Atmospheric drivers of weather-related insurance losses using ERA5 reanalysis

Weather-related hazards represent a major source of risk for the insurance sector. However, insurance risk assessment still largely relies on isolated events, single hazard analyses, and probabilistic loss metrics wich provide a limited understanding of the recurrent weather conditions that drive insurance losses. We introduce an impact-driven and decision-oriented framework to identify weather regimes directly conditioned on insurance loss occurrence, using ERA5 reanalysis data to bridge atmospheric drivers and observed impacts over France. This framework is calibrated on high-impact loss days since 1998 to extract robust weather regimes and is evaluated across the full observation period to assess their loss contribution and spatio-temporal expression. This approach reveals a limited set of recurrent, spatially organised weather regimes associated with distinct loss signatures. We provide a weather regime-based storyline linking atmospheric drivers and observed insurance losses - offering a coherent framework to interpret loss variability, supporting impact attribution, and informing risk interpretation and decision-making at the insurance portfolio scale.