Distributed watershed models, such as the Soil and Water Assessment Tool (SWAT), have proven to be an important means for tackling problems related to nonpoint source pollution. These models require soil data as one of the minimal inputs because soil controls the runoff mechanism of a watershed. The U.S. Department of Agriculture Natural Resources Conservation Service generated two national soil databases, namely the State Soil Geographic (STATSGO) and Soil Survey Geographic (SSURGO) databases, which were designed to be used for river basin resource monitoring, planning, and management. STATSGO has a county level spatial resolution, whereas, SSURGO has a farm level resolution. These two databases have been widely used as the best alternatives for watersheds where site specific soil data are not available, which is a common case. However, in the literature, information is scarce regarding affects of using one of these two databases over another on modeling sediment processes. Thus, the objective of this study was to evaluate affects of using STATSGO versus SSURGO as an input on SWAT simulated sediment yields at the outlet of, and sediment source areas within, a watershed. The evaluation was conducted in one North Dakota watershed and one Texas watershed. Obviously, these two watersheds have distinctly different climatic and hydrologic conditions. The results indicated that for each of the study watersheds, the predicted total sediment yields at the outlet using STATSGO might be comparable with the corresponding predicted values using SSURGO, the predicted source areas, however, could be very different. In addition, the prediction discrepancies as a result of using one dataset over another were larger in the Texas watershed. This might indicate that the simulation of a watershed with streamflows mainly generated from rainfall runoff is likely to be more sensitive to soil data resolution than the simulation of a watershed with streamflows predominantly generated by snowmelt runoff.
