UT is Working With Google to Map Air Pollution in Real Time

By Chris O'Connell in 40 Acres, Research on June 6, 2017 at 9:08 am | 1 Comment

Scientists at UT, led by Cockrell School professor Josh Apte and in collaboration with Google, the Environmental Defense Fund, and Aclima, a company that builds environmental sensor networks, have developed a hyperlocal air pollution monitoring technique affixed to Google Street View cars to measure and map pollution. This collaboration allows data collection every 100 feet, to detect air pollution “hot spots”—small pockets of more pollution—within a city block. What they collected over the course of one year in Oakland, California, is the most detailed and extensive dataset of hyperlocal air pollution ever assembled, giving scientists and researchers a block-by-block analysis of the air we breathe.

“The fact that there are so many hot spots sprinkled throughout a city blew me away,” Apte says. “We know it’s higher on busy streets but what we did not realize is there are pockets within a neighborhood, even within a single city block.”

The cars covered more than 15,000 miles of Oakland city streets, measuring air pollution on every single block more than 30 times.

One of the innovations of this technique is “how little data you have to collect to see something consistent and meaningful,” Apte explains. “You only need to make measurements 20 times in a city block if you space it out over course of the year. There’s a real efficiency there.” That means it’s cheaper and easier than previously thought to get measurement programs going in other cities around the world.

The story of how Apte linked UT up with Google, the EDF, and Aclima is one of fortuitous circumstances—in Apte’s words, “blind luck.”

While traveling across Asia, Apte ran into a man backpacking across the continent, and he told him of the research he’d done in India measuring air pollution in auto rickshaws. He mentioned that he’d like to expand his research into the U.S., and on a larger scale. The man told him he should put the measuring device on Google Street View cars.

“That sounds awesome, but where can I get one?” Apte responded.

It turns out the backpacker retired early from Google after running the Street View program across Europe. He offered to put Apte in touch with some people at Google. Still a PhD student, the exuberant Apte pitched a collaboration to Google, but was disappointed when he didn’t hear back. A year later, at his 10-year college reunion at Brown, he ran into an old professor, who asked him what he was up to. He told him about the Google Street View idea.

“He gives me this crazy look,” Apte remembers. “He said, ‘Call me in a couple of months.'” Unbeknownst to Apte, his professor was working with the EDF already, who, in turn, was working with Google on a project to measure methane leakage from gas pipes in different cities.

Apte was eventually put in touch with the EDF, and he persuaded the nonprofit to partner with his research team and Google. Aclima was hired to contribute the technology needed to collect data for the project, installing them on Google’s cars, which was then interpreted by Apte and his UT team. It’s a groundbreaking new concept for a few reasons, Apte explains.

“Governments have measured air pollution for a long time,” at fixed spots in different cities, he says. But because those fixed spots are static, they don’t measure the air pollution across a city, and, because they’re expensive to build, there aren’t enough of them. As a reaction, scientists have developed models to predict air pollution, but, according to Apte, those models don’t tell us what we don’t expect to see already. Using existing models but adding measurements is the key to determining what, exactly, we are breathing. 

“There are fewer assumptions if you measure rather than model,” Apte says. “You can also to check if the models are true.”

The technology is scalable, too, and Apte hopes it will spread to the developing world, where air pollution is severe. He says that even without the help of Google Street View cars, attaching measuring devices to taxis or municipal vehicles will suffice, since they don’t take the same route every day.

“In India, for example, there are fewer than 100 air pollution models,” Apte says. “That’s a tiny number, when you think about that scale of data challenge. Finding other techniques is going to be necessary.”

On a conference call with reporters Monday morning to discuss the research published today in Environmental Science and Technology, Karin Tuxen-Bettman, the program manager at Google Earth Outreach, said that the company’s role was as a tech partner, and that Google loves these types of big challenges. 

“Today’s announcement is a major step forward for addressing air quality and air pollution,” she said. “Environmental air quality affects everyone, we hope with making this data more accessible, everyone can make more informed decisions.”

Illustration and animation by Eileen Wu

Tags: air pollution, climate change, Cockrell School of Engineering, Google