The fire lookout tower a person with binoculars, scanning a mountaintop for smoke was a fixture of Western wildfire response for generations. It's largely gone now, replaced by a network built and run out of a university campus in San Diego that watches the entire state, all day and all night, without ever needing to sleep. ALERTCalifornia, based at UC San Diego, has grown from a small research project into a statewide network of more than 1,200 high-definition cameras, and it's arguably the most consequential piece of wildfire IoT infrastructure operating anywhere in California today.
The system, developed and operated at UC San Diego's Jacobs School of Engineering, Qualcomm Institute, and Scripps Institution of Oceanography, deploys pan-tilt-zoom cameras equipped with near-infrared night vision on fire lookout towers, communications infrastructure, and observation sites across the state including familiar Southern California landmarks like Cowles Mountain right here in San Diego. Each camera performs a full 360-degree sweep roughly every two minutes, capturing a dozen images per pass, around the clock. On a clear day, a camera can see up to 60 miles; at night, using near-infrared technology, that range extends to 120 miles.
What makes the network genuinely useful for real-time detection rather than just archival monitoring is the AI layer built on top of the raw camera feeds, developed by a company called DigitalPath. The software automatically scans incoming images for smoke and early fire activity and alerts public safety personnel directly no human has to be watching a specific camera feed at the moment a fire starts for the system to catch it.
Over a recent 12-month period, ALERTCalifornia flagged approximately 3,600 wildfire incidents statewide. More than half of those detections happened before a single 911 call came in from the public meaning the system caught the fire before anyone on the ground even knew it had started. CAL FIRE now uses the network across all 21 of its Emergency Command Centers, and the agency's own leadership has been direct about what that early warning actually buys them: the ability to augment a response with additional resources before the first units even arrive on scene, rather than sizing up the incident only after crews are already there.
The funding trajectory reflects how seriously the state has come to rely on the system. CAL FIRE alone contributed at least $24 million toward expanding the network over a five-year period, and major utilities including San Diego Gas & Electric, alongside PG&E and Southern California Edison have contributed camera locations and equipment of their own, recognizing that early wildfire detection directly serves their own wildfire mitigation obligations.
It's worth being honest about the system's real limitations, since overstating any single technology's capability tends to erode trust when it inevitably fails to catch something. The cameras can't see through mountains, and a fire that starts at the bottom of a steep canyon may not produce a visible smoke plume the network can detect until it's grown large enough to crest the surrounding terrain a real gap given how much of San Diego County's wildfire risk sits in exactly that kind of canyon and backcountry topography. The AI detection software also isn't perfect at distinguishing smoke from other visually similar phenomena; engineers have specifically had to train the system not to flag steam from geothermal fields as fire activity, and dust storms and certain cloud formations can still occasionally trigger false positives that require human verification before a full response gets dispatched.
In fast-moving urban fires with a lot of nearby population the way the January 2025 Los Angeles fires unfolded, for instance residents calling 911 with cell phones can sometimes report a fire nearly as fast as the camera network detects it, since dense urban areas simply have more potential human observers per square mile than California's vast, sparsely populated wildland areas where the camera network provides its most valuable and otherwise unavailable advantage.
One of the more notable design decisions behind ALERTCalifornia, according to the researchers who built it, is a deliberate choice to keep the underlying camera feeds and data publicly accessible rather than treating them as a paywalled commercial product. Anyone can view live camera feeds at the ALERTCalifornia website a choice that's turned the network into something used well beyond wildfire response alone, including monitoring atmospheric river storms, tracking California condor recovery efforts, and even catching visual confirmation of a tsunami warning following a major earthquake off Russia's coast. UC San Diego researchers have specifically pushed back against the idea that this kind of publicly funded disaster infrastructure should ever become proprietary or restricted, arguing that a system built with substantial public investment should remain a public resource rather than something locked behind a private vendor's access controls.
The network does build in privacy protections despite its public accessibility UC San Diego blurs homes, parking lots, and other features near camera installations that could otherwise be used to track identifiable people in real time, a deliberate design choice to keep the system focused on landscape-scale disaster monitoring rather than functioning as a general-purpose surveillance tool.
ALERTCalifornia's cameras get most of the public attention, but they're part of a broader sensor infrastructure supported by HPWREN, a wide-area wireless network providing connectivity throughout Southern California, including San Diego, Orange, Los Angeles, Inyo, and Riverside counties. That same wireless backbone supports research into seismic and GPS networks, hydrological sensors, oceanographic sensors, and meteorological sensors meaning the "last kilometer" wireless infrastructure originally built to get wildfire camera data off a remote mountaintop has become reusable connectivity infrastructure for a much wider range of San Diego-region environmental and disaster monitoring, including the coastal erosion sensor networks covered elsewhere on this site.
ALERTCalifornia's wildfire detection layer doesn't operate in isolation from San Diego's other public safety technology investments it's part of a broader regional pattern of using continuous, always-on sensing to compress the time between an incident starting and a trained response beginning. The same underlying logic that makes a wildfire camera network valuable catching a problem in its earliest, most manageable stage rather than after it's already grown large and dangerous shows up again in San Diego County's pioneering Drone as First Responder programs, where the goal is similarly about compressing response time through better, earlier information rather than simply adding more personnel or equipment.
Given how directly early detection translates into faster containment and reduced fire size, the trajectory for ALERTCalifornia and similar systems is continued expansion rather than plateauing new fire departments and agencies, from large regional park districts to smaller municipal fire departments, continue signing on as network partners, each contributing additional camera locations and expanding coverage into previously monitored gaps. As California's fire seasons have grown longer and its fires have grown larger over the past two decades, the case for this kind of always-on, publicly accessible detection infrastructure has only strengthened, and San Diego's role as the technical and research hub behind the entire statewide system is likely to keep growing along with it.
Frequently asked questions
More than 1,200 high-definition, pan-tilt-zoom cameras are deployed across California as of early 2026, including locations in San Diego like Cowles Mountain, with the network continuing to expand as new agencies and partners sign on.
Over a recent 12-month period, more than half of the approximately 3,600 wildfire incidents the system flagged were detected before any 911 call was placed by the public.
The cameras can't see through mountains, and fires starting at the bottom of steep canyons may not be visible until smoke rises above the surrounding terrain. The AI software can also occasionally mistake dust storms, certain clouds, or geothermal steam for smoke.
Yes. Live camera feeds are publicly accessible on the ALERTCalifornia website, a deliberate design choice by the UC San Diego team, who argue that publicly funded disaster infrastructure should remain a public resource rather than a paywalled product.
The system builds in privacy protections by blurring homes, parking lots, and other features near camera locations that could otherwise be used to track identifiable individuals, keeping its focus on landscape-scale monitoring rather than surveillance of people.
It's based at UC San Diego and funded through a combination of state support including at least $24 million from CAL FIRE over five years along with contributions from utilities including San Diego Gas & Electric, PG&E, and Southern California Edison.
Related articles
Public Safety
