Landslide susceptibility mapping using geospatial technology in the case of the Gidabo watershed, Main Ethiopian Rift

Abstract

Landslides are natural hazards that cause significant injury and loss of life. This study aims to generate a landslide susceptibility zonation map for the Gidabo watershed using geospatial technology, including geographic information system (GIS) and remote sensing data. To delineate the landslide susceptibility of the study area, ten main factors -slope, rainfall, distance from rivers, elevation, lineament density, geology, land use/land cover (LULC), soil type, drainage density, and normalized difference vegetation index (NDVI)-were integrated in ArcGIS. The weight values of each factor that causes a landslide were determined and assigned using the Analytical Hierarchy Process. The study area's landslide susceptibility map (LSM) was created using an overlay weighted sum approach and is divided into four classes: very low (22.5 %), low (32.1 %), moderate (25.2 %) and high (20.3 %) susceptibility classes. The verification and validation of the LSM were also carried out using the field verification and the area under the curve (AUC). 76 % of landslide points belong to the high susceptibility class. The overall accuracy of the method is 84.3 %, showing very good accuracy. The findings demonstrate that the slope is a significant factor influencing landslide susceptibility, with the susceptibility pattern closely following the distribution of steep slopes. The result also shows that areas such as Teferi Kela, Abera, the eastern part of Wenago, Guanguwa, and Dilla are highly prone to landslides. The LSM of this study can be used as a tool for landslide mitigation, land use management, and planning, helping to identify and prioritize areas at risk.