Summer Virtual Series Session 2

Virtual Summer Series

Event Date

Full session moderated by: Calvin Cupini, Clean Air Carolina & Ethan McMahon, US EPA

Register for Session 2 Here

Discussion at bottom of page: Please offer a question or comment to our presenters at the bottom of the page through the DISQUS platform. We will gather questions to ask of our presenters live on Thursday!

Use of a low-cost sensor network to estimate exposures and track pollutant chemistry during the 2018 Kīlauea eruption

Presented By: Ben Crawford, University of Colorado, Denver

Description: The Lower East Rift Zone (LERZ) eruption of Kīlauea Volcano (Hawai‘i Island, HI, May-August 2018) destroyed hundreds of homes and displaced thousands of people, and represented an extreme air quality event for the entire region. Exceedingly high levels of sulfur dioxide (SO2) were introduced directly into a residential area, exposing residents to potentially dangerous levels of the toxic gas; further, the oxidative transformation of SO2 to sulfuric acid (H2SO4) led to elevated levels of particulate matter (PM) to communities downwind of the eruption. In order to provide improved estimates of residents’ exposures to volcanic pollutants, we deployed a network of low-cost sensors (LCS) across the island. This consisted of ~30 sensor nodes measuring SO2 and PM; nodes were run on solar-rechargeable batteries and sent data in real time to a central server via 3G. This provided time-resolved estimates of the population’s exposures to both pollutants, as well as estimates of the SO2-to-PM conversion rate, with much finer spatial resolution than complementary measurement techniques (satellite measurements, regulatory monitoring) can provide. This work also serves as a case study for the use of LCS for measurements near an emissions hotspot as well as the rapid deployment of LCS networks during AQ emergencies; lessons learned and future challenges will be discussed.

Follow Up Questions and Answers:

  • Can you give details of what sensors you used for the various pollutants? i.e. which SO2, which PM sensors etc - Sridhar Rajagopal
    • For SO2, electrochemical B-series sensors by Alphasense. For PM, OPC-N2 by Alphasense.
  • How did you calibrate the sensors? - Thor Bjorn Ottosen
    • Co-location with regulatory-grade monitors at two different on-Island locations at two different times during the eruption. Prior to the 2018 LERZ eruption, our group published a study about SO2 sensor performance from measurements during the effusive eruptive phase.
  • Did you co-locate any of these sensors with more traditional sensors to compare the data over time? Were there any issues with drift rates, span/zero, etc. when compared? - Nicholas Dummer
    • Yes, we co-located with regulatory monitors and there was negligible drift over the timeframe we were monitoring (weeks-months).
  • What did you learn about the H2S and SO2 concentrations? - John Saffell
    • We were able to calculate an in-plume chemical reaction rate for SO2 using measurements from our network. As far as we’re aware, this is the first time this has been done using this type of approach. We’re less confident in our H2S measurements because of instrument cross-sensitivities with SO2 and lack of a good field calibration reference during the eruption.
  • What was the uncertainty of the measurements? - Rose Eilenberg
    • On average, across all sensors in our network, SO2 was within ~10 ppb of reference sensors and PM2.5 was within ~5 micrograms per cubic meter.
  • Did you monitor temperature and humidity (both ambient and within the box)? How did they vary with pm2.5 and so2? - Allison Hughes
    • Yes, we measured RH and air temp inside and outside the box. PM2.5 tended to increase with higher RH and we corrected for this using a statistical approach (details in forthcoming manuscript). The SO2 electrochemical sensors have a non-linear relation with air temp that we used in our calibration algorithms. More details on this here.
  • Ethan, how did you calculate hourly exposure to particular concentrations? Thanks! -
    • Briefly, for each sensor, we multiplied the measured hourly pollutant concentration distribution (for both SO2 and PM2.5) by the population within 5km of the sensor. Complete details are in our forthcoming publication.
  • Did people observe any health effects?
    • Anecdotally, yes, and there is still ongoing epidemiological/toxicological research being conducted by other groups, including the Hawaii Department of Health.
  • What was the monthly cost of running the sensors and how did you test the sensors to verify accuracy? - Kathleen Maltby
    • Data transmission costs on the order of ~$10 per month per sensor. Sensors were calibrated against reference instruments in the field.
  • Since COVID-19 began, has anyone published data to show reduction in PM10 from highway corridors?
    • We haven’t observed this in Hawaii; baseline values are typically very low and our sensor network was configured for vog rather than anthropogenic PM sources.
  • Have you thought about using UAV for monitoring in your Island study? - Zhongren Peng
    • Yes, that would be an interesting approach. We used ground-based sensors because this project was focused more on community exposure near ground-level over extended periods of time.
  • What was the radius of coverage by the sensors?
    • In this environment, plume chemical composition and location can vary on length scales <~5 km.
  • What does DOH station mean?
    • Department of Health
  • How did your observed data compare to modeled data?
  • How do the plastic cases hold up to H2S and UV? - Amy Heidner
    • So far so good! We did notice corrosion on metal pieces in areas that were more exposed to vog (sulfur dioxide + sulfuric acid droplets).

