HydroGIS 96: Application of Geographic Information Systems in and Water Resources Management (Proceedings of the Vienna Conference, April 1996). IAHSPubl. no. 235, 1996. 469

Application of WEPP and GIS on small watersheds in USA and Austria

M. R. SAVABI USDA-ARS, National Research Laboratory and Agricultural and Biological Engineering Department, Purdue University, West Lafayette, Indiana 47907, USA A. KLIK University of Renewable Resources, Vienna, Austria K. GRULICH Agricultural and Biological Engineering Department, Purdue University, West Lafayette, Indiana 47907, USA J. K. MITCHELL Agricultural Engineering Department, University of Illinois, Urbana, Illinois, USA M. A. NEARING USDA-ARS, National Research Laboratory, West Lafayette, Indiana 47907, USA

Abstract Geographic Information Systems (GIS) are well structured databases for handling large quantities of spatially varied data within a watershed. Coupling of a GIS with a spatially variable, physically-based, deterministic such as the USDA-Water Project (WEPP) offers many advantages. The WEPP model is a new computer program based on the fundamentals of hydrology, soil physics, plant science, hydraulics, and erosion mechanics. The WEPP model provides several major advantages over existing hydrological and erosion models; for example, it reflects the effects of soil surface conditions due to agricultural, range and practices on storm runoff and erosion. The Geographical Resources Analysis Support System (GRASS) GIS was used to obtain many of the needed input parameters for the WEPP computer model. The results indicate GRASS- GIS technology is a powerful tool and can be used to parameterize a complex hydrological model such as WEPP.

INTRODUCTION

Computer models that are process-oriented, physically-based, and therefore, mathemati­ cally mimic the spatial behaviour of a hillslope, watershed and/or basin's conditions are effective tools for examining the impact of various management decisions on water resources of a region. However, the amount of data and required parameters increase as the models become more complex. In recent years, GIS has become useful spatial data handling tools. Databases resulting from the combination of distributed information maps can support spatially distributed hydrological models. Several investigators have 470 M. R. Savabi et al. successfully integrated hydrological models with GIS (Mallants & Badji, 1991; Savabi et al., 1995). The common agreement is that GIS is a convenient and well structured database for handling large quantities of spatially varied data for a watershed and entire basin. The objective of this study is to explore the feasibility of using GIS to obtain some of the required parameters to test the WEPP (USDA, 1995) computer model for application under different crop conditions and geographic locations.

MATERIALS AND METHODS

WEPP model

The WEPP computer model can be divided into six conceptual components: climate generation, hydrology, plant growth, , management effect, and erosion. A brief description of each component is given here. The meteorological data required by the WEPP model can be generated, if not available, by a separate computer model called CLIGEN (USDA, 1995). Based on long-term statistics from historical climate data, the CLIGEN model generates daily values of required meteorological data for a station near the desired simulation location. Precipitation may be either in the form of rain or snow, depending upon the temperature. Rainfall is disaggregated into a time-rainfall intensity format for use by the infiltration and erosion components. If meteorological data for a location are available, the user can create the climate file using the observed climate and may also enter breakpoint precipitation information. The hydrology component includes simulation of storm runoff, snow melt, soil evaporation, plant transpiration, percolation, irrigation, and subsurface flow (Savabi, 1993). Excess rainfall is calculated as the difference between rainfall rate and infiltration rate. Infiltration rate is calculated using the Green and Ampt equation for unsteady rainfall. The effect of spatial and temporal variability of soil water intake is simulated by adjusting the effective hydraulic conductivity of the soil for human disturbances such as tillage, and natural phenomena such as soil surface sealing and vegetal roots. Excess rainfall is routed down slope to estimate the overland flow hydrograph using a kinematics wave approach (USDA, 1995). Peak runoff and runoff duration are used in calculating flow shear stress, transport capacity, and rill erosion. The WEPP winter component predicts frost and thaw layer development, snow accumulation and snow melt. The subsurface drainage component simulates water flow to subsurface tile drains and/or drainage ditches (Savabi, 1993). The plant growth model in WEPP assumes phenological crop development based on daily accumulated heat units, and a harvest index for partitioning grain yield. Many of the soil parameters used within the WEPP model change with time as a result of field operations, freezing, thawing, and weathering. The soil component simulates the temporal variability of soil properties such as bulk density, hydraulic conductivity, surface roughness, and erodibility parameters (USDA, 1995). The effect of various land management practices on hydrology and erosion for a site can be simulated with the WEPP model. The management component uses the data contained in the management input file to determine changes in soil physical properties and surface roughness and cover conditions due to practices such as tillage, crop harvest, grazing, and various residue management options (USDA, 1995). Application of WEPP and GIS on small watersheds in USA and Austria 471

