Five existing fire models, both experimental and theoretical, did not adequately predict rate-of-spread (ROS) when tested on single- and multiclump fires in oak chaparral in Arizona. A statistical model developed using essentially the same input variables but weighted differently accounted for 81 percent ofthe variation in ROS. A chemical coefficient that accounts for effects of fuel chemistry on ROS is applied to the model. The model provides usable guidelines for predicting fire spread in Arizona oak chaparral.

[This publication is referenced in the "Synthesis of knowledge of extreme fire behavior: volume I for fire managers" (Werth et al 2011).]