Extreme value analysis of forest fire behavior parameters

![CDATA[This paper presents the results and discussion of the application of extreme value analysis to forest fire (or wildland fire or bushfire) parameters for the determination of fire protection design. Some fire behavior models in the literature for predicting rate of spread and flame dimension have been used to determine design conditions for fire protection practice. However, the statistical approach to the selection of design bushfires has not been widely accepted. Some questions as how to apply the extreme value analysis method to the modeling parameters remain unanswered. Attempts have been made in the current study to verify and compare a number of methods in the application of the generalized extreme value (GEV) analysis to two forest fire models, namely the MacArthur model and the Project Vesta model, that have been developed for Australian landscape. Four scenarios involving different vegetation classes and weather conditions are used as case studies to demonstrate the validity or otherwise of the methods. It is found that the method of applying the GEV analysis to the output parameters is more appropriate than applying to the input parameters of the models. It is also found that from safety point of view Project Vesta model tends to produce more conservative estimate of fire behavior parameters than MacArthur model for the limited scenarios investigated.]]