Applied comparison of the erosion risk models EROSION 3D and LISEM for a small catchment in Norway

- We compared the performance of two physically based soil erosion prediction models.
- The result maps showed quite different erosion and deposition features.
- Changes in grid cell size and time resolution effect models differently.
- Similar models must not necessarily give similar results.
- The right choice of models for the right purpose is essential.

The loss of fertile soil from agricultural areas in Norway is especially harmful because of the thin layer of nutrient rich soil and the limited space where agriculture is possible. Physically based soil erosion prediction models have proved to be good tools to simulate and quantify soil erosion, but are not well established in Norway yet. Due to that this study was undertaken to further improve the knowledge about soil erosion development on agricultural areas and to better establish physically based models as an additional tool for soil research, in Norway. Two models were chosen for this study: the Limburg Soil Erosion Model (LISEM) and the EROSION 3D model. These two models were applied to the Skuterud catchment in the Ås municipality, for which measured discharge data, at the outlet, was available. The goal of this study was to investigate how the differences of two physically based models will influence the result of one and the same problem, to give an in-depth insight of what are the sources of uncertainty in modelling processes. To do that both models were calibrated by comparing the simulated hydrograph with the measured data. Special attention was given to the dependency of the model results on effects of grid cell size and time resolution. The grid cell size of the maps was easily adapted by using digital elevation models (DEM) obtained from airborne light detection and ranging (LIDAR) data. Furthermore the predicted erosion patterns were compared with an orthographic picture to validate the simulation results also in a spatial context.With both models, it was possible to simulate a satisfactory accurate hydrograph and total amount of surface discharge. However, the output maps produced by the models showed quite different erosion and deposition features.