Weather, forest vegetation, and fire suppression influences on area burned by forest fires in Ontario

Using several models of fire growth and fire suppression, this study shows how area burned by forest fires is influenced by weather and forest vegetation (or fuel) and is reduced by fire suppression. Previous attempts to characterize the relationship between fire suppression and area burned have focused on the initial attack process. Although the vast majority of forest fires are contained by initial attack forces while they are still small, most of the area burned in Canada is due to those few fires that escape initial attack and become large. This study focuses on these large fires. First, I used the concept of 'spread events' to refine the fire growth predictions of physical and empirical fire growth simulation models developed by Van Wagner (1969), and by Tymstra et al. (2005). I was able reduce prediction errors of fire sizes from overestimates several times too large to errors of 10-40% in some cases. Second, by calculating the fuel composition of area burned by real fires and comparing these compositions with those of area burned by simulated fires in Ontario using the WILDFIRE growth model (Todd et al. 1999), I showed how fires 'prefer' more flammable fuels as predicted by the Canadian Forest Fire Behaviour Prediction (FBP) System (Forestry Canada 1992). By developing a discrete event simulation model for the growth and suppression of large fires in the province of Ontario for multiple weather and fire suppression scenarios, I found that, while severe weather limits suppression effectiveness, suppression has an important effect even during severe seasons. For example, my model predicts that a 50% increase in suppression resources would reduce area burned by up to 55% during a severe fire season (1988), while the same increase in suppression resources would only reduce area burned by 0.4% during a mild season (1990). Decreases in fire suppression resources have more severe effects: a 50% decrease in fire suppression resources during a severe season would increase area burned by 65%, and even during a mild season (1990), a 50% decrease in resources would increase area burned by 40%. These findings were consistent with a simple probabilistic model of annual area burned, fire weather, and suppression, which I developed to validate the more complex discrete event simulation.