Professional Documents
Culture Documents
Assessing water erosion in small (Baffaut et al. 1997). Each hillslope, rep-
resented as a rectangle in WEPP, must
watersheds using WEPP with GIS have a representative length, width, and
slope profile as shown in Figure 1, part 3.
and digital elevation models Hillslopes drain into the top, left side, or
right side of a channel, eventually leading
to the watershed outlet.
TA. Cocbrane and D. C. Flanagan Integration of WEPP with geographi-
cal information systems (GIS) is desirable
ABSTRACT Three different approaches using geographical information systems (GIs) and to facilitate and possibly improve applica-
digital elevation models (DEMs) are described and evaluated for applying the Water Erosion tion of the model. Savabi et al. (1995)
Prediction Project (WEPP) model to assess water erosion in small watersheds. Thefirst approach conducted an initial application of WEPP
describes a typical application of the watershed version of WEPP using GIs only as an aidfor on the Purdue University animal science
construction of required input files. The second approach presents an automated method for the watershed with a GIs. In this study, they
application o f W P P through the extraction of hillslopes and channels fiom DEMs. The third used the GRASS GIS (U.S. Army 1987)
approach uses W P P model simulations on all possible flowpaths within a watershed. The three to obtain some of the physical parameters
method were applied to six research watersheds: one fiom Trqnor, la., two fiom Watkinsville, required by WEPP. However, they did
Ga., and three fiom Holb Springs, Miss. A statistical analysis for all methods and watershed not address the discretization of hill-
compared the predicted us. measured runoff and sediment yieM fiom watershed outlets on an slopes, channels, and representative slope
event- by-event basis for runoff and sediment loss. The results indicate that the automatic hillslope profiles from digital elevation models
method (the second approach) peforms as well as the manual technique (the first approach) for (DEMs).
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
all watershed. A comparison of erosion from onb hillslopesfor all three methods indicates that As illustrated in Figure 1, GIS maps
W
atershed erosion is a cause of
nonpoint source pollution that extension of the WEPP hillslope model irregular network (TIN). The most com-
can have an adverse effect on that can be used to estimate watershed mon type of DEM is grid-based, where
downstream and ecosystem water quality. runoff and sediment yield (Ascough et al. each grid point represents a cell of a
Unfortunately, it is very expensive and 1997). T h e application of WEPP to a certain size or resolution. Extraction of
impractical to monitor erosion in all watershed requires that hillslopes be features, such as hillslopes and channels,
watersheds of interest, hence, the need to
predict erosion with modeling.
The Water Erosion Prediction Project
(WEPP) model is a process-based contin-
uous simulation erosion model (Flanagan
'<I
12a. Manual method
Manual discretization of \
hillslopes, channels, and
sentative slope p r o f w
,)
and Nearing 1995). O n e advantage
WEPP has over existing models, such as
the popular Universal Soil Loss Equation
(USLE), is that deposition of sediment
I 1. GIS Maps I
.. ... -..... - . . .
G),yiEFGiq
also can be predicted (Wischmeier and
Smith 1978). In other words, soil loss on 4 -
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
integrating WEPP and GIs. These meth-
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
to the next, starting from a cell where no bottom of the hillslope. dently by applying WEPP to the actual
I
nel routing with unevenly distributed
29 Monona-Ida-Napier Corn
lateral flow to the channel. Since it was w2 la.)
(Treynor, 1985-90 series (Silt loam)
not the i n t e n t i o n o f t h e a u t h o r s to
change the WEPP code, these modifica- P1 1972-75 2.70 Cecil (Sandy loam Wheat, sorghum,
(Watkinsville, Ga.) and SCL) barley, soybean,
tions were not further pursued. T h i s clover
implies that a comparison of the flowpath
P2 1973-75 1.29 Cecil (Sandy loam Corn, bermuda
method to observed soil loss results from (Watkinsville, Ga.) and SCL) grass
the watershed outlets would only be
appropriate if no scouring or deposition WCl 1970-77 1.57 Grenada (Silt loam) Soybean, meadow
(Holly Springs, Miss.)
occurred in t h e channels. However,
erosion from hillslopes can be compared wc2 1970-77 0.59 Grenada (Silt loam) Corn, wheat,
(Holly Springs, Miss.) soybean
between all methods.
