Abstract Compound events—combinations of multiple hazards that contribute to societal or environmental risk—can significantly exacerbate disaster impacts. However, their effect on flood-related losses remains insufficiently quantified. In this study, we compile a novel, pan-European multi-hazard dataset spanning 1981–2020 at sub-national resolution. This dataset reveals that nearly 80% of recorded flood events in Europe involve compounding hazards, encompassing meteorological extremes such as heatwaves, cold spells, and windstorms, alongside anomalous river discharge linked to floods and droughts, with an increasing trend over time. Notably, the top 1% of these events, in terms of economic losses, are all compound. We introduce a novel metric to characterize the complexity of these compound events in a single index. By integrating flood loss records with data on hazards, exposure, and vulnerability, our ensemble machine learning models predict that flood losses for compound events are 3.4 times higher than for single events. Using explainable AI, we demonstrate that higher compound hazard complexity significantly amplifies flood damage, with a relative increase in impact of 0.34 units in high-vulnerability regions compared to low-vulnerability areas. To our knowledge, this study provides the first comprehensive, data-driven evidence that compound events are associated with higher flood losses across Europe. This underscores the importance of incorporating compound hazard information into risk modeling and disaster preparedness strategies and offers insights that could support the development of multi-hazard early warning systems as part of the Early Warnings for All initiative.
Ronco et al. (2025) studied this question.