Forest fire modeling using cellular automata and percolation threshold analysis

Multi-state probabilistic cellular automata are developed for forest fire modeling. We propose a forest fire dynamics model considering intensities of fires as multiple states and having the probability that fire spread depends on the states of the neighboring cells. Furthermore, the idea of percolation threshold is introduced to characterize the strength of the fire propagation. Specifically, we propose a new method to derive the critical probability, below which forest fires do not expand to infinitely large area, using percolation theory and mean-field approximation.