Assessing windstorm hazard emerging from multiple types of storms

In the context of global climate change, windstorms pose significant environmental, ecological, and socioeconomic challenges. Mountainous and forested regions of Europe, including the Veneto region in northern Italy, have been devastated by unprecedented events such as the storms in July 2023 and Vaia in October 2018, raising the question whether such events may occur more frequently in the future. The probability of observing such extremes in present-day climate can be quantified using cumulative distribution functions of annual maxima wind speeds, obtained from extreme value analysis methods. Once these are derived, however, is it near to impossible to project future changes in these distributions as extreme wind speeds are caused by storms driven by diverse synoptic conditions, the characteristics and occurrence frequency of which may change differently in response to climate change.

This study introduces a method to derive cumulative distribution functions of annual maximum wind speeds explicitly considering the intensity and occurrence frequency of multiple types of storms. Independent windstorms are separated and their maximum hourly wind speed is isolated. Storms are then organized into types based on their local wind direction using a clustering technique. We then use a multi-type Simplified Metastatistical Extreme Value distribution (SMEV) to estimate the cumulative distribution function of annual maximum wind speed for the location of interest. The study focuses on mountainous areas, seeking a simpler relation between typical wind directions and synoptic conditions.

A thorough leave-one-out evaluation with benchmark models, including the traditional Generalized Extreme Value distribution (GEV) and a single-type SMEV, is conducted on 22 mountain stations in the Veneto region (northern Italy). We show that, overall, the proposed multi-type method provides estimates of extreme return levels that are comparable with the ones of single-type SMEV and GEV. Our results demonstrate that it is possible to derive cumulative distribution functions of annual maximum wind speeds explicitly considering storms emerging from different marginal processes. This paves the way to the use of projections of large-scale atmospheric dynamics from climate models to improve our prediction of future extreme wind speeds.

 

Keywords: Windstorm; Extreme events; Wind direction classification; Multiple types; Simplified Metastatistical Extreme Value (SMEV); Mountainous areas.