Evaluating a spatially-explicit and stream power-driven erosion and sediment deposition model in Northern Vietnam

Land use change and unsustainable farm management practises have led to increased soil erosion with severe consequences on the natural resource base in mountainous Northern Vietnam. Given the often prevailing data-limited situations in these regions, simulation models can be used to evaluate alternative land use trajectories or provide decision support for soil conservation planning. In this study, we present a newly developed dynamic and spatially-explicit EROsion and sediment DEPosition model (ERODEP), which simulates soil erosion by stream power principles, sediment deposition based on texture-specific settling velocity classes, and sediment re-entrainment to move previously deposited particles back into runoff flow. ERODEP runs on a daily basis and was linked with the Land Use Change Impact Assessment model (LUCIA) building on its hydrological and vegetation growth routines. The combined modelling framework was employed for a period of four years using field datasets of a small case study watershed. ERODEP-LUCIA simulated reasonably well soil erosion and sediment deposition patterns following the annual variations in land use and rainfall regimes. Output validation (i.e. modelling efficiency = EF) revealed satisfying to good simulation results, i.e. plot-scale soil loss under upland swiddening (EF: 0.60–0.86) and sediment delivery rates in monitored streamflow (EF: 0.44–0.93). Cumulative sediment deposition patterns in lowland paddy fields were simulated fairly well (EF: 0.66), but showed limitations in adequately predicting silt fractions along a spatial gradient in a lowland monitoring site. Findings of a sensitivity analysis demonstrated the interplay of soil erosion and sediment deposition by superimposed variations in stream power, sediment velocity and vegetation related parameters. Results highlighted the potential of ERODEP-LUCIA as an integrated biophysical assessment tool for mountainous ecosystems with moderate data availability.