The erosion component uses a steady-state sediment continuity equation as the basis for the erosion computations. Soil detachment in the interrill areas is calculated as a function of effective rainfall intensity and runoff rate. Soil detachment in the rills is predicted to occur if the flow hydraulic shear stress is greater than critical shear and the flow sediment load is below transport capacity. Deposition in the rills is computed when the sediment load is greater than the capacity of the flow to transport it (USDA, 1995).

Model parameters

The WEPP hillslope profile erosion model requires a minimum of four input data files: climate, soil, slope, and plant/management. Climate input files include daily precipitation amount, duration of storm, maximum intensity of storm, time to maximum storm intensity, maximum and minimum temperatures, solar radiation, wind speed and direction, and dew point temperature. Soil input files include such soil parameters as albedo, initial soil water content, soil textures, effective hydraulic conductivity, rock content, percent organic matter in the soil, and soil cation exchange capacity (CEC). The slope input file includes physical features such as slope length, slope steepness, and profile aspect. The plant/management file requires (agriculture, range, or forest) to be identified by users. For each land use, information about the specific plants present and management practices used are needed. For instance, for cropland, information about the crop plant growth (such as planting and harvest dates), type and dates of tillage, and type and dates of residue management are required.

Watershed descriptions

Austrian watersheds Three watersheds at Mistelbach, Kuffern and Feldbach were selected for this study. For the Mistelbach and Kuffern watersheds stereo photos with a scale of 1:15 000 were available. From these photos the topography was derived for a grid of 20 m distance. This grid then was used to derive the Digital Elevation Map (DEM) by using the computer driven SCOP method (Molnar, 1992). Because of the small scale of the aerial photos, the accuracy of the DEM is ca. 30 cm. The DEM for Feldbach is based on grid data (with a point distance of 50 m) from the Federal Surveyor's Office. Using the DEM, maps with contour lines, different slope classes and slope vectors were drawn. The maps of flow vectors were used to get the watershed boundaries, the boundaries of the sub watersheds and the lengths of the longest hillslopes of each subwatershed. The watersheds were represented by several hillslopes along the longest runoff pathway (Table 1, Fig. 1). Each hillslope was subdivided into Overland Flow Elements (OFE), depending on soil type and crops. The elaboration of all three watersheds was done by the Institute for Photogrammetry and Remote Sensing at the Technical University of Vienna. The Mistelbach and Kuffern watersheds have similar soil types that are deep, well drained and have a mollic epipedon with textures of loam and loam. Most of the soils have, according to the Austrian Soil Survey Reports (Oesterreichische Bodenkartierung, 1995), "moderate" hydraulic conductivity. In Mistelbach, eight different soil types can be classified, while in Kuffern and Feldbach there are only seven and five soil types, 472 M. R. Savabi et al.

Table 1 Size, slope, length, mean annual precipitation, and daily maximum precipitation.

Mistelbach Kuffern Feldbach Size (ha) 37.46 38.95 31.14 Average slope (%) 7.8 9.6 18.1 Max. length (m) 900 690 1009 Number of hillslopes 6 5 1 Number of OFE per hillslope 10 8 9 Mean annual precipitation (mm) 539 695 744 Daily max. precipitation (100 years, mm) 49 49 59 respectively. As soil input parameters for the WEPP program, all soil data were derived from the Austrian Soil Survey Reports (Oesterreichische Bodenkartierung, 1995). The erosion parameters for the watersheds were calculated with the WEPP recommended formulas. The hydraulic conductivity values were adjusted internally by WEPP to account for the effect of tillage, soil surface sealing and macroporosity (USDA, 1995). Plant parameters were taken from the WEPP data bank, and supplemented by using data from Brown (1988).

