AXA probabilistic Severe Convective Storm model in western Europe

Severe convective storms are a common occurrence during spring and summer season in European countries. The damages caused by hail and wind gusts can be substantial to properties, especially on motor. The development of a convective storms hazard stochastic catalog is an important step for AXA to assess and mitigate this peril.
We propose a method to build a catalog of synthetic events based on multiple meteorological drivers from ECMWF-ERA5 and EUMETSAT-CMSAF. New atmospheric temporal sequences are created by reshuffling historical data, with constraints to keep physical consistency (identification of weather patterns and historical transition probabilities between them). The probability of hail occurrences is then assessed for each meteorological configuration, learning from in-situ reports (ESWD and Keraunos), with historical validation to ensure accuracy of the hail prediction. A catalog of new plausible scenarios for convective storm hazard is produced and crossed with exposure and vulnerability data to assess the subsequent risk.