AI Program Now Being Used to Spot Wildfires in California

California firefighters are leveraging artificial intelligence to detect wildfires, utilizing video feeds from over 1,000 strategically placed cameras statewide. This AI system alerts first responders when to mobilize. For instance, the ALERTCalifornia AI program, launched last month, demonstrated its potential when a camera detected a fire at 3 a.m. local time (3:30 p.m. IST) in the remote Cleveland National Forest, about 50 miles east of San Diego.

The fire could have escalated into a major wildfire as people were asleep and the darkness concealed the smoke. However, the AI alerted a fire captain, who then dispatched about 60 firefighters, including seven engines, two bulldozers, two water tankers, and two hand crews. The fire was extinguished within 45 minutes, according to Cal Fire.

Developed by engineers at the University of California San Diego, with AI from Chico-based DigitalPath, the platform utilizes 1,038 cameras installed by various public agencies and power utilities throughout California. Each camera can rotate 360 degrees, controlled remotely.

Since the AI program began on July 10, Cal Fire has reported instances where the AI alerted fire captains to fires before any 911 calls were made, though a comprehensive report is not yet available. Neal Driscoll, a professor of geology and geophysics at UCSD and the principal investigator of ALERTCalifornia, mentioned that the sample size is currently too small to draw definitive conclusions.

Cal Fire envisions this technology becoming a model for other states and countries, especially given the recent severe wildfires in Hawaii, Canada, and the Mediterranean. “It’s 100 percent applicable anywhere in the world, especially with the increasing frequency and intensity of fires due to climate change,” said Suzann Leininger, a Cal Fire intelligence specialist in El Cajon, near San Diego.

Part of Leininger’s role is to refine the AI’s accuracy. She reviews previously recorded footage from the camera network, confirming whether the AI correctly identified fires with a simple yes or no. Various phenomena, such as clouds, dust, or smoky vehicle exhaust, can cause false positives. With hundreds of specialists performing this task across the state, the AI has already improved significantly in just a few weeks, according to Driscoll.

The platform also gathers extensive additional data, including aerial surveys to quantify vegetation that could fuel future fires and map the Earth’s surface beneath the canopy. Airplanes and drones collect infrared and other wavelength data beyond human visual capabilities.

During winter, the system measures atmospheric rivers and snowpack. The UCSD team also collects data on burn scars and their impact on erosion, sediment dispersal, water quality, and soil quality. This data is available to private companies and academic researchers, potentially aiding in modeling fire behavior and developing new AI applications for environmental studies.

“We’re facing extreme climate conditions,” Driscoll said. “By sharing this data, we aim to address a problem bigger than all of us. We need to use technology to make even small advancements.”