Using Crowdsourced Data to Improve User Communication in Wind Gusts Warnings

Enhancing user communication is a central goal of the RainBoW (Risk-based, Application-oriented and INdividualizaBle Provision of Optimized Warning Information) programme, which encompasses the development of the new weather warning system for the German Meteorological Service (Deutscher Wetterdienst, DWD). A clear understanding of issued weather warnings is essential for enabling both professionals and the general public to take appropriate action in affected areas. To support this, future warnings by the DWD will increasingly focus on impact information. Crowdsourced data, which reflects how users perceive specific weather events, offers a rich and valuable resource for improving this communication.

In this study, we focus on the hazards of wind gusts, which are among the weather elements with the greatest impact on infrastructure and society. To assess the severity of such events, we propose an impact proxy for Germany, derived from crowdsourced data collected through the Warnwetter App — the official weather app of the DWD. The app allows users nationwide to report their weather interpretations, providing a broad representation of the population’s exposure to weather events. Specifically, for wind gusts, users can choose from five predefined severity levels, ranging from "weak wind" to "severe storm".

We combine these crowdsourced reports (total number of roughly 55k) with meteorological ensemble model forecast data (ICON-D2-EPS) and geographic information (such as elevation and land use). Using statistical methods, we evaluate the suitability of this data as an impact proxy for wind gusts. While we find a strong correlation between user reports and model output, the data does not readily allow for a clear classification to predict user perceptions based on forecasts. We discuss potential ways to use this approach to improve communication with the public.