Comparison of Erosion and Runoff Predicted by WEPP and AGNPS Models Using a Geographic Information System

The Water Erosion Prediction Project (WEPP) model and the Agricultural Non-Point-Source Pollution Model (AGNPS) were used in conjunction with a geographic information system (GIS) database to predict runoff and sediment discharges for Rock Creek watershed, an agricultural watershed in Ohio, USA. Observed and predicted values were compared for selected storm events in 1988 and 1990. The statistical evaluation of the WEPP and AGNPS models showed that WEPP predicted average runoff, peak runoff and sediment yield better than AGNPS. WEPP and AGNPS overpredicted peak runoff rates compared to observed data by 15.5% and 26.5%, respectively. The t-test showed that there was no significant statistical difference between measured and predicted runoff and sediment data for both models (at ∝ = 0.05 level). The average root mean square error between observed and predicted average runoff was 11.5 m s and 14 m s for WEPP and AGNPS, respectively. While AGNPS underestimated average sediment discharge by 17%, WEPP overestimated average sediment load by 37%. With careful parameterization, the study demonstrated that WEPP and AGNPS could be used to simulate runoff and sediment in agricultural watersheds.