A Study on the Effects of Sensitive Parameters Errors on Digital Soil Erosion Simulation

Abstract

A combined runoff-sediment model is used to simulated soil erosion processes and predict soil loss of land surface. The Stanford Watershed Model (SWM) is used as runoff generator for this model. Effects of sensitive parameters errors on soil erosion characteristics are investigated. These characteristics include peak erosion and mean monthly erosion rates. Three different methods are used for the analysis, namely; first-order uncertainty analysis method; direct investigation technique and mean-maximum likelihood method. The aim is to quantify sensitive parameters errors propagation and to gain an appreciation of the approximate magnitudes of model output uncertainty caused by different levels of sensitive parameters uncertainty. Model output uncertainty ranges between (4.000-83.115)% for mean monthly erosion against (4.680-83.098)% for peak erosion. Uncertainty in simulated erosion due to sensitive parameters uncertainty is subsequently analyzed. The probability of peak erosion values occurrence due to sensitive parameters error are investigation. Based on the result obtain, high and moderately parameters are identified. Appropriate conclusion are drawn and suggestion for future work are introduced.