
L. W. ZEVENBERGEN AND C. R. THORNE
rectangular matrix ofevenly spaced elevations covering the entire area in question. A 3 x 3 submatrix
(Figure is analysed repetitively throughout the rectangular altitude matrix. All topographic indices are
related to th central point of the 3 x 3 submatrix. For the analysis, Evans chose a full quadratic surface
given by the equation:
+ +Cxy+Dx+EY+F
Z=
The nine elevations of the 3 x 3 submatrix are used to determine the Six parameters, A ... F, of equation l. This
results in a surface which Will not necessarily pass through the nine original elevations. The units of the
topographic indices derived by Evans arc: aspect, degrees (0-360); slope, degrees (0—90); profile and
plan. curvatures,
where LU is the unit oflength in which elevation is measured (i.e. LU = metres if,
Z is measured in metres). Evans (1980) suggests a Wide variety of uses for topographic analysis. Of these
suggestions, that of 'predicting the process rates and discharges from their relationships to attributes of land surface form• is
ofprimary interest to the authors. For chis purposc, however, upstream drainage area at a poi is of great importance but is not
determined in Evans' analysis. The authors, therefore, undertook a review of
topographic analysis to develop a system to incorporate upstream drainage area. In the course of this
review, Other changes were made to improve the analysis.
z,
z, z,
z,
z,
z,
MODIFICATIONS
Evans* quadratic equation, equation l, need not pass exactly through the nine elevations of the 3 x 3 altitude
submatrix. If the surface represented by equation I does not coincide with the nine original elevations, it is
questionab[e whether equation I represents the land surface accurately. Therefore, the first modification is
that ofchoosing a surface which does pass exactly through the nine submatrix elevations. The appropriate
surface is produccd by the partial quartic equation:
Z = Ax2y2 + B? y + Cxy2 + Dx2 + Ey2 + F x», + Gx +
The nine parameters, A .
l, can be determined from the nine elevations of the 3 x 3 submatrix by Lagrange
Plynomials. Equation 2 is more general than equation I and can be regarded as being more 'flexible'. If the
surface indicated by the nine submatrix clevations is oflower order than equation 2 the appropriate
coefficients Will equal zero and not affect the equation (if the surface is quadratic A, B and C Will equal zero
and if the surface is a Plain ,4 .
F Will equal Zero). It cannot be stated, however, that either equation I or 2 Will fully replicate the actual
land surface between grid points. It is necessary to choose an appropriate grid mesh distancc to minimize this problem.
The relationships between the nine parameters of equation 2 and the nine submatrix elevations are:
(3)
(4)
B= + Z, -Z, -Z9)/4 -(Z2
(5)
C= -Z: + 7.3 -Z, +
(6)
D= -Zs]/L2 a
(7)
E= + ze)/2 -Z5]/L2
F = ( -Zl + Z) + z, -Z9)/4L2 (8)
(9)
6
H = (4 -Z8)/2L
{Zl 10 Z9 arc the nine submatrix clevations numbered systematically as shown in Figure 5 isl. Zthe centre point
(x = y — 0). L is the distance between matrix points in the row and column dimtions and must be in the same
lunits as Z.
The need for the second modification is dictated by the application of topographic indices to the
prediction of
of flow and sediment transport. Since the estimation of hydraulic and sediment transport
rameters (mean flow velocity, avcragc Shear stress, stream power) traditionally require slopes in
dimensionless terms (dH/dL
sin 0) rather than in terms of degrees, it is desirable to measure slope
in ,dimensionless terms. Also, changes in the hydraulic and sediment transport parameters are related to the
rate of Change in (dimensionless) slope so curvature in terms of I/LU is superior to curvature computed in
terms of degrees/LU. However, the option to retain the Other units is available in the program.
Finally, one addition is made to Evans' terrain analysis system, again to facilitate flow and sediment
nsport prediction. Using the aspect values determined for each point in the watershed, two 'upslope'
ameters are determined. These parameters are UPAREA and UPDIST. UPAREA is the drainage area
ntributing to a point's flow of water and sediment and UPDIST is the longest travel path to an upslope
ivide. These parameters are determined at every matrix point but would only have meaning when the altitude
trix includes an entire drainagc area.
