Scientists devote careers to getting wildfires’ number
After a day wrestling with the mysteries of fire, Francis Fujioka turned on the TV and watched Southern California burn into the night.
A news feed showed a line of fire climbing a hillside. It was small, moving slowly, not a threat for the moment. But Fujioka saw what viewers missed: The flames were bent in a way that suggested a tiny weather pattern had formed within the fire, and that would influence what happened next.
“I’m watching this and I’m thinking to myself, ‘There’s no way in my career -- what’s left of it -- that I’m ever going to be able to model that on a computer,’ ” said Fujioka, 61, a research meteorologist at the U.S. Forest Service’s Fire Laboratory in Riverside. “Predicting a fire’s growth is really a difficult business.”
For nearly three decades, Fujioka has fought fire not with water but with mathematics in an attempt to predict what is inherently unpredictable.
Weather, fuel and topography are the ingredients of wildfires like the ones that savaged Southern California recently. The study of the myriad ways they influence each other is an inexact science.
A fire’s predicted path of destruction using today’s forecasting tools often has little in common with what actually burns.
Fujioka heads a little-known team of researchers working to translate the physical environment of wildfires into mathematical language that, when processed by powerful computers, can predict fire behavior far more accurately than is possible today.
“While 50 years ago people could empirically describe the conditions in a blow-up fire, now we have the computational power to get inside that,” said David Weise, a Riverside lab scientist who blends forestry with statistics.
“These rules of thumb that we’ve been using for 50 years -- how do they actually play out? . . . There are still a lot of times that we get surprised by a fire.”
California has been a petri dish in the field of fire science ever since Gifford Pinchot, the first chief of the U.S. Forest Service, ordered studies of Southern California wildfires in 1898.
Many wild-land firefighting tools and procedures were either developed or refined in the state -- from remote weather stations and firebreak systems to the use of aircraft to drop flame retardant.
Mathematical computer modeling emerged in the 1970s, but its accuracy is limited by the distance between weather stations. By comparing how fires were predicted to burn with what actually burned, Fujioka and his colleagues hope to use statistics and probability to fill the gap.
Their study of last year’s Esperanza fire in Riverside County, which killed five firefighters, found that computer reconstructions of the blaze’s path based on available weather data diverged from reality by as much as 80%.
The wind direction didn’t match up with the fire’s spread, suggesting the influence of small weather patterns out of reach of the nearest monitoring stations.
“We understand more about hurricanes than we do wild-land fires,” said Shyh-Chin Chen, a research meteorologist at the lab. “We don’t understand the very small stuff that happens within complex terrain.”
Many of Chen’s neighbors in Rancho Bernardo lost their homes in San Diego County’s Witch fire. His house was spared -- the fire jumped over it.
Chen was relieved, but his curiosity lay elsewhere.
He took pictures and video of burned areas on both sides of Interstate 15.
“I wanted to know how it could jump over the interstate,” he said. “I couldn’t make sense of it.”
Mountain topography creates wind patterns that are notoriously hard to predict. The effect these microclimates have on a fire will determine its behavior -- as does weather created by the fire itself.
Early on in Orange County’s Santiago fire, crews working to cut a fire break on a ridgeline above Foothill Ranch were forced to pull back.
In less than a minute, flames raced down toward homes where engine crews kept them at bay.
Suddenly, the neighborhood was engulfed in a blizzard of hot, black smoke and white ash. The wind generated by the fire bent mature palm trees.
For the moment, one part of Foothill Ranch was trapped inside an atmosphere all its own. A few blocks away, it was sunny and calm.
The researchers’ goal is to predict a fire’s path on such a fine scale and in real time so the information can be used by firefighters on the ground.
Instead of forecasting what’s expected over miles of forest, a computer model might one day be able to predict wind patterns in a 100-meter slot canyon.
“The evolving nature of these fires is so quick -- especially under Santa Ana conditions. If they could get something that was even more than 50% accurate, that would be a real asset,” said Dick Mangan, a wild-land fire consultant who worked as a Forest Service fire manager for more than 30 years. “It’s a great concept. We need to keep pushing the envelope.”
--
