Multi-Watershed Assessment of WEPP in the Tahoe Basin

In theory physically-based models rely on directly measurable parameters and observable processes. Hence, such models should be better suited for wide spread application to multiple watersheds than empirically-based or lumped models. In recent years, however, development and testing of physically-based models has focused on calibration techniques to improve Nash-Sutcliffe efficiency values, essentially converting the model to a complex empirical model. In this study we assess the ability of the Water Erosion Prediction Project (WEPP) model to simulate snowmelt, streamflow and sediment transport in five diverse watersheds in the Lake Tahoe basin using publicly available soils, climate, elevation, and canopy cover data. The only calibration performed in the study was the determination of the linear reservoir coefficient from recession curves of one of the stream gaging stations. Soil properties were taken from the Tahoe basin soil survey. Despite distinctly different soil types, which range from shallow rocky soils with flashy runoff response to deep alluvial soils dominated by subsurface lateral flow and baseflow, the agreement between simulated and observed streamflow over the 18 year record was very good on three of the five watersheds. Water yield and streamflow was over-predicted on two of the watersheds. In these watersheds, drought flow analysis of the observed data revealed that streamflow is largely controlled by long term lags indicative of large groundwater storage. Since this over-prediction largely occurred during low flow periods this error did not affect the predicted sediment load from these watersheds. Although stream channel erosion was not simulated, the timing and magnitude of sediment yield for each of the watersheds matched well with estimated upland erosion contributions in observed data. Simulated and observed snow water equivalent also matched very well at eight SNOTEL stations ranging from low elevation transient snowpacks to high elevation, deep snowpacks. The study reveals that although there is an important need for better methods to characterize the geohydrology of a watershed, the WEPP model proved to be a viable, robust tool that could be transferred without modification to multiple watersheds without any significant drop in accuracy.