For visual assessments of water erosion wc3 1970-77 0.65 Grenada (Silt loam) Corn, wheat,
from small watersheds, detachment or (Holly Springs, Miss.) soybean
deposition values for each grid cell within
a watershed can be estimated by using the
flowpath method in a slightly different 2 - ~ - -----~ - ~-. -
--.--I __"
**/
manner. In this case, the WEPP hillslope 1 I/','
1.8
model is applied to each flowpath using
hillslope method
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
the flowpath's length, width, and slope
--
Application
0.2
The three methods were applied to six
research watersheds. The years of simula- 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
tion, watershed area, and soil and crop
types are presented i n Table 1 . T h e Observed sediment loss (tlhalevent)
WEPP data files for the Watkinsville and
Holly Springs watersheds are available via Figure 4. Total sediment loss from each watershed (normalized by watershed area and
number of events).
the Internet at: http://topsoil.nserl.pur-
due. edu/ wep p main/ we p p. h t ml . Addi-
tional information and measured data those conditions. Hence, default soil of these maps is believed to comply with
from the Watkinsville watersheds were parameters from the WEPP soils database U.S. Geological Survey (USGS) map
provided by Smith et al. (1978). Descrip- were used for the simulations, and appro- accuracy standards, which states that not
tions of the Holly Springs watersheds priate channel parameters were chosen to more that 10% of points tested shall be in
were provided by McGregor et al. (1969). represent the permanent grass waterways. error by more than one-half the contour
Validation of the WEPP model for these The three methods were applied using interval. The Treynor watershed was aeri-
watersheds was performed by Liu et al. W E P P Version 98.4. For these water- al surveyed with ground controls to create
(1997). sheds, each hillslope was represented by a DEM. Even though a smaller resolution
These watershed applications of WEPP o n e soil type a n d crop/management could be used, the quality of the data and
were used as templates for the simulations description. A 1-m resolution was used the size of the Treynor watershed suggest-
with the manual method. Management for the Watkinsville and Holly Springs ed that a 5-m grid was appropriate for
and climate data used for the Treynor watersheds, and a 5-m resolution was this modeling exercise.
watershed simulation was obtained from used for the Treynor watershed. DEMs
a previous application of WEPP to this from Watkinsville and Holly Springs were Analyses
watershed by Kramer (1993). However, created from field survey contour maps
the optimized soils and channel parame- (0- and 1.524-m contours, respectively,) All watershed simulations were con-
ters used by Kramer (1993) were found which were determined t o be precise ducted with WEPP running in a continu-
to be uncharacteristic for that region and enough for a 1-m grid. Vertical accuracy ous mode. T h e WEPP event-by-event
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
Figure 4 shows the results of a compar- yield normalized by number of events using the total runoff volume leaving
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
Treynor, W2 Total 59620 50880( 15) 54580(8) 60790(2)
* Means were tested using student t-test, Duncan’s, and Tukey’s comparisions of means, all of which gave the same decision results for hypothesis testing at
a = 0.05. The means of the observed and methods are not statistically different at the 95% confidence level unless otherwise marked.
t Observed and method means were statistically different at the 95% confidence level.
Table 4. Nash-Sutcliffe model efficiency coefficients for event sediment loss predicted hillslope method. It is apparent that the
by the three methods. three Holly Springs watersheds have
Nash-Sutcliffecoefficient (NS) channels that are in an advanced deposi-
tional mode, whereas, channels in the
Watersheds I Manual I Hillslope I Flowpath
Watkinsville watersheds may not be in
Watkinsville, P1 0.50 0.39 0.48 such a dramatic depositional mode.
Watkinsville, P2 0.67 0.87 0.74 Comparisons of hillslope sediment yield
HoUy Springs, WC1 0.84 0.54 -0.63 between the three methods suggest that
Holly Springs, WC2 0.79 0.79 0.60 they predict similar results; however, a
Holly Springs, WC3 0.23 0.42 0.13 visual observation of Figure 5 suggests
Treynor, W2 0.36 0.40 0.47 t h a t predictions from the flowpath
method are consistently lower than the
hillslope method. This may be due to the
large amount of computational weighting
involved in the flowpath method.
manual method
Sediment yield results are comparable
between the manual and hillslope meth-
ods because channels were modeled in
both methods. Figure 4 shows that both
manual and hillslope methods performed
similarly for total sediment loss in all
watersheds. The regression lines indicate
both a good fit and low variance for the
manual and hillslope methods.
The slope and intercept of the regres-
sion fit show that the hillslope method
predicts slightly higher values than the
manual method. T h e figure also shows
that for lower observed sediment losses,
both methods over-predict; whereas, for
P1 P2 wc 1 WCL wc3 w2 I large sediment losses, the methods under-
watershed
I predict observed values. From the NS
Figure 5. Comparison of hillslope sediment yield between methods (channels excluded). values and percent deviations of watershed
F O U R T H Q U A R T E R 1999 683
Table 5. Sediment yield and percent deviations for measured events.*
I Sediment yield in kg, (% Deviationfrom observed)
I Observed I Manual I Hillslope I Flowpath
Watkinsville, P1 Total 1 84490 137430(26) 161210(13) 100060(46)
36 measured events Mean 5125 3817 4478 2779
S.D. 10645 8779 11314 6676
Range 53-50000 0-41100 0-54300 0-32600
Watkinsville, P2 Total 28640 21690 (24) 28500 ( 4 ) 17050 (40)
48 measured events" Mean 597 452 594 355
S.D. 2051 1043 1688 1167
Range 0-12810 0-5860 0-10651 0-7717
Holly Springs, WC1 Total 104870 108200 (3) 217770(1 1 6) 306270(1 92)
207 measured events" Mean 507 523 1052 1480
S.D. 3803 2830 5192 7805
Range 3-53300 0-35600 0-59900 0-89700
Holly Springs, WC2 Total 40150 6261O(56) 64490 (60) 82930(1 06)
199 measured events** Mean 202 315 324 417
S.D. 1601 1415 1486 2132
Range 1-22152 0-16899 0-17820 0-25646
Holly Springs, WC3 Total 29520 43900(49) 37630 (27) 43770 (48)
187 measured events** Mean 158 235 201 234
S.D. 82 1 1133 929 1170
Range 1-7476 0-13672 0-10871 0-13679
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
Treynor, W2 Total 11 81150 1674140(42)
Means were tested using student t-test, Duncan's, and Tukey's comparisions of means, all of which gave the same decision results for hypothesis testing at
= 0.05. The means of the observed and methods are not statistically different at the 95% confidence level unless otherwise marked.
