GIS-Based Landslide Susceptibility Mapping Using Frequency Ratio and Shannon Entropy Models in Dejen District, Northwestern Ethiopia

Abstract

A GIS-based study has been carried out to map areas landslide susceptibility using both frequency ratio (FR) and Shannon entropy (SE) bivariate statistical models. A total of 270 landslides were identified and classified randomly into training landslides datasets (70%) and the remaining (30%) of landslides datasets were used for validation purpose. The 11 landslides conditioning factors like slope, elevation, aspect, curvature, topographic wetness index, normalized difference vegetation index, distance from road, distance from river, distance from faults, land use, and rainfall were integrated with training landslides to determine the weights of each landslide conditioning factor and factor classes using both frequency ratio and Shannon entropy models. The landslide susceptibility maps were produced by overlay the weights of all the landslide conditioning factors using raster calculator of the spatial analyst tool in ArcGIS 10.4. The final landslide susceptibility maps were reclassified as very low, low, moderate, high, and very high susceptibility classes both FR and SE models. This susceptibility maps were validated using landslide area under the curve (AUC). The results of AUC accuracy models showed that the success rates of the FR and SE models were 0.761 and 0.822, while the prediction rates were 0.753 and 0.826, respectively.