Towards interactive real-time forecasts of weather-related fire brigade operations

Severe weather events like thunderstorms or extra-tropical cyclones can lead to large numbers of fire brigade operations. For example, heavy precipitation can lead to local flooding of basements or underpasses, while wind gusts can lead to treefall and damages to buildings and structures. Such events disrupt the routine operations at fire departments, because a large number of calls need to be handled within a short period of time. However, if suitable warning information about approaching weather events is available in time, there is potential for preparatory actions, for example calling in additional staff or preparing equipment and vehicles. Traditional weather warnings provide information about atmospheric hazards and possibly qualitative statements on expected impacts of the event. However, a quantitative estimate of weather impacts is difficult, because effects of local exposure and vulnerability needs to be taken into account. For example, the extent, intensity and duration of the weather event needs to be combined with information on the numbers and characteristics of affected buildings, trees and infrastructure. Because quantitative impact information is usually not provided in warnings, the estimation of weather impacts on fire brigade operations is left to the experience of emergency managers. Data-based statistical models for for weather-related fire brigade operations could help emergency managers in the decision-making process in an operational setting.

In this study we aim to develop and evaluate models and interactive visualization tools for weather related fire brigade operations in a co-design approach together with fire brigades from different regions of Germany. Fire brigade operations will be modeled based on weather information from nowcast products and ensemble forecasts, as well as exposure information from open data sources like OpenStreetMap. Here, we present methodologies for interactive computation and visualization of probabilistic hazard and impact information using logistic regression models. The aim is to allow for a flexible selection of the spatial and temporal scale of forecast information. This can help emergency managers to plan operations, for example based on specific shift schedules.