Is an expert-based model able to map erosion in an agricultural catchment?

Intensive agricultural practices on sensitive soils induce high erosion rates in central Belgium. Expert-rules models quantify runoff and erosion at watershed scale avoiding over-parameterization. The aim of this study is to test the ability of an expert-based model, LandSoil, to quantify runoff and to locate erosion and deposition areas in a small agricultural watershed in Belgium. For that purpose, Chastre catchment behavior is modeled using LandSoil during 3 years. Measurements of runoff and observation of spatial erosion/deposition patterns permit to assess the reliability of the model. Runoff modeling gives satisfactory results with good linear adjustments (r² of 0.94, Nash-Sutcliffe criterion of 0.92). However, 3 events tend to greatly overestimate runoff production. Graduated rulers and runoff samplings demonstrate that the model is able to provide a coherent pattern of erosion/deposition. The study highlights more sensitive effect of land use compared to landscape design. Grass strips induce a deposition of eroded particles when slopes are gentle (<2%), wood strip decreases connectivity by cutting erosion stream and deposit thicker sediment layers (up to 5 cm). This modeling validation in the Belgian loess context allows us to use expert based models in other similar environments and to estimate the effect of landscape management scenarios.