In a significant leap forward for wildfire detection technology, Google Research has announced the development of FireSat, a purpose-built network of satellites aimed at identifying and monitoring wildfires at their earliest stages. This innovation comes at a crucial time, as wildfires become increasingly common due to hotter and drier climates worldwide.
Christopher Van Arsdale, Climate & Energy Lead at Google Research, emphasised the importance of early detection: "Until now, firefighters have had to rely on satellite imagery that's either low resolution or only updated a few times a day, making it difficult to detect fires until they've grown larger than a soccer field."
FireSat aims to change this by providing high-resolution imagery updated globally every 20 minutes. The system can detect fires as small as 5x5 meters, roughly the size of a classroom. This dramatic improvement in both resolution and frequency could enable authorities to respond to fires before they become destructive.
The project is a collaborative effort, with Google.org providing $13 million in funding to the Earth Fire Alliance, a nonprofit established to launch the FireSat constellation. Additional support comes from the Moore Foundation.
Key to FireSat's capabilities are custom infrared sensors developed by the Google Research team, in partnership with Muon Space and the Environmental Defense Fund. These sensors are specifically designed to detect small-scale fires with unprecedented accuracy.
The system employs artificial intelligence to rapidly compare any 5x5 meter spot on Earth with previous imagery, while also considering factors such as nearby infrastructure and local weather conditions. To validate the detection model and establish a baseline dataset for the AI, the team conducted test flights over controlled burns.
Muon Space plans to launch the first satellite in early 2025, with the full constellation to follow in subsequent years. Once operational, FireSat will provide near real-time information about the location, size, and intensity of early-stage wildfires, enabling faster and more effective responses from firefighters and emergency responders.
Beyond immediate firefighting applications, FireSat's data will contribute to a global historical record of fire spread. This valuable information will help Google and scientists better model and understand wildfire behavior, expanding on existing fire simulation work conducted in partnership with the U.S. Forest Service.
The FireSat project is part of Google Research's ongoing efforts to address the impacts of climate change through technology. Since 2020, the company has been using AI models to provide wildfire alerts and information in over 20 countries. Earlier this year, Google also released FireBench, an open-source machine learning benchmark dataset for wildfire research.