Using expert judgment to model initial attack fire crew effectiveness

Abstract

An expert judgment elicitation methodology was developed and used to encode subjective assessments of fire crew effectiveness from experienced initial attack crew leaders. During structured individual interviews, experts from four Canadian forest fire management agencies provided assessments of the probability of fire containment (POC) by a 'medium' (5- to 7-person) initial attack crew for 35 initial attack scenarios that varied in terms of fire size and head fire intensity. This repeated-measures data was used to develop individual, logistic response curves for 34 of the experts. Analysis of the coefficients of these response curves showed that fire size, fire intensity, and the interaction between size and intensity significantly influenced the POC assessments. Using data for seven ancillary variables concerning the background and experience of the experts, it was found that agency had the greatest influence on the POC estimates. Random coefficient regression modeling was used to develop composite probability of containment models for the entire data set, each agency, and suppression with and without bucketing.