Geospatial multi-criteria decision analysis for vegetation health index mapping and landslide susceptibility assessment

Abstract The Himalayan region’s complex terrain and ecological fragility make it increasingly vulnerable to vegetation stress and landslides, driven by both natural and human factors. This study focuses on a 2663.92 km² area encompassing Rudraprayag, Ukhimath, Gopeshwar, and Joshimath, aiming to develop a Vegetation Health Index (VHI) and explore its link with landslide susceptibility. Six key parameters such as NDVI, NDMI, soil, rainfall, slope, and road proximity were selected for their relevance and tested for multi-collinearity. All variables showed VIF values below 5, confirming their independence and suitability. Expert weights were applied on six critical parameters using the Analytic Hierarchy Process (AHP) with a Consistency Ratio of 0.093 establishing reliability. The GIS-weighted overlay technique was used to create the spatially varied Vegetation Health Index (VHI) map. Validation using sensitivity analysis, LAI regression (R² = 0.62), and AUC_ROC analysis (0.423) revealed 57.7% occurrence of previous landslides in low VHI areas, establishing a negative relationship between vegetation status and landslide susceptibility. The hybrid geospatial-statistical model is a reliable predictor of ecologically and geologically sensitive regions and an effective tool for land degradation mapping and disaster risk reduction. The results provide vital data for future mountain environmental planning, slope management, and policy formulation.