Geospatial Modeling of Landslide Susceptibility Using Statistical Index and Frequency Ratio Methods in the West Seti Watershed, Nepal

Abstract

The Hilly and Himalayan region of Nepal is highly sensitive to landslides and debris flows. Many people lose their lives and settlements every year during the rainy season due to landslides in the Nepalese Himalaya. This study aims to assess the landslide susceptibility in the West Seti watershed of Sudurpashchim Province, Nepal. The landslide susceptibility map was prepared by using frequency ratio and statistical index models. The work began with the preparation of a landslide inventory map by using Google Earth and Sentinel-2 imagery. The old landslides were used as training data (70%), and recent landslides that occurred post-2018 were mapped for testing data (30%). A total of,029 old and new landslide locations were identified in the study area. The eleven conditioning factors of topographical (slope, elevation, plan curvature, profile curvature, aspect, topographical wetness index), geological (geological map, hydrological (rainfall, distance to drainage, stream power index), and (l) land use/land cover, distance to road) weresed for landslide susceptibility analysis after conducting a multicollinearity test. The area under the curve method was used to validate model performance. The landslide susceptibility index was divided into five susceptibility classes, namely very high, high, medium, low, and very low. The very high to very low susceptibility percentages obtained from the model were 9%, 21%, 23%, 27%, and 10%. Similarly, the very high to very low class susceptibility obtained from the SI model were 13%, 24%, 20%, 25%, and 17%. The results indicate that the success and prediction for the FR model were 62.7% and 55.4%, while the success and prediction rate for the SI model were 74.7% and 71.2%. The SI model provided better success and prediction rates in comparison to the FR model. These findings are useful for the local authorities in the study area for disaster risk management planning and safer land-use planning.