Landslide Susceptibility Analysis in the Ramban Basin, Jammu and Kashmir: A Statistical Approach for Comprehensive Susceptibility Assessment

Abstract

The Ramban district, located in the northern Himalayas inside the Union Territory of Jammu and Kashmir, is well-known for its increased vulnerability to landslides. This region consistently faces a repetitive occurrence of landslides that result in loss of life and cause significant harm to both cultivated and non-cultivated regions, essential infrastructure, and properties. Given the need to reduce these negative effects, the creation of a landslide susceptibility map is seen as a vital approach. This study systematically utilized a Geographic Information System (GIS)-based Information Value Model (IVM) to methodically gather a thorough inventory of 796 landslides. The landslide inventory was divided into training datasets (70%) for model prediction and testing datasets (30%) for reliable model validation. The analysis encompassed thirteen contributing elements to landslides, namely altitude, slope, aspect, curvature, distance to drainage, distance to structural lineaments, geomorphon, land use land cover, lithology, relative relief, stream power index, and topographic wetness index. The IVM approach was subsequently utilized to determine the weight of each element and factor class based on their correlation with training landslides. The combination of these efforts led to the creation of a map that shows the likelihood of landslides occurring. This was accomplished by combining the weights of all factors contributing to landslides using the raster calculator in GIS. The IVM's ability to predict landslide susceptibility was thoroughly evaluated by calculating the area under the curve (AUC) of the receiver operating characteristic (ROC) curves for both the training and testing datasets. The model demonstrated a notable AUC accuracy of 71.4% for the success rate and 70.5% for the prediction rate. Surprisingly, when the Landslide sensitivity Map (LSM) was superimposed, 91.3% of the landslide pixel area was categorized as having a very high or high susceptibility to landslides. The results reveal important insights, identifying places with a very high level of vulnerability. Specifically, areas with extremely high susceptibility and high susceptibility make up 13.9% and 23.5% of the total area, respectively. Areas with moderate susceptibility cover 29.1% of the entire territory, while areas with low and extremely low susceptibility together account for 33.5% of the total area. Landslide susceptibility mapping improves our comprehension of the probability of slope instability and serves as a vital tool for making well-informed decisions in land use planning and methods to reduce risk. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.