Brief description

Temporal variability and spatial heterogeneity of surface runoff generation triggers the dynamic of source areas of sediment and sediment associated nutrients. Reliable modelling of hydrological exceptional situations i.e. snowmelt is of high importance for the quality of nonpoint source pollution modelling. Data from the research catchment "Schäfertal" demonstrate the individuality of snowmelt event in terms of runoff coefficient and delivery ratio. The winter erosion and nutrient load model system considers these dynamic aspects by coupling a hydrological model with a nutrient turnover model and a nutrient load model. Additionally, snowmelt erosion is simulated with a new developed model that is firstly applied on a catchment scale. A sensitivity analysis of this model part demonstrates the plausibility of the model approach and the overall robustness of the model system. The results of the long term hydrological modelling from 1991 to 2003 are reliable and form the base for the simulation of six snowmelt events which were observed in the "Schäfertal" catchment. The estimated total runoff volumes for these events match the observation well. The modelled overland runoff coefficients vary from 0.001 to 0.72. The significant spatial differentiation of the results is caused by a heterogeneity of soil moisture and the occurrence of soil frost. The mean values of cell erosion, which were modelled with one set of parameters for all six events range from 0.0006 to 0.96 t ha -1. The total modelled erosion for the events with unfrozen soil and low amount of snowmelt surface runoff is of a factor 50 below those with partly frozen soil. Beside these distinctions the major differences are caused by flow accumulation in shallow depressions in variable parts of the catchment. However, the validation of these results on the single event scale is restricted due to limited spatial data. Total simulated sediment yield at the catchment outlet can reach up to 13.84 t but underestimates the observed values with the exception of one event. Oversimplification of the modelled channel processes may be a reason. The temporal variability and spatial heterogeneity of the surface roughness parameter which was identified to be sensitive, also causes uncertainty in the parameter estimation. Despite these findings, the new model system was applied successfully on the catchment scale and the delivered results are reliable.