Short-term health effects of traffic-related air pollution exposure in multi-modal commuting in Chengdu, China

Presented By: Yisi Liu, University of Washington

Background: Modern mobility can include mixtures of transportation options, which potentially impact pollution exposures and health.
Objectives: To investigate variations in traffic-related air pollution (TRAP) exposures in different transportation modes and related cardiopulmonary health effects in Chengdu, China.
Methods: This was a randomized double-blind crossover intervention trial. Each of the twenty-one eligible subjects enrolled into the study completed eight two-hour trips on script route between November and December in 2019. Subjects travelled with each of the four modes (walking, bus, subway and car) twice, where they used effective masks once, and another one with sham masks. The order of the two masks was randomized and double-blind. Each trip was separated by at least one day. During the travelling, personal ultrafine particles (UFP), PM2.5, black carbon (BC) and noise levels were monitored using portable sensors. Blood pressure, fractional exhaled nitric oxide (FeNO) and spirometry were measured right before and after each trip.
Results: Mean age of the 21 subjects was 27.4 years; 15 were female and 6 were male. Walking exposed subjects to the highest median PM2.5 levels, while taking a bus had the highest average PM2.5, BC and UFP concentrations. During the transportation, per 1ug/m3 increase in BC, FeNO increased 2.0 ppb (95%CI: 0.2, 3.0). Increased PM2.5 concentrations and lung-deposited surface area of UFP were associated with decreased forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC). Compared to travelling with sham masks, people wearing effective masks had higher FEV1/FVC ratios (0.75%, 95% CI: 0.1% -- 1.4%).
Conclusion: Results from this study suggest that TRAP had adverse effects on respiratory health among healthy young adults. The Intervention of using face masks could prevent the adverse effects of TRAP on respiratory system.

Follow Up Questions and Answers:

