Predicting land use and soil controls on erosion and sediment redistribution in agricultural loess areas: model development and cross scale verification

This study quantifies soil and land use controls on sediment mobilisation and redistribution in cultivated loess soil landscapes, as these landscapes are frequently used for intensive cultivation and are highly susceptible to erosion. To this end we developed and verified a process based model named CATFLOW-SED at the plot, hillslope and catchment scales. The model relies on an explicit representation of hillslopes and their dominant physiographical characteristics which control overland flow formation, particle detachment and sediment redistribution (transport and sedimentation). Erosion processes are represented by means of the steady state approximation of the sediment continuity equation, their interaction is conceptualized based on the sediment transport capacity of overland flow. Particle detachment is represented by means of a threshold approach accounting for the attacking forces of rainfall and overland flow which need to exceed a threshold in soil erosion resistance to mobilize soil particles (Scherer et al., 2012). Transport capacity of overland flow is represented as proposed by Engelund and Hansen (1967). Top soil particles and aggregates are detached and transported according to their share in the particle size distribution. Size selective deposition of soil particles is determined based on the sink velocity of the various particle size classes. CATFLOW-SED was verified on the plot, hillslope and catchment scale, where either particle detachment or lateral redistribution or sedimentation is the limiting factor, to check whether the respective parameterizations are transferable for simulations at the next higher scale. For verification we used the Weiherbach data set providing plot scale rainfall simulation experiments, long term monitoring of sediment yields on a selected hillslope as well as observed sediment fluxes at the catchment outlet. Our findings corroborate that CATFLOW-SED predicted the sediment loads at all scales within the error margin of the measurements. An accurate prediction of overland flow turned out as being necessary and sufficient to guarantee spatial transferability of erosion parameters optimized at smaller scales to the next higher scale without need for further calibration. Based on the verified model setup, we investigate the efficiency of land use management to mitigate measures in erosion scenarios for cultivated loess landscapes