The first FireSat satellite has launched and made a connection with Earth. The FireSat system is a collaborative effort between Google Research, Muon Space, Earth Fire Alliance, Moore Foundation, and numerous other agencies, and it has a single, deceptively simple purpose: to detect wildfires before they become too hard to contain and control.
Wildfires have been a constant problem for agencies. Early detection is vital, but fires can often start in subtle ways; by the time anyone notices the growing blaze, it’s too late to stop. Just take the wildfires in Los Angeles earlier this year as an example. Apps have been created to crowdsource fire detection, and the traditional method of watching for wildfires is through satellite imagery.
However, smoke, debris, and other obstacles make satellite imagery unreliable at best. It is often only updated every 12 hours, too. FireSat builds on existing infrastructure to create what its team calls a “wildfire boundary map.” This system will enable people to view wildfires through services like Google Maps.
FireSat utilizes AI to search for fires. The way it works is by comparing a current image of a given location with previous imagery. “Differentiating between real fires and random ‘noise’ in the environment was a challenge,” Chris Van Arsdale, co-founder of FireSat, says. “We had to determine where to draw the line between what is a real fire versus things like sensor issues or misaligned pixels.”
The new model is particularly effective. Previous detection methods often only picked up on wildfires after they were a few acres in size, but FireSat is quite a bit more capable. Erica Brand, one of the project managers for FireSat, gave this example. “One of our team members lit a barbecue and a firepit in his backyard so we could fly the plane over and test it. And the sensors were able to pick it up.”
FireSat is still several years away from being fully functional, but the team believes it will be able to accomplish things no wildfire detection system has been able to before. According to Google, the full system will be able to detect a wildfire about the size of a classroom within 20 minutes. This will give response teams significantly more time to contain the blaze before it spreads to residential and commercial areas.
