Soil erosion estimation based on rainfall disaggregation

Abstract in Undetermined
Soil loss estimation remains one of the most difficult research tasks all over the world. Current simulation tools are still not detailed enough to allow for realistic scenarios to handle soil erosion problems. A common problem is the lack of rainfall data at a sufficient level of detail. The present study uses a cascade disaggregation model to generate short time scale rainfall data, needed to calculate the erosivity index in erosion modeling. The model is used to determine the spatial soil loss rate by the Revised Universal Soil Loss Equation and a GIS approach. Comparison between observed and generated data in terms of erosive rainfall characteristics shows that the erosivity factor is over-estimated. This is caused by an overestimation of short rainfall events. Consequently, different duration limits beyond which erosive events could be considered within the generated series were used to estimate the model performance curve. This provided a suitable duration limit needed to reproduce the observed erosivity. The results showed that generated series only considering rainfall events superior than 90 min are appropriate. This procedure provides a soil loss rate less than 10% under-estimation. Moreover, using Masson, Wischmeier-Smith's and recent erosion limit intervals gave a realistic spatial erosion distribution. The results are promising and can be used to better manage erosion-prone soils.