The Western Himalayan region, comprising Himachal Pradesh, India, manifests complex topography and faces significant damage due to geohazards, particularly landslides. Parts of the Chandra-Bhaga basin are especially vulnerable to increased risk of landslide activities. These events result in casualties, fatalities, and huge losses to the economy. Identifying landslide-prone zones in this region is therefore essential for effective hazard mitigation and land-use planning. The present study aimed to generate landslide susceptibility maps (LSMs) of the Chandra-Bhaga basin using discrete and collective bivariate and multi-criteria decision-making (MCDM) approaches, including frequency ratio (FR), and analytical hierarchy process (AHP), respectively. The FR method was employed to quantify objective statistical relationships between conditioning factors and the occurrence of landslides, while AHP was applied to incorporate expert-based multi-criteria evaluation. The FR–AHP hybrid model was adopted to reduce subjectivity in factor weighting and enhance predictive performance. A landslide inventory comprising 1,137 historical landslide events (2016–2022) was prepared using field observations, government records, and high-resolution satellite imagery. In addition, 13 landslide conditioning factors were selected based on regional geomorphic relevance and data availability. LSMs were generated using FR, AHP, and FR-AHP models, and their predictive performances were evaluated using the area under the receiver operating characteristic curve (AUC-ROC). The FR model achieved an accuracy of 84% (AUC: 0.84), while AHP predicted landslide point positions with an accuracy of 81% (AUC: 0.81). The LSM based on FR-AHP exhibited the highest accuracy of 90% (AUC: 0.90), highlighting FR-AHP as the most accurate model for landslide hazard prediction, prevention, and mitigation. This study presents the first comprehensive basin-scale landslide susceptibility assessment of the Chandra-Bhaga basin integrating FR, AHP, and FR-AHP models, as confirmed by an extensive review of existing Himalayan and regional literature. However, the analysis is based on static conditioning factors and historical landslide inventories, and does not consider the temporal evolution of controlling factors or dynamic landslide-triggering processes; future studies should therefore integrate time-series data and machine-learning-based approaches to enhance predictive capability. • Chandra-Bhaga basin mapped for landslide susceptibility using FR, AHP, and FR-AHP. • First application of integrated FR-AHP for LSM in the Chandra-Bhaga basin. • Eleven key landslide conditioning factors selected for basin-scale modelling. • Integrated FR-AHP model achieved the highest performance (AUC–ROC = 0.90). • Hybrid model reduced AHP subjectivity using FR prediction ratios. • Outputs support better land-use planning and hazard mitigation in the Himalayas.
Dhiman et al. (Wed,) studied this question.