  • How long is your monitoring data? How big a sample is needed to validate your models? - Zhongren Peng
    • Answered live 
  • What is the implications of the seemingly contradictory correlation of sham masks and better health outcomes - Amy Heidner
    • Answered live
  • I would like to ask that how do you deal with the negative values of AE51? - Bai Li
    • The AE51 does have many negative values especially when there was sudden temperature change. We used the ONA (optimized noise-reduction algorithm) recommended by EPA to smooth the data and deal with those negative values.
  • There has been made an EKG watchband made recently for the iWatch that gives time-resolved measurements. - Charls Davidson
    • Yes, the iWatch is exciting! We are looking for opportunities to use iWatch in short-term health studies.
  • I believe the PUWP uses Shinyei PPD42NS - how was the quality of PM data (esp between PM10, PM2.5 ) compared to calibration sensors? - Sridhar Rajagopal
    • For the PUWP monitor, we use Plantower sensors. Here is the calibration work just published. The Chengdu study used the same sensor as used in the published study. Zusman, Marina, et al. "Calibration of low-cost particulate matter sensors: Model development for a multi-city epidemiological study." Environment International 134 (2020): 105329.
  • How long were the subjects exposed to? Did you encounter a situation where a subject moved to different places aside from the study area?. If yes, how did this affect the overall estimates? - Emmanuel Appoh
    • Subjects were exposed 2hr each time. Yes, there were times that subjects were in another city on business trips. We deal with such situations and other potential confounders by the pre and post trip health measurement, where the pre-trip health measurement is the way to deal with the uncontrolled exposures. So in the analysis, we used the differences between the pre and post measurements in the health estimation, which is ideally the effect only from the 2hr travelling on road.
  • Can you explain the exhaled  NO (Nitric Oxide) measurement? - Peter Fleming
    • The exhaled NO (FeNO) is a measure of airway inflammation. It has been used increasingly in both research as well as clinical settings, especially for people suffering from Asthma. A FeNO level larger than 35 ppb is considered as Asthma not under control. Here is an article talking about FeNO: Montuschi, Paolo, et al. "Diagnostic performance of an electronic nose, fractional exhaled nitric oxide, and lung function testing in asthma." Chest 137.4 (2010): 790-796.
    • It sounds as if the FeNO testing must be done for individual patients in a clinical setting, rather than any ambient stationary monitors, then.
    • Yes, it should be in an indoor environment with comfortable temperature and humidity, but not exactly a clinical setting. Actually asthma patients could test for their FeNO daily at home. What we did in the study was to conduct the pre and post health measurements all in a CDC office (which was also the starting point of each trip). - Yisi Liu
  • Is it better to replace the concentration used to calculate the inhalation dose with the (Inhalation concentration - Exhaled concentration)?- Bai Li
    • It is a good point to get better estimations of the internal exposure for subjects. I think the difficulty is how can we effectively measure the exhaled concentration and inhaled concentration? I guess it would be very hard to separate the inhaled air and exhaled air at the breathing zone.
  • How will you consider the background concentration of air pollutants? - Tie Zheng
    • Yes, some studies tease out background concentration of air pollution to count air pollution only from traffic. There are several ways to do that. One is to use the measured concentration minus the concentration measured by EPA station at the background site. The other way is to subtract a certain percentage (5% was used in some studies) from the measured air pollution.  We actually did not tease out the background concentration in the analysis, as we are more interested in the health effects of commute (the total air pollution concentration exposed on road), instead of air pollution explicitly from traffic.
  • My experience with PM2.5 sensor using light scattering technique, e.g. Plantower, is the influence of humidity on the mass measurement reading. What is your experience? - Charles Lo
    • Yes, light scattering sensors are impacted by the humidity. We did all the field work within two months (Nov., Dec.) in the winter time, when the humidity was relatively stable with less variation between days. We also had the humidity in the statistical model to kind of control for the effect of humidity. Another way would be calibrating the sensor in chambers with different levels of humidity to get the calibration equation under different humidity; also, some monitors add components at the inlet of the sensor to dry air flowing into the sensor (but this might be hard for low-cost personal/portable monitors). Our lab is right now working to calibrate the sensor under different humidity levels, and there have been publications on this.
  • What is the accuracy of sensors when colocated with AQM? What are the calibration methods used? 
    • Yisi Liu: There is the article about the calibration and accuracy for the Plantower sensor we used in Chengdu study: Zusman, Marina, et al. "Calibration of low-cost particulate matter sensors: Model development for a multi-city epidemiological study." Environment International 134 (2020): 105329. Basically the Plantower sensor works well. The two calibration methods normally used are collocation (sensors calibrate at the regulatory monitoring stations) and lab-calibration with regulatory-grade instruments.

Part 1 Group Question & Answer

TD Environmental - summary of recent air sensor projects

Presented by: Tim Dye, Founder

High-Density Deployment of PM2.5 Sensors in the Maywood Environmental Justice Community

Presented By: Jennifer DeWinter, Sonoma Technology & Felipe Aguirre, Comite Pro Uno

Description: The city of Maywood, California, located in southeastern Los Angeles County, is a community of nearly 30,000 residents in 1.2 square miles. Maywood is surrounded on the north and east by multiple air pollution sources, including the I-710 freeway, rail lines, and local industrial sources. The I-710 freeway—as well as the two main arterial roads in the community, Atlantic Bvld. and Slauson Ave.—have significant heavy-duty diesel truck traffic. Seven of the nine census tracts in Maywood are in the highest decile on CalEnviroScreen 3.0 scores; the other two are in the top quintile. To understand the spatial variation of PM2.5 concentrations in the community and engage residents on air pollution issues, we deployed 20 Purple Air sensors at community residences for one year. We will report on how PM2.5 concentrations vary within the community, how concentrations compare to those at nearby communities, and how concentrations vary with proximity to local sources such as freeways, arterials, rail yards, and point sources. The Purple Air measurements will also be compared to measurements from nearby regulatory monitoring locations that are measuring PM2.5 and other air pollutants. We will report on feedback from the community members and lessons learned.

