Process Representation in Watershed-scale Hydrologic Models: an Evaluation in an Experimental Watershed

Hydrologic response varies within a watershed as a function of topography, soil, and land cover. Spatial and temporal data from experimental watersheds may provide information on where, when, how, and why the response varies. This study examined the hydrologic response of an agricultural watershed, FD-36, in the Appalachian Valley and Ridge physiographic region. FD-36 is characterized by shallow, fragipan soils in near-stream areas and deep, well-drained soils in upland areas. Three computer simulation models – Areal Nonpoint Source Watershed Environmental Response Simulation (ANSWERS-2000), Soil and Water Assessment Tool (AVSWAT2000), and Soil Moisture Distribution and Routing (SMDR) – were used to simulate the surface hydrologic processes in FD-36. The three models vary in their temporal and spatial process representations. AVSWAT2000 and SMDR are daily time-step models while ANSWERS2000 runs at a one-minute time step. Spatially, ANSWERS-2000 and SMDR divide the watershed into grid cells; AVSWAT2000 uses hydrologic response units (HRUs). Of the three models, temporal output from AVSWAT2000 matched measured stream flow most closely (r = 0.67). ANSWERS-2000 and AVSWAT2000 both reacted to variations in land cover and soils, whereas SMDR did not. ANSWERS-2000 and AVSWAT2000 indicated the majority of high runoff depths from croplands on near-stream, fragipan soils. Overall, AVSWAT2000 was determined the most favorable for depicting hydrological processes in FD-36, although spatial representation of runoff processes in this model may need further refinement. Veith and Srinivasan are agricultural engineers, and Gburek is a hydrologist, all at the U.S. Department of Agriculture, Agricultural Research Service, Pasture Systems and Watershed Management Research Unit, University Park, PA 16802.