Time Series Analysis: Comparing a Forest- Fire Cellular-Automata Model to Real Forest-Fire Datasets

This dissertation compares the statistics of two forest-fire databases from Catalonia and Ontario (Canada), with the results from a forest-fire cellular-automata model. Strong similarities are found regarding the frequency-area statistics, following a power-law distribution in all cases. In both model and reality, around half of the overall burned area are found to be caused by fires less than 10-15% the biggest recorded ones. Some guidelines are derived regarding validation of forest-fire models, forest-fire prescription and fire-extinction priorities.