Poster Presentation Abstracts

Policy and Air Quality Management

Comparative Assessment of Pollutant Concentrations and Meteorological Parameters Between Houston, TX Region and the Rio Grande Valley Region of South Texas

By: Amit Raysoni, The University of Texas Rio Grande Valley

Summary: Air pollution exposure burden within an urban airshed varies considerably and there exists spatial heterogeneity in the pollutant concentrations between multiple urban sites. This research proposal aims to characterize the levels of these variations between the central ambient monitoring sites located in the Houston region and the Rio Grande Valley region of South Texas. We characterized the levels by using Spearman Correlation which tests for the strength of the association between two ordinal variables. This research helps understand the pollution levels in the two regions.

Advances in Voluntary Consensus Standards for Air Sensors

By: David Elam, TRC Companies

Summary: Although EPA provides definitive guidance for the establishment of ambient air quality monitoring networks used for compliance purposes, low-cost sensors are often deployed with little consideration of EPA’s siting and quality assurance requirements.  Although there is overwhelming interest in the use of low-cost sensors to address the full range of environmental areas, voluntary consensus standards (VCS) for evaluating, deploying, and reporting low-cost sensor data are limited.  As discussed at ASIC 2018, low-cost environmental sensors will achieve their full potential when VCS specific to them are available that address sensor evaluation and validation; deployment and siting, data assimilation, integration, and exchange; and reporting.

Since ASIC 2018, there have been significant advances in the development of VCS for air sensors.  Specifically, ASTM is working on three VCS of relevance to air sensors:

•    ASTM D1357-95 ( Standard Practice for Planning the Sampling of Ambient Atmosphere) and ASTM D3249-95 (Standard Practice for General Ambient Air Analyzer Procedures) were reauthorized with the goal of modifying them to address developments in the sensor field.
•    ASTM Work Item 64899 (Performance Evaluation of Ambient Air Quality Sensors and Other Sensor-based Instruments) was initiated to define standard procedures for evaluating sensor performance.

This presentation will examine the content of these standards and work items, describe the ASTM standard development process, outline the procedures that are required to keep standards active, and explain how members of the air quality monitoring community can contribute to VCS development.

Why More Air Quality Data (Alone) Won't Lead to Better Policy

By: Emilia Tjernstrom, University of Sydney

Summary: Using observational data to draw inference about the causal effects of a given policy is challenging but crucial. Basing policy recommendations on associational studies is fraught with issues. I will illustrate common challenges to causal inference and show how the wrong analysis could lead to the wrong policy recommendation. Finally, I will discuss ways to address the key challenges, using as a case study from Kenya as motivating example. 

Our study is designed to estimate the causal effect of air quality on primary school exam performance in Nairobi, Kenya. Using novel data from 23 low-cost sensors across Nairobi and a quasi-experimental design, we will estimate both the cumulative and the contemporaneous effects of particulate matter on test performance. I will discuss how our research design overcomes issues of selection bias, site selection bias, and bias due to unobserved confounders.

To give one example of selection bias, we may expect schools that are closer to pollution sources to be located in poorer, less-resourced neighborhoods, or to be affected by other environmental or economic factors that also influence cognition and educational performance. A naive comparison of students with higher exposure to students with lower exposure would fail to account for these unmeasured confounders. This would then bias any estimates of the relationship between pollution and test performance. I will illustrate how this kind of bias cannot simply be "controlled for," and how in fact certain controls may exacerbate the bias.

Application of air quality sensors relevant to remote sensing and health effect assessments

By: Saliou Souare, Direction of Environment and Establishment Classified/ air quality monitoring center 

Summary: African is the continent with the fastest rate of urbanization. In order to properly inform urban development, encourage greener policy actions and mitigate the effects on human health, access to pollution data (historical and real-time) is critical. The availability of timely, accurate, and highly spatially resolved pollutant data can increase awareness of global air quality and help design effective plans to mitigate its impact, particularly on human health. 
Senegalese Government being aware of the need to improve the quality of life of people; created in 2009 an air quality monitoring centre. Network of 6 fixed stations for measuring air quality for the Dakar region.
The applications that these sensors can provide us are and how will air pollution data promote healthy cities :
•    Support decision makers to develop evidence based air quality management policies (e.g. Clear Air Strategy, State of Environment Report, Substance Relaease Regulation, Urban planning),
•     Assess whether additional industrial activity in an area should be approved or operating within national limits,
•    Ensure pollutant concentrations remain below levels that are considered safe for human exposure,
•    Assess impacts of local emission sources on air quality and identify hotspots – dynamic and static,
•    Facilitate predictive air modeling and continuous monitoring to inform the public, manage traffic sources, assess and report on the impact of sustainable development projects (e.g emobility, solar grid, waste management, etc).
•    Seed entrepaneurship by integrating air quality data into mobile and stationary advertising platforms ( e.g air purifiers sales, outdoor gear or activities) and other innovative business opportunities. 

The Characterization of Low-cost Optical Particle Counters During a Concurrent Evaluation at Two Supersites in London

By: Stefan Gillott, King's College London

Summary: Interest in low-cost optical particle counters (OPC’  s) continues to grow due to their potential to augment research and regulatory air pollution monitoring. New paradigms are available with high density networks and personal exposure measurements using wearable sensors. Evaluations of sensor performance have so far shown inconsistent   results against reference instruments primarily from their sensitivities to meteorological conditions and physical and chemical properties of the aerosol. These sensitivities and the temporal and geographical variability in aerosol composition mean OPC performance may differ between locations and over time. Therefore, limiting the applicability of any calibration carried out away from the target location, or when the OPC moves through different microenvironments. To help quantify these uncertainties, three OPC’s (OPC-N3; PMS6003 and SPS30) were characterised at two supersites in London (Kerbside and Urban Background) concurrently. PM mass reported by each OPC is compared against a reference instrumentation (FIDAS) as well as aerosol size distribution and chemical composition at both sites. This approach allows for the regional meteorology and physical and chemical properties of the aerosol to be similar and therefore isolate the key factors that contribute to OPC discrepancies in different locations. As low-cost OPC’s are routinely deployed in different environments this work will bring new insight and support future studies.  

Wildfire Smoke Adjustment Factors for Low-Cost and Professional Air Quality Monitors with Optical Sensors

By: William Delp, Lawrence Berkeley National Laboratory

Summary: Wildfire smoke is increasingly presenting as an acute air quality hazard to communities around the world. Low-cost particle sensors and monitors can be used to track plumes and quantify the hazard both outdoors and within buildings, but calibration/adjustment factors are needed for accurate assessment. This poster presents adjustment factors (AFs) for several consumer-grade and portable professional devices using data from fires in 2018. During the Northern California Camp Fire, PM2.5 was measured in a well-ventilated lab within the impacted area using a Thermo TEOM-FDMS (Federal Equivalent Method for PM2.5), a Grimm mini-WRAS (particles from 10 nm to 35 m), two professional photometers (TSI DustTrak, Thermo pDR1500), and four monitors with mass-produced optical sensors (AirVisual, PurpleAir, Air Quality Egg, and eLichens). Median AFs for infiltrated PM2.5 were WRAS=0.85, PDR=0.53, DT=0.25, PA=0.48, AVP=0.59, AQE=0.46, and ELI=0.60. We also compared public data from 53 PurpleAir PA-II monitors to 12 nearby regulatory monitoring stations impacted by the Camp Fire and PA devices near stations impacted by smoke from the Carr and Mendocino Complex Fires in California and the Pole Creek Fire in Utah. Camp Fire AFs varied by day and location, with median (IQR) of 0.48 (0.44-0.53). Adjusted PA-II 4-h average data were generally within ±20% of PM2.5 reported by the monitoring stations. Since the LCMs uniformly overreport PM2.5 this leads to discrepancies in the Air Quality Index (AQI) between the LCMs and the official monitoring stations. Adjustment improved the accuracy of the predicted AQI hazard level e.g. from 14% to 84% correct in Sacramento during the Camp Fire.

PM2.5 Air Quality and Children’s Exposures in Mongolia

By: Zhiyao Li, Washington University in St. Louis

Summary: Mongolia has the lowest population density of any country, yet its population centers experience poor wintertime air quality because the cold climate drives strong ground-level inversions and pervasive solid fuels use for distributed residential space heating. Ulaanbaatar (UB, pop. ~1.5MM) has robust air monitoring but measurements outside of UB are sparse. Project objectives include quantifying PM2.5 spatiotemporal variability in Bayankhongor (BKH, pop.~30K) and assessing children’s PM2.5 exposures in kindergartens, maternity wards, and pediatric hospitals.

Pilot studies commenced in Nov. 2019 to inform the Feb. 2020 deployment of larger-scale indoor and outdoor low-cost sensor (LCS) networks. Outdoor PM2.5 measurements in BKH include a USEPA Federal Equivalent Method beta attenuation monitor (BAM) collocated with three devices each of five LCS types to evaluate LCS accuracy, precision, and ruggedness. For the first ten weeks of data, three device types exhibited high data capture and high correlation with the BAM for temperatures as low as -30C and hourly PM2.5 ~10-300 µg/m3 (5th-95th percentiles). These three LCS device types were biased high by 25-40% with a discernible temperature-dependent bias. The pilot study will continue through the springtime to capture Gobi Desert dust impacts on LCS-to-BAM relationships.

Kindergarten measurements focus on children’s exposures and, to the extent practicable, the efficacy of interventions such as air purifiers and mechanical ventilation.  This indoor pilot study includes two BKH and four UB locations.  Each site has one device each of five LCS types (with three of each device at one of two locations), and one device outdoors.  Indoor PM2.5 and indoor/outdoor PM2.5 temporal profiles are observed to vary by kindergarten and type of intervention.

The presentation will summarize results from the pilot studies and first two months of full network deployments, and how these data are being utilized in public communications.

Performance Targets

Integrated Performance Evaluation Systems for PM2.5 Monitors and Sensors in Taiwan

By: An-chi Huang, ITRI, Taiwan

Summary: In Taiwan, the PM2.5 annual air quality standard is 15 μg/m3 and the annual average value is around 15 to 20 μg/m3. Considering health exposures, pollution hot spot tracing and public education, Taiwan EPA cooperates with ITRI in developing PM2.5 monitors/sensors performance evaluation platforms and a series of QA/QC activities. The aim is to specify commercial available monitors/sensors for its bias, precision, data correlation, and data recovery, etc. under various test protocols. For automatic continuous monitors, data quality is evaluated by field collocation comparison with filter-weighted manual reference methods. For sensors, the evaluation platform includes two individual testing systems. One is conducted in the field as collocation comparison with qualified reference monitors and the other is operated under well-controlled wind tunnel conditions, while parameters can be adjusted with wind speed, temperature, relative humidity, and PM2.5 concentration up to 300 μg/m3.  Sensor testing platforms are established and operating since early 2018.

Some research observations may find out the disadvantages of this “Low cost” air quality sensors compared to monitors.  References have shown that the measurement is susceptible to critical climate surroundings, such as the high concentration of PM or severe, extreme temperature and humidity. We found from laboratory test under simulated climate condition of 25 °C 80 RH%, similar to common weather pattern in Taiwan, the slope of data linear regression between sensors versus monitors can be higher than 2 indicating potential bias interference of humidity on sensor performance especially in high concentration level. Using KCl as PM2.5 source, bias can be dramatically different while humidity increases. 

Monitoring data quality of a mobile air quality sensing network using regulatory reference stations

By: Brian LaFranchi, Aclima Inc.

Summary: Large-scale networks of mobile sensing platforms can now measure a suite of pollutants with the potential to provide information about exposure and emissions at hyperlocal spatial scales, capturing unexpected features that are the least straightforward to predict. Data of sufficient quality is necessary to have confidence in the presence and magnitude of these features. Dynamic and rapidly changing pollutant concentrations and atmospheric conditions experienced onboard mobile platforms result in an additional layer of variables that needs to be taken into account when assessing data quality for mobile applications.

We have deployed a network of mobile sensing devices to produce hyperlocal air quality data throughout the entire San Francisco Bay Area. In addition to rigorous calibration and testing of each sensor using directly collocated reference instruments prior to deployment, we utilize stationary regulatory reference stations within air districts in California as checks on sensor calibrations while deployed within our mobile collection network. These comparisons can serve both as quality control checks of individual devices as well as for quantifying and minimizing inter-network bias between independent datasets. 

In this session, we present a detailed analysis of mobile vs. stationary collocations across our fleet in the San Francisco Bay Area, using regulatory site data from the regional agency (note that Aclima’s technologies and data may be subject to a variety of proprietary and intellectual property notices). We explore the results of these collocations as a function of pollutant type, distance, site location (e.g. near road or not), and other variables that may impact the comparison. 

