Climate Hazard Mortality: Diagnosing Trends and Outliers

Abstract

Climate hazards like floods and heatwaves kill many people; their deadliness may shift with the frequency or intensity of extreme weather, population growth, or migration. Assessing preparedness and projections requires understanding historical mortality trends, though this is challenging due to the variability in, and complex causal chains of, events' deadliness. This study diagnoses mortality trends in EM-DAT, the largest public disaster mortality database, using a nonstationary Poisson-Generalized Pareto statistical model. Several trends in frequency and/or intensity of deadly hazards are identified and discussed. Most notably, Asian floods and storms became less deadly and frequent due to reduced vulnerability via increased adaptive capacity, which is conservatively estimated to have saved 350,000 (95% CI: 220,000–560,000) lives. Storm Daniel in 2023 is also found to be a once-in-two-centuries outlier event indeadliness for an African flood or storm. These results highlight the importance of continued disaster monitoring as climate hazard mortality evolves.