FireSat and AI cameras are catching wildfires before anyone dials 911

[Nina24]

A growing network of AI powered satellites and ground cameras is changing how quickly wildfires get spotted, with California's Cal Fire reporting its camera network now detects about half of all wildfires before a single 911 call comes in. One recent fire near Fresno was spotted roughly 20 minutes before the first emergency call reached dispatchers, a gap that can matter enormously for how fast crews actually get deployed.

The most ambitious project is FireSat, a planned constellation of 50 satellites built specifically to detect wildfires anywhere on Earth within 20 minutes of ignition, even something as small as a beach bonfire. Google.org provided $13 million in seed funding through the nonprofit Earth Fire Alliance, and the first satellite, built by Muon Space, launched from Vandenberg Space Force Base and has already made contact with Earth. Once fully deployed, FireSat is designed to compare any 5 by 5 meter patch of ground against previous imagery every 20 minutes globally, factoring in nearby infrastructure and local weather to decide whether something is actually a fire.

On the ground, California's Alert California system already runs more than 1,200 cameras used by Cal Fire, while startup Pano AI has deployed over 1,400 high resolution cameras across 17 states, reporting roughly 90% detection accuracy from its network. Utilities are adopting the technology too, with Portland General Electric running Pano AI cameras across Oregon and planning to expand further this fire season.

Separately, researchers at West Virginia University are working on a more experimental step forward, satellites with onboard AI that can not only detect a fire but reposition themselves in orbit to keep tracking it as it spreads, rather than relying on satellites in fixed orbits that simply pass overhead on a schedule. Existing satellite propulsion systems already support that kind of maneuvering, they're just not currently used for this specific purpose.

Climate change is making the underlying problem worse regardless of how good detection gets. With some western US regions now experiencing roughly two additional months of fire weather each year compared to the 1970s, meaning faster detection is buying valuable response time rather than solving the actual underlying trend

AI eyes in the sky: new satellites and artificial intelligence are transforming wildfire detectionwww.theguardian.com