Abstract. Simulating fire spread and identifying potential propagation pathways in the Wildland–Urban Interface (WUI) are critical for wildfire prevention, emergency preparedness, and firefighting—especially in the wake of catastrophic events such as the 2025 Los Angeles wildfire, the 2019–2020 Australian bushfires, and the 2023 wildfires in Greece. Despite growing awareness of wildfire risks near urban boundaries, lightweight, high-resolution 3D simulation tools remain limited, hindering scenario-based planning and rapid response. To address this gap, we present a voxel-based 3D wildfire propagation simulator developed in Python. The simulator integrates LiDAR-derived voxel models of urban environments, GIS-informed fuel characterizations, and high-performance parallelism via the Taichi framework. Fire dynamics are modeled across 3D voxel grids using a hybrid of physics-based and empirical approaches, incorporating key parameters such as wind speed, fuel type, and moisture content. Critical processes—including inter-voxel heat transfer, crown fire spread, and surface fireline intensity—are captured to simulate realistic fire behavior. Simulation results are exported in standard 3D formats for immersive visualization in platforms such as Blender and Unity. A case study using LiDAR data from Newcastle, Australia demonstrates the tool’s real-world applicability. Designed for modularity and extensibility, the simulator supports model replacement, parameter tuning, and integration with diverse spatial datasets. It also serves as a scalable framework for high-fidelity modeling of inter-voxel mass and energy transfer in complex urban environments, enhancing decision-support capabilities. Additionally, the tool generates synthetic fire spread data, enabling the training of generative AI models and integration with broader urban and environmental simulation platforms.
Xu et al. (Tue,) studied this question.