Using a Mobile Platform to Assess Sensor Performance

By: Caroline Parworth, Aclima Inc.

Summary: The development and deployment of mobile platforms of small-scale sensors facilitates the mapping of hyperlocal air quality. These pollutant measurements at hyperlocal spatial scales fill the measurement gaps of traditional stationary sites, providing insight into improving emission inventories and pollution exposure estimates as well as supporting air quality modeling. Interventions and policy decisions based on mobile measurements that aim to improve air quality require data of sufficient quality. Characterization of sensor performance has been a key component of our research and design process. Assessing sensor performance in mobile applications has unique challenges in that pollutant concentrations and atmospheric conditions are dynamic and rapidly changing compared to what might be experienced during a traditional stationary validation exercise. Our team has been using mobile laboratories (vehicles equipped with reference-grade instruments) to evaluate the performance of our sensor-based devices. In this presentation, we will discuss the assessment of sensor performance based on in situ mobile calibration and third-party verification (noting that Aclima’s technologies and data may be subject to a variety of proprietary and intellectual property notices).

Preliminary Chamber Evaluations of Low Cost IAQ Monitors 

By: Elliott Horner, UL Environment and Sustainability

Summary: The ability to monitor indoor environmental parameters in real time provides substantial benefits both for managing indoor air quality (IAQ) and investigating IAQ complaints.  Numerous devices monitoring IAQ, with much lower price points than reference grade monitors, have been introduced commercially in recent years.  However, there have not been concomitant performance validations of many of these low cost monitors (LCMs).  We applied ISO 16000-9 environmental chamber technology to challenge several LCMs with the purpose of determining whether their sensitivity extends into a IAQ useful range.   
Multiple LCMs were set up in a controlled environmental chamber, which was then dosed with several analytes of interest.  Readings from the LCMs were compared either to recorded measurements from reference grade instruments or to results from laboratory analysis of air samples collected from the dosed chambers.  Comparisons were also made among the LCMs.   
One model of LCM was challenged with four levels of formaldehyde.  
•    In two sets of background readings from the 12 LCMs (where formaldehyde levels were <2 µg/m3 as measured per ISO 16000-3), three of 24 LCM readings were >10 µg/m3 (max 25.6 µg/m3).  
•    In four sets of dosed challenges of the 12 LCMs, values from LCMs ranged from the following: 306 to 448 µg/m3 against 671 µg/m3 laboratory measurement; 40.0 to 90.1 µg/m3 against 127 µg/m3, 18.4 to 67.9 µg/m3 against 74.1 µg/m3; and from <2 µg/m3 to 25.6 µg/m3 against 12.1 µg/m3 (four LCMs ranged from 2.1 to 25.6 µg/m3 and eight LCMs at <2 µg/m3).    
Although perhaps acceptable for screening level assessments, the variability in the low µg/m3 range of LCMs for formaldehyde measurement should be recognized when used for IAQ work since concentrations below 50 µg/m3 are of interest.  These preliminary results need to be confirmed and expanded, but the environmental chambers offer a platform to conduct further challenges.

How do air sensors specifically respond to road traffic pollution? – a trial in a road tunnel

By: Ian Longley, NIWA

Summary: There currently exist a range of low-cost air quality sensors that claim to detect common ambient air pollutants. These are increasingly being deployed on many environments from sparsely populated rural areas, to dense urban areas, with very different emissions source profile and it its assumed that these sensors have a similar response regardless of where they are deployed and what the sources pollution are.
This assumption is not straightforward when dealing with particulate matter where the dominant measuring principle is light scatter which has a different sensitivity to different sizes of particles. Anecdotal evidence has shown that low-cost particle sensors that perform well in wood-smoke dominated areas, don’t replicate their performance when in traffic dominated environments.
During November and December 2018 we conducted a trial of some medium and low-cost air sensors in and around a busy road tunnel co-located with regulatory-grade instrumentation. We selected a road tunnel because it offers the advantage of a wide range in concentrations every 24 hours, a realistic pollution mix dominated by road traffic emissions and good traffic metadata. Sensors included in the trial were NIWA’s ODIN (Plantower based) and MetOne’s ES-642 optical particulate matter sensors as well as SPEC’s NO2 gas sensor.
Our key findings were that the SPEC NO2 sensor performed remarkably well when compared to a regulatory-grade chemiluminiscence instrument while inside the tunnel but it was severely influenced by ambient temperature when operating outdoors. The optical particulate matter sensors agreed well with each other and with a Grimm 11E aerosol spectrometer but they reported roughly 50% of the readings of a regulatory-grade Beta-gauge instrument. The PM sensors didn’t degrade their response when deployed outdoors.

Lessons Learned from Deploying High Density Low-Cost Sensor Networks in Cities

By: Sean Wihera, Clarity Movement Co.

Summary: We are interested in participating in the Performance Targets topic through a presentation or panel. The rapid adoption of low-cost sensors and cloud-based data management tools by government agencies and air districts have demonstrated the value of high resolution air quality data and its implications for pollution research and management. As an early technology provider in the field since 2014, Clarity Movement Co. has deployed more than a thousand Clarity Node sensors in 70+ cities throughout 30 countries and 6 continents. In this presentation, we will share key insights on sensor performance and network design gained from these real world projects. In particular, we will highlight how different use cases have unique performance requirements low-cost sensor networks. In considering the goals of a project, a singular indicator, such as sensor accuracy represented by R^2 values or spatial coverage represented by number of measurement points, is not enough to evaluate the network’s performance. Users must take a multi-dimensional approach to evaluate the requirements for factors such as reliability, accuracy, consistency, ease of deployment, and cost of ownership to design a low-cost sensor network that maximizes the network’s value for their distinct project goals. Drawing from our past projects, we will outline the key requirements and use cases that we’ve identified and review a few case studies through this framework of performance evaluation. The goal of this presentation is to help the user understand how to implement a low cost sensor network most compatible with their unique project needs.

Long-term Field Evaluation and Application of Low-Cost Air Monitoring Sensors in a California Community

By: Zemin Wang, University of California, Los Angeles

Summary: More affordable and portable low-cost air sensors are gaining popularity for quantifying the spatiotemporal variability of particulate matter (PM). However, the performance of these sensors under real-world conditions requires field evaluation. In this study, we evaluated the long-term performance of a popular low-cost PM sensor, PurpleAir II, over two years in a California Community in various conditions, including low traffic, high traffic, indoor cooking, and wildfires. 
We have deployed 30 PurpleAir sensors (12 outdoor and 18 indoor) at an apartment complex located in the vicinity of both sides of the I-405 freeway in Los Angeles, CA since May 2017. Meantime, we administered questionnaires for household cooking behaviors and ventilation conditions. The reference instrument, DustTrak, were collocated with the PurpleAir sensors in various environmental conditions. We analyzed the qualification of data collection and the correlations between PM concentration and several environmental factors. During our sampling campaign, the Woolsey fire occurred 20 miles northwest from the study site. 
Over the sampling period, the outdoor sensors had an average of 41% data missing, while the indoor sensors had an average of 15% data missing. PurpleAir II sensors showed better correlations with DustTrak when measuring PM2.5 (R2=0.99) compared with measuring PM10 (R2: 0.24-0.83), and the absolute values of most PurpleAir Sensors were always higher than DustTrak. The results show that the PM2.5 Indoor/Outdoor ratio during cooking hours (14.3) was substantially higher than that during non-cooking hours (1.5). During the Woolsey fire, we found a threefold increase in outdoor PM levels and a strong correlation between indoor and outdoor PM (R2: 0.59-0.64). Overall, we conclude that PurpleAir sensors are effective in detecting PM emissions from various indoor and outdoor sources, but data loss and environmental conditions could induce some limitations in their long-term performance. 

Exposure & Health

Sensor Based Monitoring in three National Parks on the Island of Hawaii: Results from the 2018 Kilauea Eruption

By: Barkley Sive, NPS Air Resources Division 

Summary: Hawaii Volcanoes National Park (HAVO) attracts more than two million visitors each year and is unique in the national park system because it experiences extremely high concentrations of sulfur dioxide (SO2), far higher than any other national park and urban areas. The NPS operates a real-time sensor based SO2 advisory system to notify and protect the public, park staff and other affected parties from SO2 and particulate exposures that routinely exceed the EPA standard of 75 ppb. The current HAVO SO2 advisory system network consists of 9 SO2 monitoring sites that utilize pairs of electrochemical sensors, 7 with wind speed and direction sensors, and 3 with PM sensor based measurements.  As a result of the 2018 volcanic activity of Kilauea in the lower East Rift Zone, additional sensor based monitoring was deployed in HAVO, in addition to Kaloko-Honokōhau National Historical Park (KAHO) and Pu'uhonua o Hōnaunau National Historical Park (PUHO) on the island of Hawaii.  The NPS installed 2 portable SO2 sensor systems at the Kahuku Unit of HAVO; which included the integration of 3 PurpleAir PM sensors. A total of 5 portable CO2 sensors were deployed at HAVO, KAHO and PUHO to determine the viability of these measurements as a tracer for plume encounters with elevated SO2 and high PM2.5 episodes. Whole air sampling for volatile organic compounds was also conducted at HAVO, KAHO and PUHO to characterize the volcanic and anthropogenic emissions for trace gases.  Additionally, 2 E-BAMs and 3 additional PurpleAir sensors were also co-located for 25 hours at KAHO in order to test the agreement of the instruments prior to deploying separately at KAHO and PUHO.  Results from the 2018 eruption event and the utility of the sensor based advisory system at HAVO and the additional sensor based monitoring at KAHO and PUHO will be presented.

High-Temporal Resolution Personal Exposure Pilot Study in Inland Southern California

By: Cesunica Ivey, University of California, Riverside

Summary: We present the results from a pilot personal exposure study that included 18 participants from inland Southern California. Participants carried a light-weight pack for seven days that housed a low-cost, wearable monitor for particulate matter concentration, relative humidity, and temperature, and a GPS data logger. Using DBSCAN, Google Maps, and speed measurements, we assigned one of seven microenvironment classifications to each cluster of PM2.5 measurements. We expected to see that participant personal PM2.5 exposure would follow the trend of ambient PM2.5, which was observed to be lower at night (inactive period) and higher during the day (active period) during our study. In contrast, the personal PM2.5 exposure measurements were more temporally and spatially variable. When examining hourly temporal variability, participants from Moreno Valley had fluctuating PM2.5 levels, while participants from Riverside were exposed to higher PM2.5 concentrations during the inactive period and lower PM2.5 concentrations during the active period. From the microenvironment analysis, several participants were exposed to higher concentrations at home compared to non-residential locations. We also carried out a spatial and temporal comparison between ambient PM2.5 and personal PM2.5. For the majority of the one-week measurement period, most participants had lower personal PM2.5 exposure compared with the corresponding ambient concentrations. However, during brief periods some participants had very high personal PM2.5 exposure levels compared with the corresponding ambient concentrations. The results here support that idea that personal exposure closely correlates with personal behaviors and activities in inland Southern California. Our study highlights the potential biases in exposure studies for acute health effects analyses. 

Associations between air pollutants, housing characteristics, demographics, indoor environmental factors and asthma outcomes in Chicago, IL

By: Insung Kang, Illinois Institute of Technology

Summary: In this work, we present preliminary results on associations between air pollutants from both indoor and outdoor, housing characteristics, demographics, indoor environmental factors, and adult asthma outcomes in Chicago, IL. We assembled air quality monitoring boxes for indoor and outdoor to measure size-resolved particles (0.3-10 µm), ozone (O3), nitrogen dioxide (NO2), formaldehyde (HCHO), carbon dioxide (CO2), carbon monoxide (CO), as well as temperature and relative humidity. 41 homes were monitored for 7 consecutive days on a quarterly basis from July 2017 to August 2018. In addition, 54 adults with asthma completed a baseline survey at the beginning of the study as well as monthly follow-up asthma control test (ACT) surveys. Participants were divided into two groups based on their average ACT score: a well-controlled group (ACT>19, n=33) and a poorly-controlled group (ACT≤19, n=21). Preliminary results indicate a negative association between ACT scores and indoor NO2 concentrations across the study population, with a Spearman correlation coefficient (rs) of -0.32 (p=0.024). For only those homes with observable dampness (49% of the homes), negative associations between ACT scores and several pollutant measures were observed, including indoor NO2 (-0.47; p=0.030), indoor O3 (-0.70; p=0.001), and I/O O¬3 ratio (-0.46; 0.036). Additionally, white participants were less likely to have poorly controlled asthma than other races, with an odds ratio (OR) of 0.12 (95% CI 0.03-0.50). Self-reported air freshener use, observed dampness, and occupant density were each associated with a higher risk of poorly controlled asthma, with OR of 4.09 (95% CI 0.99-16.93), 3.05 (95% CI 0.85-10.91), and 2.53 (95% CI 0.82-7.86), respectively. This study provides a unique opportunity to explore the utility of air sensors for health studies and further analysis will reveal the more interesting outcomes.

