Wildfire Hazard Mapping Using Cellular Automata

Since fuel load is a major factor influencing wildfire risk, the standard approach to build related hazard maps is mainly grounded on land-cover data. However, the risk level is also influenced by other factors interacting nonlinearly, such as wind, fuel moisture, ignition sources and topography. For these reasons, an increasingly used approach for the computation of hazard maps involves the explicit simulation of the fire dynamics. This paper exploits a novel CA model for wildfire simulation to evaluate fire risk within a Monte Carlo approach. The adopted CA model has the ability to provide accurate burned areas, taking much less computing time than a typical vector approach for wildfire simulations. The improved accuracy and efficiency were obtained: (i) relaxing the restriction to a few pre-defined directions of spread, which characterizes most of the techniques for simulating wildfires on a raster space; (ii) using an adaptive time-step duration, which allows for avoiding unnecessary computation. The preliminary tests presented in this paper indicate that the model under study can be a suitable component of a tool for wildfire risk assessment.