Story at a glance

  • Data refinery company Descartes Labs uses satellite imagery to monitor wildfires.
  • The technology runs on AI to interpret if there is a fire and where it is located.
  • The major advantage is that the company can alert authorities to fires before they grow.

Wildfires have ripped through California for the past few years, and they aren’t expected to slow down. Studies have found that seasonally warm days in California have increased by about 1.4 °C since the 1880s, resulting in hot and dry conditions that promote forest fire, particularly in the North Coast and Sierra Nevada regions of California. 

To combat a growing crisis, the tech industry has responded with innovative tools to help lessen the destruction of wildfires. One company is setting out to catch wildfires as early as possible by going to outer space. 

Santa Fe-based Descartes Labs is developing new artificial intelligence (AI) technology to analyze satellite imagery as a way to catch wildfires before they expand. Named a promising start-up by CNBC, the company specializes in geospatial analytics that run on data collection and analysis. Its wildfire detector uses AI algorithms to analyze images from NOAA satellites that track potential fires on Earth. 

Indicators of a fire in Descartes satellite images are shown below as thermal infrared data with high concentrations of heat and smoke. This enables the company to catch wildfires early in hopes of preventing casualties and destruction. 

Different algorithms track specific characteristics of a potential fire. In a partnership with the Los Angeles Times, Descartes Labs runs a livestream of current fires in California and other West Coast states, but can report fires as far East as West Virginia. Some of the details the algorithms track are the perimeter of the fire, locations in and near the fire, thermal hotspot measurements, and even air pollution. 

To eliminate errors, Descartes models what the land looks like from space without an active wildfire to control for false alarms. Objects like solar panels, bodies of water, and even rocks can give off thermal readings that might seem like a fire. Having a basis of comparison of Earth allows Descartes scientists to distinguish a fire from a hotspot. 

The company intends to go a step further in wildfire management. In a Medium post, Descartes Labs scientist Clyde Wheeler says their technology “heralds the beginning of a digital transformation in how we identify wildfires, suppress them, and evacuate people at risk.” 

Published on Dec 06, 2019