An interpretable wildfire spreading model for real-time predictions

Forest fires are a key component of natural ecosystems, but their increased frequency and intensity have devastating social, economic, and environmental implications. Thus, there is a great need for trustworthy digital tools capable of providing real-time estimates of fire evolution and human interventions. This work develops an interpretable, physics-based model that will serve as the core of a broader wildfire prediction tool. The modeling approach involves a simplified description of combustion kinetics and thermal energy transfer (averaged over local plantation height) and leads to a computationally inexpensive system of differential equations that provides the spatiotemporal evolution of the two-dimensional fields of temperature and combustibles. Key aspects of the model include the estimation of mean wind velocity through the plantation and the inclusion of the effect of ground inclination. Predictions are successfully compared to benchmark literature results concerning the effect of flammable bulk density, moisture content, and the combined influence of wind and slope. Simulations appear to provide qualitatively correct descriptions of firefront propagation from a localized ignition site in a homogeneous or heterogeneous canopy, of acceleration resulting from the collision of oblique firelines, and of firefront overshoot or arrest at fuel break zones.