Spatially distributed data for erosion model calibration and validation: The Ganspoel and Kinderveld datasets

Recent experience in distributed modelling has revealed that the performance of process-based erosion and hydrological models are extremely sensitive to parameter estimations and that predictions are often poor. It is also observed that quite different parameter sets may lead to very similar results and that no 'best' parameter set can be identified. In this study, we describe a dataset that offers possibilities for improved evaluation and parameterisation of spatially distributed soil erosion models. The dataset combines rainfall, runoff and sediment discharge data collected at the outlet and field surveys within the catchments that describe soil surface characteristics and soil erosion features. This offers clear advantages over traditional model evaluation as not only the simulated overall system response, integrated over time and space, but also the simulated internal system dynamics can be compared with measured data. The paper discusses and illustrates the use of the dataset to narrow uncertainties associated with model predictions.