Leveraging mobile monitoring, modelling, and satellite remote sensing to estimate the health burden of air pollution on the hyper-local scale: case study for the California Bay Area

By: Veronica Southerland, George Washington University

Summary: Estimates of the health impacts associated with ambient air pollution in the United States are typically reported at the state or county level, masking potential heterogeneity in impacts at finer spatial scales.  The spatial distribution of air pollution health impacts at finer scales can reveal neighborhoods that may be experiencing greater than average exposure. Estimating air pollution health impacts at the hyper-local scale is now possible with pollutant concentrations derived from satellite remote sensing and mobile monitoring. The Environmental Defense Fund and Google Earth conducted mobile monitoring of black carbon and nitrogen dioxide (NO2) in the Bay Area using Google Street View (GSV) cars outfitted with air pollution monitors. We estimate health impacts at 100m resolution in the Bay Area using epidemiologically-derived health impact functions with concentrations measured by mobile monitors and other pollutant concentration datasets. For estimating the health burden from fine particulate matter (PM2.5) we use satellite-derived PM2.5 estimates. For NO2, we estimate health burdens using concentrations from GSV mobile monitoring and from a land use regression (LUR) model. We explore how health impacts differed when using mobile monitoring versus satellite and LUR concentration estimates, and when using varying disease rate data. We find using highly resolved baseline disease rates and mobile monitoring concentrations reveals spatial variability is obscured when more coarsely resolved data inputs are used. Both mobile monitoring and LUR modeling capture intra-urban spatial variation in pollutant-attributable health outcomes, though estimates using mobile monitoring concentrations identify a larger magnitude of intra-urban variation. Hyper-local estimates may help decision-makers understand how air pollution affects neighborhoods and populations and where to target interventions to maximize health benefits and reduce disparities.

Development and Application of a Personal Exposure Device for Daily PM2.5 Monitoring

By: Wen-Cheng Vincent Wang, Research Center for Environmental Changes, Academia Sinica, Taiwan

Summary: Low-cost sensors are being increasingly distributed in the places of interest, not only in the atmospheric environment but also in living micro-environments. The small and inexpensive sensors are being designed for this coming type of environmental researches. The integration of these various sensors has become a new challenge. In this work, we focus on the development of a personal exposure device, called AS-LUNG-P, short for Academia Sinica-Lung Portable Version to evaluate whether the device is applicable for daily exposure monitoring.
AS-LUNG-P integrates low-cost sensors of PM2.5 (Plantower PMS3003), temperature, relative humidity, GPS, and motion. The device is a portable size of 13 cm * 6.8 cm * 4 cm with a weight less than 177 g. The choices of 15-sec, 30-sec, 1-min, and 5-min data intervals can be used for different study designs. Real-time data can simultaneously be transmitted with the built-in WIFI system back to the cloud database and be recorded to an inserted SD-card for the back-up.
Performance evaluation of the portable AS-LUNG was conducted in the laboratory. The PM sensor performance is compared to one research-grade instrument (Grimm 1.109 , GRIMM Aerosol Technik Ainring GmbH & Co, Ainring, Germany) side-by-side in the chamber tests with good correlation coefficients (0.99 − 0.81). The response of the PM2.5 sensor can be programmed to be corrected based on the regression functions when the data are transmitted back to the cloud database. 
Traditionally, personal exposure studies were conducted to recruit volunteers to carry heavy and expensive instruments. To convince people to carry the expensive but vulnerable instruments has been difficult in recruitment. This personal exposure device solves the problem and initiates the new frontier of exposure researches. Our results indicate the low-cost sensors having great potentials for atmospheric aerosol research to assess PM2.5 exposures in high spatiotemporal resolution with much lower costs.

Laboratory evaluate the particle size, particle composition, temperature and humidity effect on low-cost PM sensors's performance 

By: Yangyang Zou, The Ohio State University

Summary: Low-cost (<$250) particle sensors and consumer devices offer the possibility of inexpensive and ubiquitous monitoring and control of airborne particles in indoor environments.  However, the reported performance of these low-cost sensors varies widely.  In order to understand the performance of these sensors and the conditions under which we can rely on them for indoor environmental monitoring and control, we investigated the effects of concentration, particle size and particle composition on the performance of eight commercially available integrated devices and bare sensors. We conducted both laboratory and field studies using sources that included atomized (NH4)2SO4, propane torch-generated black carbon, polystyrene latex spheres (PSLs), burnt toast smoke, outdoor air, and incense. Low-cost sensors were collocated with the reference instruments and several different features of the outputs were compared between the sensors and the reference instruments. Most sensors were able to detect particles smaller than 1 micron both in the field and the laboratory.  In general, performance increased with particle size in the range from 100 nm to 700 nm, with detection ratios from 10%-60% for most sensors. For similar-sized particles, the PSLs were detected most effectively, followed by (NH4)2SO4, and black carbon.  Future research will examine the effect of temperature and relative humidity on sensor performance.

The Impacts of Dust Particles on Human Lung Cells - an Analysis at the Single Cell Level

By: Karin Ardon-Dryer, Department of Geosciences, Atmospheric Science Group Texas Tech University

Summary: Aerosols particles (Natural and anthropogenic) are a key component of our atmosphere, their presence defines air quality levels and they can affect our health. Small particles penetrate into our lungs and this exposure can cause our lung cells to stress and in some cases leads to the death of the cells and to inflammation. During dust storm events there is an increase in particle concentration, many of them are breathable particles that can penetrate deep into our lungs. Exposure to dust particles can lead to respiratory problems, particularly for people with asthma. Therefore, during and after a dust event the number of people who are hospitalized with inflammation and respiratory problems increase. However, the exact mechanism that causes these health problems is still unclear. In this project, we are investigating the impacts that dust particles from different sources and of different concentrations (doses) have on human lung cells, performing a new and unique analysis at the single cell level. To accomplish this, each individual lung cell is continuously tracked after being exposed to dust particles. We monitor the behavior of the cell over time, identify the cells time of death and type of death (e.g. cell explosion). With this analysis, we can quantify cell death as a function of dust concertation (doses); to our surprise, an increase in cells death was not observed only as a function of an increase of dust concertation. In addition, we noticed that the way particles come in contact with cells, by sticking to or being engulfed by, and the interaction duration has an effect; cells that interact with dust particles for a longer period died earlier compared to cells with a shorter interaction period. These findings will help us to better understand the health related consequences of exposure to dust storm events and serve as a baseline for when evaluating other aerosol. 

Source Characterization & Identification

Atlanta Rail and Port Sensor Project: An Air Quality Characterization Pilot Study

By: Ryan Brown, US EPA Region 4

Summary: As part of the Atlanta Rail and Port Sensor (RAPS) project, the US Environmental Protection Agency (EPA) Region 4 and Georgia Environmental Protection Division (EPD) are conducting a pilot study for measuring particulate matter and black carbon concentrations around the largest railyard in Atlanta, Georgia using lower-cost air sensor technology. The objectives of the project are to evaluate air quality near the railyard and to evaluate the utility of air sensor technologies in understanding near-source air quality patterns. Ten PurpleAir particulate matter sensors, two Davis Instruments meteorological stations, and two Aethlabs microaethalometers were used to collect over six months of air quality and meteorology data at various distances and directions around the railyard. Pre and post-performance testing for the sensors was conducted at Georgia EPD’s multipollutant air monitoring site in the Atlanta area. The Atlanta RAPS project was conducted on a limited budget and existing EPA and GA EPD staff resources. This presentation will cover the project procedures, best practices, and challenges, as well as lessons learned for conducting lower budget near-source-community air quality studies. Additionally, we will present our initial findings and analysis from Atlanta RAPS project.

Verification and testing of mobile air-quality monitoring equipment

By: Chen Fan Lun, Industrial Technology Research Institute

Summary: The Environmental Protection Administration in Taiwan has deployed over 8,000 low cost sensors in industrial parks, communities, and transportation areas for long-term environmental monitoring. In view of the mobile pollution of auto and motorcycle exhaust in Taiwan's air, the Environmental Protection Administration, for the sake of grasping mobile pollution sources, has applied low cost sensors in monitoring autos and motorcycles at idle speed, finding that their PM2.5 emission is subject to the influence of atmospheric diffusion significantly. Meanwhile, the height of the installation sites of sensors affects its effectiveness significantly, with the density of PM2.5 detected by sensors at the height of 1.5 meters only half that of those at 0.5 meters. In comparing the emissions of motorcycles, automobiles, and diesel-oil cars, there is no conspicuous PM2.5 emission by motorcycles and autos with regular maintenance, But high PM2.5 concentration of 31μg/m3 and 300μg/m3, respectively, of autos and motorcycles without regular maintenance. Even with regular maintenance, diesel-oil cars still pollute 70μg/m3 PM2.5. Meanwhile, four sets of low cost sensors were installed on the four sides of car roof. It was found that with the car in movement, density of PM2.5 detected by the sensors were affected significantly by wind speed and driving speed. In order to rule out the effect of wind speed, rapid movement, and vibration caused by driving car, mobile low cost sensors feature flow-velocity control and vibration proof and is connected to TSI-8533 for inlet-air sampling. Test in movement shows that correlation coefficient (R2) between TSI-8533 and mobile micro sensors reached 0.86~0.64, with median bias error standing at -12.5%. It was observed that whenever there were high value for data collected by mobile sensors, high-polluting vehicles could be found in the neighborhood, which testifies to the efficacy of the mobile sensors. 

Predicting traffic-related air pollution using feature extraction from built environment images 

By: Arman Ganji, University of Toronto

Summary: This study develops a set of extraction algorithms that work based on image processing and single view metrology to extract built environment features from Google satellite and street view images, reflecting the micro-characteristics of an urban location as well as the different functions of buildings. These features were used to train a Bayesian Regularized artificial neural network model to predict near-road air quality based on measurements of ultrafine particles and black carbon in the City of Toronto. The resulting models (adjusted R2 of 75.87% and 79.10% for UFP and BC) were compared with similar ANN models developed using the same predictors, but extracted from traditional GIS databases, and exhibited a higher predictive power (adjusted R2 of 58.74% and 64.21% for UFP and BC for the models with GIS variables), thus highlighting the higher classification accuracy of the proposed methods compared to GIS layers that are solely based on satellite images. A comparison with a traditional land-use regression model also demonstrates that the Bayesian Regularization neural network, with its low sensitivity to the network architecture and balance between output variability and acceptable bias, can achieve a higher predictive power.

Making sense of mobile measurements - a hybrid modeling approach to quantitative source characterization

By: Benjamin Fasoli, University of Utah

Summary: Spatially resolved air quality measurement methods including mobile systems and dense spatial networks are enabling new methods for characterizing intra-city pollution sources. Beginning June 2019, Salt Lake City is hosting two Google Street View vehicles outfitted with research grade instruments measuring PM2.5, NO, NO2, BC, CO2, CH4, CO, and O3. These data supplement other Salt Lake City efforts to deploy dense networks of stationary and mobile air quality measurements. We apply statistical techniques to identify instrument failures and decompose measurements into components that represent hyper near-field enhancements and background. Here, we demonstrate a hybrid modeling approach using machine learning paired with the STILT particle dispersion model to explore relationships between observed tracer enhancements and source-segregated emissions. 

Advance reverse trajectories for Source Identification from small sensor triggers

By: Andres Quijano Garcia, Envirosuite Corp


One of the main advantages the advancements on sensor technology offer is the possibility of identifying potential emission sources from boundary monitoring. One of the challenges is that More often than not, airflow in industrial situations is subject to interference from structures. For that reason, popular trajectory models used (e.g. HySplit) cannot accurately depict particle motion in complex flow situations. In addition, the temporal resolution of the data often used in the trajectory model is too coarse to resolve the flow in these situations.

A new pseudo 3D model based on the Navier Stokes equations, is being developed by the Envirosuite team. These are equations which can be used to determine the velocity vector and pressure field that applies to a fluid, given some initial conditions. With this approach we can build a 3D kinematic trajectory model (D’Abreton, 1996; D’Abreton and Tyson; 1996). The model is Lagrangian, with motion being described in terms of air parcels moving with air streams.
This model allows near real-time solution of complex flow from a single onsite weather station and sensor data a trigger event. The complex flow model can be used to drive a particle trajectory model to determine near-field transport in disturbed flow. Our experience so far demonstrates that emission sources can be identified from a combination of trajectories. 

