Development of a New Stochastic Event Set for European Wind Storms using GCM Output.

Catastrophe models are used by the insurance industry to assess the risk from mid-latitude winter storms, a major driver of financial losses across Europe. A major component of these models is the stochastic event set, a catalogue of thousands of storms of sufficient spatial coverage and resolution to be used to support robust risk analysis for a (re)insurer’s property or motor portfolios. The stochastic hazard model must provide a realistic and physically consistent representation of the current storm climatology impacting northern and western Europe. Aon’s Impact Forecasting team have developed a stochastic event set by extracting synthetic events from the output of a Global Circulation Model (GCM). This approach has several advantages as the extracted events are physically consistent, being the product of the physics of the GCM, resulting in a robust storm climatology and clustering depiction.

This study presents a comprehensive approach to calibrate and validate a set of downscaled synthetic storms against gust data from meteorological stations. The storms have been extracted from the LArge Ensemble of Regional climaTe modEl Simulations for EUrope (LAERTES-EU) dataset, providing over 12,000 years of synthetic climate data. The extracted event catalogue includes 62,500 possible winter storm events.  The original spatial resolution (~27 km) has been downscaled to 3km. Firstly, a gust climatology of the downscaled storms is constructed and compared against a corresponding gust climatology synthesised from the historical observations of meteorological stations across Europe. A quality-controlled selection of weather stations is used to build the historical event set - spanning between 30 and 60 years, depending on the station. The differences between the synthetic gusts and historical gusts are quantified, analysed and used to build correction coefficients applied to calibrate the synthetic events set.