Geospatial Integration of Landslide Susceptibility and Road Network Vulnerability in Mountainous Terrain

Abstract

Transportation networks in hilly regions are often subject to landslides, which induce large socio-economic losses and impede emergency response and regional communication. Hence, it is important to identify landslide prone locations and evaluate the vulnerability of road infrastructure for sustainable transportation planning and catastrophe high risk reduction. This research develops a geospatial machine learning framework for landslide susceptibility mapping and road network vulnerability assessment in Kaski, Nepal, a mountainous region characterized by complicated topography, intense monsoonal rainfall, and fast land use change. A landslide susceptibility model was formulated by employing the Extreme Gradient Boosting (XG-Boost) method with 12 causative elements, including geomorphological, topographic, hydrological, geological, and anthropogenic variables. The Variance Inflation Factor (VIF) was used to check multicollinearity; all variables were under the permitted limit (VIF < 5). The model showed good predictive performance with 93% accuracy, recall, and F1-score, 98% AUC, and a Kappa value of 0.861. The resulting landslide susceptibility map generated shows that most of the study region (59.69%) falls under very low susceptibility; 12.18% is under high to very high susceptibility with localized hotspots. Furthermore, the feature significance analysis showed the major factors influencing the incidence of landslides were land cover (13.9%), rainfall (13.5%), elevation (12.5%), NDVI (12.2%), and slope (11.7%), whereas curvature was the least relevant component. Spatial overlay of the susceptibility map with the road network suggests that about 693 km (23.3%) of the entire road length falls under high and very high landslide-vulnerable zones. The most vulnerable roads were unclassified roads, with 34.1% of their total length in high and very 

high susceptibility groups, followed by primary routes (21.7%) and residential roads (8.3%). The results emphasize the significant vulnerability of major road infrastructure and the urgent need for mitigation measures at a site-specific level. The developed geospatial framework is an effective and flexible method for landslide vulnerability assessment, which facilitates informed decision-making for sustainable infrastructure development in landslide-prone mountainous regions.