An earlier version of the model was used on some of the Next generation emissions measurement (NGEM) test completed at the Rubber Town Project and published on paper “Rubbertown Next Generation Emissions Measurement Demonstration Project”

Evaluation of Low-Cost Particulate Matter Sensors for Measuring Wildfire Smoke

By: Dena Vallano, Region 9, U.S. EPA 

Summary: Until recently, most wildfire air quality impacts were determined by permanent stationary regulatory monitors that are used to calculate the Air Quality Index (AQI). Low-cost particulate matter (PM) sensors have found widespread use by the public in smoke impacted areas but have not been evaluated at the high smoke concentrations frequently encountered near wildfires. We collocated three low-cost PM/air quality sensor systems (Aeroqual – AQY1, PurpleAir - PAII-SD, Sensevere - RAMP) with reference PM monitors near three wildfires in the western U.S. and one prescribed fire in the eastern U.S. (max PM = 295 µg/m3). The sensors were moderately - highly correlated with the reference monitor (hourly averaged r2 = 0.52-0.95). All sensors overpredicted PM2.5 concentrations, with an average normalized mean bias of 41%, 62%, and 40% for AQY1, PAII-SD, and the RAMP respectively. 

Calibration factors for individual fires varied, likely due to the different concentration ranges observed at each fire. By combining all datasets, a smoke specific calibration factor was developed that reduced the normalized root mean square error to less than 35%. The calibration factors varied among the sensors, demonstrating the impact of the physical configuration of the sensor and the algorithm used to translate the size and count information into PM concentrations. These results suggest the low-cost sensors tested here can fill in the large spatial gaps in monitoring networks near wildfires with errors of less than 10 µg/m3 in the hourly PM2.5 concentrations when using a sensor specific smoke calibration factor.

This abstract was reviewed by EPA and approved for publication, it may not necessarily reflect official Agency policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. 

Community impacts of aircraft traffic

By: Elena Austin, University of Washington

Summary: The Mobile ObserVations of Ultrafine Particles Study (MOV-UP) is a two-year project funded by Washington State to analyze air quality impacts of ultrafine particles from aircraft traffic for communities near and underneath Seattle-Tacoma International Airport (Sea-Tac) flight paths. The study assessed ultrafine particle concentrations (UFPs) within 10 miles of the airport in the directions of aircraft flight. The primary aims were to (1) assess the concentrations of UFPs in areas surrounding and directly impacted by aircraft traffic; (2) distinguish and compare UFP concentrations attributable to aircraft-related and other sources and; (3) coordinate with local governments, and share results and soliciting feedback from stakeholders. A project advisory committee (PAC) was formed to ensure strong participation of community and local representatives at the onset. Hypothesis development, identification of knowledge gaps and dissemination efforts were conducted with full participation of the PAC.  

Mobile monitoring was conducted to maximize the area monitored within the constraints of the instrument and staffing budget. Over the course of 2018, a fixed route around Sea-Tac was sampled using two simultaneous mobile platforms on 63 distinct days over 4 seasons. To distinguish potential sources of UFP, multivariate eigenvalues generated by a Principal Component Analysis (PCA) were used to discriminate between two important sources, aircraft traffic and roadway traffic. These two features accounted for 61% of the overall variability in the pollutant dataset. Total concentrations of UFP alone (10 - 1000 nm) did not distinguish roadway and aircraft features. However, key differences were identified in the particle size distribution and the black carbon concentration for roadway and aircraft features. The MOV-UP study findings demonstrate two clear and consistent spatial features of ultrafine particles independently associated with vehicle traffic and aircraft emissions. 

Development of Analytical Tools to Support AB 617 Communities

By: Jeremy Herbert, California Air Resources Board

Summary: The AB 617 Community Air Protection Program aims to reduce cumulative exposure from criteria pollutants and toxic air contaminants in California’s most impacted communities.  This program requires the ability to understand pollution sources or source categories within affected communities in order to inform community assessment, identification, and selection and to support development and implementation of air quality monitoring and community emission reduction programs within selected communities.  

A variety of data analysis techniques and tools can be used to better understand a community’s emission sources.  Choosing the best approach is largely dictated by the availability of air quality data and the representativeness of these data for the intended use.  This includes consideration of the methods or sensors used to collect the data. 

This presentation will discuss technical tools to analyze air quality data such as simple trend analysis, wind/pollution roses, to more complex, receptor-based source apportionment techniques such as positive matrix factorization (PMF).  These techniques can be used to identity emission sources or categories of sources contributing to cumulative air pollution exposure in a community. Current and future communities can benefit from these tools to develop, modify or track progress of their emissions reduction programs. 

New Ambient Sample Capture Apparatus Provides Speciation Methods using Next Generation Air Monitoring Networks 

By: Ken McGary, Apis-US, Inc.

Summary: Performance of modern low-cost sensor networks is fast approaching the realm of full-on reference instruments. Electrochemical and PID gas detection levels are now often in the low single-digit-ppb domain, and inexpensive optical particle counters can show impressive correlation to reference-grade analyzers. However, a major drawback remains the inability to speciate detected pollution levels of both Volatile Organic Compounds VOCs) and Particulate Matter (PM).

Apis solves this problem with our Data Fusion platform, using sensitive broadband sensors, sophisticated anomaly detection, and our new Capture Box accessory to collect ambient air samples for lab analysis using tried and trusted methods. For VOCs this means triggered subatmospheric pressure sampling with SUMMA canisters and GC/MS back in the lab. For particulates, a triggered hi-volume sampler such as the Tisch Environmental HiVol can be used to collect samples for various physical or chemical characterizations. 

In either case, the time-consuming hit-and-miss nature of ambient sampling is refocused on sampling during pollution “events” recognized in real time, vastly increasing the efficacy and cost-effectiveness of these expensive analytical procedures. These selective sampling techniques are already paying dividends in a Colorado sensor network deployment near oil and gas drilling sites. We are now also deploying VOC capture stations in a variety of industrial and community projects, and are looking for new partners to further expand the usefulness of this new paradigm based on well-established analytical methods.

Total non methane hydrocarbon emissions in Los Angeles: low cost air quality sensor network and relation to urban oil drilling

By: Kristen Okorn, University of Colorado Boulder

Summary: Past studies have shown causation between oil drilling and total non methane hydrocarbon (TNMHC) emissions, which can cause discomfort to humans and contribute to climate change. In the studies conducted, a network of low cost air quality monitoring sensors was deployed in each of two communities surrounding oil drilling sites in Los Angeles, California to determine emissions on localized spatial and time scales. Linear regression was used to fit the sensor data to reference data, ensuring a higher degree of accuracy. Higher TNMHC levels were observed at monitor locations closest to the oil drilling sites as well as at control sites situated near major roadways. Sensor selectivity and cross sensitivities may also be explored. This data serves to clear up the muddled sources of TNMHCs in complex urban environments, where emissions from traffic and other human activities can be difficult to separate from drilling and other industrial processes. Sharing localized pollution data with community members can help them understand their risk level and enables them to protect themselves from future emission events. Studies with low cost air quality monitors make it possible to assign individual exposure levels within a small spatial radius. 

Use of the Vehicle Add-on Mobile Monitoring System (VAMMS) for Smoke Monitoring 

By: Maiko Arashiro, U.S. Environmental Protection Agency 

Summary: Many factors such as pollution sources, topography, and weather conditions can affect the spatial variability of fine particulate matter (PM2.5 or particles smaller than 2.5 aerodynamic micrometers) over a given region. In mountainous regions, wintertime inversions coupled with smoke sources provide scenarios where PM2.5 levels can be highly spatially variable. To characterize this spatial variability, the U.S. Environmental Protection Agency (EPA) has developed a user-friendly mobile monitoring system designed to map PM2.5 concentrations from any vehicle. The Vehicle Add-on Mobile Monitoring System (VAMMS) is a self-contained suitcase-sized package that runs on rechargeable battery power. The VAMMS is equipped with an isokinetic probe, magnetic probe mount, GPS module, and a mid-cost particulate matter sampler (pDR-1500, Thermo Scientific). The VAMMS data are processed, analyzed, and visualized using custom R scripts as well as through a web-based interactive data visualization program ( The full system is designed to facilitate usability for air monitoring professionals and has been used to evaluate PM2.5 spatial variability in three field locations. During the summer of 2019, the VAMMS was utilized to investigate wildfire smoke spatial patterns by researchers in Missoula, Montana and by the Decker Fire’s Air Resource Advisor in areas near Salida, Colorado. Additionally, the VAMMS was utilized at the Hoopa Valley Tribal Reservation in northern California to study PM2.5 spatial patterns, heavily influenced by wood smoke emissions, during 2019-2020 wintertime inversions.
The views expressed in this abstract are those of the authors and do not necessarily represent the views or policies of the U.S. Environmental Protection Agency.

A VOC sensor suite for ppb-level detection and rough speciation

By: Jiayu Li, Carnegie Mellon University

Summary: Volatile organic compounds (VOCs) are harmful to human health and the environment. Current VOC measurement techniques are expensive and complicated. Low-cost VOC sensors suggest a promising future to provide additional and complementary information to conventional VOC measurement techniques. The goal of this study is to realize ppb-level detection and VOC rough speciation. We will fabricate a VOC sensor suite containing eight commercial VOC sensors, including metla oxide sensors (MOS), electrochemical sensors (EC), and photoionized detectors (PID). From our preliminary study, VOC sensors operating on different working principles responded differently to the same pollution event, which is likely due to their different selectivity. Taking advantage of such differences, with laboratory and field calibration, we will lower the limit of detection (LOD) by signal conditioning and advanced machine-learning calibration algorithms. Furthermore, we will compare sensor responses across classes of VOCs as a function of their composition. It is unlikely that a low-cost sensor will have sufficient sensitivity to identify individual VOCs at sub-ppb concentrations. Instead, our approach will be to develop calibrations for groups of VOCs that are chemically similar and emitted from similar sources. We will start by building calibration models for two important classes of VOCs, BTEX (benzene, toluene, ethylbenzene and xylene) and biogenic (isoprene and monoterpenes) species. We will then move on to other groups of VOCs including oxygenated secondary VOCs, small aldehydes (e.g., formaldehyde and acetaldehyde). Machine-learning algorithm, principle component analysis (PCA), and analysis of variance (ANOVA) will be used to realize the rough speciation considering the different selectivity of various sensors. We will present the performance of the VOC sensor suite and developed algorithms, demonstrating their advantage for ppb-level VOC detection and rough speciation.


Evaluating multiple calibration methods and characterizing measurement uncertainty for the Breathe London sensor network

By: Daniel Peters, Environmental Defense Fund

Summary: Calibration of low-cost air quality sensors is often necessary to improve measurement accuracy and to ensure that data quality is sufficient for intended applications ranging from scientific research to policy analysis. The Breathe London project, convened by Environmental Defense Fund (EDF), operates a network of 100 fixed sensor pods in greater London. EDF and Breathe London partners including the University of Cambridge have undertaken extensive efforts to calibrate deployed pods using multiple approaches. These include reference site collocations, transfer standard collocations with pods in the field, and a novel network-based calibration algorithm developed by Cambridge. The parallel development of multiple calibration approaches on the Breathe London network allows the unique opportunity to evaluate the performance of different calibration approaches on a large network. This presentation will demonstrate efforts to quantify the performance of multiple calibration methods and characterize the NO2 and fine particulate matter (PM2.5) measurement uncertainty associated with pods calibrated using different approaches. The results provide insight on calibration method selection and have the potential to inform procedural design for future sensor networks.

Assessment of correction methods for BC artifacts in AE51s facing rapid temperature changes

By: James Ross, Lamont-Doherty Earth Observatory of Columbia University

Summary: Monitors for personal sampling face rapid change in temperature as subjects change locations. During our recent study of bicyclists we have learned that the AE51 (AethLabs®), which we rely on for its high-quality BC data and ease of use, produces predictable positive BC artifacts when subject to rapid temperature drop, for example, when one walks outside on a cold day. These artifacts can be as high as 4 µg/m3 , enough to obscure the true signal. Conversely, when temperature rises suddenly (e.g., returning indoors), there is a negative artifact of similar extent. The artifact is most extreme when internal temperature of the unit is changing most rapidly and subsides as it stabilizes (over 2-4 hours). This effect can be seen with T changes of less than 5°C. The size of the artifact is proportional to the change in temperature. 40% of sampling sessions in our 2019 campaign included swings of >5° when bicyclists went in- and outdoors. Since much of cyclists’ exposure comes during their ride, and the time riding is insufficient for the unit’s internal temperature to stabilize, these artifacts must be addressed to accurately assess exposure. Here we present lab and field data characterizing the artifacts and methods to correct them. The underlying driver of the artifact is found in the sensor data (reference and sensing channels). Typically both sensors respond inversely to temperature in a fairly linear manner, the reference sensor more so. We have investigated a simple correction using best fits of sensor response to temperature, which removes most of the artifact; however, error at the beginning of the warming or cooling episode is generally large enough to make these unsatisfactory. We have also investigated empirical corrections based on the artifact characteristic of each unit, determined in lab tests with no BC. These can be adapted for the overall temperature change and speed of cooling/warming (how well-insulated the unit is) and appear more promising.

