Evaluation of Streamflow Estimates Using Different Soil and Land Use Geospatial Datasets.
Gary Heathman and Myriam Larose. USDA-ARS, PO Box 3939, West Lafayette, IN 47996-3939
The integration of geographical information systems (GIS) and hydrologic models provides the user the ability to simulate watershed scale processes within a spatially digitized computer based environment. Such model simulations have become increasingly popular within the scientific community for several reasons, most notably: 1) the increased availability of input data sets via the internet, 2) modeling is usually more cost effective and less labor intensive than conducting field research and, 3) the use of spatially distributed data rather than point-scale measurements. Fundamental to optimal model performance is the quality, consistency, and structure of the spatial data sets used as model input. In particular, soil type and land use information are essential data elements in hydrologic modeling. This investigation was conducted to evaluate the use of SSURGO and STATSGO soil classification schemes and the National Agricultural Statistics Survey (NASS) and GAP Analysis Project (GAP) land use data sets in the watershed scale model referred to as the Soil and Water Assessment Tool (SWAT). Performance of the model was tested on the Cedar Creek Experimental Watershed (CCEW) in northeastern Indiana, one of twelve benchmark watersheds in the U.S. Department of Agriculture, Agricultural Research Service (USDA ARS) national Conservation Effects Assessment Project (CEAP). Results show that for the seven year streamflow record a combination of the STATSGO and NASS data sets gave the lowest RMSE (0.7067) and highest ENS value (0.7245). However, on a monthly basis, a combination of the GAP and STATSGO data sets resulted in the lowest RMSE (3.912) and highest ENS value (0.775). Results of this study indicate that the estimation of streamflow by the SWAT model is most sensitive to land cover input datasets.