Limitations of Fire Behavior Models

Computer models are considered valuable tools to assist fire managers with the decision making process. However, quantification of natural processes related to changing weather patterns is difficult at best. Historically, computer generated models have had difficulty simulating several important conditions related to fire behavior. Four of these very influential, but elusive quantifications are:

1. Horizontally discontinuous fuels
2. High frequency variability in winds
3. Interaction of multiple fire fronts
4. Flame attachment to slopes and fire moving through constricted canyons (Finney 2003).
5. Behavior of crown fires
6. Conditions that might lead to fire whirls.

Recent advances in personal computers and GIS technologies are creating opportunities for natural resource agencies to acquire and develop more sophisticated fire behavior models. Still, supercomputers designed to process trillions of calculations per second become bogged-down in the minutiae involved in predicting fire behavior via weather-related variables.

It is understandable that the inherent dynamic nature of atmospheric conditions presents unusual difficulties related to both data collection and input, and the computational downtime associated with new or changing data sets in the fire behavior modeling process. Following are discussions of the inherent difficulties involved with incorporating weather and topographical effects into fire behavior models.

Weather effects in models

The effects of certain weather conditions on wildland fires in ideal situations are fairly well understood. Localized on-site observations of current conditions have served to assist firefighters in predicting short-term fire behavior for many years. Models have incorporated pertinent weather and environmental variables with varying degrees of implementation and utility. Accuracy of forecasts is, however, plagued by the inherent difficulty that arises from quantifying the dynamic nature and relationship of weather conditions to fire behavior. Oftentimes, weather data is either too local for application to a large fire, or too remote for predictions at a localized situation.

It is unrealistic to expect time-sensitive weather predictions to be generally available with such specificity as to address the multiple conditions that may be present at, or near, a wildfire. While local topography and fire/weather relationships have plagued the utility of regional forecasts, the advent of GIS and advancements in fluid dynamics modeling software has shown promise in the realm of scalable weather/topography interactions (Forthofer et al. 2003)

Topography effects in models

Topographical influences incorporated by fire behavior simulation and prediction models have become fairly advanced. The advent of GIS and satellite-based terrain mapping has streamlined the data entry process and provided a high degree of error control. Such in-depth topographical mapping provides reliable probabilities concerning fire spread and direction. Alternatively, the downside of such data intensive methods is that while accurate, they often serve to complicate modeling intentions and diminish the feasibility of such applications in time sensitive settings.

Empirical laboratory testing has allowed operational models to be formulated based on effects generated by a gamut of environmental variables. These models function fairly reliably in low wind, flat terrain situations. Alternatively, physical models have been developed in response to erratic fire behavior that occurs in other than “ideal” conditions. These models base their predictions on physical combustion process formulas designed to mathematically compute the probable fire behavior in a variety of situations. The major drawback to these models is the immense computational downtime. These models are so data driven and relational calculation intensive that supercomputers fail to provide timely results (Achtemeier 2003).