Data-driven disaster resilience assessment: a case study in the Spanish transportation system

Abstract

Critical infrastructures (CIs) form the backbone of societal functionality but are increasingly threatened by natural disasters. Developing disaster management tools, particularly those leveraging Information Technology for Development, can enhance practitioners’ ability to prepare and respond effectively to such events. Analysing their resilience is crucial, nonetheless, current studies often rely on biased data and specific regions, limiting their generalizability. To address this gap, this work applies machine learning (ML) to obtain a national resilience index from Spanish meteorological and insurance data. By employing the best-performing ML unsupervised methods within the Spanish transportation scenario, the data were clustered by disaster magnitude and losses, with weighted and moving averages used to generate the index. The resilience disparities and areas needing improvement were highlighted and confronted with specific actions promoted by municipalities. The results enhance the understanding of CI resilience and are applicable to real-world cases, supporting increased transportation resilience in critical regions.