Using CCHE1D to improve stream flow routing capabilities of the WEPP model

Accurate assessment of forest management practices on sediment yield and delivery requires the use of physically based simulation models. Upland erosion rates are commonly simulated with the Water Erosion Prediction Project (WEPP) model. Since the WEPP was designed to evaluate erosion rates from fields and small watersheds without stream channels, it does not simulate hydrodynamics or sediment transport (especially important in watersheds larger than approximately one square mile). To resolve this problem, WEPP was linked via an interface program to CCHE1D, a one-dimensional, channel network, hydrodynamic-sediment transport model. The linked model was run with data from Caspar Creek Experimental Watershed, California, a long-term, U.S. Forest Service, experimental watershed. Calibration of the hydrodynamic and sediment transport capabilities of CCHE1D were completed separately and only the hydrodynamic results are presented here. A test simulation with 13 consecutive days of rainfall resulted in predicted runoff volume that was in very good agreement with observed runoff volume (less than 1% difference). However, due to limitations of the output data from the WEPP model, only daily, rectangular hydrographs could be generated. After routing the flows, daily peak discharge rates had an average relative error of 97% and a Nash-Sutcliffe coefficient of 0.41. These results are a marked improvement over the current method that WEPP uses for routing flows; with an average relative error of 1,110% and a Nash-Sutcliffe coefficient of -70.1. These results suggest that by using an appropriate hydrodynamic model to route flood-flows, the accuracy of peak flow estimates is increased by over two orders of magnitude.