Modeling smoke-driven evacuation routes during wildfires: Architecture and implementation of the s-ERP system

The rise in frequency and intensity of global wildfires presents increasing risks to public health and safety, especially through the respiratory and cardiovascular issues resulting from smoke exposure and elevated temperatures. Although immediate evacuation is crucial for protecting individuals, existing evacuation planning tools lack integration with smoke behavior simulations, which limits their effectiveness. This paper introduces s-ERP (smoke-based Evacuation Route Planning), a real-time decision-making system designed to enhance evacuation outcomes during wildfire events. s-ERP constitutes a cost-efficient integrated framework that analyzes spatiotemporal data by developing synergies with various open external tools to proactively determine both safe evacuation routes and the optimal departure time. This tool comprises three core modules: (i) the Smoke Dispersion Module that forecasts the progress of smoke behavior by adopting a Gaussian plume model; (ii) the Evacuation Route Planning Module that implements an algorithm identifying alternative safe routes, while minimizing exposure to smoke-affected areas; and (iii) the Web Application, consisting of a RESTful API with multiple endpoints, which manages data flow and enables operational use by stakeholders. The proposed system is validated through extensive experimental evaluations under realistic wildfire and smoke conditions and in the context of a real scenario of distributed emergency management. The empirical findings indicate that s-ERP mitigates evacuee exposure to harmful smoke in comparison to conventional methods and improves route reliability under wildfire dynamic conditions. These results demonstrate the system’s capability to deliver timely, adaptive and health-oriented evacuation strategies.