Follow Up Questions and Answers:

  • Aside from the 2 week collocation with a BAM prior to sampling, did you do any other collocations to assess the correction? - Beth Friedman
    • We also compared to collocation of subset of sensors with a T640 during deployment and compared sensor-to-sensor relationships throughout the project. A post-study deployment may also occur.
  • Could you talk more about the correction method you used for the Purpleair ? - Yngyangzou
    • The correction equations we developed were based on a simple linear regression of the data from each sensor data from a beta attenuation monitor (BAM) during a two week collocation of the sensors with the BAM prior to deployment in Maywood. We will also explore the correction factors and results if temperature and relative humidity are added as variables to the regression.
  • How did you account for local vs background concentration? - Thor-Bjorn Ottosen
    • We’ve evaluated each sensor measurement relative to the sensor wide mean to identify what may be local increases in PM2.5 at each location. We also plan to do additional time series based analysis to quantify local versus regional PM2.5.
  • Instead of seeing the influence of "very local pollution sources" for the hotspots, is it possible that you are seeing instead the influence of local "cleaning"? For example, washing sidewalks increases airborne moisture which may decrease the PM values. - Amy Heidner
    • so you’d need highly specific knowledge regarding the location to truly interpret data. - Rebecca E. Skinner
  • What is the range of cost to say it's a low-cost sensor network, does that implies, Sensor cost or whole device(Hardware, App, data portal)? - Sairam D
    • Knowledge of local micro-climates is essential for data interpretation. These can be anthropogenic, or geographic.  In the first presentation, about vog during the 2018 eruption on Hawai'i, you can see the latter (that purple "leeward eddy '' which pushes the H2S droplets back toward the Kona Coast.) - Amy Heidner
  • What is the location on the road? Upwind or downwind ? - Gerry Bemis
    • We don't have the data we'd really need to answer regarding upwind vs downwind at each of the PurpleAir locations. Generally, winds are from SW during spring/summer and from NE during fall/winter in Maywood. October is mixed in terms of wind directions. So likely experiencing both upwind and downwind along 61st locations. - Jennifer DeWinter
  • What is the accuracy of the PurpleAir? - Amy Heider
    • In our project, the sensors were well correlated with the BAM and showed moderate-good accuracy (R2 of ~0.8, slope of ~2 and intercept of ~4µg/m3). Additional evaluations of PurpleAir accuracy, that are consistent with our findings, are available via AQSPEC (
  • Do you have a Particle number data and is this more useful?
    • Low cost sensors will only count particles larger than about 0.3 microns. As most of the particles in ambient air are smaller than 0.1 microns, counting would provide only limited information. - Brain Stacey
  • What was the error of the measurements? Was the 1-2 ug/m3 measured difference between the sensors meaningful? - Laura Rosales
    • The 1-2 µg/m3 difference was statistically significant, but this is aided by the number of data points. Sensor-to-sensor bias was low (~1-2%) for these particular devices highlighted in the presentation as hotspots. Overall, sensor to sensor bias pre-deployment ranged from 1-10%.
  • What was the average height for mounting your purple air monitors? - Allison Hughes
    • 6 feet was the average
  • How were the PurpleAir sensor locations determined? Was it based on where the volunteers that meet the installation requirements live? Did you ever exclude volunteers who live too close to each other in order to distribute sensors more evenly over the space? - Yoo Min Park
    • It was based on wanting to have monitors spaced around the city and based on the requirements for their placement. We never excluded people because they live close to each other. We have several places where the monitors are close to each other.
  • Were the residents who hosted the sensor involved in the data interpretation? - Yoo Min Park
    • Yes, the residents participated in community meetings to discuss the data and interpretation. They were also involved in taking follow up action such as petitioning the city to ban fireworks.
  • Are there any regulatory demands from the community for the industrial area north of Mayward? - Aaron Rojas
    • The northside of Maywood abuts directly the city of Vernon and there is a lot of industry both on that side and of course in Vernon. The residents there have requested a wall be built around Maywood to exclude all the different plants that exist on our northern border. Also, many projects that are built-in Vernon have little or nothing to say about them because they are controlled by the city of Vernon. We also have Exide a battery recycler in the city of Vernon and a chemical plant next to Maywood elementary school.
  • Any mitigation strategies being considered? If so, which? Juan Sanchez
    • Our group (Comite Pro Uno) is working on various projects like greening our city and building a green wall between Maywood and Vernon, and an urban gardening system for the residents of the city.
  • If the PM data was complemented by a traffic study, this would be excellent source attribution for local traffic exacerbation of emissions pollutants. - Rebecca E. Skinner
    • The results from this project may be used to identify locations for collecting truck and traffic counts in Maywood in a subsequent upcoming project.
  • Is there any performance difference in using Low-cost sensors for measuring Indoor Air Pollution vs. Outdoor. - Praful Dodda
    • Please check with AQSPEC regarding indoor sensor performance but there may be similar biases with regulatory-grade instruments and perhaps RH or temperature.
  • For the Maywood project an additional example of how community members are using the data from the PurpleAir sensors is that after seeing the impact fireworks had on the local air quality, a community member contacted the Mayor of Maywood and shared those results with him and was able to get banning fireworks on the agenda for a recent City meeting. From what I understand during the public comment portion of the meeting community members called in and gave personal accounts of how their health was impacted by the diminished air quality from the fireworks. The city is now in the process of putting in an ordinance banning fireworks. - Jenny Lentz
    • See above response from Coalition for Clean Air regarding use of the project findings.
  • For the Maywood Project: How do you measure highway and train yard PM? If PurpleAir measurement does not give an indication? - Ken Szutu
    • It is possible that some fraction of the PM observed by each PurpleAir is due to these sources. We may apply a time series based approaches to try to differentiate.
  • AQ-SPEC evaluation on the Purple Air PAII can be found here:

