Context and objectivesArtificial neural networks (ANN) are used with success to solve classification problems wherein 1) - the underlying statistical distributions of the output data are unknown and 2) - a considerable amount of data is to be processed. The study aims to evaluate the performance of ANN for the estimation of land degradation as a function of factors of influence, such as population density, soil type, transport infrastructure.
Expected scientific results
A crust risk model and map was produced via ANN. The accuracy was thematically and spatially higher than the results obtained through the simple linear regression method.