Application of the WEPP model to some Austrian watersheds.

Abstract

Abstract This study aims to evaluate the performance of the hillslope and watershed version of the Water Erosion Prediction Project (WEPP) model by comparing measured with simulated amounts of runoff and soil loss/sediment yield. The predictive capability was investigated for two spatial scales: for natural runoff plots with an area of 60 m2 and for a large (0.66 km2) agricultural watershed, both located in the eastern part of Austria. As a first step, the basic model concepts and fundamental equations were discussed. In a second step, a one-dimensional sensitivity analysis was performed. This showed the high sensitivity of infiltration-related parameters in runoff and soil loss assessments. This behaviour is of particular importance because of the close relationship between these two parameters. Other crucial parameters included rainfall intensity and duration, soil surface roughness and slope steepness. The third step of this study was the application of the WEPP model to different spatial scales. The calibrated model was used to simulate runoff and soil loss from natural runoff plots operated since 1994 and 1997 at three sites in the eastern part of Austria. With the uncalibrated model, runoff and sediment yield from a large 66 ha agricultural watershed were calculated and compared with actual 5-year measurements. Judged by the Nash-Sutcliffe efficiency parameter and the Willmott's index of agreement the runoff predictions were more accurate when the model was calibrated. For the watershed with very low sediment yield data, WEPP simulated values were well correlated to measurements. A comparison between two growing seasons of simulated and measured soil profile water contents indicated that WEPP overestimated soil water contents in the spring and underestimated them in the summer. This can be attributed to the too high evapotranspiration rates calculated by the model. Uncalibrated use of soil erosion models is not advisable because the spatial variability of runoff and erosion processes within watersheds can strongly influence the simulation results.