LANDSLIDE SUSCEPTIBILITY ASSESSMENT OF CHENAB VALLEY AREA

Abstract

The purpose of this dissertation work is to produce a reliable Landslide Susceptibility Mapping (LSM) using Frequency Ratio (FR) and Certainty Factor (CF) models with the aid of ArcGIS software for the Chenab Valley, J&K, India. Remote Sensing (RS) and Geographical Information System (GIS) are powerful tools for assessing landslide hazards and are being used extensively in landslide research for many years. In this study area, about 365 landslides have occurred. These landslide sites can be seen on satellite images and also on Google Earth. Information on many of these landslides is also available on the GSI portal. A landslide inventory map had prepared mainly based on landslide occurrence of the study area. Then, 256 (70 %) landslides were randomly selected for modelling, and the remaining 109 (30 %) landslides were used for the model validation. To produce a landslide susceptibility map using FR & CF models, 13 prominent landslide contributing factors were selected based on hydro-geomorphological characteristics. These landslide contributing factors are the slope, aspect, altitude, geology, distance to lineament, distance to fault, distance to road, distance to drainage, Topographic Wetness Index (TWI), rainfall, curvature, Normalized Difference Vegetation Index (NDVI) and earthquake. These factors were mapped with the aid of ArcGIS software and classified into significant classes using the natural break (Jenks) method. After that, the landslide susceptibility maps had prepared using landslide contributing factors based on the FR and CF models. Based on these maps, the area was further classified into five significant classes, i.e., very low, low, moderate, high, and very high categories based on the severity of landslides. The FR & CF model results showed that approximately 5.9 % & 12.89 % of the study area fell into the very high zone of landslide. The landslide susceptibility maps can be helpful to select a site, predict future landslide chances, and mitigate landslide hazards in the study area. Finally, the accuracy of the landslide susceptibility maps developed from the two models were validated using Area Under the Curve (AUC) analysis. The validation showed that the successor rate curve accuracy of the two models was 89.80 % for the FR and 90.12 % for the CF. The prediction rate curve accuracy of the two models was 88.80 % for the FR model and 88.43 % for the CF model. As per the results, both models showed an approximately similar level of accuracy.