Using expert knowledge to model forest stand vulnerability to fire

The present study analysis the possibility of using virtual reality models, Internet questionnaires and regression methods for pairwise comparison to acquire information from experts in forest fires, and analyse it in a quantitative way. Two models were developed for estimating the vulnerability of forest stands to fire, depending on the stand characteristics, using two different sets of images. The models were based on the relative priority that experts gave to pairs of stand images with respect to their fire vulnerability. The results indicate that the abundance of an understory and the structural irregularity of the stand (expressed as coefficient of variation of diameters) have an important effect on the stand vulnerability as perceived by the experts. The vulnerability decreased with decreasing amount of ground vegetation and diminishing variation in tree size.