Geospatial Analysis of California Wildfires Shows What Actually Burned
Geospatial analysis processes vast amounts of data in order to gain fresh insights on large-scale events, like crop failures or natural disasters.
This month’s wildfires in northern California were the worst in the state’s history, causing at least $1 billion in damage and 42 deaths. At least 8,400 homes and other buildings were destroyed, and at one point, classes were canceled for 260,000 students in nearly 600 schools.
Yet details about the exact extent of the damage have been spotty, particularly during the height of the conflagration, when obtaining clear satellite imagery was difficult due to thick plums of smoke emanating from the flames.
But a Santa Fe, New Mexico-based tech company says it has a solution. Using tools from the emerging field of geospatial analytics, Descartes Labs analyzes vast amounts of satellite data in order to gain fresh insights about large-scale events, like crop failure or natural disasters.
The company says it gathers some 10 terabytes of near real-time data from observational satellites per day, or about 5,000 hours of standard video — a volume that requires massive computing power to absorb and analyze.
With the northern California fires, it found, despite the vast amounts of confirmed loss of property and real estate, that only 4.2 percent of the severely damaged area in Sonoma and Napa counties was actually developed — and that over half of the severe burning took place in Northern California’s vast, low-lying shrub land. The numbers reflect conditions as of Oct. 12.
“It was astounding that this was only 4.2 percent of the burn area,” Caitlin Kontgis, a member of Descartes Labs’ applied science team, told Seeker.
That insight is an example of what the company says is the promise of geospatial analytics: the ability to digest vast amounts of information collected by satellites into useful, actionable intelligence.
“Essentially what we’re trying to build is a data refinery for geospatial data where any researcher can come to get access quickly,” Kontgis said. “That might sound trivial — unless you’ve tried to deal with satellite imagery before.”
The team collects mountains of visual, infrared, thermal, and other kinds of data from groups like NASA, the European Space Agency, and a myriad of smaller private companies. By churning through that data, the group can arrive at its own estimates of crop yields for corn or soybeans, for example, or help rescue workers track certain aspects of a natural disasters.
Descartes Labs was recently selected by the US Defense Advanced Research Projects Agency, known as DARPA, for a $1.5 million dollar three-year contract to create crop failure warning alerts based on satellite imagery of the Middle East and North Africa.
“What’s innovative about what we’re doing is storing and processing all that data in-house,” Kontgis said.
In the case of the California fires, the company used data from the European Space Agency’s Sentinel-2 Earth observation system land and the US Department of Agriculture to get a better grasp of what exactly has been destroyed.
While photographic imagery from satellites provide part of the picture by detailing plumes of smoke, the Descartes Labs analysis used infrared and shortwave infrared bands to uncover the severity of the burn on the ground and to peer through the plumes. The researchers then overlaid those findings with a land-use maps to analyze the type of areas that had been damaged.
The company estimates that over 96,000 acres were burned in Napa and Sonoma counties, with severe burning over 45,000 acres. Only 4.2 percent of the severely burned area, however, was developed. And, while much has been made of the damage to California wine-makers, the study found that only 152 acres — or 0.3 percent of the damaged area — was used for growing grapes.
WATCH: Should We Control Wildfires?