Smokey Bear certainly had his work cut out for him last year. In what was one of the worst wildfire seasons in U.S. history, more than 92,000 wildland fires burned over seven million acres in the US, according to the National Interagency Fire Center. This year, dry conditions that are expected to prevail in the Southeast and West promise to keep him busy once again.
There may be good news on the horizon, however, as scientists work to develop computer models that will accurately predict when, where and how wildfires spread. Researchers at the Lawrence Livermore (LLNL) and Los Alamos (LANL) national laboratories have been using hightech virtual fires to learn about the real ones. "One of the interesting things about this type of work is that, because were doing it in the virtual world of a computer, we can burn the same area over and over and over," says Michael Bradley, a senior atmosphere scientist at LLNL who is coordinating the labs effort.
Firefighters normally use weather information and fire behavior models to figure out how to control blazes, but sometimes these simple models dont work so well. Last year, for example, a prescribed burn unexpectedly went out of control at Los Alamos (see sidebar below, right), incinerating thousands of acres and destroying hundreds of homes.
But researchers have been able to learn from such catastrophes. By inputting data about a 1996 wildfire near Malibu, California that took firefighters by surprise, they have used a computer to recreate the blaze and see if they could predict how it spread. "After doing the simulation and talking to firefighters who were actually on the scene, they confirmed that our model basically duplicated the fire behavior for this fire," says Bradley.
The computer models work by taking into account detailed maps and satellite images of forest terrain, the types of fuels (both living and dead vegetation) to be found in a particular forest, and atmospheric information such as wind and moisture conditions. Using the worlds biggest computer, these elements are combined with complex models of how the air turbulence created by a fire interacts with it.
One of the complications faced by firefighters as they do their jobs is that big fires create their own microclimates. As the heat created by the fire is released, oxygen gets sucked in and the climate conditions within the fire can be dramatically different from whats going on outside it. The new computer models can incorporate this information.
Scientists say theyll eventually be communicating real-time information gleaned from the models to fire crews on the field via satellite. This will help them make quick decisions and hopefully allow them to stop fires before they rage out of control.
Now that theyve studied past fires that have already burned, Bradley and his colleagues are currently building computer models that will predict future wildfire behaviors. For prescribed burnsthe fires intentionally started by forest managers to keep forests healthy and safethese models are almost ready to work. "This type of tool, once it reaches maturity and is able to predict fire behavior accurately, will help firefighters plan their fire fighting activities in a much safer way," says Bradley.
Elsewhere on the web
Fire detection around the world
Historical US wildfires
Wildfires by state during 2000
Worldwide fire events 2001
National Fire News
NOAAs Fire Weather Information Center
Satellite observations of forest fires
Fire weather forecasts
USDA Forest Service
"Letters from Elizabeth Millers dad, who fights fires while flying a helicopter"