Evaluating crown fire rate of spread from physics based simulations to field data

Abstract

Wildland fire behavior models are commonly used to augment expert opinions, experiments and field observations by both the research and management communities. However, modelling wildfires is challenging in part due to complex set of coupled processes that drive the properties of a spreading wildfire. Further these processes occur over a vast span of spatial and temporal scales that further complicate the development and validation of models. Due to these complications there has been a variety of model types developed for a variety of specific applications. Regardless of the type and purpose of a model, well quantified fire behavior data from wildland fires and field and laboratory experimental fires are necessary for a variety of reasons including the calibration of empirically based models, the evaluation of physically based or theoretical models, and to provide model developers with potential areas to improve model performance by identifying inadequacies in the code. Here we utilize a compiled data set of crown fire rate of spread from Alexander and Cruz (2006) to evaluate published crown fire rate of spread predictions from two physics-based fire behavior models HIGRAD/FIRETEC developed at Los Alamos National Laboratory and the Wildland Urban Interface Fire Dynamics Simulator (WFDS) developed by the National Institute of Standards and Technology and the USDA Forest Service. Our preliminary results suggest that physics based models reasonably predict the crown fire rate of spread given the current data set. In addition we discuss the sensitivity of physics based models to a variety of parameters which likely influence crown fire rate of spread.