LANDSLIDE SUSCEPTIBILITY MAPPING BY USING INFORMATION VALUE AND WEIGHT OF EVIDENCE METHODS IN THE SEGO CATCHMENT AREA, SOUTHERN ETHIOPIA

Abstract

This research investigates landslide susceptibility within the Sego catchment, a region situated in Ethiopia's main rift valley, approximately 24 kilometers from Arba Minch. The primary objective of this study is to assess landslide and susceptibility mapping in this region via the information value and weight of evidence methods. The factors were derived from various sources, including digital elevation modal data, field surveys, satellite imagery (Landsat 8 and Sentinel 2), and tropical rainfall mission machinery rainfall data. Two bivariant statistical model information values and weights of evidence were used to calculate the probability of landslide occurrence and determine the influence of each factor. A comprehensive field investigation and Google Earth analysis identified 125 landslide occurrences. A total of 94 landslides (75%) were used for model development, whereas the remaining 31 landslides (25%) served as a validation dataset. This study considered several factors contributing to landslides, including elevation, slope, aspect, curvature, land use land cover, proximity to streams, normalized difference in vegetation index, geology and rain fall. The resulting susceptible maps revealed varying degrees of influence from different factors. The landslide susceptible map, which is produced using information values, is categorized into five levels in the study area; very low, low, moderate, high and very high susceptible zones. The areal coverage is 13.5%, 24.7%, 26.5%, 23.8% and 11.5%, respectively and 103.62km2 of the study area falls in the very high and high susceptible zone. By employing the weight evidence method, the landslide susceptible map resulted in: very low (16.5%), low (27.8%), moderate (25.3%), high (17.3%) and very high (13.1%) and 88.88km2 falls in the very high and high susceptible zone. The validation results demonstrated that the weight of the evidence model outperformed than information value model, with area under the curve values of 80.8% and 75.4%, respectively. According to my study result; planting trees, constructing retaining walls, improving drainage systems and creating public awareness can reduce the occurrence as well as damage of landslides in the study area.