LANDSLIDE SUSCEPTIBILITY MAPPING USING GIS-BASED FREQUENCY RATIO, SHANNON ENTROPY, INFORMATION VALUE AND WEIGHT-OF-EVIDENCE APPROACHES IN PART OF KULLU DISTRICT, HIMACHAL PRADESH, INDIA

Abstract

Landslides are serious geological hazards that cause significant damage and casualties in India, creating a high need to identify landslide-prone areas and their associated causative factors, for efficient risk reduction strategies by authorities. This study aims at evaluating the effectiveness of four GIS-based statistical approaches namely, Frequency ratio (FR), Shannon Entropy (SE), Information Value (IV) and Weight-of-Evidence (WofE) for the landslide susceptibility mapping of a region in Kullu district, situated in the state of Himachal Pradesh, where a high surge in tourism and development since the past decade has been witnessed. The causative factors considered as input in this study are slope, aspect, curvature, lithology, distance to roads, distance to faults/lineaments, distance to drainage, land use/land cover and elevation. Since the existing landslide inventory maps from Geological Survey of India and past literatures do not cover the whole study area, an updated inventory has been prepared from visual interpretation of Google Earth Images (2001-2019) and use of a Scarp Identification and Contour Connection method (SICCM) ArcGIS toolbox. The compiled landslide inventory data was randomly divided into training (70%) and validation (30%) datasets. The correlation between past landslide locations and each landslide-influencing parameter has been carefully evaluated using the statistical models. Four landslide susceptibility maps resulted from this research work which were then validated and compared using the three different metrics namely, Landslide Density Index (LDI), Relative Landslide Density Index (Rindex) and Area Under Curve (AUC) of Receiver Operator Characteristics (ROC) to find out the most suitable methods for susceptibility mapping in this geographical extent. FR and SE depicted highest fitness and predictive ability respectively. The resultant maps can be useful for future land use planning and disaster mitigation measures.