;:,\ ^\«^_"^\^-'

v / \ %

srway

jfigSmall grain (Winter) |gMnf• •, "" 1km ^Small grain (Spring) j—]A|fe|fa [ | Vineyard fallsunflower -^ Hillslope Boundry

Fig. 1 Land use (crops) in the Mistelbach watershed, Austria. Application of WEPP and GIS on small watersheds in USA and Austria 473

Four scenarios were simulated: (a) conventional tillage and management in the whole watershed; (b) conventional tillage and a 12 m filter strip at the bottom of each hillslope; (c) conservation tillage; and (d) conservation tillage and a 12 m filter strip. The filter strip was assumed to be planted grass. For land use, the actual situation during the summer 1994 in these regions was recorded and used for the simulation. For each watershed the land use was mapped (Fig. 1). At each site, storm events with 60 minutes duration and frequencies of 100 years were used. The single storm mode of the WEPP model was used for the simulations. The 100 year storm was assumed to occur in the middle of the growing season (July 15). Erosion and storm runoff was simulated for the storm events from each hillslope. The weighted average of soil loss and storm runoff from each hillslope within each watershed was calculated.

US watershed The Allerton watershed located southwest of Monticello, Illinois on the University of Illinois's Allerton farm was used for this study. Rainfall, runoff and sediment were monitored during 1950-1981. Watershed IB has an area of 18.2 ha with two nested watersheds 1B1 (13.4 ha) and 1B2 (10.2 ha). The soils are predominately Drummer silty- loam and Flanagan silt loam. These soils are moderately permeable, dark-coloured prairie soils (Fig. 2). The GRASS-GIS database of the Purdue University, Department of Agricultural and Biological Engineering (AGEN) was used to obtain WEPP required data and parameters (US Army, 1987; Mitchell et al., 1993). The grid resolution was 20 m. The average slope of the entire watershed is about 1.0% (Fig. 2). The pedon data of the Soil Interpretation Records (SIR) of Illinois and the National Soil Survey Laboratory (NSSL) were used to obtain the required soil parameters for the different soil types in the watershed area (Fig. 2). Meteorological data provided by the University of Illinois were

N A

!

• '. • ' G''' ' •• •: ' . " ris :J k • - • ?.-""' \ ' ' • • .{•<••• • i

' • .. . • i X • * •

• Gaging Station VX"" ' ..../'..Counter line

• Flanagan soil •Drummer soil CZH Sunburysoil

Fig. 2 Allerton watershed near Monticello, Illinois, USA. 474 M. R. Savabi et al. used to create the WEPP climate input file. To create the WEPP management input file, information was obtained from the University of Illinois Allerton farm database (crops, tillage activity and harvest dates of 1950-1981). A watershed basin analysis program applied to a DEM was used to obtain informa­ tion on the shape of the watershed, aspect (flow direction), slope, and water flow (accumulation) within the watershed boundary and on the main stream in the watershed (Fig. 2). As shown in Fig. 2, there is one distinct flow path within the watershed that can be represented by five hillslopes based on slope, flow direction, soil type and cropping. The areas that canbe represented by eachhillslope are 10.28,1.98,1.13,2.83, and 1.98 ha for hillslopes 1 through 5, respectively (Fig. 2). The lengths of the channels are 115 and 230 m for channels 1 and 2, respectively. The WEPP watershed computer model was used to predict the storm runoff of watershed IB, and the predicted storm runoff was compared with the observed values that were measured at the outlet of the watershed.

40 - • Conventional M Conv.+filter • Conservation H Conser.+filter mm ) 30

20 Runof f (

10 7"i 1 i ] i Stor m Mistelbach Watershed Kuffern Watershed Feldbach Watershed Fig. 3 Simulated storm runoff resulting from a rainfall of 60 minutes dur< 100 year frequency.

