GIS-Based Landslide Susceptibility Assessment Using a Statistical Approach: A Case Study of Srinagar-Bandipora Highway, Kashmir Himalaya India

This study aimed to generate a landslide susceptibility zonation (LSZ) map along the Srinagar-Bandipora highway in North Kashmir, India. The relative effect (RE), a bivariate statistical method, was employed to determine the relationship between landslide causative factors (CFs) and landslide occurrence. For this purpose, a landslide inventory map database was prepared from primary and secondary sources. During a comprehensive fieldwork, a detailed landslide inventory was generated. Consequently, the generated landslide inventory was divided into two datasets: Training samples (70%) and Validation samples (30%). After examining the correlation between landslides and ten potential CFs in a GIS, a final LSZ map was created. The resulting landslide susceptibility map recognized four vulnerable zones viz: Very High (21.08 km2), High (35.41 km2), Moderate (51.73 km2), Low (93.81 km2), and Very Low (53.61 km2). In addition, the ROC approach was utilized to validate the LSM. The prediction rate of REM-generated LSM shows an area under curve value of 92.80%. Thus, the RE method exhibits good predictive performance, resulting in a reliable landslide susceptibility map. As a result, LSM agrees with the historical landslide location data. According to the RE method, nearly 22.10% of the study area falls into high to very-high zones. Furthermore, landslides are significantly impacted by slope, lithology, land use/cover, proximity to roadways, lineaments, and drainage. Moreover, engineers and land use planners will be able to utilize the final landslide susceptibility zonation (LSZ) map as a foundation to create landslide control measures before construction on slopes in the study region.