GIS-Based Assessment of Transportation Network Resilience under Hazard Scenarios

Natural hazards such as landslides and floods can disrupt alpine transportation corridors far beyond the directly affected sites, cutting off critical access routes, delaying emergency response, and amplifying cascading socio-economic impacts. However, hazard susceptibility mapping and transportation resilience analysis are still often conducted as separate exercises. This study therefore proposes a GIS-based framework combining hazard susceptibility mapping with network resilience analysis. Landslide and flood susceptibility maps for Zell am See and Saalfelden (Pinzgau, Salzburg) were generated using a patch-based 2D convolutional neural network (CNN) with 15×15-pixel contextual inputs, after logistic regression screening to remove redundant factors. Node importance was evaluated via a principal component analysis (PCA)-derived composite of betweenness, straightness, and degree, followed by role-based classification and staged hazard simulations. The CNN achieved high accuracy (AUC = 0.89 for landslides and 0.90 for floods), with hazard zones strongly matching historical events. Simulation results show that removing just 10% of high-risk nodes can reduce average straightness by over 30% in Zell am See, while Saalfelden’s network degrades more gradually. The framework identifies hazard-exposed Fragile Hubs as priority targets for monitoring or reinforcement and highlights the resilience advantage of Robust Cores. This approach offers a transferable tool for multi-hazard transport resilience planning in alpine regions.