Enhancing North Atlantic Hurricane Damage Prediction Through Integration of Hazard, Exposure, and Vulnerability Data

Hurricanes are among the most destructive natural hazards worldwide, posing significant risks to communities and economies. The Saffir–Simpson hurricane wind scale is widely used to communicate hurricane magnitude, but it relies solely on wind speed and has limited predictive skill of potential damages. In this presentation and in a recent paper, we introduce a novel statistical modelling approach that integrates publicly available hazard, exposure, and vulnerability data to more skilfully predict the financial impact of impending landfalling North Atlantic hurricanes.

By applying optimal weights to hurricane hazard, exposure, and vulnerability attributes, our model significantly improves damage predictions, reducing root mean squared error from over $35 billion USD when using the Saffir–Simpson hurricane wind scale to just $7 billion USD when using our new model. This new simple model greatly outperforms conventional single-parameter damage estimates e.g., hurricane Vmax and central pressure (Cp). We also propose a new ' Predictive Hurricane Damage Scale' that indicates Hurricane magnitude as a function of damage. This new scale facilitates clearer communication for financial industries of potential damages from an impending hurricane, whilst being open source. This framework not only enhances understanding of past hurricane impacts but can also help policymakers and stakeholders prepare more effectively in the days preceding a hurricane landfall. The approach underscores the importance of open-source exposure and vulnerability data, which is a necessity for quantifying risk.