Google Aims for Global Wildfire Detection Every 20 Minutes with FireSat Satellites

Google Research is building a system to detect wildfires anywhere on Earth at 20-minute intervals through the FireSat satellite constellation. By launching numerous low-cost small satellites and using AI to compensate for precision, it aims to overcome the limitations of existing methods.

Detection Performance and Technical Approach

FireSat aims to identify small wildfires as small as 5x5m (25m²). This is a more ambitious figure than the minimum size (50–100m²) that fire authorities can extinguish. Google initially trained its AI models using data from controlled burns flights and is distinguishing false alarms and improving algorithms through imagery captured over California. In particular, AI is used to distinguish harmless heat sources, such as agricultural burning or barbecue grills, from actual dangerous wildfires.

Operational Plan and Social Application

Google expects to achieve the goal of monitoring the entire Earth at 20-minute intervals by around 2030. The collected data will be used to secure initial response times for firefighters, predict fire spread, and determine locations for firebreaks during the urban planning phase.