Flood risk to population in Italy: a synoptic overview based on fatality data.

Floods continue to pose a threat to human life despite the major advances in managing flood risk. To quantitatively assess potential loss of life, we present a statistical approach to model societal flood risk from sparse, point information on fatal floods in Italy. At this end, the Zipf distribution and its parameters were estimated over a regular grid covering the national territory. Despite the difficulties in modelling sparse data, we tested the approach using a detailed record of 667 fatal floods at 652 different sites for 345 major or minor meteorological events during the 78-year period 1946-September 2023. The results provide new insights into the spatial variability of societal flood risk across the national territory and into the expected return periods associated with different numbers of fatalities. Model validation indicates a slight decrease in the return periods of fatal floods in the validation dataset. This suggests a possible increase in the frequency of localized events, such as flash floods or high-volume torrential floods causing at least one fatality, potentially linked to changes in rainfall regimes rather than to urbanization. The findings can contribute to refining institutional flood-risk zoning and to supporting mitigation strategies aimed at reducing loss of life. The proposed framework may also be applied to other countries where sufficiently detailed information on fatal flood events is available.