As devastating wildfires rage throughout the American West, fire agencies across the region are turning to cutting-edge technologies from supercomputers generating near-real-time fire maps to fireball-dropping drones to enhance the way they respond to these disasters.

Fires are still won and lost through grueling work on the field and relatively low-tech tactics such as burning strategic areas close to the edge of an active fire to slow or stop its progress and spraying water and foam to slow the blazes. The best tools are often simple ones: saws, bulldozers, water hoses.

However, with climate change contributing to more frequent, more severe and larger blazes that threaten humans, infrastructure and natural resources at unprecedented levels, response and suppression methods need to evolve. Innovations are providing firefighting crews with extra tools to detect, contain and even extinguish fires faster and with greater safety.

Spotting fires faster

One common problem with wildfires is that they are spotted too late and already widely spread out. Currently, most fires are reported by civilians or airplane pilots. That spotty approach, however, can lead to fires that go on for hours or, sometimes, days before resources are mobilized.

A Santa Fe-based startup called Descartes Lab is trying to cut into that delay by training its artificial intelligence to detect budding blazes. The company’s A.I. software scours images delivered every ten minutes from two U.S. weather satellites, in search for hotspots such as smoke or shifts in thermal infrared data that could mean a fire may have broken out.

From there, several algorithms—each looking for different properties of a wildfire—are run to determine if a fire is present. If the algorithms reach consensus, the system sends a text alert to the state's fire managers, providing the longitude and latitude coordinates of the blaze and how to get there.

"That's really helpful, particularly at night or when you're on top of a peak 20 miles away and it's hard to determine where the fire is actually on," says Donald Griego, New Mexico state forestry department bureau chief.

Descartes is now testing its wildfire detector by sending alerts to the department’s officials. Initial results are promising: since its launch two years ago, the system has spotted more than 6,000 fires, some as small as ten acres and in as little as nine minutes.

Similar early-detection technology is being tested in California’s wine country. In Sonoma County, several local agencies began installing a system of tower-mounted cameras as part of a system named ALERTWildfire .

Surreal image under the plume of the #CaldorFire from our Leek Springs camera pic.twitter.com/imsZclDfJ2

— ALERTWildfire (@AlertWildfire) August 17, 2021

The devices scan and photograph fire-prone areas and every ten seconds send images to the county’s fire emergency center, where they are closely observed by dispatchers. Cameras are also linked to an A.I. software that compares all incoming images with historical pictures of the same locations. If anything looks out of place, emergency crews are immediately notified and sent to verify first-hand.

In the weeks since the system was fully activated in May, it has beat civilians’ reports by as much as ten minutes—a small time frame but one that can mean the difference between a tiny cluster of flames and a runaway wildfire.

Computing a blaze’s path

The one thing that makes wildfires so dangerous is their wildness. Once ignited, they can spread at varying speeds and change direction in a matter of seconds, making blazes tricky to predict. Most agencies do it manually, looking at the weather, terrain and the dryness of the vegetation. But coming up with calculations can take up to a day—an eternity when facing a fleet-footed fire.

Now, fire agencies are getting some help from a powerful new tool. FireMap , an artificial intelligence-based platform developed by WIFIRE Lab , a spin-off of the San Diego Supercomputer Center (SDSC) at the University of California, San Diego, can create in minutes a predictive map of the fire’s expected trajectory.

The system builds on a combination of deep learning techniques to crunch real-time data about weather, topography, the dryness of vegetation and more from satellites, on-the-ground sensors, utility cameras and, more recently, a fixed-wing aircraft outfitted with infrared radars.

“We bring all this information together and feed them into models that can tell us where the fire will be, its rate of spread and its direction for up to six hours,” explains Ilkay Altintas, chief data scientist at SDSC and WIFIRE Lab’s principal investigator.

Those predictions help incident commanders make critical judgment calls, such as where to send their limited firefighting personnel and whether to issue evacuation orders, according to Ralph Terrazas, a battalion chief at the Los Angeles Fire Department.

“It gives us, as responders, a fundamental edge to make better, quicker, and more educated decisions,” says Terrazas, who first came across FireMap in 2015 and has now made the system an integrated part of his department’s fire protocols.

Today, LAFD and a number of other fire departments across Southern California routinely put the fire-predicting system to work for fighting the area's increasingly dangerous wildfires; about 130 other groups are testing the technology.

The power of drones

From portable quadcopters to fixed-wing platforms, drones are showing they have key advantages over conventional human-piloted firefighting aircraft.

Airplanes and helicopters used to survey wildfires and drop retardant can’t fly after dark and in smoky conditions, or in too cramped of a space. Flying over raging fires also puts pilots and crew at risk. About a quarter of all wildland firefighter fatalities are related to aviation, according to the U.S. Forest Service.

As fires raged across the West last summer, two dozen remotely controlled devices equipped with thermal imaging cameras peered through the smoke, capturing high-resolution footage and other real-time data that informed responders in their suppression efforts.

“The drones provided an opportunity to gather critical information for decision making at a time when we wouldn’t be able to do it any other way,” says John Kennedy, the director of one of three branches that battled the Grizzly Creek wildfire in Glenwood Canyon, Colorado, last August.

Joining the effort was Kelly Boyd, a drone specialist with the Unaweep wildland fire module, a seven-person crew that’s called to help on incidents during fire season across the Upper Colorado River region. He brought along the Ignis system, a funnel-shaped device developed by Drone Amplified , a Nebraska company, in partnership with the Department of Interior, which mounts to the underside of a drone and can drop 450 small incendiary balls in about four minutes.

Known as dragon eggs, these ping-pong-ball-like spheres are filled with two chemicals that react after they hit the ground, starting what firefighters call prescribed burns—small fires purposely set in the path of an approaching fire to deny it fuel to spread.

At the Grizzly Creek fire, the eggs Boyd dropped ignited the canopy along the southeast edge of the 32,000-acre blaze. The flames charred dry timber and foliage, creating a scorched barrier that connected two other containment lines, making for a much more effective barrier to contain the fire.

Airborne firebombing “is swift, efficient and versatile,” says Boyd, who notes unmanned aerial ignition also helps reduce the risk of using helicopters above dangerous terrain and often can be more precise.

Asked what he expects to see in the coming years, Boyd suggests high-altitude drones that cruise above fires for days to send back a continuous stream of video, and remote-controlled aircraft that carry in supplies to help douse flames long after sunset, when manned vehicles are grounded.