Empirical water erosion modelling for soil and water conservation planning at catchment-scale

Empirical water erosion models may simplify the erosion processes, but are straightforward, require few input data and can be useful in pinpointing areas exposed to high erosion risk. For soil and water conservation planning, the capacity of the empirical models to locate the erosion patterns correctly may be more useful than the ability to accurately quantify erosion losses. The aim of this study was testing two empirical water erosion models, the Universal Soil Loss Equation (USLE) and the Morgan, Morgan and Finney model (MMF), in predicting spatial erosion patterns in a small (310 ha) catchment at Alora, southern Spain. The available data set comprised climatic data, a topographic map, a soil map and a land use map. Both models were incorporated in the ILWIS (Integrated Land and Water Information System) GIS. The predicted spatial patterns of average annual erosion were compared with the results of an erosion survey of the same catchment. The erosion losses predicted by both models seemed wrong: the USLE resulted in high erosion rates (190 Mg ha(-1) y(-1) on average), whereas the MMF predictions were low (6.3 Mg ha(-1) y(-1) on average). The erosion spatial pattern predicted by the USLE showed poor agreement with the observed pattern (error occurrence 69%). The MMF erosion pattern, however, showed better agreement with the observed erosion pattern (error occurrence 57%). The MMF model performed better when erosion was splash detachment limited (error occurrence 44%) than when it was transport limited (error occurrence 77%). It is concluded that the MMF model is the better empirical water erosion model to predict spatial erosion patterns at catchment-scale.