Building damage impact forecasting for winter windstorms in Switzerland

Weather extremes can have high socio-economic impacts. Better impact forecasting and preventive action help to reduce these impacts. In Switzerland, the winter windstorms caused high building damage, felled trees and interrupted traffic and power. Events such as Burglind-Eleanor in January 2018 are a learning opportunity for weather warnings, risk modelling and decision-making.

We have developed and implemented an operational impact forecasting system for building damage due to wind events in Switzerland. We use the ensemble weather forecast of wind gusts produced by the national meteorological agency MeteoSwiss. We couple this hazard information with a spatially explicit impact model (CLIMADA) for building damages due to winter windstorms. Each day, the impact forecasting system publishes a probabilistic forecast of the expected building damages on a spatial grid.

This system produces promising results for major historical storms when compared to aggregated daily building insurance claims data from a public building insurer of the canton of Zurich. The daily impact forecasts were qualitatively categorized as (1) successful (2) miss or (3) false alarm. The impacts of windstorm Burglind-Eleanor and five other winter windstorms were forecasted reasonably well, with four successful forecasts, one miss and one false alarm.

 The building damage due to smaller storm extremes was not as successfully forecasted. Thunderstorms are not as well forecasted with 2 days’ lead time and as a result the impact forecasting system produces more misses and false alarms outside the winter storm season. For the Alpine-specific southerly Foehn winds, the impact forecasts produce many false alarms, probably caused by an overestimation of wind gusts in the weather forecast.

The forecasting system can be used to improve weather warnings and allocate resources and staff in the claims handling process of building insurances. This will help to improve recovery time and costs to institutions and individuals. The open-source code and open meteorological data makes this implementation transferable to other hazard types and other geographical regions.