Health Notes

Citizen scientists take to the air

By Emily Spacek

PurpleAir, AirU and the future of air quality monitoring.

It’s the tail end of Utah’s inversion season—a time when one might assume winter’s passionate concern regarding air quality would simmer down. But this year, things are different.

More people seem to be paying attention to the government-provided air quality apps, which relay data from strategically placed monitors throughout the Valley. And curiosity is growing in regard to the newfangled affordable monitors and the University of Utah’s air quality “maps.”

Exposure to both indoor and outdoor pollution sources is responsible for the premature death of 5.5 million people worldwide annually, according to The Global Burden of Disease Study. (See“Smaller, Deeper, Deadlier,” by John deJong, February 2020 CATALYST.) Among the most dangerous commonly monitored pollution particles, fine particulate matter (PM2.5) has the greatest adverse health effects. In northern Utah, elevated PM2.5 levels have exceeded national ambient air quality standards ranging from 24-hour periods to weeks.

As of 2015, fewer than a third of all counties in the United States had ozone or particle pollution monitors. In this regard, Utah is making significant headway. Several local innovative projects, including programs from the University of Utah, the Utah Division of Air Quality and the community-based air quality network PurpleAir, have widely expanded the spatial coverage of air quality measurements across the Wasatch Front. But how useful is all of this data for local residents?

The Utah Division of Air Quality (DAQ) relies on air quality measures from sparsely distributed monitoring stations that may not accurately represent the pollutant gradients both in spatial and temporal distribution across our cities. Within the Salt Lake valley, for example, differences in elevation, emissions, atmospheric chemistry and daily wind patterns can result in varying PM concentrations at the neighborhood level that the nearest state monitoring station does not pick up on. Furthermore, the stations only provide hourly updates on PM levels. According to University of Utah research, this gap in temporal resolution is concerning because even short-term spikes in pollutant levels increase the incidence and severity of asthma and cardiac events.

DAQ uses high-quality, costly instruments that meet federal monitoring requirements. But they are not focused on hyperlocal air quality. The need for a new way to track local air quality has led to a recent movement: community-based, citizen science.


PurpleAir: DIY monitoring

Back in 2013, Adrian Dybwad became concerned about the PM2.5-laced dust produced at Draper’s Geneva Rock Products pit near where he spent a lot of time outdoors hang gliding. At the time, the closest DAQ monitoring station was in Herriman, 12 miles away. Recognizing the problem of sparsely distributed state monitors, Dybwad became focused on resolving gaps in the air quality information grid.

From there, he purchased his own small, inexpensive particle sensor and began working out how to connect it to the internet. His theory: A network of low-cost air quality sensors could help create a higher resolution and locally representative air quality image to help average citizens make informed choices on when and where to expose themselves to outdoor air.

Partnering with others interested in air monitoring, Dybwad created successful protocols, began distributing sensors  and, finally, in 2015 founded his company, PurpleAir.

Now the company has over 600 sensors in place at schools, businesses and on resident properties in Utah as well as thousands more across the country and worldwide. At $180 for the indoor model and $229 for the outdoor, now those with access to a power source and Wifi connection can install their own sensors to monitor, in real time, measurements of PM1.0, PM2.5 and PM10 in and around their home.

“Everyone understands that air quality can change quickly. If you’ve ever stood around a campfire, you’ll know firsthand because when the wind is blowing toward you and you’re in the smoke, you have to move pretty quickly to get out of it. So, the big deal with this is that it’s data in real time. Plus, you can take part in the collection for an important air quality database,” Dybwad explained to me over the phone last month.

For individuals, knowing when our home air is most polluted or when to avoid going outside because the air around us is spiking is important for our own safety. But this idea of participating in a larger, basically community-run database is also key. Every 120 seconds, your personal sensor’s data will report to the PurpleAir Map, a free interactive online source that anyone in the community can access (the map can be found on the homepage of

Maybe the air quality in your neighborhood spikes every Saturday night in the wintertime. Maybe you know that a neighbor hosts a weekly bonfire out in their yard. While the levels reported by the DAQ might not instigate a halt to citywide wood burning, you can now see the effect that burning is having on your hyperlocal air. Community engagement at this level can contribute to informed, mindful decision making for the benefit of the collective.

Sometimes access to real time data can even be lifesaving. Dybwad says PurpleAir’s largest customer base is actually in California. During some of the wildfire events, the PurpleAir website experienced a 100-fold increase in their viewership.

The company is currently working on adding sensors to measure ozone and volatile organic compounds, which will provide additional important pollution information.

AirU: expanding the network

Early on in PurpleAir’s development, the University of Utah became interested in its sensors and their potential to give better estimates of air quality at the neighborhood scale. Kerry Kelly, assistant professor of Chemical Engineering and the associate director of the Program for Air Quality, Health and Society, became involved with an air quality project working with schools to teach educational modules on air quality and even teach young students how to build their own air quality sensors.