Empowering Community Based Organizations to Improve Air Quality in NYC Environmental Justice Communities

Presented By: Jalisa Gilmore, New York City Environmental Justice Alliance & Leslie Velasquez, El Puente Green Light District

Description: The clustering of facilities such as bus depots, waste transfer stations, food distribution hubs, and highways in low-income communities and communities of color unfairly concentrates air pollution and their health consequences. As a result, negative health outcomes such as asthma, heart disease, and cancer rates are significantly higher in environmental justice communities such as the South Bronx, East Harlem, and North Brooklyn.

The New York City Environmental Justice Alliance’s Community Air Mapping Project for Environmental Justice (CAMP-EJ) is a grassroots air quality monitoring research project led by low-income communities and communities of color in New York City’s most environmentally-overburdened neighborhoods. CAMP-EJ empowers four community-based organizations in the South Bronx and Brooklyn to measure and understand their local air quality, and to use this data to drive advocacy campaigns to address the disproportionate impacts of PM2.5. Utilizing hand-held air quality monitors, communities are able to collect and visualize the air quality data in real-time using the AirCasting platform -- an open-source platform for sharing environmental data. 

Our presentation will focus on a citywide, environmental justice perspective of fine particulate matter (PM2.5) pollution in New York, and how low-cost air monitors can serve to educate and empower communities, while also informing advocacy for policies that improve air quality and reduce exposures for all New Yorkers. We will then place the findings, challenges, and lessons learned from mobile and stationary PM2.5 monitoring within the neighborhood context of North Brooklyn where the work of CAMP-EJ is being led by a local community-based organization, El Puente.

