Fire Behavior Models: Successes and Required Improvements
Mark A. Finney
Rocky Mountain Research Station
Fire Sciences Laboratory
Fire modeling has historically been directed toward calculating fire behavior in one dimension under uniform and constant conditions. Although these conditions are rare or limited in nature, models for surface fire spread, spotting distance, and crown fire transition have been used by fire management organizations to assist with decision making on a wide variety of practical fire behavior problems. The one-dimensional models have also been extended to two spatial dimensions for fire behavior simulation over time based on empirical models for fire shape. Despite the many assumptions of independence and uniformity, these models have found acceptance for practical uses because:
1) they have attainable input requirements that are robust given the uncertainty in estimating or measuring them in the field,
2) provide understandable relationships between the main environmental input variables and the predicted fire behaviors, and
3) produce results that are consistent with experience given the uncertainty in the inputs and observations.
Despite their wide utility, however, these models are poorly suited to explaining or predicting fire behavior in a number of common and dangerous situations. The following fuel or environmental conditions are examples of conditions that are not addressed, or addressed only in a restricted sense. They involve rapid transitions in fire behavior, abrupt thresholds in fire activity, strong feedbacks between the fire behavior and environmental conditions, or dependencies on previous states of fire behavior:
1) Horizontal and vertically discontinuous or clumpy fuels (e.g. pinon-juniper woodlands, multi-strata forests, crown fire and transition)
2) High frequency variability in winds causing non-steady behavior (e.g. 10-1 to 100 Hz)
3) Interaction of multiple fire fronts (e.g. mass ignition, junction zones)
4) Flame attachment to slopes and fire moving through constricted canyons
More advanced modeling of the combustion and heat transfer physics is required to address these problems. The role of convection and radiation in the various situations will need to be considered. Experimental data on some of these problems is possible to obtain in laboratory conditions that can help drive model development. Other problems must be approached mainly through modeling with occasional comparisons with field observations. Even with highly sophisticated and improved models, however, field applications will always require simplification to accommodate the uncertainty and error in available input data.