Empirical models of forest fire initial attack success probabilities: the effects of fuels, anthropogenic linear features, fire weather, and management

Canadian fire managers seek to contain fires below some target size (here 3 ha) by initial attack (IA). Suppression failures occur when fire size at IA exceeds this target (a response failure) or if an initially small fire cannot be contained below it (a containment failure). We examined the effects of cause, season, forest fuels, anthropogenic linear features, weather, and fire management (response time, size at IA) on the probability of these two types of suppression failures, using multiple logistic regression on 1196 fires that occurred within the boreal mixedwood forest of northeastern Alberta during 1995–2002. The frequencies of containment (7%) and response failures (10%) were similar, but the latter accounted for 85% of the area burned. Response failure probability was greater for fires caused by lightning than those caused by humans and increased with longer response times, local abundance of black spruce in summer, and pine fuel under severe fire weather. We found no effect of linear features or other fuel types. Containment failure probability was related to size at IA and fire weather conditions. Our models suggest that a reduction in area burned might be possible if additional fire-specific factors affecting response failure probability could be incorporated into operational decisions.