TOPOGRAPHIC INDICES
topographic indices are found by differentiating equation 2 and solving the resulting equation for the nttal
point ofthe 3 x 3 submatrix (x = y — 0). The slope is the first derivative ofZ with respect ofS, where S is
50 L W. ZEVENBERGEN AND C. R. THORNE
PROGRAM CODING
A program written in FORTRAN 66 uses the previously outlined procedure to determine aspect, slope, profile
curvature, plan curvature, drainage area, and upslope distance for points within an altitude matrix. The altitude
matrix must be rectangular with unknown altitudes, those altitudes outside the catchment under analysis, given a
zero value. The matrix need not have a northern orientation, as the angle the matrix is rotated from north is a
required input datum. There are three options available to the user: (I) Slopes can be calculated in either
dimensionless terms or degrees; (2) and (3) profile cutvature and/or plan curvature can be calculated in umts of
1/mLU or degrees/mLU.
ANALYSIS
A subprogram, OUTPUT, produces line printer maps of the Six topographic indices plus altitude. Figures
2 to 8 are from an analysis of the Goodwin Creek watershed in Panola County, northwest Mississippi. The
82 square mile (21 km2) watershed was hand digitized at a
ft (61 m) grid mesh from a I
scale, 2 ft
(061 m) contour interval topographic map producd by the U.S. Army Corps of Engineers. Altitude and slope are
mapped showing 10 per cent intervals of the maximum value (Figures 2 and 3). Profile curvature and plan
curvature are also mapped at 10 per cent intervals of the (absolute) maximum with a minus sign overstrike for
10.30 2. aoa
c. N' e.c•o c. o•.
Figure 2. Linc pnntcr map of alti:ude Figure 3. Line printer map or s10*
L. W. ZEVENBERGEN AND C. R. THORNE
values less than zero (Figures 4 and 5). Aspect is mapped showing eight directions (Figure 6). A good
representation ofthc drainagc network is shown by the map of upstream area (Figure 7), also mapped at 10 pet cent
intervals. Use of a 10 per cent interval can be inadequate for this mapping because a small drainage network (with
respect of the catchment size) often captures most of the upstream area and runoff. It appears•, that upstream area
varies somewhat geometrically with distance so the square-root of upstream area is also.
o.cca
0.090
mapped at 10 per cent intervals (Figure 8), which gives a more detailed representation of the drainage
network. A similar map (not included) to Figure 8 is also produced for the drainage distance parameter.
These line printer maps give quick visualization of the range and distribution of the topographic parameters.
Direct quantitative analysis from these maps would be dificult as each symbol represents a tange of values
(i.e. each symbol in Figure 3 represents slopes over a range of about 1'3 per cent).
•oe ua•a e;
Figure 6. Line printer map of aspEt Figure 7. Line primer map of upstream atea
'a
L W. ZEVENBERGEN AND C. R. THORNE
00 "ow.aaa•oeoox uoacoo
DISCUSSION
The use ofa partial quartic or quadratic equation for land surface analysis assumes that the surface represented by
the altitude matrix has a continuous first derivative. Although this is a reasonable assumption for most land
surfhces there are instances whcn this assumption can cause problems. When a land surface contains an abrupt
break in slope, such as a river bank, special treatment must be made for grid points which fall within such areas.
TOPOGRAPHY ANALYSIS
The elevations of grid points which fall between river banks should be adjusted to a height slightly lower
than the top of bank elevation. This avoids the
of extremely high values of slope and curvature and
spurious values of
This is especially a problem if the grid mesh distance is greater than the river
width. When this hapFns, only occasional grid points fall between the river banks causing an apparent
hole in the land surface. [f this occurs, it is likely that the drainage area and drainage
routine Will fail.
