Today we may take it for granted that tools are available that can estimate how a fire, unplanned or prescribed, will spread across a landscape. It is not an exact science because there are far too many variables than can realistically be accounted for, at least with the technology available today. But in 1972 when Dick Rothermel and others developed the Forest Service’s first quantitative, systematic tool for predicting the spread and intensity of forest fires, it introduced a new era in fire management. And surprisingly, it is still the main tool being used today. Many researchers have produced alternative models, but none have made it into the hands of firefighters on a widespread basis.
After Mr. Rothermel developed the mathematical model, others used the information to make the concept more user-friendly and to analyze complex scenarios. Behave, software burned onto a custom made chip in a hand-held Texas Instruments 51 calculator, and later BehavePlus for personal computers, became must-have tools for fire behavior analysts. FARSITE added the ability to predict spread across variable terrain, vegetation, and weather. Rare Event Risk Assessment Process (RERAP) estimates the risk that a fire will reach a particular place before it dies. FireStem estimates tree mortality based on fire behavior and intensity. And there are many others.
When Mr. Rothermel began researching the behavior of wildland fires, he had just been downsized from a shuttered Department of Defense program that had been attempting to develop a nuclear-powered airplane.
Below is an excerpt from an excellent article by Gail Wells for the March, 2008 edition of Fire Science Digest, a publication of the Joint Fire Science Program.
[Jack] Barrows, [the first director of the fire laboratory in Missoula when it opened in 1960], went looking for researchers. He learned that General Electric was closing a laboratory in Idaho Falls where engineers had been working on a defense project to develop a nuclear-powered airplane. The government scrapped the program in 1961, and a handful of highly trained engineers and scientists were suddenly up for grabs.
“GE wanted to see that we got as good a placement as we could,” Rothermel recalls. “So we all wrote resumes, and Jack got hold of these, and he said it was like a Sears and Roebuck catalog of people.” Barrows hired four of the GE scientists: Hal Anderson, a physicist; Stan Hirsh, an electrical engineer; Eric Breuer, a technician; and Dick Rothermel.
Their hiring represented a departure from Forest Service custom. Up until that time, fire research had been pretty much the domain of foresters, who are used to looking at their work through the lenses of biology and silviculture. Gisborne was a forester; Barrows was a forester. But Barrows recognized that fire is a physical process, and that physical scientists and engineers could contribute much to the emerging science of fire behavior.
Rothermel, then barely into his 30s, was glad to join Barrows’s staff. He had a bachelor’s degree in aeronautical engineering from the University of Washington. During the 8 years since he’d graduated, he had worked in the engineering of nuclear systems in Albuquerque and then in Idaho. (Rothermel later went on for a master’s degree in mechanical engineering from Colorado State University.)
“I had the option of staying on [at GE] and working on a lot of programs, but with the cancellation of the atomic-powered airplane, nothing sounded that appealing,” he says. “And then I heard about this laboratory, and they said they had two wind tunnels and a combustion lab where you could control the atmosphere, temperature, and humidity. I thought, “Wow, that’s an opportunity!” Rothermel worked with Hal Anderson to get the new lab’s equipment calibrated and running smoothly. Then they began a set of experiments in the wind tunnel and combustion chamber, testing the effects of wind and moisture on various fuels and determining how fast a fire would spread under different conditions.
Given their training, it made sense to Rothermel and Anderson to approach the task as an engineering problem. Says Rothermel: “The idea was, if we could develop a way of describing the fuels, the weather, the topography, and something about the fire, and be able to put that into what we call a mathematical model, and if we described all these things properly, the model would integrate it and produce answers. It would tell you the resulting fire intensity, rate of spread, flame length, these sorts of things.”
Rothermel, Anderson, and Bill Frandsen, another physicist on the project, adapted an approach developed by an early Forest Service fire researcher, Wally Fons, which turned on the concept of conservation of energy. A fire spreads by igniting a series of little fires in the fuel ahead of it. The ignitions are driven by convection, radiation, and conduction. Even if it’s unknown which mode is operating in a given instance, the rate of heat transfer can be measured. The researchers reasoned that if they knew how much fuel was ahead of a fire, how big and how densely packed the fuel particles were, and how much moisture the fuel contained, then they could figure out how much energy would be needed to transfer enough heat to bring the fuel up to the ignition point. They could then calculate the rate of ignition that would carry the fire as it spread. The model would also have to account for the critical variables of wind speed and slope of the ground.
Because of the limitations of wind tunnels and combustion chambers, the model is forced to make certain assumptions that don’t hold in real life. For example, it assumes that the fuel is continuous and evenly distributed and burns uniformly. It further assumes that the fire is carried primarily by dead plant material and that only moisture will stop it.
The Rothermel model “describes very well a fire burning in a field of wheat,” says Bret Butler, a mechanical engineer at the Fire Sciences Lab whom Rothermel hired in 1992. “As you get further away from that uniformity, the less accurate it becomes.”
More significantly, the researchers had no basis for modeling the endless spatial variability that actually exists in a forest. So there was no way to simulate a fire’s movement through clumpy, discontinuous trees and shrubs. There was also no way to model a crown fire, one that leaves the surface and moves up into the crowns of trees. These were significant and universally acknowledged shortcomings.
Fire research scientists throughout the world are working on developing more accurate surface-fire spread models, but at this point all of them are too complicated to be used in an operational system. The beauty of Rothermel’s model, says Butler, “is that it’s simple—it can be run quickly with a low-capability computer.”
(end of excerpt)
What made me think of Mr. Rothermel was a graphic distributed on Twitter today by the National Weather Service. It is a fancy, colorized version of the figure in his 1972 paper that depicts how heat is transferred in a fire.
But of course Mr. Rothermel’s contributions are far more complex than this graphic.
Below is a screenshot from his paper where he describes Propagating Flux, just one of many elements of his mathematical fire spread model.
And here is his summary of equations for the model:
Epilogue 1: The current administration has expressed a desire to zero-out the budget for the Joint Fire Science Program, the organization that published the 2008 article.
Epilogue 2: Mr. Rothermel was one of the 655 attendees at the Fire Continuum Conference in Missoula last month.