Cloud enabled on-road mobile observational platform using Google Street View cars in Salt Lake City, Utah: Instrumentation, methods and hyperlocal hot-spot identification

By: Ryan Bares, University of Utah Department of Atmospheric Sciences

Summary: Two Google Street View cars were instrumented with an expansive array of air quality and trace gas instrumentation and deployed for a full calendar year of on-road sampling in the greater Salt Lake City area. These measurements include: carbon monoxide, carbon dioxide, methane, nitrogen oxide and dioxide, ozone, particulate matter, ultra-fine particles, brown and black carbon, and meteorological observations.  A cloud enabled data collection system was developed to record and transmit data in near real-time with locally hosted displays in the vehicles and remote dashboards with instrument diagnostics and mapping features.  In this poster we outline the instrumentation, the cloud enabled data collection and transferring system, as well as overview methods and initial results of hyperlocal hot-spot detection. 

Low-Cost Air Sensor Performance Evaluations in Research Triangle Park, NC 

By: Samuel Frederick, Oak Ridge Associated Universities (ORAU)

Summary: In recent years, numerous low-cost air quality sensors have emerged which offer consumers increasing access to higher spatially and temporally resolved measurements of air pollutants while encouraging scientific engagement. These devices often measure multiple pollutants, adding both to their potential usefulness and complexity. With new devices constantly entering the market, the prospect of their wide adoption necessitates evaluation of sensor performance and accuracy. This work summarizes the results for evaluations of several low-cost air quality devices at the Ambient Air Innovation Research Site (AIRS) in Research Triangle Park, North Carolina. At least three of each sensor model were deployed simultaneously in order to evaluate precision between sensors of the same type. Sensor measurements at 1-hour averages were analyzed for accuracy against Federal Reference or Equivalent Method (FRMs/FEMs) instruments and pollutants evaluated include particulate matter (PM), Ozone (O3), Nitrogen Dioxide (NO2), and Carbon Monoxide (CO). The influence of environmental conditions, including temperature and relative humidity, have been explored. These results provide valuable information regarding the reliability of low-cost air sensor data when compared to regulatory methods.  
Although this abstract was reviewed by EPA and approved for publication, it may not necessarily reflect official Agency policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

The use of low-cost particulate matter sensors to predict photovoltaic soiling 

By: Sarah Toth, CU Boulder and NREL

Summary: Ambient particulate matter in urban environments is dynamic and heterogeneous, so understanding photovoltaic energy loss due to soiling is challenging. Silicon reference cells were deployed in an urban-industrial area in Colorado co-located with both a GRIMM and Dylos measuring ambient particulate matter concentrations. Regressing measured soiling ratios against cumulative sums of ambient coarse particulate matter (PM10-2.5) since the last precipitation event and ambient fine particulate matter (PM2.5) since the first day of deployment produces a root mean squared error of approximately 0.013 with GRIMM data and 0.014 with Dylos data. This model partially addresses the challenge of quantifying the relationships between ambient air quality and photovoltaic soiling, and proves that low-cost particulate matter sensors are a promising and feasible option to implement at photovoltaic system sites where soiling needs to be quantified. 
A another study of urban air quality utilizing historical data from PurpleAir sensors as well as BAM, TEOM, and other high-quality monitoring equipment spread throughout the Los Angeles, CA basin will illuminate how well these sensors can augment the existing high-quality measurements on a larger scale. The deployment of photovoltaic systems in the Los Angeles valley is widespread due to the high solar resource availability in that area as well as progressive state incentives toward renewable energy. However, this area is also subject to high levels of urban pollution as well as entrained desert dust due to the climatology and typical land-use change of the region. This makes the study in Los Angeles a very unique and important contribution to the study and modeling of photovoltaic soiling using low-cost particulate matter sensors, especially since photovoltaic power losses from soiling have been reported to be as high as 70%; a higher impact on annual photovoltaic performance than cell degradation.

Evaluating spatio-temporal accuracy of LUR models using low-cost sensor network data 

By: Vijay Kumar, Clarkson University

Summary: Land-use regression (LUR) models allow for prediction of air quality based on local terrain, air properties, and human activities and are popularly used for accurate exposure assessments.  These models are often built using data collected over short durations and from sites that have low spatio-temporal resolution.  The data collection constraints limit model accuracy and create uncertainty in model applicability for long-duration prediction and wide spatial application. Recent deployment of low-cost sensor networks has provided high spatio-temporal resolution data that allows for fundamental testing of LUR models.  Here we use long-term PM2.5 data generated from the Array-of-things (AoT) network in Chicago to test the temporal stability and durability of LUR models.  We also apply this model to low-cost sensor data in other locations to determine the spatial applicability of these models.  In this presentation, we will discuss our approach to handling the large data sets, the model development techniques, model validation methods, and the results of our study.

Innovative Technologies and Applications

How is air quality data validated and calibrated using machine learning before publishing on the AirVisual platform?

By: Chrystal Gaither, IQAir


The AirVisual platform aims to centralize and aggregate as much air quality information as is currently available in once place, in order to provide the most comprehensive overview of global air quality.

Data sources which are reported through the AirVisual platform include sensor data, from governmental monitoring stations (typically considered high-cost "reference monitors"), as well as low-cost sensors such as public AirVisual Pro stations and PurpleAir sensors.

All data which is published through the AirVisual platform is subject to data validation, and this process differs between these two sources of sensor data.

AirVisual's data validation system is cloud-based and driven by machine learning, and all measurements are passed through this system before publishing to our platform.

Governmental "reference" sensor data

Although high-cost governmental sensors are typically considered the most accurate and reliable source of measured air quality data, sometimes these sensors also report anomalies or inaccurate data. Reasons for this may include temporary periods of maintenance or defects, or even temporary hyperlocal emission sources nearby the sensor.

Low cost sensors

Measurements from low-cost sensors are also subjected to a data calibration and correction process, in addition to the validation process described above, which will identify and discount anomalous readings.

Application of Electrochemical Analysis Methods to Diagnose Atmospheric Impacts on Electrochemical Gas Sensor Response

By: Anna Farquhar, Aeroqual Ltd.

Summary: Amperometric electrochemical gas sensors are widely used for the detection of gases in air quality instruments. Such sensors generate output currents that are proportional to the target and cross-interference gas concentrations under diffusion controlled conditions. The sensor is also influenced by atmospheric conditions particularly temperature and relative humidity. Changes in temperature or humidity can cause both transient and slow drift in the output current. This presentation will discuss the use of electrochemical methods to characterize commercial electrochemical sensors before, during, and after a sudden change in the atmospheric condition. In situ electrochemistry allows the determination of how important sensor characteristics (active surface area, electrolyte resistance, electrode contamination etc.) change with atmospheric conditions in real-time. This work gives a better understanding of how electrochemical sensors react to their environment, and enables the development of diagnostic tools to evaluate sensor health and predict sensor response under changing climatic conditions.

Air pollution mapping using drone based sensor platform

By: Ardevan Bakhtari, Scentroid

Summary: With the progressing landscape of urbanization, increasing concerns on air pollution in large cities and near industrial sites are forwarding the development and practical applicability of unmanned aerial vehicles (UAVs). Modern UAVs feature a stable and high precision spatial-temporal platform allowing for the collection of air samples. Advantages of UAVs when compared to conventional technologies include exceptional maneuverability, collection of high-resolution spatial data from a set of on-board sensors, and the ability to collect air samples from sources up to 150 m in height – enabling organizations to avoid putting workers at risk while maintaining the integrity of capture data from pollutant-emitting stacks. Modern UAVs are equipped with multiple sensors and thermal imaging cameras, which collect and record data in real-time during flight. The data is used to create 3D pollutant maps (eg. heatmaps), track particulate matter dispersion, set alarms when exceeding threshold values, and allow for the comparison of results from multiple runs. Modern UAV platforms are also capable of flying using pre-configured flight routes in a grid-based, stop-and-go, and/or EN 16841-2 sampling approach. The technology has shown promising results from users that are interested in monitoring fugitive emissions, flare emissions, leak detection along pipelines, landfill and odour emissions, compost emissions, and identifying ambient concentrations at varying heights. Seemingly, the next progression step is the utilization of UAVs for monitoring pollutants in urban metropolises.  

Emerging Sensor Technologies and Source Identification with Low-Cost Sensor Networks

By: Austin Heitmann, Montrose Environmental

Summary: As the interest in low-cost gaseous sensors increase so has the interest in developing the sensing technology.  While some groups look to develop lab-on-a-chip type sensors that can monitor multiple compounds at once by shrinking existing laboratory equipment down or introducing completely new techniques that implement ground breaking biological based sensing.  Other technologies simply integrate current market available technologies with traditional EPA methodology sampling, while this does not provide real-time data it does provide precision and accuracy that no sensor can match.  Using these sensors allow for higher temporal and spatial resolution over the use of traditional reference methods.  While the data quality does not mirror the higher priced laboratory grade equipment, the qualitative trends and significant events can provide valuable data in identifying the source of the compound of interest.  The incorporation of an anemometer that will supply wind speed and direction data can facilitate correlating a pollutant concentration spike with real world events.  This presentation will discuss emerging sensor technology and dissect the advantages of sensor networks.  Case studies will be reviewed from field collected data.

Mobile Sensing: A Quick-Start Guide to Equipping Vehicles with Low-Cost Air Quality Sensors

By: Berj Der Boghossian, Air Quality Sensor Performance Evaluation Center, South Coast Air Quality Management District

Summary: Commercially-available low-cost air quality sensors (LCS) are becoming pivotal for developing new air quality networks while also expanding existing networks and advancing regional spatiotemporal mapping of air quality. These stationary LCS networks, in addition to fixed governmental air monitoring stations, allow for large areas of unmonitored air quality to exist. Mobile monitoring is proving to be increasingly vital to provide a deeper analysis of these unmonitored areas. Recently adopted California State legislation (i.e., Assembly Bill 617) now requires air quality monitoring in communities heavily impacted by nearby air pollution sources. The Air Quality Sensor Performance Evaluation Center (AQ-SPEC) at the South Coast Air Quality Management District has begun developing mobile platform vehicles equipped with regulatory-/research-grade analyzers and best available technology systems. The AQ-SPEC scope of the mobile platform development centralized on creating a Mobile Sensor Testing Protocol from which new and innovative methods of equipping vehicles with LCS emerged. A standardized “quick-start guide” to equipping a vehicle with LCS to accurately and properly sample and measure air pollutants while on a mobile platform was created to aid citizen scientists with studying their daily exposure to air pollution on their own.

Using modular, low-cost systems in multiple verticals to provide reliable, actionable air quality data

By: Cameron Schorg, Lunar Outpost


Lunar Outpost's flagship product, the Canary-S (Solar), is a solar powered air quality monitoring system designed to be class leading in size and cost, reliability, flexibility, and scalability. 

With cellular communication, these systems can be placed nearly anywhere to provide measurements on Particulate Matter and targeted gases. The system automatically chooses its network and whether to use 2G, 3G, or 4G based on detected signal strength, ensuring the best connection possible.

As the only air quality monitoring system in its class to allow data integration into the platform of your choice, the Canary- S allows fully customizable data solutions and puts data ownership and control in the customer’s hands.

The quality and quantity of actionable data to cost ratio makes the Canary-S a good addition to almost any air quality monitoring project.

For this presentation we will look at case studies from verticals like smart cities, oil & gas, and emergency response. 

Investigating Sensor Sensitivities and the Potential for Meaningful Low-Cost VOC Measurements

By: Amanda Gao, Massachusetts Institute of Technology

Summary: Though commercially-available low-cost sensors (LCS) that are sensitive to volatile organic compounds (VOCs) exist, their use in air quality and atmospheric chemistry research has been hampered by the inherent complexity of ambient VOC mixtures and by limitations in LCS technologies. Effectively, any single VOC sensor acts as a “broadband” sensor with poorly-characterized sensitivities to a poorly-characterized range of VOCs and generates a signal that is exceedingly difficult to interpret.  Improving our understanding of these sensors and their sensitivities, using both laboratory and theoretical methods, could inform novel applications of LCS in atmospheric chemistry and air quality contexts. In this work, we consider two commercially-available VOC LCS: Alphasense’s p-type metal oxide (MOx) sensor and ETO-B1 ethylene oxide electrochemical (EC) sensor.  We explore and characterize the responses of these sensors to several different VOCs, as a function of operating temperature (MOx) or bias voltage (EC), using both laboratory experiments and simple models. We then investigate the response of these sensors to various VOCs and mixtures, with a focus on atmospherically-relevant species and concentrations. Finally, we explore the potential for leveraging different sensor sensitivities to infer VOC composition in a way that one broadband sensor alone cannot provide.

