Improved calibration of Green-Ampt infiltration in the EROSION-2D/3D model using a rainfall-runoff experiment database

Abstract

Abstract. Soil infiltration is one of the key factors that has an influence on soil erosion caused by rainfall. Therefore, a well-represented infiltration process is a necessary precondition for successful soil erosion modelling. Complex natural conditions do not allow the full mathematical description of the infiltration process and additional calibration parameters are required. The Green-Ampt based infiltration module in the EROSION-2D/3D model is adjusted by calibration of the skinfactor parameter. Previous studies provide skinfactor values for several combinations of soil and vegetation conditions. However, their accuracies are questionable and estimating the skinfactors for other than the measured conditions yields significant uncertainties in the model results. This study presents new empirically based transfer functions for skinfactor estimation that significantly improve the accuracy of the infiltration module and thus the overall EROSION-2D/3D model performance. The transfer functions are based on a statistical analysis of the rainfall-runoff simulation database, which contains 273 experiments compiled by two independent working groups. Linear mixed effects models, with a manual backward elimination approach for the predictor selection, were applied to derive the transfer functions. Soil moisture and bulk density were identified as the most significant predictors explaining 79 % of the skinfactor variability, followed by the soil texture and the impact of previous rainfall events. The mean absolute percentage error of the skinfactor prediction was improved from 192 % using the currently available method, to 66 % using the presented transfer functions. Error propagation of the predicted skinfactors into the surface runoff and soil loss on the hypothetical slope showed significant improvement in the EROSION-2D/3D results. A first validation of real rainfall-runoff events indicates good model performance for events with a higher total precipitation and intensity.