Comparative landslide susceptibility assessment using information value and frequency ratio bivariate statistical methods: a case study from Northwestern Himalayas, Jammu and Kashmir, India

Abstract

In the northwestern Himalayas, including Jammu and Kashmir (J&K), frequent landslides pose significant risks, necessitating proactive zoning to mitigate potential damage through effective land-use planning. Fourteen causative and two triggering factors, such as slope, aspect, curvature, relative relief (RR), terrain ruggedness index (TRI), geomorphon, dissection index (Di), lithology, structural tectonic, drainage density (Dd), stream power index (SPI), topographic wetness index (TWI), land use land cover (LULC), road density (Rd), earthquake density (Ed), and rainfall density (Rd), were selected based on terrain conditions to assess landslide susceptibility. Utilizing frequency ratio (FR) and information value (IV) approaches, a comprehensive landslide susceptibility mapping (LSM) study covered 54,922 km², incorporating 6669 landslide instances. This dataset was split into 70% (4659 landslides) for modeling and 30% (2010 landslides) for validation. The landslide susceptibility map, classified into five categories (very low, low, moderate, high, and very high), delineates varying proportions of the study area. Using the FR approach, these zones cover 12.9% (7063 km²), 25.7% (14,101 km²), 25.6% (14,049 km²), 24.7% (13,586 km²), and 11.1% (6123 km²) of the area, respectively. Meanwhile, employing the IV approach, the coverage percentages are 5.7% (3119 km²), 11.0% (6063 km²), 20.1% (11,057 km²), 38.9% (21,373 km²), and 24.1% (13,310 km²). Validation using receiver operating characteristic curves revealed high correlations for both FR (AUC: 0.809) and IV (AUC: 0.778) models, indicating their effectiveness. The FR model, characterized by simplicity and higher accuracy, outperformed the IV model, offering valuable insights for local, regional, and governments in land-use planning, disaster prevention, and mitigation efforts.