A Comparison of Feature and Expert-based Weighting Algorithms in Landslide Susceptibility Mapping

Determination of factor weights for landslide susceptibility mapping problem should be performed by some intelligent approaches instead of personal choices when a large number of factors are available. In this study, the quality of factors and their effects on the production of landslide susceptibility maps were assessed using Chi-square and Fisher weighting methods. Process of factor weight determination was automatized employing feature weighting algorithm with user-based Analytical Hierarchy Process (AHP) approach. In order to produce the most accurate and precise susceptibility maps, factors were integrated into the GIS environment using the factor-weighted overlay method. In this study, Arakli district of Trabzon Province, Turkey is considered as the study area. The primary focus in this study is to determine the weights of landslide causative factors using Chi-square and Fisher algorithms. On the other hand, AHP method was used as a benchmark method to compare and validate the performances of the landslide factor weights. All weighted factor sets were tested on factor-weighted overlay method for producing landslide susceptibility maps. The quality of susceptibility maps was assessed using overall accuracy measure and success rate curve analysis. Results showed that the weights determined by Chi-square and Fisher methods outperformed the conventional AHP method by about 6%.