STEAR: Spatio-Temporal Exposure-Aware Routing for Airborne Releases

During a hazard scenario (chemical, nuclear, smoke), users aim to use their smartphones for evacuation instructions. However, current mobile apps do not allow advanced evacuation primitives, such as the computation of single target or multi-target evacuation routes to pick up children or elderly, find evacuation shelters or collection point based on available transport means or availability (e.g., car, bicycle, walking). More importantly, these do not provide recommendations based on predicted conditions on how the phenomenon will unfold. In this paper, we present \emph{HazardEvac}, a platform for time-dependent, dose-aware evacuation route planning under hazardous airbone releases. Our platform allows real-time computation of cumulative exposure to airbone releases a human will incur during evacuation on a suggested route. HazardEvac simulates an airbone release in real-time using the HYSPLIT-based plume generation, concentration-field mapping on a routing grid, and an exposure-aware graph-search algorithm that computes evacuation routes by jointly considering movement cost and time-varying contamination levels. The platform supports different release scenarios, including chemical and radiological incidents, with particular emphasis on radiological dispersion. These capabilities are combined in an interactive interface that visualizes evolving plume conditions and the resulting evacuation route. We demonstrate the practicality of the platform using simulated release scenarios and time-indexed concentration fields, showing that HazardEvac can support adaptive evacuation decisions under dynamically evolving conditions.