Supercomputer to Help Fight Wildfires
Inspired partly by the deaths of 14 firefighters, scientists are trying to use the power of supercomputers--once used only to simulate nuclear bomb blasts--to help foresters battle wildfires.
“The ultimate goal is to save lives and property,” said Rod Linn, who is doing the research at Los Alamos National Laboratory’s theoretical division.
Rugged terrain and capricious wind can combine to make a wildfire veer and turn unexpectedly.
Linn and his mentor, laboratory fellow Frank Harlow, want to prevent another tragedy like the one on July 6, 1994, in which 14 firefighters were killed on Colorado’s Storm King Mountain by wind-driven flames that suddenly raced up a steep hillside.
They plan to use the laboratory’s supercomputers, which can run thousands of complex equations simultaneously, to predict how fires will behave.
“We can’t predict when certain leaves or branches will catch fire,” said Linn, but they do want to be able to predict in general how a particular fire might spread.
Originally, the laboratory’s supercomputers were programmed to simulate the effects of nuclear weapons explosions. Researchers now are applying those principles to other problems.
Harlow said the computers have helped develop better internal combustion engines, improved metal casting and designed better ways to extract geothermal energy, and can be used to study economic trends.
There is a computerized fire prediction model available now for firefighters, called Farsite, that weighs general weather and terrain variables. But it is compatible with personal computers, not the big high-speed supercomputers, and is not as sophisticated as the Los Alamos team has envisioned.
Once their program is refined, firefighters could send information about a wildfire to the supercomputer through any available telecommunications link.
Harlow said the supercomputer also could tap a database to retrieve a three-dimensional map of the burning area.
The computer would then use factors including vegetation, moisture content, and recent and long-term weather patterns, to create a model or virtual environment.
Anyone directing firefighting operations could then run simulations on that model to test different strategies.
So far, the research has produced simulations in real time, meaning it takes a full hour to calculate one hour of simulated fire behavior. The next challenge is to accelerate the process.
“These computers can pull up all the data and commence calculating, and then calculate it much faster than the fire is actually progressing,” Harlow said.
“The goal is five to six times faster than real time,” Linn said.
Their program probably won’t be ready for use for another year at the earliest. But Linn, who is doing his work through New Mexico State University’s mechanical engineering program, said he hopes to have a working model by December.
Harlow and Linn got a close-up look at fire behavior in April, when a 17,000-acre fire burned to the boundaries of the Los Alamos lab.
“The jury’s still out on whether we can make this succeed or not. But if there’s anything we can do that really helps mitigate fire danger, we’ll sure do it.”
