CoreLogic Eurowind™ Model: Capturing Current and Future Risk
The analyses in this white paper leveraged CoreLogic’s Eurowind™ Model. This model provides a highly detailed view of risk to help insurers understand the spectrum of potential European windstorm losses. This probabilistic model quantifies the risk from windstorms across 24 European countries and the North European offshore region (Figure 2).
Figure 2: CoreLogic’s Eurowind™ Model geographical coverage. Source: CoreLogic, 2024
BUILT ON A LONG HISTORICAL RECORD OF GOLD-STANDARD OBSERVATIONS
At the heart of the Eurowind™ Model is a stochastic event set based on actual European windstorms from the mid-20th century to current day. The stochastic event set within the Eurowind™ Model was created by applying a proprietary physical perturbation scheme to historical event wind footprints. First, meteorological agencies in charge of more than 4,000 stations spanning Europe provided gust observation data from 1960 to 2022, which were then used to create the Eurowind™ historical storm footprint catalogue. The events in this historical storm footprint catalogue are then perturbed to create a final stochastic event set.
The stochastic perturbation scheme was designed to address uncertainties in the historical events by altering their physical characteristics and in doing so, generating thousands of synthetic storms that are meteorologically credible and physically possible. Perturbations are made to each storm’s location, orientation, spatial extent, intensity, and duration. Hence, a given parent historical event has an associated ensemble, or family, of offspring stochastic events (each assigned a frequency which decreases with intensity). By analysing historical families, it is possible to generate counterfactual loss time-series for a given span of historical seasons.
Maximisation of the period spanned by the historical storm catalogue is key to accurately modelling current climate risk. The 63-year long Eurowind™ catalogue corresponds to the period over which the highest quality wind observations are available. Within the Eurowind™ Model exist four views of risk providing users different scenarios under which European windstorm activity may differ. The four views of risk are the Analytical Model, the Empirical Model, the North Atlantic Oscillation (NAO) positive phase and NAO negative phase.
CoreLogic’s Eurowind™ Analytical model
The Eurowind™ Analytical Model represents CoreLogic’s recommended view of risk. In this stochastic model, the “gold-standard” accuracy of the wind observations is supplemented with robust climatological information from a 1,200-year Earth System Model (ESM) simulation and the European Centre for Medium-Range Weather Forecasts (EMCWF) approximately 1,800-year seasonal ensemble prediction dataset, called SEAS5.
The Analytical Model incorporates a frequency model based on the climatological simulations. It is a major challenge to accurately assign frequencies to the extreme events in the historical storm catalogue. The Analytical model assigns such events lower frequencies than the simple empirical estimate of 1/63.
The Analytical Model is further enhanced by incorporating downscaled ESM footprints in the peripheral regions where observation density is relatively sparse, in addition to providing a mechanism to include more extreme events than have been observed to date. Clustering of storms is accurately modelled using ESM simulations.
The combination of historical and climatological data used to build the Analytical Model leads to an accurate depiction of European windstorm risk under current climate conditions and is fit-for-purpose in assessing risk over the coming years.
CoreLogic’s Eurowind™ Toolbox
The Eurowind™ Model includes three further views of risk to enable the modeller to gain detailed insights into the impact of European windstorms on their insured portfolio. They are as follows:
Empirical Model - this provides a stochastic event set solely based on perturbations of the 63-year historical set.
NAO positive phase – a stochastic event set, conditioned on the Analytical Model, corresponding to the positive phase of the NAO, a major driver of natural variability in European windstorm activity.
NAO negative phase – a stochastic event set conditioned on the Analytical Model corresponding to the negative phase of the NAO.
Separate event sets for the positive and negative phases of the NAO are provided due to the system’s influence on European windstorm activity. The NAO is a large-scale atmospheric oscillation which manifests as the seesaw of pressure anomaly difference between the Azores High and Iceland Low. Its strength is expressed in the form of the NAO Index (NAOI), which in one of its most common formats is based on measurements in the north-eastern and south-eastern parts of the North Atlantic Ocean, typically at Reykjavik and Lisbon.
The above sets, along with the transparent link between historical and stochastic event sets within CoreLogic’s Eurowind™ Model provide an invaluable toolbox for internal validation, what-if scenario modelling and regulatory stress-testing alike.