USA — The Moose Creek district ranger for the Nez Perce-Clearwater National Forest had spent five days trying to manage the lightning-sparked Johnson Bar Fire with direct attack.
Dropping buckets of water from helicopters was not working. Firefighters were building lines as fast as they could, but they couldn’t catch the fire.
Conditions were becoming dangerous. The Nez Perce didn’t need another firefighting tragedy. Two summers earlier, it had lost 20-year-old Anne Veseth to a falling tree. Her memory was front and center as Hudson decided to pull the firefighters off a direct attack about 20 miles northeast of Grangeville.
Hudson needed to understand what he was up against. He’d called Byron Bonney, a retired fire staff officer on the Nez Perce forest. Bonney knew large fires. When he arrived Aug. 7, the two men developed a long-term plan and entered it into the Wildland Fire Decision Support System, a Web-based fire modeling program designed by scientists to help managers make tactical and strategic decisions.
Hudson and Bonney plugged in data. The model gave them a good sense of how the fire might spread over the next few days. Combined with the veterans’ knowledge of the forest, the model helped identify a network of roads, trails and ridges that could be used to build containment lines or conduct burnouts.
THE FIRE MODEL
Fire modeling tools rely on information from the National Weather Service, detailed maps of fuel layers in forests and other factors. They estimate how fast the fire will burn and how it will spread in relation to vegetation, trees, homes and other properties.
For Hudson and Bonney, the WFDSS program calculated the fire’s potential spread within a 26,000-acre planning area where firefighter actions could slow or stop the fire. The modeled fire behavior informed them on the potential effects on threatened values: homes along the Selway and in nearby Lowell, a rustic lookout, the historic Tahoe Trail, habitat for fish, and timber and replanted forests.
“Once the fire has escaped initial direct attack, the goal is to protect the values at risk and contain the fire,” said Hudson.
Hudson called in the Incident Management Team, an interagency group that manages large fires. The IMT set up camp Aug. 8 at the Kooskia airport, 20 miles west of the fire.
Winds were pushing the fire north. Winds were gusting 35 mph on the ridges, triggering an Aug. 12 flare that doubled the size of the fire in one day. People living in the 30 homes along the Selway already had been evacuated.
The IMT kept the fire from spreading and establishing itself on the other side of the river. The WFDSS analysis was helping guide their decisions.
With the fire spreading down the slopes of the Selway and Middle Fork Clearwater River, the managers decided to perform burnouts using the rivers as barriers.
It worked. Welcome rains helped tame the fire. Firefighters were able to establish containment lines.
Fighting the Johnson Bar Fire cost $12 million.
THE FIRE SCIENTIST
Mark Finney works in Missoula, Mont., for the Forest Service’s Rocky Mountain Research Station and the Missoula Fire Sciences Laboratory, overseeing a team of scientists who research fire behavior, soil heating and the effects of fire on the ecosystem.
Finney has a doctorate from the University of California, Berkeley in wildland fire science. He studied prescribed fires and their effects on coastal redwood forests. He fought fires during his undergraduate summers at Colorado State University, helping pay for his studies. Finney spent two years as a fire ecologist studying redwoods at Sequoia National Park, where he’d watch wildfires during the day, and then head back to his computer to write code to model the movement of fires he’d witnessed.
He understood early in his education in fire ecology that it was important to get fire back in the ecosystem. If he could figure out a way to predict which fires would threaten an area and which fires could burn without harm, it would help people make better decisions.
Finney arrived in Missoula in 1993 and did the best he could with the computing power of the 1990s. To predict a fire’s spread, scientists need to know about fuels and vegetation. Where are the grasses? What are the different shrubs? What trees are around? What is the density? Where are the valleys, ridges and streams?.
Finney and his colleagues began their modeling by tracking fires in Glacier National Park in Montana and Yosemite National Park in California in 1994. The park rangers had geographical information systems and had accumulated data on fuels and vegetation patterns.
“I went to the fires when they began and went back to my computer model and punched in the information and faxed a copy of the projections about the fire’s spread to the fire management officer at Yosemite,” said Finney. “He did not know what to make of it.”
A few days later Finney got a phone call.
“The fire did just what you predicted it would,” the fire management officer told Finney.
The next day, Finney flew back to California and helped map the fire over the next few weeks.
He did the same thing at Glacier. He drove to the park in early July when lightning struck deep in the forest. Finney’s computer model, which he called FARSITE for Fire Area Simulator, showed the fire manager what resources were threatened and what firefighters could do about it.
His prediction was the opposite of how the fire teams were thinking the fire would spread.
It was David Mihalic’s first year as park superintendent at Glacier. Based on Finney’s calculations, Mihalic decided to let the fire go and turn it into a “prescribed natural fire.” The park was thick with smoke. Many people thought that Mihalic was crazy. The Howling Fire burned 15,000 acres; it has been written up in fire textbooks.
“It was a significant event for our whole program,” said Finney. “People became more open to the idea that we don’t need to suppress every fire.”
THE NEW FIRE RESEARCH
FARSITE morphed into WFDSS. Two decades of faster computing later, WFDSS has proved invaluable.
But Finney and his colleagues have come to learn the weakness in the models. “There is a lack of understanding of how fire spreads,” said Finney.
Finney knows the ingredients necessary to spread a fire: heat, fuel and oxygen. But the physics responsible for wildfire spread are not well known.
So in his lab, Finney and his team are working on understanding how fire spreads. They can control certain factors, such as wind and humidity, but they had problems finding just the right fuel beds. In the past they used wood and grass. More recently they have been cutting cardboard into specific configurations and finding that they are able to eliminate the variability in the fuel beds.
Over and over again they set fire to various fuel beds to record the action of the flames.
“Finally, we were able to see things in the flame structure and behavior that people never recognized as a cause of a fire spreading,” said Finney.
The scientists have found regular and predictable patterns in the fire caused by “buoyant instabilities” as heat creates a buildup of low-density fluid.
“Flames and gases produced by combustion are very hot and low-density,” Finney explained. “When it heats up, it rises like a hot air balloon.
“Finney wanted to see what flames look like during a forest fire. He’s been using digital cameras placed at various angles at the edges of fires.
“When you look at a fire you see a sawtooth appearance – there are peaks, troughs and jagged edges. These peaks and troughs are caused by the buoyant instabilities we identified in our lab experiments. It is the pattern of flame behavior that is responsible for its spread,” he said.
He believes this finding will lead to better predictions and better firefighter training and techniques. Things are moving fast, but the new information has not been incorporated into the models.
“For the first time in history,” said Finney, “we can link behavior we see in spreading fires to the physical mechanism of the spread.”