Humidity Effects on Indoor and Outdoor PM Sensors - 8 Universities' Combined Study

By: David Pariseau, Particles Plus, Inc.

Summary: This presentation will be based on combined data collected by participating universities throughout South East Asia, comparing data from indoor and outdoor, collocated sensors to study the effects of humidity control on the sample data. 

Air Quality Monitoring and Comparison on Olympic Stadiums of five continents using low cost sensors

By: Edurne Ibarrola, Kunak Techonologies S.L

Summary: World Athletics started in 2018 to create a real-time air quality network with global coverage to help athletes choose the best times to train and compete and to help organizers to protect the health of athletes, understanding air quality impacts on people’s life quality. The objectives were to (1) monitor environmental conditions in Olympic stadiums and marathons, (2) study the air pollution impact in athlete’s health and performance, (3) raise awareness on air quality issues and make better decisions, (4) offer an added value to athletic competitions in terms of sustainability, increased protection of the health of participants and social awareness.
Kunak-Air 10 is provided with CO, NO2, NO, O3, NOx, PM10, PM2.5, PM1, and meteorological sensors. Five different K-Air 10 were set up in Olympic stadiums: Addis Ababa (Ethiopia), Mexico DF, Monaco, Sydney and Yokohama. In this context, the temporal patterns for gases and particles were studied during a period of one year (December 2018-November 2019). 
For gaseous pollutants, higher concentrations of O3 appear in Monaco and Yokohama, while the highest concentrations of NO2, NO and NOx appear in Addis Ababa and Mexico DF. Yokohama and Monaco stadiums are notable for the low pollutant concentrations reported (except for O3). The impact of traffic emissions is detectable in all the stadiums during the mornings, whereas distinct evening traffic peaks are observed in some of them.Tthe highest particle concentrations were recorded at the Addis Ababa stadium and Mexico DF.
The main conclusions are: all the stadiums appear to be impacted by vehicular traffic emissions; the highest air pollutant concentrations were recorded in Addis Ababa and in Mexico DF; the daily time patterns of air pollutants provided useful information to identify optimal periods for training or competition in each stadium; and periods and days of the week which were most advisable for outdoor sports activities in each of the stadiums were identified.

Monitoring of personal location and air pollution source activity to investigate linkages between behaviors and exposure outcomes on short timescales

By: Evan Coffey, University of Colorado Boulder


It’s widely understood that personal exposure to air pollutants often occurs heterogeneously through time and space. Therefore, strategies to alleviate overall exposure should take into account where and when potential reductions can be made. It follows from this that human behaviors (involving a location, time and activity), however challenging to change, play a crucial role in exposure outcomes. This work introduces and describes a low-cost activity monitoring system and data toolkit which integrates user Global Positioning System (GPS) information (~5meter resolution), user proximity to identified point sources (<1m), source activity (when sources could be emitting pollutants) and personal exposure measurements with the goal of exploring relationships between behaviors and exposure outcomes on short (<5min) timescales.

Results from the Prices, Peers and Perceptions (P3) study, a 2-year, 600-household improved cookstove intervention in northern Ghana, are presented to feature how the low-cost monitoring system can be utilized to measure and fuse together participant location, cookstove use and exposure to particulate matter (PM2.5) and carbon monoxide (CO) during 48hr in-home exposure assessments. This case study highlights the potential value of such a system to researchers designing interventions or establishing best practices aimed at more confidently attributing exposure outcomes to specific actions.

Harnessing the Potential of Low Cost Particle Sensors for Use in Hierarchical Air Quality Networks

By: Hamesh Patel, The University of Auckland

Summary: Poor air quality is responsible for an extra 8 million premature deaths worldwide, with 92% of the world’s population living in regions where air quality fails to meet recommended guidelines. Particulate matter is a toxic air pollutant that is detrimental to human health and has a degrading effect on the environment. Particulate matter comes in many different shapes, sizes and chemical compositions making it extremely complex to measure. Councils and regulators use specialized instrumentation to measure particulate matter which is very expensive to acquire, set-up and maintain. As a consequence regulatory networks have a limited temporal and spatial resolution which can impede an accurate assessment particulate matter on human health and the environment; it is already thought that current levels of particulate matter are significantly underestimated.
Improvements in technology have seen cheap particulate sensors entering the market. However, questions have been raised regarding their reliability, accuracy and long term performance. This work characterized low and medium-cost particulate instruments alongside regulatory instruments deployed under real-world environmental settings to gain a better understanding of instrument performance and the impact of changes in environmental variables in order to successfully correct data and reduce error. The results demonstrate a reduction in data quality as the notional quality of the instrument decreases. An addition of a mini-cyclone to a low-cost instrument improved data quality. Instruments are affected by environmental factors such as specific humidity and temperature with varying results across different instruments and particle size ranges. A range of empirical correction factors were calculated and utilized  to successfully correct data and reduce error for all bar one instrument.

A new optical aerosol sensor and its calibration with traceability to SI

By: Jaakko Yli-Ojanperä, Vaisala Oyj

Summary: Many of the publications and reports dealing with aerosol sensors concentrates on PM2.5 and PM10 measurement performance of the sensors in the field, which is the primary application of the sensors. A notably fewer amount of publications and work concentrate on the laboratory calibration of aerosol sensors. Most of them present normalized and/or non-traceable calibration results, which could be due to the calibration infrastructure lacking traceability to SI. In order to present true values of essential and inherent properties of an optical single particle counter type sensors or instrument, such as the detection efficiency as a function of particle size, number concentration response and size response, traceable calibration using monodisperse aerosol is required.

In this presentation we introduce a new optical aerosol sensor and calibration results of three such optical aerosol sensors resulting from a traceable calibration. The sensor utilizes single particle detection and analysis in order to determine the count size distribution of ambient aerosol. Values for PM2.5 and PM10 are calculated based on the distribution and reported to the end user. Results will include detection efficiency as a function of particle size, number concentration response and size response of the sensors.

Breathe London 3: Source characterization and emission ratios estimation from hyperlocal measurements of air quality using a low-cost sensor network

By: Olalekan Popoola, Department of Chemistry University of Cambridge UK

Summary: Exposure to air pollution is the leading environmental health risk factor globally, resulting in 7 million premature deaths annually. Protecting global populations from the detrimental effects of air pollution requires policies and regulations on both national and international levels. 
Air quality measurements are required in order to understand pollutant emission sources for the development of appropriate intervention policies, and are also needed to monitor their effectiveness once implemented.
The Breathe-London project ( combines state-of-the-art measurement technology with over 100 low cost sensor nodes (each measuring NO, NO2, O3, and CO2, PM2.5 and PM10) along with two Google Street View cars instrumented with reference grade air quality instruments. 
We will present results from the static network, showing how fast response measurements (1-minute data) can be used in separating local and large-scale emissions. By including CO2, a robust analysis technique will be presented that allow direct measurements of pollutant emission ratios. In this presentation these emission ratios will be compared to outputs from a state-of-the art air pollution model (ADMS-Urban).
The presentation will include results and early conclusions from the Breathe-London study.

Regulatory Comparison of Low Cost OPC Instrumentation for Ambient Particulate Mass Concentration Measurement

By: Jen Brown, Met One Instruments Incorporated

Summary: This experimental study compares regulatory PM1.0, PM2.5, and PM10 mass concentration measurements with results from real-time optical particle counter (OPC) mass measurement devices.  Total mass in the form of PM1.0, PM2.5, and PM10 was continuously measured during the event using US-EPA FEM designated beta attenuation mass monitors “BAM”.  Three “BAM” units were collocated with three OPC  mass measurement devices at each of three different locations:  Elizabeth NJ, Salt Lake City UT, and Riverside CA.  Data comparisons include hourly and 24-hour averages.  Analysis of data from May 2019-Jan 2020 reveal excellent PM10 linear correlation coefficients of 0.73, 0.90, and 0.89; PM2.5 linear correlation coefficients of 0.66, 0.78, and 0.69; and PM1.0 linear correlation coefficients of 0.50, 0.65, and 0.59 for each respective site (NJ, UT, CA).  OPC versus “BAM” slopes differ between mass size concentrations at each site, as well as when comparing the same mass size cut point between different sites.  This provides evidence of the importance of a k factor to accommodate for dissimilar ambient conditions at different sites.  Flow is critical to the accuracy of optical particle counters; flow stability is vital to minimize drift.  The OPC mass measurement units maintained excellent intra-model precision across the study, and this is attributed to its enhanced internal flow system.  Meteorological parameters have been investigated to explain occasional large shifts from FEM reported particulate concentration measurements.

Determining Spatial and Temporal Decorrelation Scales Using Stationary and Mobile Collocated Sensor Data

By: Nicole Goebel, Aclima, Inc.


Deployment and maintenance of a large-scale network of mobile sensing platforms has allowed us to measure a suite of pollutants at hyperlocal spatial scales throughout the Bay Area. These spatially resolved measurements have the potential to complement and bridge spatial measurement gaps between the network of temporally resolved reference stations, such as those maintained by air quality agencies throughout California and the United States. Assessment of the agreement between mobile and stationary data sources is essential in order to utilize and integrate data collected by these complementary platform types. 

Collocation is defined as the temporal and spatial proximity of sensors. In order to refine what is meant by “proximity,” we investigate agreement between measurements from a network of Bay Area Air Quality Management District reference stations and Aclima’s network of small-scale mobile sensors collected at a range of temporal and spatial differences across a suite of pollutants. 

In this session, we will compare measurements from these data collection platforms that are characterized by different temporal and spatial scales in order to identify the time and distance scales at which measurements decorrelate, while also taking into account measurement error from different measurement sources. Potential use cases for the outcomes of this analysis include in-the-field sensor monitoring and calibration, measurement validation and optimization, and sampling strategy (noting that Aclima’s technologies and data may be subject to a variety of proprietary and intellectual property notices).

Open Source Edge Computing Platform for Air Quality Applications

By: Sai Yamanoor, DesignAbly


Citizen Science or Community Driven efforts can provide data that can be used to predict or identify common and repeatable events. For e.g.: Images collected from community experiments have been used to identify certain insect species etc. Also, satellite images have been used to predict start of wildfires. Likewise, it is possible to use data community collected data to predict events in Air Quality Monitoring or detect anomalies. 

Powerful computing resources are required to build and train a model to predict events. With the recent advancements in the field of Machine Learning and Computing power, it is possible to run neural networks on hardware that costs less than $20. This enables conducting analysis on the collected data at the source instead of transferring it to a centralized “data lake.'' It also reduces costs as it helps avoid certain cloud computing and related infrastructure costs. 

This concept of conducting the data analysis at the source instead of the cloud is called Edge Computing. Edge computing can be a great resource in citizen science experiments because it helps avoid “reinventing the wheel”. In citizen science applications, the arrival of lost cost edge computing hardware enables running community driven experiments at a low cost. This also enables verify or re-emulating results observed in pre-existing datasets. 

In our presentation/poster, we are proposing an open source edge computing platform for citizen science experiments. The platform is centered around the Artemis module that costs US$8. The platform comes with interfaces to accommodate sensors with different outputs like UART, I2C etc. We would also like to demonstrate a use case for the platform using an existing dataset.  

Cloud Remote Senor Monitoring for Industrial Hygiene

By: Tim Quinn, SGS Galson Labs


Learn the basics on cloud based Industrial Hygiene sensor monitoring for the protection of workers, featuring electrochemical sensors; CO, CO2, H2S, SO2, O3, NO, NO2, VOC's and Particles 1, 2.5 and 10.  This new technology resolves the issue of "not being there at the right time" and works well for Indoor Air Quality assessments, nuisance odors complaints and many outdoor applications in construction, wild fires, hospitals and others.

This presentation will explain in detail "what is the cloud," which companies provide this service and how this make life better for companies who protect the air workers breath. Cloud monitoring allows you to get the job done remotely from wherever you are in the world.  

We will provide an in-depth discussion on how to harness the power of data analysis.  With continuous monitoring, workers can view sensor readings from the cloud once a minute and compare device data via a dedicated portal web dashboard. View device locations on a map and see the latest real-time data feeds and create personal e-mail and text message notifications to flag when particular devices exceed certain air quality parameter thresholds.

