Modelling hail hazard over Italy with ERA5 large-scale variables

Hail is a meteorological phenomenon with adverse impacts that affects multiple socio-economic sectors such as agriculture, renewable energy, and insurance. Nevertheless, the understanding of the favourable environmental conditions for hail formation and the models’ inadequacy to represent these phenomena have been limited by the scarce temporal and spatial coverage of hail observations. This is a major concern for the mitigation of hail-related risk in sensitive regions such as Italy, which is one of the more hail-prone areas in Europe. In this work, we present a hail model that has been developed to describe the hail hazard over Italy. This model relies on several ERA5 large-scale meteorological variables and convective indices that are combined following the statistical method described in Prein and Holland (2018). The identification of the best set of variables to be used as predictors in the hail model has been performed by a systematic machine learning procedure based on a genetic algorithm. The hail model estimates the hail probability over Italy in the 1979–2020 period, on the ERA5 spatial grid resolution (∼ 30 km). The output of the hail model has been used to characterize the seasonality and long-term variability of hail events in Italy. Furthermore, the categorical verification of the hail probability over the Friuli Venezia Giulia region has revealed that the hail model is able to effectively estimate the hail occurrences in specific Italian regions.