A cellular automata fire model represents ‘elements’ of fire by autonomous agents. A few simple algebraic expressions substituted for complex physical and meteorological processes and solved iteratively yield simulations for ‘super-diffusive’ fire spread and coupled surface-layer (2-m) fire–atmosphere processes. Pressure anomalies, which are integrals of the thermal properties of the overlying heated plume, drive the surface winds around and through the fire. Five simulations with differing fuel and wind conditions were compared with fire and meteorological data from an experimental grassfire (FireFlux). The fire model accurately simulated bulk patterns of measured time-series of 2-m winds at two towers and observed fire behaviour (spread rate, flaming depth and heat released). Fidelity to spatial windfields in the vicinity of the fire was similar to results from full-physics fire models for other grassfires. Accurate predictions of fire spread depend critically on accurate wind speeds and directions at the location of the fire. Simulated fire–atmosphere coupling using FireFlux data increased wind speeds across the fire line by up to a factor of three. With its computational speed relative to full-physics models, the fire model can inform full-physics modellers regarding problems of interest. Although the fire model is tested for homogeneous fuels on flat terrain, the model is designed for simulating complex distributions of fire within heterogeneous distributions of fuels over complex landscapes.
Stochastic cellular automata model for wildland fire spread dynamics
