study explores real-time prediction of wildland blazes
Frontline fire fighting could soon go high tech. In the not-so-distant
future, analysts using supercomputers may be able to send real-time
maps and predictions of a wildfire's next moves to wildfire
incident management teams hundreds of miles away. That crucial information
NCAR’s Janice Coen. (Photo by Carlye
could be passed on to Palm Pilots and other wireless devices in
the hands of frontline firefighters deciding how best to battle the
With millions of dollars in property lost and millions of acres in
the American West burned by devastating wildfires in recent years,
researchers at the University of Colorado at Denver, the University
of Kentucky, Texas A&M University, Rochester Institute of Technology,
and NCAR are working together to develop state-of-the-art information
technology tools and apply them to wildland fire.
This team has been awarded $2 million by NSF to develop a dynamic,
data-driven system that will predict wildfire behavior and progression.
The four-year project will use the most recent advances in computer
speed and power, high-speed information networks, satellite and sensor
monitoring, mathematical theory, and meteorology to develop tools
to warn firefighters about where a fire may go and what sudden changes
in wind or fire behavior might occur.
Called the Data Dynamic Simulation for Disaster Management, the effort
is part of NSF's Information Technology Research Program. The
team is headed by CU-Denver mathematics professor Jan Mandel, who
will work with a coupled weather and wildfire computer model developed
at NCAR to build a software system that will use data from
the fire scene to determine
wildfire-spread scenarios and probabilities.
The grant will allow the team to create a system where multiple sensors
placed around a wildfire will continuously gather and transmit data
on temperature, wind direction and speed, and the moisture in grass
and sticks. A supercomputer will use the mathematically based wildfire
model to continuously send
maps and forecasted fire locations to the front lines in real time, allowing
a fire manager to see minute-by-minute predictions or anticipate
where fire growth will occur along the fire line.
Among its potential benefits, the system could allow fire managers
to plan the most effective and efficient actions—for example,
by foreseeing situations where weather, the terrain, fuels, and winds
created by the fire would combine for a fire that would grow rapidly
unless more resources are used to stop it early on. The system might
also identify situations where wildfires can be allowed to spread
harmlessly under controlled conditions for hazardous fuel reduction
and natural resource benefits.
"In the past, running a model on a computer meant starting
a simulation and then waiting for the results," says Mandel. "It
is time to change the way scientific modeling is done. In a
movie, you may see a computer on a starship, and the computer takes
into consideration new information as soon as it comes. This is how
computers work in the imagination of movie directors, and this is
how people expect computers should work. Our project will help make
this a reality."
Wildland fires are driven by complex phenomena that are poorly understood.
Scientists at NCAR have a history of coupling numerical regional
weather simulations with fire-spread models to advance the understanding
According to NCAR scientist Janice Coen, "There are many things
about wildfires that aren’t understood scientifically. There
are also many technological challenges involved in simulating phenomena
that change very rapidly, and in quickly transmitting data from remote
locations into a model running many possible scenarios on a supercomputer
far away. You have to deliver this information rapidly, reliably,
and in a meaningful way—with images, not words—
through secure means to people who may be far from telephones."
The efforts of the team will eventually be put to the test: in four
years, they are scheduled to take the technology to a real wildfire.