AI-Assisted Satellites to Detect Wildfires Faster; Google Funded

Google Research has partnered with fire community leaders to launch a new purpose-built satellite constellation designed to detect and track small wildfires within 20 minutes.

Called FireSat the system will use AI and satellite imagery to spot fires as small as a classroom – around 16 by 16 feet. Once detected, the technology will provide near real-time information about the location, size and intensity of early-stage wildfires so firefighters can respond quickly and effectively.

Due to hotter and drier climates, wildfires are becoming increasingly common across the world. Currently, firefighters 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.

To develop this new system the Google Research team worked with Muon Space and the Environmental Defense Fund to develop custom infrared sensors for the satellites that could better detect small-scale fires. 

To spot a fire FireSat will use AI to rapidly compare any 16 by 16-foot spot on Earth with previous imagery, while also combining factors like nearby infrastructure and local weather. 

To validate the detection model for smaller fires and establish a baseline dataset for the AI, the research team flew sensors over controlled burns. 

Related:IoT Satellite Connectivity Platform Enables Early Wildfire Detection

The first satellite in the constellation will be launched early next year by Muon Space. The full constellation will follow in the coming years, Google said. 

The initiative will be led by the nonprofit Earth Fire Alliance, which was set up to launch FireSat, with support from the Moore Foundation and $13 million in funding from Google.org. 

In addition to supporting emergency response efforts, FireSat’s data will be used to create a global historical record of fire spread, helping Google and scientists to better model and understand wildfire behavior. This will expand Google’s existing work on fire simulation with scientists at the U.S. Forest Service.

Google has been working on wildfire detection for several years. Since 2020 through wildfire boundary tracking, Google has used AI models to create alerts and features that provide helpful information to people in close proximity to a wildfire. This is now available in over 20 countries. 

Earlier this year the company released FireBench, an open-source machine learning benchmark dataset for wildfire research.

“As we work to expand our wildfire detection, data and AI play an important role in providing critical information to emergency responders and opening up new opportunities to scientists. We’ll continue to partner with the fire community in the face of the growing threat of wildfires,” Google said. 

Related:AI-Enabled Wildfire Monitoring Sensors Exhibited at CES 2024

This article first appeared in IoT World Today's sister publication AI Business.