Chaotic dynamics of forest fires

In the thermodynamic limit, a probabilistic cellular automaton can be approximated by a deterministic nonlinear map. Here we construct such a inap for the forest fire problem. The construction is based on the results of the Monte Carlo simulation, performed on a square lattice of million cells. The results of the calculation are analyzed by means of the Hoshen-Kopelman algorithm (HKA). The only parameter of the inap describes the probability that a tree appears at an empty cell during one time step. The obtained inap is non-differentiable at the percolation threshold. The Lyapunov exponent for the map is positive. Also, we found the cycle of length three by means of the method of symbolic dynamics. The results are illustrated by the experimental data on the forest fires in Canada in years 1970-2000. Although these data are far from thermodynamic limit, their qualitative character is reproduced for smaller lattices.