Modelling Sediment Input to Surface Waters for German States with MEPhos: Methodology, Sensitivity and Uncertainty

Soil erosion on arable land and on steep vineyards is a major problem in the state of Hesse (21,115 km?) in central Germany. The aim of a joint study between the Research Centre Jülich, the Hessian Agency for the Environment and Geology and the Hessian Ministry for the Environment, Energy, Agriculture and Consumer Protection was to delineate parcels which are severely affected by erosion and to identify sediment source areas. For this purpose, the ABAG, an adaptation of the USLE approach to German conditions, has been employed with the best available data sets on K‐, C‐, R‐ and LS‐factor. Model results at the field scale show that soil losses in Hesse vary between <0.5 and >15 tonnes/hectare/year. The mean loss amounts to ca. 4.3 tonnes/hectare/year. The sediment delivery ratios for 450 sub‐catchments range between 0.5 and 78% with a mean of 18%. Further analysis showed that LS‐ and C‐factor are of highest sensitivity for the model output. Therefore, the effects of alternative algorithms or sources for LS‐ and C‐factor on the results were assessed. An uncertainty analysis based on Gaussian error propagation and Monte Carlo simulation showed that the uncertainty of model results induced by input parameters is 1.7 tonnes/hectare/year or 34% of the mean annual soil loss. The model results are a good basis for further works concerning a soil erosion atlas and internet‐based soil data viewer.