Landslide susceptibility mapping using GIS-based statistical models and Remote Sensing in the Kathmandu valley, Nepal

Abstract

Landslides are geological hazards that typically occur on both temporal and spatial scales in different settings, causing significant loss of life and property. The occurrences of landslides in Nepal are increasing in recent years claiming loss of lives and properties. The coupling effect of Asian monsoon and seismo-tectonic activities along with anthropogenic activities are the major cause for the landslide generation in the Nepal Himalaya. A systematic landslide research is essential to prevent or control the issues generated by landslides, including widespread damage of buildings and structures, property, cultivated areas, and loss of life. This study aims to perform a GIS-based landslide susceptibility mapping of the Kathmandu valley using bivariate statistical approaches (Frequency Ratio and Information Value) and heuristic approach. The landslide inventory data base was prepared from 2010 to 2021 using Google Earth Pro where 105 landslides were identified. Predisposing factors were categorized into different classes (Aspect, Slope, Geology, Curvature, Landuse, Distance to road, Distance to drainage, Rainfall, NDVI, and Relative relief) for the suitability mapping. The landslide susceptibility classes for all three methods were divided into three classes as low, medium, and high. Furthermore, the Frequency ratio (FR) and Information Value (IV) methods were validated through Area Under Curve (AUC) approach. The results indicate that the FR and IV approaches have predictive rates of 70.16% and 81.43%, respectively. This study is useful for geohazard assessment for infrastructure planning and land use zoning.