Google Research has partnered with leaders in the firefighting community to launch a new dedicated satellite constellation called FireSat, designed to detect and track small wildfires within 20 minutes. This system will utilize artificial intelligence and satellite imagery to spot fires as small as a classroom and provide near-real-time information on the location, size, and intensity of early wildfires, enabling firefighters to respond quickly and effectively.
Image source note: The image was generated by AI, with authorization from service provider Midjourney
FireSat will use AI to rapidly compare any 16x16 feet location on Earth with previous images, while also considering nearby infrastructure and local weather conditions. To validate the small fire detection model and establish a baseline dataset for AI, the research team flew sensors over controlled burn areas.
The first satellite of the constellation is scheduled to be launched by Muon Space early next year, with the entire constellation being completed in the coming years. The project will be led by the Earth Fire Alliance, a nonprofit organization established for the launch of FireSat, and is supported by the Moore Foundation and a $13 million grant from Google.org.
In addition to supporting emergency response efforts, FireSat's data will be used to create a historical record of global fire spread, helping Google and scientists to better model and understand wildfire behavior. Google has been committed to wildfire detection for years, and earlier this year, the company released FireBench, an open-source machine learning benchmark dataset for wildfire research.
Google emphasizes that data and AI play a crucial role in providing critical information to emergency responders and opening up new opportunities for scientists. Facing the increasing threat of wildfires, Google will continue to collaborate with the firefighting community.