Modelling soil erosion in tropical environments using remote sensing and geographical information systems

Abstract

Soil erosion probability maps were produced under various case scenarios by accounting for uncertainties in the data and in the decision rule, using the Universal Soil Loss Equation (USLE), remote sensing and geographical information systems (GIS). This objective was realized by applying the Bayesian Probability Theory within IDRISI, a raster based GIS. The outcomes were two continuous probability soil erosion maps ranging from zero to 1. Comparing these maps with an earlier study indicates that accounting for the uncertainties has, in general, decreased the probability of soil erosion. Based on average readings for specific sites on the maps, increases in erosion risk under the second case scenario have had the highest impact on the highlands that is in the central, eastern, and northern regions of Langkawi Island, Malaysia. Assuming a 10% risk, this impact has increased by 11.98, 11.83 and 5.741% for high, medium and low soil erosion risk areas on the island respectively.