Reducing air pollution with smart design

Originally published on CITRIS and the Banatao Institute By Nate Seltenrich.

Air pollution is one of the greatest threats to human health worldwide, responsible for millions of premature deaths each year. Emissions from vehicles, power plants, wildfires, and other sources can contribute to a range of serious health problems, both in places like Beijing and Delhi and across the comparatively safe United States. Now a Berkeley startup with a global vision wants to help.

As the building block of Clarity Air Monitoring, the Internet of Things-enabled Clarity Node is a comprehensive, weatherproof, air sensing device that can be deployed in less than 10 minutes.

Developed through the CITRIS Foundry startup accelerator program and led by 2016 Atmospheric Science alumnus David Lu, Clarity seeks to help cities test and improve their air-pollution policies — and ultimately reduce citizens’ exposure to harmful pollutants — through robust networks of small, low-cost sensors. Unlike most portable, affordable air-quality monitors, they’re able to maintain high accuracy over time, too, thanks to a unique ability to self-calibrate using machine learning over the cloud, says Director of Business Development Sean Wihera, who has previously worked at CITRIS and the Banatao Institute, Berkeley Lab, and the Berkeley Energy and Climate Institute.

Clarity currently has demonstration networks operating in 22 cities in 12 countries — including about a dozen sensors in the Bay Area that measured the effects of last fall’s wildfires. The company’s longest-running and, to date, largest deployment is in Mexico City, where it hopes to add another hundred or so units by the end of the year. Future prospects include London and Singapore.

For more on what Clarity hopes to achieve and how, we spoke with founder and CEO Lu.

To begin, what’s wrong with existing air-quality monitoring networks?

Although governments have the best intentions to address air pollution, their efforts can be hindered by outdated technology. The major problem with current sensor technology is that it’s very expensive, really bulky, and heavy. What Clarity is trying to do is not to replace these systems, but to supplement them, so that we can expand government monitoring efforts significantly. Our aim is to help deploy networks of air-quality sensors that can deliver information on a hyper-local level. Governments can use this abundant data to inform their decisions about air-pollution management and improve their policies.

So because your sensors are so much smaller and cheaper, you can install many more of them across a single city and thus generate more localized data?

Exactly. Because of the cost and size of the equipment that governments typically use, the number of monitors is very limited, especially at street level. A lot of current monitoring is designed to measure ambient air pollution across a large region, and doesn’t represent what we are breathing daily on the street. The advantage of such a hyper-local monitoring network is not only to add data points, but also to measure what people are really exposed to. Instead of mounting a sensor on the top of a building, we can actually put it on the ground to capture local air-pollution sources.

With street-level data, it would seem that relevant policy and interventions would also have to target specific streets and blocks.

One of the major uses of this data for the government is to assess the impact of specific policies or protocols used to control air pollution. Essentially, it can help governments measure the impacts of their policy, and see if it’s working as intended.

For example, in Mexico City they are using our monitors to measure the impact of a school-bus policy. They introduced a program to encourage kids to take the bus to school instead of having their parents drop them off in a private car. The government wants to see whether this policy has actually made an impact on air-pollution levels around the school. Without a sensor network like Clarity’s, they would not be able to assess the impact of this policy, to determine whether or not it’s working as intended. Our solution can help a government address this challenge by providing hyper-local, real-time feedback on pollution levels. Essentially, you are taking an x-ray to the city to see exactly what’s going on.

What have you learned over the last few months through your Bay Area sensor network? What did you see with the October fires?

With a forest fire, we can quickly deploy more sensors to the area. We can respond faster than existing monitoring networks and can also capture the regional variety in the smoke. For example, we saw that Walnut Creek fared much worse than Berkeley. In Emeryville, the air pollution [level] was actually okay. In Oakland, some regions were substantially better than others. This is where we see the value of our system. We can deliver local air-quality information so that you can know what it’s like outside your door and when it’s safe to bring your loved ones out for a walk or to play in the park. That’s really what we are envisioning for the future.