An empirically derived high-resolution spatial windstorm risk model: an application for the Netherlands

Extratropical cyclones generate hazardous weather conditions, such as windstorms, which can result in significant societal and economic impacts. This study proposes a transferable empirically derived high-resolution winter windstorm risk model, which is developed in the Netherlands but the methodology is widely applicable to other regions with similar wind climates, exposure characteristics, and building stock, assuming adequate local data is available. Using downscaled reanalysis data (2.5 × 2.5 km 2 ) from 1979-2019, we apply extreme value theory to estimate wind gust return levels across the country. These hazard estimates are integrated with detailed residential building data, asset-specific vulnerability curves, and a claim ratio function (number of claims/exposed assets) to model insured property losses. A key innovation is the combination of probabilistic physical hazard modelling at a high resolution with empirically derived loss and claim functions to provide risk maps and metrics such as expected annual damage (EAD). Validation using twelve historical storms provide a useful first-order assessment of model performance against aggregated reported insurance losses. The outputs such as return-period losses, the EAD estimate and map, provide a representation of potential national losses and its spatial distribution and a crucial understanding into windstorms' long-term financial effects. These data provide actionable insights on adaptation and risk management investments for insurers, policy makers, and planners, supporting prioritizing adaptation measures, more accurate insurance pricing, and resilient infrastructure planning. • We develop a generic, transferable model for assessing winter windstorm risk. • High-resolution wind hazard maps pinpoint the hotspots, clearly exposing the areas most vulnerable to extreme wind speeds. • The risk model integrates high-resolution hazard, exposure, and vulnerability to deliver Expected Annual Damage (EAD). • We validate the risk model using observed national insured losses. Comparing these with aggregated reported losses provides a useful first-order assessment of model performance. • This risk model supports targeted climate adaptation, insurance pricing, and resilience planning.