Abstract. Two versions of the SWAT-model with different dominating runoff generation processes have been applied. One version comprises the original available SWAT version where only the basic input data are used. In the second version SWAT has been modified, by the integration of an impermeable layer at the subbasin level, in order to better reflect the boundary between soil and bedrock that results in increased lateral flow in low mountain ranges. As well, since conventional German soil maps do not describe soil horizons beyond 2 m depth, we also added a 4 m fixed depth in the lowland areas in order to reflect the deep loess deposits in this region. The decision for the location of the impermeable and the additional loess layer was based on a GIS analysis of additional geologic information.

This study revealed that both model versions produced acceptable and comparable results regarding the evaluated goodness of fit measures. The GLUE analysis showed that the SWAT model set up with additional information about the distribution of impervious soil layers and the loess depth in the lowlands produced the highest simulation quality and the lowest uncertainty. Moreover, SWAT II version was able to better represent the spatial extend of the dominating runoff processes best. This leads to the conclusion that the SWAT II version is better suited for scenario analysis than the original model version.