DigitalGlobe Turns to the Crowd to Analyze Post-Typhoon Images
Since Super Typhoon Haiyan struck the Philippines last Thursday, the world has witnessed scenes of devastated villages, destroyed homes, and washed-out roads and bridges.
The pictures are heartbreaking and have motivated people around the world to try to help. Usually that means donating money or supplies, but DigitalGlobe, a satellite imagery company based in Colorado, is taking a novel approach. The Longmont, CO-based company is making “before” and “after” images of the effected areas available to the public in an attempt to help analyze the damage in ways that might benefit the ongoing relief effort and the upcoming reconstruction.
DigitalGlobe (NYSE: DGI) satellites passed over the Philippines immediately after super typhoon Haiyan passed through and hurried to get the shots to responders and online observers.
“Within hours of the data coming out of the satellites, we were able to make the data available to the public,” said Luke Barrington, senior manager of research and development at DigitalGlobe.
The company has been making use of its archive in the aftermath of natural disasters for a while. DigitalGlobe’s FirstLook subscription service is available to first responders, and it is used for about 200 disasters a year, Barrington said. FirstLook was put to use in the days before Haiyan struck to track the typhoon as it moved across the Pacific, and was available for agencies like the International Red Cross and the Philippines government, he said.
But DigitalGlobe also put the photos online, in the hope that volunteers will help identify and geo-tag damaged structures, roads, or bridges. Barrington is part of the team leading the effort, and he believes it is one of the first times volunteers from around the world have been able to utilize their time and knowledge to help coordinate the response to a disaster of this scale.
The technology DigitalGlobe is using for the effort was developed by Tomnod, a startup Barrington cofounded. DigitalGlobe bought Tomnod, which was based in San Diego, for an undisclosed price earlier this year.
“Tomnod is a crowdsourcing app for getting the public’s help in analyzing huge amounts of satellite imagery,” Barrington said. He believes it is well suited for a disaster like Haiyan.
With Haiyan, Tomnod is giving users high-resolution images of small bits of the Phillippines’ coastline and interior. The pictures not only make the devastation apparent, they also are detailed enough that people can identify the wreckage of houses, roads, bridges, or large buildings.
Users can log in and start geo-tagging the damage. Tomnod not only collects their input, but it is smart enough to compile tags from multiple users to verify their work.
“It’s not what any one person says. You don’t want to take the word of a single person on the Internet,” Barrington said. “To extract insight, we run an algorithm that evaluates the input and determines who is reliable and what the consensus is. It then takes the improved data and disperses it.”
Through Thursday afternoon, Barrington said Tomnod had identified more than 10,000 points of interest. There had been more than a quarter million page views of Tomnod’s site, and he said users had covered about 100,000 square kilometers.
The crowd is solving what had been a problem for DigitalGlobe, hopefully to the benefit of victims.
“We’re capturing data and information faster than we have the ability to analyze it,” Barrington said. “We have all of these amazing images, but there are just more pixels than we have time to deal with ourselves.”
Putting photos of disaster damage online might seem at first like a macabre high-tech version of rubbernecking when passing a bad accident on the highway. But Barrington is persuasive in arguing Tomnod could grow into a very valuable tool, even if it is so new it isn’t quite clear what its best use will be.
“We’re not 100 percent sure yet,” he said. It could be used to allocate emergency personnel in the early days of a relief effort, or to determine which remote areas need heavy equipment. Or it could help catalogue buildings and infrastructure that need to be rebuilt.
“We’re still trying to find out what the best fit is,” he said.
Tomnod began as a project at the University of California, San Diego. Its goal—to find the long-lost tomb of Genghis Khan by having volunteers pour over satellite imagery of Mongolia—sounded like the plot of a Hollywood adventure movie adapted to the age of social media. The company’s name is Mongolian for “big eye.”
Tomnod grew into a seven-employee startup by the time it caught DigitalGlobe’s eye.
DigitalGlobe is one of the world’s largest commercial providers of satellite imagery. It operates a constellation of satellites and provides high-resolution imagery to clients including the military and Google. The Longmont-based company has a market cap of $2.77 billion, and it had the clout to survive a takeover bid from GeoEye, one of its few rivals. DigitalGlobe ended up buying its competitor in 2012 for $1.4 billion.
According to Barrington, it was the story of an industry leader turning to a small startup to address a unique pain point. For DigitalGlobe, it was a big data problem unique to its field, he said.
Getting the most out of detailed satellite images historically has been labor intensive. While the technology had advanced greatly, the analytic work is much like it was back when the CIA began using photographs from spy planes to try to find missile launch pads during the Cold War.
“Until now, it’s always required a team of in-house GIS (geographical information systems) who knew that data formats, who knew how to look through the pictures and extract the info,” Barrington said.
But as satellite images became widely available to the public through apps like Google Earth and Google Maps—which relies heavily on DigitalGlobe images—people began searching for their homes, local landmarks, or attractions they had visited or would like to.
Barrington and his Tomnod co-founders thought it made sense to capitalize on that public curiosity and realized there was potentially an untapped well of expertise.
DigitalGlobe had also realized there was a demand for its pictures in the media. When an event grabbed headlines around the world—like the Navy SEAL raid of Osama bin Laden’s compound in Pakistan—DigtialGlobe began releasing its images online.
Disaster recovery could be an ideal use for Tomnod—and unfortunately this is likely to be a growing need in the future.
“If we never had to turn it on for another disaster, that would be great. But they keep coming, so we need this technology in place,” Barrington said.