Predicting soil erosion and sediment yield at regional scales: Where do we stand?

Assessments of the implications of soil erosion require quantification of soil erosion rates (SE) and sediment yield (SSY) at regional scales under present and future climate and land use scenarios. A range of models is available to predict SE and SSY, but a critical evaluation of these models is lacking. Here, we evaluate 14 models based on 32 published studies and over 700 selected catchments. Evaluation criteria include: (1) prediction accuracy, (2) knowledge gain on dominant soil erosion processes, (3) data and calibration requirements, and (4) applicability in global change scenario studies. Results indicate that modelling of SE and SSY strongly depends on the spatial and temporal scales considered. In large catchments (>10,000km2), most accurate predictions of suspended sediment yield are obtained by nonlinear regression models like BQART, WBMsed, or Pelletier's model. For medium-sized catchments, best results are obtained by factorial scoring models like PSIAC, FSM and SSY Index, which also support identification of dominant erosion processes. Most other models (e.g., WATEM-SEDEM, AGNPS, LISEM, PESERA, and SWAT) represent only a selection of erosion and sediment transport processes. Consequently, these models only provide reliable results where the considered processes are indeed dominant. Identification of sediment sources and sinks requires spatially distributed models, which, on average, have lower model accuracy and require more input data and calibration efforts than spatially lumped models. Of these models, most accurate predictions with least data requirements were provided by SPADS and WATEM-SEDEM. Priorities for model development include: (1) simulation of point sources of sediment, (2) balancing model complexity and the quality of input data, (3) simulation of the impact of soil and water conservation measures, and (4) incorporation of dynamic land use and climate scenarios. Prediction of the impact of global change on SE and SSY in medium sized catchments is one of the main challenges in future model development. No single model fulfils all modelling objectives; a further integration of field observations and different model concepts is needed to obtain better contemporary and future predictions of SE and SSY. © 2013 Elsevier B.V