Integration of geospatial analysis, frequency ratio, and analytical hierarchy process for landslide susceptibility assessment in the maze catchment, omo valley, southern Ethiopia

Abstract

This investigation was conducted in southern Ethiopia's Maze watershed in the Omo River Valley. Frequency ratio (FR) and analytic hierarchy process (AHP) techniques were used to assess landslide susceptibility in the region. Identifying causative components and landslide inventory data achieved the goal. Remote sensing and on-site investigations found 793 landslide polygons. To assess vulnerability, the landslide inventory information is categorized into two groups: the training dataset (70%) and the validation dataset (30%). This study examined “slope, aspect, curvature, lithology, land use and cover, normalized vegetation index, and proximity to fault lines, rivers, and distance to road as landslide controlling factors”. The spatial analysis capabilities in Arc GIS were used to overlay the weights of all landslide-causing components to create the susceptibility map. A final landslide susceptibility map is produced using FR and AHP methods and categorized as “very low,” “low,” “moderate,” “high,” and “very high.” The frequency ratio method divides the region into susceptibility classes by frequency. The very low, low, medium, high, and very high susceptibility groups cover 25%, 20%, 18%, and 19% of the territory. The analytical hierarchical process technique shows that 3%, 7%, 26%, 36%, and 28% of the area are very low, low, medium, moderate, and very high landslide susceptibility. The receiver operating characteristic curve was employed to validate the area-underlayer susceptibility maps. The success rates were determined using the FR and AHP approaches, resulting in AUC numbers of 0.873 and 0.87. Similarly, the prediction rates were determined to be 0.81 and 0.80. The landslide susceptibility maps will significantly influence land resource allocation.