"Number of observed sediment yield events from watershed P2, WC1, WC2, and WC3 were less than measured runoff events for the same watersheds.
Means of hillslopesfor the three methods were tested using student t-test, Duncan's, and Tukey's comparisions of means, all of which gave the same decision
results for hypothesis testing at a = 0.05. The means were not statistically different at the 95% confidence level.
t Total values may be higher than in Table 5 because these include predicted values that were not measured.
WC 1, the manual method seemed to per- bly from problems in representing plant of severe erosion within a watershed.
form best; for watersheds P2, WC3, and and cover relationships. The channels of However, caution should be used when
W2, the hillslope method appeared to be these watersheds also acted as sediment interpreting quantitative results because
best. In other watersheds, both methods traps that, together with the over-predic- spatially-distributed results within water-
performed equally well. Ir is possible that tion of small events, would help to ex- sheds have not been validated (spatially-
the differences in the accuracy of the pre- plain large percent deviations of total sed- distributed erosion measurements for
diction5 for WCI can be attributed to its iment loss obtained. watersheds over time are not available or
coin pl icated topography and channel O n e of the biggest advantages of the are limited to small data sets). Applica-
structure, which the manual method flowpath method is that hillslope bound- tion of the flowpath method has other
modeled in greater detail. The other wa- aries need no delineation, which removes limitations, too. O n e is its computational
tersheds are either smaller, such as WC2 the burden of defining hillslopes. Another intensity. Due to the relatively high reso-
and WC3, or have less topographical or advantage is the ability to obtain spatially- lution of DEMS of all watersheds (1 m
channel structure complications. distributed results of erosion or deposi- for smaller watersheds and 5 m for the
The Holly Springs watersheds show an tion within a watershed, as Figure 6 shows large Treynor watershed), the number of
over-prediction of sediment yield (Table in the Treynor watershed (W2). These flowpaths in each watershed also is very
5 ) . This is consistent with observations results can be visually displayed o n an high. For example, the Watkinsville P2
made by Liu et al. (19!17), who suggest event-by-event or on a cumulative basis. watershed required 4,671 runs of the hill-
that this is due to a n over-prediction bias T h e visualization approach in t h e slope version of WEPP Other watersheds
for low-ma~nitudt.erosion events, proba- flowpath method can help identify areas required fewer runs, but the number of
units. Journal of Soil and Water Conservation
Detacirment (kgt/mA 5 1(5):427-433.
rn -6 - (-4) ESFU. 1998. Arc View 3.0a Users Manual. Environ-
m-3 - (-2)
-4 - 1-31
mental Systems Research Institute: Redlands CA.
Flanagan, D.C., and M.A. Nearing. 1995. USDA-
Water Erosion Prediction Project: Hillslope
-2 - (-1) profile and watershed model documentation.
NSERL Report No. 10. USDA-ARS National
-1 - 0 Soil Erosion Research Laboratory: West
l----10- 1 Lafayette, IN.
Garbrecht J., and L.W. Martz. 1997. TOPAZ: An
automated digital landscape analysis tool for
topographic evaluation, drainage identification,
watershed segmentation and subcatchment pa-
rameterization: Overview. ARS-NAWQL 95-1,
USDA-ARS: Durant, OK.
Garbrecht J., P.J. Starks, and L.W. Martz. 1996.
New digit a1 1an ds c a p e par am e t e r i z R t i o n
methodologies. In: Proceedings of the 32nd An-
nual Conference and Symposium on CIS and
Water Resources, TPS-96-3. American Water
Resources Association: Herndon, VA.
Jenson, S.K., and J.O. Dominique. 1988. Extracr-
ing topographic structure from digital elevation
Figure 6. Sediment detachment and deposition from flowpath application of WEPP on data for geographical information system analy-
Treynor watershed W2 over a 6-yr period. sis. Photogrammetric Engineering and Remote
Sensing 54(11):1593-1600.
Copyright © 1999 Soil and Water Conservation Society. All rights reserved.
flowpaths still were several thousand. watersheds requires development of an Kramer, L. 1993. Application of WEPP 93.005 to