This talk is for the early technology adopter to the novice, learn how to work smarter with the cloud!

Air quality mapping with automotive air quality sensors

By: Herve Borrel, Airlib Inc.

Summary: Automotive air quality sensors have been used for more than 20 years to improve air quality inside cars. Each year about 8 million more are installed on new cars. It is probably the most massive pool of air quality sensors on earth, and they move all over city roadways, day and night. Airlib has developed a way to collect this data and treat it to extract information on VOC and NO2 levels. The method relies on a multitude of measurements from a multitude of sensors, rather than individual measurements. These sensors have a response time of the order of the second and are therefore well suited for the detection of pollution variations in traffic. The Airlib data engine creates and updates a measurement grid as new cities or new areas of a city are covered. The system is fully automated. Resulting AQI levels are calculated based on contributions from a multitude of sensors. The relative inaccuracy of individual sensors is compensated by the number of measurements. A key benefit of the method is that the mapping is quasi continuous, as opposed to a snapshot of the pollution at one specific time. The method therefore creates near real-time maps, as well as predictive maps for each time of day. Pilots have been running for more than a year in several cities. Some of the key results are presented here: VOC maps, NO2 maps, variations during the day, statistical distribution of pollution peaks, resulting average human exposure. The main strength of the method is its scalability at viable cost. Airlib could map the top 50 cities in the US within a few years.  

Developing Innovative Fire Weather Sensor with Air Quality Capabilties

By: Jacob Holle, Intellisense Systems Inc


Intellisense Systems Inc. (ISI) is advancing the development of a new Fire Weather Observing Sensor (FWOS) system to meet the need of highly portable, high performing weather system. The FWOS will be a field deployable fire weather observation sensor system capable of remote unattended operation. The devices will be placed throughout forest areas and areas prone to wildfire outbreak. The suite of sensors measures critical fire weather parameters to detect the outbreak of a wildfire and report back to decision makers, first responders, and researchers. During fire management by local authorities, the FWOS system will provide valuable real-time weather data for immediate action and future analysis. In partnership with the USDA Small Business Innovative Research (SBIR) team, ISI is enabling detection and reporting of critical fire weather parameters, visual imagery, and particulate matter in a small, easy-to-use unit capable of data storage and dissemination. By working alongside the USDA with an SBIR Phase I, ISI is excited to bring a critical capability to help mitigate the wildfire challenges faced today. 

The FWOS system leverages ISI’s existing low size, weight, and power technology used in the Micro Weather Sensor (MWS), which was deployed by CalFire, BLM, and other government agencies during recent year fires. After seeing the install challenges for the current large weather systems during the 2017 California fires, ISI’s team began developing user-driven fire enhancements to the MWS. The new parameters include: solar radiation, fuel moisture, and air quality (PM 2.5). All information is shared via two-way satellite that leverages a free data plan, eliminating a key budgetary barrier for users. 

The presentation will highlight ISI’s development path for FWOS, share details of participating in USDA’s SBIR program, and working through systematic barriers for introducing innovative technology solutions to standardized system expectations. 

Use of Networked Sensors

November 2018 California biomass burning as measured by A network of PurpleAir units

By: Karin Ardon-Dryer, Department of Geosciences, Atmospheric Science Group, Texas Tech University

Summary: Recent advancement in technology and a rise in public awareness have led to an increase in the popularity of low-cost sensors. These sensors can be easily deployed and used by community members, creating large networks of air quality sensors whose data is accessible to everyone. One of these low-cost sensors is the PurpleAir PA-II unit that measures Particulate Matter (PM). Previous studies have compared PurpleAir units to other reference monitoring systems, and though the results have shown good correlations between the data of the different systems further investigation is required to evaluate the response of PA-II units under extreme conditions. In this study, we evaluated the performance of the PA-II units deployed in California during the November 2018 biomass burning events. California has a large number of PA-II units (>1000); our study examined spatial and temporal changes in PM2.5 and PM10 (Particulate Matter with aerodynamic diameter <2.5 and <10μm, respectively) as measured by several hundreds of PA-II during the biomass burning events.

A synthesis of CO2 emissions at sub-urban scales using inverse models, sensors, and satellites

By: Alexander Turner, University of California, Berkeley

Summary: Carbon dioxide (CO2) emissions in urban areas comprise a large fraction of the anthropogenic source to the atmosphere, even at the global scale.  As such, there is much interest in being able to both accurately predict their emissions with inventories and evaluate those predictions with atmospheric measurements.  Here we present results from an inverse modelling study estimating CO2 emissions at 1-km spatial resolution in the SF Bay Area.  The work combines measurements from a dense network of sensors and high resolution emission inventories.  The sensors part of the Berkeley Environmental Air-quality and CO2 Observation Network (BEACO2N) that includes ~70 sensors spaced roughly 2-km apart.  The bottom-up inventories include a realistic representation of traffic and a biosphere inferred from new satellite measurements.  We use a Lagrangian particle dispersion model to link the emission inventories and the measurements, allowing us to evaluate the emission inventories with measurements.  This work will present a synthesis of our modelling and measurement work in the SF Bay Area.

Exploring efficient calibration and collocation methods for low-cost sensor networks

By: David Ridley, California Air Resources Board


In 2018 the California Air Resources Board (CARB) awarded grantees across the state with $10 million to build capacity for air quality monitoring and education. Many of these communities have begun monitoring using low-cost sensor technologies. CARB has collocated over 100 sensors measuring particulate matter (PM2.5), ozone, and nitrogen dioxide to better understand the uncertainty in the sensor measurements to assist the grantees. Evaluation has taken place primarily at the downtown Sacramento regulatory site, but also under different meteorological conditions across the state.

We explore different sensor adjustment processes, dependent on the level of pre-deployment calibration, to maximize the accuracy and utility of the sensors. The uncertainty in the resulting measurements is quantified as a function of both the agreement with the regulatory instrumentation and the conditions during collocation. For the optical particle counters, we test different strategies for calculating the PM2.5 and find significant improvement relative to the out-of-the-box performance. Overall, we find that the sensor performance can be close to instrumentation with Federal Equivalent Method designation for ozone and PM2.5 with sufficient calibration and automated error detection.

Finally, through partnership with Aclima, CARB is exploring low-cost sensor network adjustments using mobile monitoring vehicles as a calibration transfer from instruments at a regulatory site. This process reduces the burden of periodically relocating sensors to well-maintained regulatory-grade equipment sites.

Bleed Orange, Measure Purple: A Low-Cost Particulate Matter Sensor Network on the University of Texas at Austin’s Campus

By: Hagen Fritz, University of Texas at Austin

Summary: Low-cost sensor technologies for ambient air quality characterization are being developed at and incorporated into research at an increasingly faster pace. While not able to detect minor fluctuations in pollutant concentrations or provide ground-truth estimates of pollutant concentrations, low-cost sensors are useful at detecting general air quality trends. In addition, the cheaper price for these instruments allows researchers to easily scale their efforts up in order to create vast networks of sensors. Sensor networks can give insight into how pollutant profiles evolve over different timeframes and identify where potential pollutant sources. 
In this study, we monitor the particulate matter (PM) concentration across the University of Texas at Austin’s (UT) main campus by setting up a network of 16 Purple Air PA-II outdoor PM sensors. The UT campus is unique in that it is surrounded by two highly trafficked roads to the north and south and a major interstate highway to the east. The network was set up so as to include a large portion of sensors in the interior of the main campus, sensors on the perimeter, and two sensors upwind and downwind of the major interstate. Purple Air sensors were initially collocated at a site free from major PM sources with a MetOne BAM-1022 Beta Attenuation Mass Monitor for an initial comparison and compared again against an Aeroqual S500 portable monitor once placed on campus. These comparisons are used to help derive correction factors that can be applied to the raw data from the purple air sensors to further improve the instrument’s accuracy. Data are then used to identify PM sources on campus, identify areas on campus that might be more polluted than others, and to determine to what extent the major roadways around campus influence the interior PM concentration. 

Practical network correction of low-cost PurpleAir PM2.5 sensors in Phoenix, Arizona

By: Ian VonWald, ORISE participant hosted by US EPA

Summary: Regulatory air quality monitoring networks are often too sparse to capture fine scale variability in fine particulate matter (PM2.5) produced by localized sources. Networks using low-cost air sensors have the potential to identify pollution hotspots. However, these sensors may be inaccurate or imprecise out-of-the-box, and performance may have seasonal or age-related drift. Unfortunately, corrections to low-cost sensors based on comparisons to collocated regulatory monitors are often labor- and time-intensive.
Here, we explore practical strategies for correcting a network of 20+ low-cost PurpleAir PM2.5 sensors deployed over an area of ~1000 sq km in Phoenix, Arizona starting in June 2019. The deployment includes 5 sites where PurpleAir sensors are collocated with regulatory monitors and a mobile reference monitor. Before network deployment, all sensors were collocated with regulatory monitors. The goal of this work is to determine a correction equation to improve sensor performance that can be developed easily. We investigate the differences between corrections developed during collocation periods before and later in the deployment, the necessary collocation duration for the PurpleAir sensors, the variation in performance as a function of sensor age, and whether a separate correction equation is needed to adjust sensor data during times of high PM2.5 concentrations (e.g., recreation wood burning or monsoon storms). These results will provide practical guidelines for correcting PurpleAir sensor networks and will support future research questions regarding the transport of woodsmoke and PM2.5 in Phoenix.
Disclaimer: although this abstract was reviewed by EPA and approved for publication, it may not necessarily reflect official Agency policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

Characterization of electrochemical gas sensors for ambient air quality measurements

By: Jari Suikkola, Vaisala

Summary: During the recent years, there has been a growing interest in establishing networks of supplementary air quality sensors to increase the spatial resolution in air quality measurements. In supplementary air quality sensors such as Vaisala AQT400 series instruments, the gas measurements are usually done by using electrochemical gas sensors as these are a cost-efficient method of measuring ppb-level concentrations of pollutant gases. However, these sensors are highly affected by temperature and humidity, they present cross-sensitivities to other gases, and their performance changes over time. In order to get reliable readings out of an electrochemical gas sensor, the instrument manufacturer needs to understand these phenomena to compensate for their effects in the instrument algorithms.

In this presentation, we show how these electrochemical gas sensors have been tested and characterized in Vaisala laboratories, and what the results of these tests are. We have tested the temperature and humidity dependency of the baseline and sensitivity of the sensors, and the short-term drift of the sensor readings. In addition, we have studied the effects of long periods of high humidity on the sensors, as this is linked to the sensors breaking down in certain areas such as South-East Asia, and this is something we are aiming to prevent. In addition to the laboratory test results, we show what level of performance is achieved in outdoor testing when these laboratory test results have been applied in Vaisala AQT420 measurement algorithms.

Supporting Emissions Reduction Planning in Richmond, CA using a Novel Sensor Data-Driven Modeling Approach

By: Julia Luongo, Ramboll Shair

Summary: California’s regional air quality has drastically improved since the 1970's, but environmental justice communities are still disproportionately impacted by local mobile and industrial emission sources. Assembly Bill (AB) 617 was signed into law to address local air pollution in disadvantaged communities - shifting the focus from regional to community-scale. Under AB617, communities who are disproportionately affected by air pollution have the opportunity to form Community Emissions Reduction Plans (CERP) through the Steering Committee process. During development and prior to adoption of the CERP, the plan must be approved first by the local air districts and then by the California Air Resources Board (CARB) and the CARB Governing Board. Therefore, CERPs must include detailed research into existing air quality conditions and emission sources. In Richmond, CA, a real-time data-driven modelling tool was implemented to better understand localized air quality impacts and likely sources across the city. Here we present how Groundwork Richmond and the Richmond Steering Committee use a 50-node sensor network, coupled with an innovative modelling approach, to drive evidence-based emissions reduction strategies and decisions towards improving air quality in Richmond.