Follow Up Question & Answer


  • Who do you bring the data to and what do you ask them to do with it?
    • Jalisa: The data can and will be used in a number of different ways. The data will be used to support already existing advocacy campaigns related to air quality that our members are leading. For example, campaigns around transportation equity, waste equity, and the need for increased green infrastructure in environmental justice communities. Advocacy is typically targeted to elected officials and City agencies. This includes but is not limited to NYC Mayor de Blasio, the NYC Department of Mental Health and Hygiene, and NYS Department of Environmental Conservation. This advocacy ranges from calling for more regulatory monitors in environmental justice communities as well as advocating for less GHGs and co-pollutants in our communities entirely through our NYC Climate Justice Agenda. 
  • How is the community trying to affect change?
    • Jalisa: Community Air Mapping Project for Environmental Justice (CAMP-EJ) is a grassroots, community-led participatory research project born out of the shared concern from our members about air pollution in their neighborhoods. CAMP-EJ focuses on empowering communities to address air quality issues due to polluting environmental sources in their community, and raise awareness around the resulting higher rates of negative health outcomes linked to PM2.5 pollution, including asthma, heart disease, and cancer. We do this through community awareness, community engagement, and using data as an advocacy tool.  Similar to all NYC-EJA’s work we are working to address the disproportionate amount of polluting infrastructure and its associated health impacts in environmental justice communities by using research and advocacy to improve environmental policy in NYC.

Building Youth Capacity for Community Science and Policy Advocacy in Dearborn, MI

Presented By: Natalie Sampson, University of Michigan-Dearborn & Karima Alwishah, University of Michigan-Dearborn

Description: Environmental health inequities are well documented in the U.S., but little attention is given to potentially disproportionate exposures experienced among Arab American communities. EHRA emerged to address cumulative air pollution exposures in Dearborn, MI, where nearly 48% of residents identify as Arab. In response, Environmental Health Research-to-Action (EHRA) is a community-based partnership focused on building skills and multigenerational knowledge in environmental health, community/citizen science, and policy advocacy. Since 2017, EHRA has conducted two 2-week summer academies for youth (16-18 years old) (with another planned in 2019). With action-oriented training and mentorship from government and community leaders, EHRA Fellows have presented on air pollution issues and policy solutions to the Mayor and City Council of Dearborn, Michigan's Attorney General, a U.S. Representative, and other residents and leaders in the region. Learn more here: Fellows report an overall positive experience and feedback on the academy’s interactive pedagogy, refinement of career goals, and marked increases in related knowledge of environmental health science and policymaking. In January 2020, EHRA is also publishing the Dearborn Air Quality and Health report which documents air pollution sources and exposures, related health concerns, and policy solutions. EHRA is beginning to build its capacity for using air sensors as a tool for community mobilization. While much environmental justice research and organizing is conducted by and with youth across the country, few have published on effective models like EHRA.

Follow Up Questions and Answers:

  • This made me think of neighborhoods where all the shopkeepers wash the sidewalks at the same time every late afternoon- must be reflected in local AQ data if collected. - Rebecca E. Skinner
  • In LA for sensor installs and maintenance the Coalition for Clean Air is using the following safety measures:  requiring mask wearing of all parties present, maintaining social distancing during the visit, and using hand sanitizers throughout.  In-person Community Outreach, Meetings, & general Engagement, have been replaced with mailings, phone calls, and zoom meetings (using a paid account so that community members without internet access will be able to call in).  We're also exploring the use of Esri Story Maps as a means of sharing the project and its results in real time, with community members in an interactive way -- the drawback is that story maps require internet to run, so we are limited in terms of how we can remotely share them with community members who don't have internet access. - Jenny Lentz
    • Thanks for this comment. We are grappling with all of this, too. We did use paper maps in our academy and do have a StoryMap version of them on our website: Obviously, COVID-19 and digital access are major issues to consider. We welcome suggestions! 
  • Do you have an example of the students presentation of the data? - Laura Rosales
    • We have the PPTs but I don’t know if we have any recordings. Youth presented to high schools students, community members, and elected officials. While their presentations were well done, the resulting conversations were as or more powerful. Happy to follow up on how this worked (

Part 2 Group Question & Answer


The air and climate implications of energy access in Northern India: A case study in two villages

Presented By: Naomi Zimmerman, University of British Columbia