Democratising air quality data at nearly no cost

Exposure to air pollution, particularly from PM2.5, is the dominant environmental health risk worldwide and the exposure risk is more critical in low- and middle- income countries for several reasons, including rapid urban development, the use of solid fuels for cooking and home heating, and the use of older vehicles that lack modern pollution controls.

Reducing air pollution exposures relies on evidence-based policies. These policies rely on measurements of air pollution levels with high precision and accuracy. Additionally, measurements need to be chemically specific so that policymakers can identify the most important sources of air pollution. However, research- and regulatory-grade air pollutant monitors are expensive. The high cost of air quality monitors makes it nearly impossible for many countries to regularly monitor pollutants, leaving policy makers unable to enact evidence-based policies to reduce exposure.

Air pollution monitoring is extremely limited in Africa. Overall, there is a lack of air pollution monitoring infrastructure in many African nations, though there has been some success in deploying low-cost air pollutant sensors in a few cities. While these low-cost sensors provide valuable data on PM2.5 concentrations, they lack compositional information that can be used to understand emissions sources.

In an effort to better understand air quality across the globe, my lab at Carnegie Mellon University in Pittsburgh, Pennsylvania has been working with colleagues at CMU-Africa to develop low-cost methods to quantify PM2.5 composition. One approach that we have developed uses cellphone cameras to measure air pollution. Since cellphone cameras are ubiquitous, this approach has the ability to provide air pollution data at extremely low cost.

Our cellphone-based approach targets a component of PM2.5 called black carbon (BC). BC is the sooty black material that you might see coming out of the tailpipe of a diesel truck. Quantifying BC concentrations is important for several reasons. Knowing the fraction of PM2.5 that is BC can give a window into the contribution of combustion to local air pollution in a city. BC may also pose higher specific toxicity than PM2.5 overall, making it more harmful to public health than other types of PM2.5. Lastly, BC is considered a “short term climate forcer” because of the way it absorbs light and consequently warms the atmosphere. For example BC deposited onto snow will cause that snow to melt faster.

Our process is unique because it can be used to determine black carbon concentrations from a variety of different samples. All we need is a cellphone camera, a custom-designed reference card, and particles captured onto a filter. In a recent study, we worked with U.S. embassies to apply our method in multiple African cities. U.S. Embassies around the world collect air quality data using an instrument called a Beta Attenuation Monitor (BAM). The BAM uses a filter tape, and collects particle deposits onto the tape. These tapes were then sent to our team for analysis.

After photographing the BAM tape on top of the reference card, we can apply an image processing algorithm to each photo to extract the red scale value of the photo. This value lets us identify the black carbon concentration during the hour of the day the tape was collected. By extracting black carbon concentrations, we can determine which parts of the pollutant are coming from fossil fuels as opposed to biomass burning or wood smoke.

To determine BD concentrations, the filters are placed on a reference card and photographed. A photograph is composed of pixels, and each pixel is a combination of RGB channels. Having performed colour calibration of the original image Albert’s team could determine the black carbon concentrations of a filter sample  using the greyscale boxes which have a red value that corresponds to a unique black carbon concentration

We collected BAM tapes from U.S. embassies in Côte d’Ivoire, Ghana, and Ethiopia and compared their particulate matter to that collected from a site in Pittsburgh. In general, the African locations were significantly more polluted than Pittsburgh and other locations in North America or Europe. BC concentrations in the African cities were 10-20 times higher than the concentrations in Pittsburgh. These higher concentrations suggest that combustion emissions contribute significantly to human health risks from air pollution in these locations.

This process is a new way to think about low-cost analysis because the tapes are already being collected, so the marginal cost for our analysis is near zero. This method can democratize air quality data because any number of research groups around the world can collect tapes from other embassies and do their own analysis for practically no cost.

Beyond working with the embassies, I believe it is important to work with stakeholders in each country, whether that be other academics or just interested local parties, who want to use the data to advocate for improved public policy.

Our findings underlined the need for more air quality monitoring in developing countries. The black carbon levels in the Sub-Saharan African countries were as much as four times higher than those collected in Pittsburgh. Our findings allow us to provide insight into seasonal differences (wet-dry or summer-winter) differences in BC levels and to determine diurnal trends. For example, we saw in Accra that black carbon peaked early in the morning, every day of the week. Knowing this, can help in identifying and mitigating isolated sources.

Additionally, with the growing work to monitor air quality from outer space, more ground data is needed to calibrate satellite performance. Using this method, we can likely grow the number of locations where we can compare the satellite to data on the ground, ultimately making more data available to more countries around the world.

While our work continues in Pittsburgh, my colleagues at CMU-Africa are active in a network called AfriqAir. The hybrid air quality monitoring network introduces low-cost sensors and reference-grade monitors across eleven African countries and trains people to gather and analyse the data.
I am confident that developing countries can make positive changes to reduce harm from PM2.5 more easily with low-cost ways to monitor air quality. My lab plans to keep working with embassies, and hopes to also find a way to extract the filters in a solvent to uncover what else PM2.5 is composed of throughout the day.

Albert Presto is a Research Professor of Mechanical Engineering at Carnegie Mellon University and the Director of the Center for Atmospheric Particle Studies – one of the world’s leading research centres focused on particulate matter and its effects on global climate and public health