Towards improved assessment of windstorm damage risk through Impact Model calibration

Winter windstorms rank as one of Europe's deadliest and most damaging natural disasters. To model the impacts of these windstorms, surface wind data can be incorporated into climate risk models to derive estimates of natural hazard-related impacts on natural or socio-economic systems. In CLIMADA, risk from a natural hazard can be modelled as the convolution between three components - hazard, exposure, and vulnerability.  The vulnerability component links the hazard and exposure components to give total impact that can be approximated through functional relationships called vulnerability curves (or impact functions in CLIMADA). Advancing the science of impact estimation from windstorms is imperative for mitigation and management of changing climate risks, and this relies on appropriate calibration of the vulnerability curve. To this end, in this study, we calibrate a popular impact function from Schwierz et al. (2010) using impact data from two types of sources: open-source (EM-DAT/XWS) and proprietary (PERILS). Results indicate substantial differences between the calibrated vulnerability curves and highlight the importance of the type of recorded disaster data used in calibration. Furthermore, for each of the aforementioned calibration cases, we discuss the uncertainties associated with the use of different cost functions and optimization techniques in the calibration process. The study brings forth how data and method choices influence vulnerability curves, helping better understand modelling uncertainty and support the development of more reliable tools for climate risk assessment and adaptation.