Kelly was also working with students at the University to develop a comprehensive community-based air quality monitoring network.  The team, known as AQ&U, began to integrate data from their sensors (which they called AirU) with data from the network of their partner, PurpleAir, and data from DAQ state monitoring stations and from the University’s Atmospheric Science Department monitors. Their goal: to create a real-time air quality map.

With funding from the National Science Foundation to expand their personal sensor network, AQ&U began seeking residential hosts throughout the valley. According to Kelly, the program had nearly 500 community members sign up to host sensors, though the team only had around 130 to give out. Still, with their integration of data from their own sensors and all of the valley’s other community-based sensors, the team felt they had an accurate scan of the valley.

The project’s funding cycle ends after this summer, but the team hopes to find new ways to continue. Just within the past year, they have made significant strides, particularly with the addition of ozone sensors.

“If we can get both PM and ozone monitors here in Utah, it will give us a significant map of our two biggest air quality problems,” Kelly says.

Additionally, the team has already completed a number of studies using AirU data and is in the midst of additional health-related research.

“Our map has proved to be a pretty amazing resource for health researchers. We have a group studying asthma and trying to relate asthma exacerbation to particulate matter concentration. We’ve been talking to people who study diabetes. We even have some environmental justice students working with our teams and air quality sensors to test if sound walls along I-80 would decrease freeway air pollution in the Glendale area,” Kelly says.


How do the low-cost sensors work and compare to state monitors?

When PurpleAir first launched and sensors were distributed around the valley, people were experiencing data that was double what the state monitors showed. They complained to both the state and PurpleAir. Where did that difference come from and has the issue since been resolved?

The initial difference between the sensors occurred because of differences in how the sensors work. The state’s $30,000 monitoring stations measure the particle pollution by their mass in the air. The inexpensive sensors use light scattering to measure particle pollution.

Here’s how Kelly from AirU described light scattering: When we have a poor air quality day, we can’t see clearly across the valley because of particles in the air that are scattering light. With so many particles in the air, so much light is scattered that you can’t make out the mountains. Low-cost sensors use the same principle and shine a laser on particles that flow past to measure scattered light and then converts that to particulate mass, based on averages.

The challenge, Kelly explains, is that measuring scattered light is not always consistent with particle mass because it changes with different particle properties and sizes. In the wintertime, particles tend to be small and highly scattering. If you were to take the raw measurements from either the AirU sensors or the PurpleAir sensors, they would overestimate particle pollution by a factor of about one and a half to two in most wintertime conditions. In a dust storm, when particle properties change completely, the sensors actually underestimate particle pollution.

Is this over- or under-estimation significant? If the raw, uncorrected data says 35 microns of PM2.5, which puts you in the EPA’s red zone, the numbers might actually be 18 which is considered reasonably acceptable. That is quite a difference.

“If you don’t use the correct data, you get a relative. That means that if it shows PM levels are going up, then PM levels are probably increasing. If you see them going down, they’re likely decreasing. This, you can know. But to be aware of the exact levels of pollutants, you have to have a correction.”

The University of Utah set out to develop a conversion factor to correct the mass concentration data measured by PurpleAir sensors to best fit the PM composition of wintertime air in Salt Lake City. Now, on the PurpleAir website, users have the option to apply the AQ&U conversion to PM2.5 data on the PurpleAir map to ensure the accuracy of the data.

Dybwad (PurpleAir) accepts the University’s conversion and is working on additional conversion factors for other environments. From nearly the beginning, though, he was confident in the accuracy of his sensors after sending them off to South Coast Air Quality Management District, one of the big air quality regulatory authorities in Southern California. The agency performed several tests on the PurpleAir devices and found that the sensors produced accurate detections of small particulate matter even when compared to federally approved sensors.

The bottom line seems to be that yes, the data from these low-cost sensors, especially after applying correction factors, is reliable and even serves as a check on the measurements of official state sensors.

Aside from knowledge about your personal home and neighborhood air, as the two networks grow along the Wasatch Front there is plenty to be noticed about the region’s air quality as a whole that even state monitors do not visualize well.

In fall of 2018, the Pole Creek wildfire in Juab, Sanpete and Utah counties sent smoke north. The maps showed that particulate matter levels were very high, from Provo to the Salt Lake Valley’s southern end. In the northern end of the valley, though, the air quality was fine.

On the fourth of July, the maps display huge differences in particle pollution associated with fireworks in the base of Salt Lake City versus in the Avenues and east of Ninth East where fireworks are prohibited.

And most recently, the moderate to mild inversion that occurred on Martin Luther King, Jr. Day was confirmed through the sensor networks that showed air quality was worse lower in the valley and much better higher up.

Curiosity and enthusiasm for learning aside, the hoped-for outcome from this community-based air quality network is that people will begin to govern their behavior according to the data. You bring your umbrella if it’s raining. Orange or red alert? You put on your good climate citizen cap and take mass transit, consolidate your trips, skip the jog—do all the things you know will help prevent smog soup or protect your heart and lungs.

And if your indoor air monitor gives you a scary score, we have some thoughts on that on, too. (See page 20-21, this issue.)


Emily Spacek is a staff writer at CATALYST.





This article was originally published on February 26, 2020.