A comparative spatial statistical approach for landslide susceptibility analysis in Getara and surrounding areas, South Wollo, Northern Ethiopia

The Getara area, located in the Northern Ethiopian highland, is characterized by frequent and active landslides that pose risk to villages, infrastructures, farmlands, and the natural ecosystem. Landslide susceptibility assessment was essential due to the absence of any existing documentation on landslides that could serve as a baseline for mitigation efforts and hazard management. The main objective of this study was to produce landslide susceptibility maps using three statistical methods (frequency ratio (FR), information value (IV), and Weight of evidence (WOE)) and compare their prediction accuracy. To do this, ten landslide conditioning factors: slope, aspect, curvature, geological material, lineament, stream, spring, stream power index (SPI), topographic wetness index (TWI), and land use land cover (LULC) were selected and a total of 480 landslides polygons (75% training and 25% validation) were prepared from google earth and field work. The numerical weight of each conditioning factor was assessed by the three statistical methods in ArcGIS 10.8. Based on this, geological material, lineament, stream, spring, LULC were found to have higher numerical weights in relation to contribution of conditioning factors to landslide occurrence in the study area. Landslide susceptibility maps were generated by raster summation of numerical weightage of all conditioning factor and classified in to three zones of low, medium and high susceptibility. To measure and compare the accuracy of the landslide susceptibility maps produced by the three methods, landslide pixel density (LPD) and Point landslide density index (R -index) validation techniques were used. Based on both validations, it has been observed that there was consistent high concentration of landslide pixel/points (high LPD and high R – index) in high susceptibility zones across all three susceptibility maps. This affirms reliability of all the methods applied and accuracy of produced landslide susceptibility maps. However, the IV method was found accurate and effective in differentiating low and high susceptibility zones relative to FR and WoE. Although the IV map shows slightly better accuracy, all FR, IV, and WoE susceptibility maps was found valuable for disaster risk reduction and land-use planning in the study area.