Saving forests with the Big Data fire brigade
By Tony Rehagen
Of all the deadly ramifications of climate change, wildfires might be the most immediate. Though humans start most fires, hotter and drier conditions have led to longer “fire seasons” and a higher number of blazes that spread farther and faster. The area affected by wildfires in the U.S. has doubled over the last three decades, swallowing billions of dollars in property damage and more than 3,000 lives each year.
It’s impossible to predict when and where a brushfire will break out. But what if there were a way to use technology and monitoring to predict where a current fire will move, enabling firefighters to clear its path and contain it? That’s the thinking behind WIFIRE, a system that was deployed as part of a fire-intelligence pilot program in September 2019 and used to fight all Southern California fires last fall.
The problem: The untamed path of fire
What makes a wildfire so frightening is, well, its wildness. Once ignited, brush fires can spread in different directions and at varying speeds depending on the type and density of vegetation, how damp that fuel is, and the weather: temperature, humidity, and wind speed. All those variables make blazes difficult to predict — especially in real time, when human lives and property may be in harm’s way.
“You can’t control the fire, but you can manage it,” says Ilkay Altintas, founder of WIFIRE Lab and chief data science officer at the San Diego Computer Center at the University of California San Diego. “You need real-time info on that fire as it changes and becomes more destructive. You have to understand current conditions and load all that data into a fire behavior model to see where it’s going to go.”
The solution: The Big Data fire brigade
Bringing all that data together used to take at least a couple of hours of work. But when facing a fleet-footed fire, an hour is an eternity. Altintas and her team cut into that delay by creating a system that takes data from fire-management and fire-science sources and quickly turns it into actionable real-time intelligence.
“The more data sharing we have, the better we can solve other climate-related issues.”
A cyberinfrastructure of high-speed fiber optics and fixed wireless networks, WIFIRE feeds weather information to Altintas’s San Diego supercomputer from hundreds of remote stations across the U.S. The system’s artificial intelligence software also examines hi-res satellite images of the terrain and vegetation surrounding a fire to account for flammability. WIFIRE plugs all of this data into a modeling program that enables firefighters to predict how an existing fire will spread. It also produces possible scenarios that will allow officials to spot and prepare for potential future blazes — all with the click of a button.
The challenge: An unquenchable thirst for data
As with any AI, WIFIRE’s algorithm is only as effective as the amount of data it can access. WIFIRE is financially supported by the Los Angeles city and county fire departments, the Ventura County Fire Department, and the Orange County Fire Authority. But in California alone, there are dozens of fire-fighting agencies, organizations, and institutions that collect information that could be used to hone WIFIRE’s models.
“Our challenge is to create convergence between all those groups and open them up,” says Altintas. “And there are so many things we could do to help fire scientists solve these problems.”
The big picture
Fires aren’t going to be completely prevented — in fact, to some extent, wildfires are necessary for the environment’s growth and replenishment. But with the help of WIFIRE and other forms of AI, firefighters and scientists in Southern California can react quickly and minimize the human toll of a blaze.
Altintas hopes more agencies across the globe adopt WIFIRE’s technology — and not only to fight fire. “The more data sharing we have, the better we can solve other climate-related issues,” she says — including floods and mudslides. “The future is fighting all hazards, not just fire.”