Choice of an appropriate grid mesh distance is clearly an important consideration. Evans (1979) states
that analyses of two land surfaces
be directly compared if the grid mesh or the elevation precision differ. Also,
choice of an appropriate grid mesh is derxndent on the variation of slope and aspect (profile and plan curvature)
and on the precision of elevation data. Too large a grid mesh Will miss the topographic features ofa land surface
while too small a grid mesh Will require excessive, precise data giving only minor Jand features.
APPLICATIONS
As shown by Evans (1980), there are many, widely varied uses for topographic analysis. Aspect and slope
were used by Foyster (1973) in a grid square technique for mapping evapotranspiration. Anderson and Burt
(1978) and O'Loughlin (1981) showed that the presence of hollows and spurs
affects rates of hillslope
discharge. Burt and Butcher (1985) related a variety of topographic indices to soil water potential, saturation
depth, and slope discharge. Thorne and Zevenbergen (1986) used a combination of drainage area,
and
planfotm curvature to predict areas of initial gully incision in croplands. Although upstream area appears
as a variable and is found to be significant in Butt and Butcher's (1985) analysis, their method for obtaining
this topographic index was not automated, as it is in the authors' analysis.
CONCLUSION
The topographic analysis presented here represents an initial analysis of a land surface, with line printer maps
provided as convenient visual output. Any
analyses, whether statistical, hydrologic, or
geomorphic, an be accomplished by subprograms written by the user. The program is written in
standard FORTRAN 66 and has been run on an ICL
series computer using both the FORTRAN 66 and 77 compilers. For
compatible systems, a copy of the program can be obtained by sending a magnetic tape to either author at
the U.K. address.
ACKNOWLEX)EMENTS
This project is funded through a specific cooperative agreement between the USDA-ARS Sedimentation
Laboratory, Oxford, Mississippi and Colorado State University, Fort Collins, Colorado. The authors
gratefully acknowledge the support, advix and assistance of Dr. Neil Coleman, Dr. Earl Grissingcr, and
Mr. Joe Murphy, ofthe USDA-ARS, and Mr. Pete Forsythe and Dr. Bill Mildner, ofthe USDA—SCS. The
analysis benefited from helpful discussion with Dr. lan Evans, University of Durham.
REFERENCES
Anderson, M. G. and Burt, T. P. 1978. 'The role of topography in controlling throughflow generation', Earth Surface Processes. 3(4), 331-334.
i Burt. T. P. and Butcher. D. P. [985. 'Topographic controls of Soil Moisture Distributions•, Journal ofSoil Science,
469—486.
Evans, l. S. 1972. 'General geomorphometry, derivatives of altitude and descriptive statistics', in Chorley, R. J. (Ed.), Spatial Analysis in
Geomorphology, p. 36, Methuen, London.
1) Evans S. 1979. 'An integrated system of tetrain analysis and slope mapping•. Finat report on Contract
US.
Army European Research
London.
Ewns, l. S. 1980. 'An integratd system of terrain analysis and
mapping•, Zeitschriftfür Geomorphologie, N.F., Supplememband. 36,
274-295.
Foyster, A. M. 1973. 'Application of the grid square technique to mapping of evapotranspiration', Journal of Hydrology, 19, 205—226.
üLoughlin, E. M. 1981. 'Saturation regions in
and their relationships to soil and lopgraphic
Journal of
Hydrology, S3, 229-246.
56
L. W. ZEVENBERGEN AND C. R. THORNE
Thorne, C. R. and Zevenbergen, L. W. 1986. 'Ephemeral gullies as sourxs of sediment•, Proceedings ofthe Fourth Federal I
Sedimentation Conference, U.S. Govt. Printing Ofice, 3-152-3-161.
Tobler. W. R. 1969. 'An analysis of a digitatized surface', in Davis, C. M. (Ed.), A Study OJ Land Type, Univ. of Michiga
Geography. 59• 76.
Young. M. 1978. •Statistacal charactenzation of altitude matrices by computer'. The fifth progress report on
DA-ERO- 591-73-
U.S. Army European Rescarch Office, London.