100

CO 80 c o 60 - • Conventional '(/) • Conv.+filter • Conservation o 40 • Conser.+filter LU

O 20 if)

Mistelbach Watershed Kuffern Watershed Feldbach Watershed Fig. 4 Simulated soil erosion resulting from a rainfall of 60 minutes duration and 100 year frequency. Application of WEPP and GIS on small watersheds in USA and Austria 475

RESULTS

Soil erosion and storm runoff resulting from heavy rainstorms of 100 year occurrence are shown in Figs 3 and 4. Although less than one percent of cropland in Austria is under conservation tillage, the results of this study indicate that a combination of conservation tillage and grass filter strip will result in significant reduction of storm runoff and soil loss from the selected watersheds. The steeper slopes of the Kuffern watershed compared to Mistelbach resulted in higher erosion in spite of the smaller (Figs 3 and 4). By adding a 12 m wide filter strip at the bottom of the slope, the eroded material leaving the watershed can be reduced significantly. This practice is more effective in Mistelbach than in Kuffern. Conservation tillage and a combination of conservation tillage and filter strips are equally effective in reducing erosion in both watersheds. Extremely high rates of erosion were predicted for the Feldbach watershed. The results are the values for the watershed as a whole; at individual locations the soil losses are substantially higher and with the use of GIS and the WEPP model the critical areas can be identified. Annual WEPP simulated and measured storm runoff are compared in Fig. 5. Although the discrepancy between model simulated and measured annual storm runoff is more than accepted for some years, the results indicate that the GRASS- GIS is a powerful tool to obtain parameters for the WEPP watershed model. If the digitizing effort is not too high compared to the expected results, coupling the GIS and WEPP model will assist in configuring the watershed and producing the soil and slope files for the WEPP model.

O T- CM CO xf m m m m to o> o> o> Year Fig. 5 Measured and WEPP predicted annual storm runoff for Allerton watershed near Monticello, Illinois, USA. 476 M. R. Savabi et al.

CONCLUSIONS

Spatially distributed data required by the WEPP model impose a major constraint on model application to a given environment. The first level of integration referred to as "ad hoc integration" was used in this study Tim & Jolly (1994). In this method, the model required data are extracted from GIS by running different GIS utility commands and the model is run separately from the GIS and model results are evaluated or compared with measured data independently of GIS. The results indicate that Digital Elevation Map (DEM) and GRASS-GIS technology are powerful tools and can be used to parameterize a complex hydrological model such as WEPP. However, we recommend that a complete integration of the WEPP model with GIS be developed. The integrated system can be readily used with the WEPP model for various applications and should make the WEPP model easy to use.

REFERENCES

Brown, R. H. (ed.) (1988) CRC Handbook of Engineering in Agriculture. CRC Press Inc., Boca Raton, Florida, USA. Mallants, D. & Badji, M. (1991) Integrating GIS and deterministic hydrological models: a powerful tool for impact assessment. Proc. EGIS1990, 672-679. Mitchell, J. K., Engel.B. A., Srinivasan,R. & Wang, S. S. (1993) Validation of AGNOPS for small watersheds using an integrated AGNOPS/GIS system. Wat. Resour. Bull. 29(5), 833-842. Molnar, L. (1992) Principles for a new edition of the digital elevation modeling system SCOP. Int. Archives of Photogrammetry and Remote Sensing, Commission IV, vol. 19, Part B4, 962-962. Washington, DC. Oesterreichische Bodenkartierung (1995) Soil Maps of the Mistelbach District, 1:25 000. Austrian Department of Agriculture, Vienna, Austria. Savabi, M. R. (1993) Modeling subsurface drainageand surface runoff with WEPP../. Irrig. and Drainage Engng ASCE 119(5), 801-813. Savabi, M. R., Flanagan, D. C, Hebel, B. & Engel, B. A. (1995) Application of WEPP and GIS to small watershed. J. Soil and Wat. Conserv. 50(5), 477-483. Tim, U. S. & Jolly, R. (1994) Evaluating agricultural nonpoint-source pollution using integrated Geographic Information Systems and hydrologic/waterquality model. J. Environ. Qual. 23, 25-35. US Army (1987) GRASS-GIS Software and Reference Manual. US Army Corps of Engineers, Construction Engineering Research Laboratory, Champaign, Illinois, USA. USDA (1995) Water erosion prediction project: hillslopeprofile and watershed model documentation. USDA-ARS, NSERL, Report no. 10 (ed. by D. C. Flanagan* M. A. Nearing). USDA-ARS, West Lafayette, Indiana, USA.