Evaluation of plot runoff and erosion forecasts using the CSEP and medrush models

Three models have been applied to the GCTE common data sets. The simplest CEP-EROS model has been applied to five areas (US-C5, W12 and R5; Port-10 and 13) with substantial calibration, a modified version which includes subsurface flow (CEP-TOPER) has also been applied to five catchments (US-CS, W12 and R5; Canada Gu4 and Gu5) and the MEDRUSH model has been applied to US-RS only, with no calibration owing to time constraints. For the R5 catchment, a forecast has also been made for the distribution of erosion over the catchment. Results for the US site are broadly consistent with expectations, with lower erodibility in clays and for uncultivated areas. The Portuguese sites, however, show higher runoff from the natural vegetation (site 13), but dramatically lower erodibility. Lower runoff thresholds would not be consistent with the observed runoff rates, but would bring erodibilities closer to values which might be compared to the US sites. The results from MEDRUSH on the US-RS catchment have been obtained with prior estimation of parameter values, but no calibration. They show good agreement on the number of runoff events, but suggest that the overland flow velocities used are too low, with runoff events lasting too long.