Integrated Topological, Physical, and Functional Vulnerability of Road Networks to Compound Seismic-Landslide Hazards in the East African Rift

Road networks in tectonically active regions face compound seismic and landslide hazards that can simultaneously disrupt multiple links. However, existing vulnerability assessments typically examine physical, topological, and functional dimensions in isolation and address only single hazard types. This study presents an integrated three-dimensional vulnerability framework combining graph theory, geographic information systems, and multi-criteria decision analysis to assess the compound seismic-landslide vulnerability of northeastern Ethiopia’s federal road network. The framework quantifies topological vulnerability through graph-theoretic metrics (edge betweenness centrality and redundancy), physical vulnerability through the expected level of seismic acceleration and landslide susceptibility zoning, and functional vulnerability through road classification, traffic volume, and travel time. Indicator weights were derived from 12 domain experts using the Analytic Hierarchy Process, with hazard intensity receiving the highest global weight (0.35). The results reveal that road segments with comparable overall vulnerability can exhibit markedly different combinations of physical, functional, and topological contributions, demonstrating that network criticality arises from the interaction of these dimensions rather than from a single dominant factor. Furthermore, the Kombolcha-Bati link records the highest vulnerability index (VI = 0.694), reflecting compound seismic-landslide exposure. Notably, a main access link ranks among the 10 most vulnerable segments, highlighting the importance of considering vulnerability beyond the trunk road network. These findings offer a spatially explicit, transferable decision-support tool for infrastructure risk management in hazard-prone low-income regions, replicable in contexts where comparable road authority data or open-data proxies are accessible.