California is expanding a system that uses artificial intelligence and high-resolution surveillance cameras to detect wildfires early, with AI now the first to spot more than half of the fires that break out in the state.
Alert California, a wildfire monitoring system run by the University of California, San Diego, operates about 1,300 remote surveillance cameras in high-risk areas across the state, The New York Times reported on the 12th.
The cameras are mounted on mountaintops, communication towers and rooftops, places with wide sightlines, and monitor surrounding forests in real time. AI analyzes the footage to detect smoke and signs of fire, and fire officials use the results to judge whether a wildfire has started and how large a response it requires.
Detection has improved sharply since AI was added to Alert California in 2023. AI is now the first to find 51% of California's wildfires, according to Neal Driscoll, the UCSD professor who directs the program. Many fires are detected before residents call 911.
When AI flags a possible fire, fire officials review the camera footage to determine whether it is an actual wildfire or merely smoke or fog. They then decide whether to deploy crews and aircraft based on the size and location of the blaze. Firefighters on the scene can also use the live video to track how the fire is spreading. An official at the California Department of Forestry and Fire Protection said the Alert California system has changed how the state responds to wildfires, adding that it provides immediate access to the field imagery and information needed to make response decisions.
The cameras are equipped with solar panels and batteries so they can operate in mountainous terrain where power is hard to supply. Footage is transmitted over satellite internet, and some cameras rotate 360 degrees every two minutes to film their surroundings. Others can be remotely repositioned or zoomed in on specific points by fire officials.
The system has also proved its value in areas hit by major fires in the past. In Napa and Sonoma, where dozens died in major wildfires in 2017, cameras newly installed in 2019 helped detect the Kincade Fire early and support a remote response. That fire grew large but caused no deaths.
Alert California began in 2001 as a network of remote sensors designed to detect seismic activity around San Diego. Wildfire monitoring was added later, and the recent combination with AI turned it into a real-time fire detection system.
The program runs on an annual budget of up to $14 million, most of it covered by California fire authorities and major utilities. Operators plan to expand the network from about 1,300 cameras to 1,800 to reduce blind spots.
Early detection through AI cannot prevent all wildfire damage, but it allows for a faster response than in the past. The Times reported that the Palisades Fire in Los Angeles last year was also captured by surveillance cameras but that high winds and extremely dry conditions kept the system from stopping the damage from spreading rapidly. Even so, it reported, California fire officials expect AI and real-time cameras to play an important role in shortening initial response times and limiting the spread of damage.