Time evolution and timescale dependence of PurpleAir correction methods from multi-season collocation studies

By: Karoline Johnson Barkjohn, US EPA

Summary: We evaluate the field performance of the low-cost PurpleAir PA-II air sensor using long-term collocation studies in multiple locations. The PurpleAir sensors are part of a large and growing national and international network of air pollution monitors. The PA-II houses a pair of Plantower microelectronic optical particle counting sensors and the data reported includes size-segregated particle number concentrations and a derived mass concentration of particulate matter with diameters less than 2.5 microns (PM2.5). We discuss analysis of PA-II field data collocated with Federal Equivalent Method (FEM) PM2.5 monitors for durations of months to years. The collocation data serves as the basis for an assessment of the accuracy, precision, and stability of the PA-II data, and, together with the FEM reference data, we establish an empirical correction algorithm that we apply to the raw PA-II PM2.5 data to bring it into better agreement with the FEM PM2.5 measurements. Preliminary analysis of hourly data from multiple collocation sites, with FEM PM2.5 ranging from 0-40 µg m-3, shows there is a root mean squared difference (RMS) of 9.8 µg m-3 between PA-II and FEM PM2.5 values, where PA-II is biased high by 7.5 µg m-3. Correcting for the high bias and ambient relative humidity results in a 64% improvement in RMS. We will discuss these results and will incorporate new collocation data to evaluate the PM2.5 comparisons at both the hourly and daily timescales, as well as evaluating consistency in the PM2.5 and number concentration data for collocated PA-IIs wherever possible. We will discuss the stability of the correction algorithm over time, and among the collocation sites, to assess the sensitivity to varying chemical composition. 

Although this abstract was reviewed by EPA and approved for publication, it may not reflect official Agency policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

Application of a Single Correction Equation Reduces Error and Regional Bias of PurpleAir PM2.5 Measurements Across the U.S.

By: Karoline Johnson Barkjohn, ORISE fellow hosted by US EPA Office of Research and Development

Summary: PurpleAir sensors are widely used by individuals, community groups, and other organizations including air monitoring agencies. Previous performance evaluations have studied a limited number of PurpleAir sensors in select geographic areas or laboratory environments; these results may not be translatable to areas with different environmental conditions, varying aerosol compositions, or a wider range of concentrations. Here, we evaluate the performance of PurpleAir sensors operated by air monitoring agencies and collocated with regulatory-grade monitoring instruments, at monitoring sites across the United States. In total, >12,000 24-hour averaged collocated PurpleAir and Federal Reference Method (FRM) or Federal Equivalent Method (FEM) measurements were identified across diverse regions of the U.S., including 14 states (AK, AZ, CA, DE, CO, FL, GA, IA, KS, NC, OK, VT, WA, and WI). Performance based on 1) the raw data, 2) a simple U.S. linear correction, and 3) a multilinear regression including temperature and relative humidity (RH) has been considered for 24-hour averaged data. Preliminary results suggest that the PurpleAir reported PM2.5 data output overestimates PM2.5 by ~60% in most states. For some states, where bias remains after applying a linear correction, a correction including temperature and RH reduces the remaining bias. An evaluation of the daily Air Quality Index (AQI) category derived using the raw and corrected PurpleAir PM2.5 compared to the AQI from regulatory data is also presented. The results are synthesized to determine the feasibility of a broadly-applicable correction equation that could be applied to PurpleAir reported values.

Although this abstract was reviewed by EPA and approved for publication, it may not reflect official Agency policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. 


Calibration of PM2.5 concentrations of low-cost PM sensors by using collocated BAM-1020 monitors

By: Kuang-Hua Hung, National Chiao Tung University,Institute of Environmental Engineering

Summary: Low-cost PM2.5 sensors can be used to detect PM2.5 concentrations with high spatial and temporal resolutions as compared to the air quality monitoring stations. However, the data reliability is doubtful due to the effects of PM2.5 concentrations and ambient conditions on the accuracy of PM2.5 sensors. This study aims to calibrate the PM2.5 sensor data using the hourly data of BAM-1020 monitor as the reference. The multi-variable regression method was applied for PM2.5 sensor calibration based on PM2.5 concentrations and ambient conditions. In this study, three to five PM2.5 sensors were collocated with BAM-1020 at each of Taoyuan, Taichung (Zhongming), and Tainan monitor stations for more than one years. The results show that the intra-model variability (IMV) values of the PM2.5 sensors at Taoyuan, Taichung (Zhongming), and Tainan monitor stations are 15.17, 16.83 and 11.21% respectively, which meet the Taiwan EPA criteria of 20%. That is, the precision of the PM2.5 sensors is good. The average limit of detection (LOD) values of the sensors were calculated to be 10.3±1.6 μg/m3 at these stations. It was found that the PM2.5 sensors over-measure PM2.5 concentrations due to significant influence of the relative humidity (RH) on sensor readings. As RH increases, the PM2.5 sensor readings also increase. Therefore, the PM2.5 sensor data do not correlate well with the BAM-1020 data with the R-squared (R2) of only 0.64 (Taoyuan), 0.74 (Zhongming) and 0.75 (Tainan), respectively. After correction using the multi-variable regression equation, the correlation of the PM2.5 sensors and BAM-1020 data is improved with the R2 of 0.76 (Taoyuan), 0.83 (Zhongming) and 0.82 (Tainan), respectively. The mean normalized bias (MNB) and Mean normalized error (MNE) are also decreased substantially from +43.1~+54.6% to ±10% and from 57.2~69.8% to less than 30%, respectively.

New, Reliable Low-Cost Pollution Sensor Nodes Drive Applications in Regional Networks, Personal Exposure, Vehicle and Indoor Air Monitoring

By: Matthew Johnson, Airlabs

Summary: Recent developments in low cost sensing technology enable new applications, giving a much more detailed view of air pollution exposure. Airlabs has developed the AirNode, a robust, low-cost sensor node for NO2 and O3 based on proprietary metal oxide sensors with onboard offset calibration, and also including low cost sensors for PM, CO2, T and RH. These nodes have been deployed in fixed and mobile settings. The performance of the nodes has been characterized by colocating them at established monitoring stations in Copenhagen, Rotterdam, Roskilde and London. In addition to monitoring outdoor air quality, we have deployed the AirNode in a study of professional drivers in London and a London office building. Over a 2 month period in 2019, AirLabs engaged in a study to measure the impact of in-vehicle air pollution in a range of use case scenarios within London. The study measured the level of exposure that certain vehicle drivers and passengers could experience while going about their working day and evaluated the impact of mitigating solutions. In addition, the study shows the analytical benefits that can be achieved through the deployment of large numbers of compact, low cost AirNode sensors within a given environment. These results have not been presented before; here we will present the first results of a series of air quality studies that take advantage of the AirNode’s small size, durability, low cost and data quality. 

Assessing Urban air quality project

By: Monika Vadali, Minnesota Pollution Control Agency

Summary: Minnesota Pollution Control Agency has deployed a network of 45 multi pollutant air quality monitoring sensor pods across every zip code in the cities of Minneapolis and St Paul. Each pod measures CO, SO2, NO, NO2, O3, PM2.5, PM10, in addition to RH and temperature, for a total of 225 gas sensors and 45 PM monitors. There is one federal monitor within this study area that measures all these pollutants. In the 12 months of monitoring, the sensor data clearly shows that there are differences in air pollution from one zip code to the next. Even within a zip code with multiple sensors, we have observed that there are differences. This is a classic example of how a denser distributed network can help communities identify and address local issues of pollution and have more data to take action if required. We know that even small differences in pollutant levels can cause serious health effects for vulnerable and sensitive populations, especially small children and the elderly. Some areas within the Minneapolis – St Paul study area also have environmental justice areas with populations of color and people of lower income groups.   
The deployment of the sensors, especially selection of locations, has been entirely driven by community input. In Minneapolis, all the sensors are on wooded street light poles in residential neighborhoods and in St Paul, the poles are located on light poles in public school parking lots. All of the data is also being displayed on a public web page, updated on a weekly basis. Communities are looking at this data and we are strive to display quality assured data but it has been a challenge. 
This presentation will discuss the sensor and federal monitor comparisons, sensor to sensor performance comparisons, data quality issues with the sensor measurements and challenges of presenting the 1 minute data to communities and discussing health impacts. Sensor data from schools has been used to analyze pollution patterns around drop off and pick up times.

Youth Education & Development

BackpAQ: Enabling students to build and deploy their own portable air quality monitors.

By: Andrew Clark, Sustainable Silicon Valley

Summary: This session will describe BackpAQ, an innovative new air quality monitoring project from Sustainable Silicon Valley. The program objective is to enable students to build and deploy community-based mobile air quality (AQ) monitors that leverage new low-cost sensors. These handheld units can be readily assembled by advanced middle-school and high school students and other STEM-oriented youth who are motivated by interest in obtaining, understanding and sharing hyper-local air quality data. The battery-powered devices display realtime PM and AQI data while also sending data to the cloud where more detailed analysis and sharing can take place

There remains a critical need for more ground-truth generated hyper-local AQ data to support investigation of suspected criteria pollutants which can be contributors to asthma and other respiratory and lung diseases in the community, especially among young people. 

Students building and carrying their devices through their daily lives will deepen their engagement with and understanding of air pollution and the importance of hyper-local monitoring. Students will also receive training in and hands-on experience with advanced data science tools. It is hoped that this initial cohort will become ambassadors to their peers across the region, attracting more student, school and city participation.

Enhancing Community Science through the Development of Resources and Tools for Youth Engagement 

By: Ashley Collier-Oxandale, South Coast Air Quality Management District

Summary: Established in 2014, the Air Quality Sensor Performance Evaluation Center (AQ-SPEC) at the South Coast Air Quality Management District evaluates commercially available low-cost air quality sensors and shares the results in publicly available reports. Complementing these ongoing evaluations, AQ-SPEC engages in projects involving the development of support and resources to aid communities and members of the public interested in using low-cost air quality sensors. Throughout this work our team has had the opportunity to engage with youth, which has provided unique insights. In the US EPA funded STAR Grant project:  “Engage, Educate and Empower California Communities on the Use and Applications of Low-Cost Air Monitoring Sensors”, AQ-SPEC has deployed approximately 400 low-cost sensor devices across 14 California communities. One of these community partners established a collaboration with a team of students from a local college. AQ-SPEC provided support and resources to this team resulting in useful data analysis and capacity building for the community partner. This collaboration also helped to inform the development of tools and resources at AQ-SPEC and provided a model of support for community-based science. For another project, AQ-SPEC is developing a Library Program that will maintain and lend out low-cost air quality sensors. Initially, this program will only be open to educational audiences, such as K-12 teachers and students. Beginning with this focus will help to ensure that the necessary supporting materials are available to encourage the effective use of sensors. Currently we are working with partners to develop curriculum and activities that will be tested with students. We will present these examples as well as a several others that highlight the value of youth engagement and illustrate how this work benefits the goals of the AQ-SPEC Program. 

Engaging School Communities in Air Sensors

By: Aubrey Burgess, Denver Department of Public Health and Environment

Summary: The City and County of Denver’s Love My Air program is creating a citywide air quality (AQ) monitoring network to provide real-time AQ data—utilizing low-cost cutting-edge air pollution sensor technology to make it useful for widescale deployment and replicability. Love My Air aims to empower school communities to reduce air pollution and limit exposure through behavior change, advocacy, and community engagement. By 2021, 40 Denver Public Schools (DPS) will have sensors and dashboards that relay hyper-local, real-time data. The data dashboards are visible via a large TV display inside each school and at School communities (10 currently) use this data to learn AQ basics, how their behaviors impact local AQ, and how it impacts their health. 
Love My Air Denver aims to explore innovative opportunities to engage school communities. For example, AQ curricula are paired with hand-held sensors allowing students to conduct their own citizen science projects. Students also have the opportunity to engage in several art projects, including working with a local artist to paint their school’s sensor and designing customized anti-idling campaign signage. Anti-idling signs designed by students are printed professionally at the City’s Public Works Sign Shop. 
Another core value of Love My Air Denver is to utilize community members at experts. Parents, teachers, nurses, and principals have been involved in the development process from designing the dashboards to developing curricula. The vision is to share best practices and lessons learned throughout this process to make this program replicable by other municipalities. 

Beyond Books: Lessons Learned from a Regional Air Sensor Loan Program 

By: Benjamin Weiss, US Environmental Protection Agency Region 5

Summary: The explosive growth in the public’s interest of sensor technology and acutely localized air quality has uncovered several barriers and limitations to using sensors and achieving useful outcomes. The unfamiliarity with the wide variety of lower cost air sensors, sensor performance, data analysis and interpretation, and data communication can be obstacles to educators and environmentally-focused community and public health representatives.  Often these interested parties have contacted U.S. EPA’s regional offices for clarification and direction.  In the application of these sensors for various educational and awareness purposes, partners have acknowledged various reasons for not using these tools, including uncertainty of which sensors to purchase in a market saturated with sensors, only needing sensors for a one-time situation, a hesitancy to make an investment in these devices until they can test them out, and lacking the financial resources to purchase the equipment.  EPA Regions 5 has been educating interested partners in the aspects of a successful sensor project to temper expectations, and at the same time developing an air sensor loan program to remove some of these barriers to realize their goal.  The air sensor loan program has provided educators with a new resource for teaching about air quality and has supported projects throughout the region including student citizen science efforts, science fair projects, summer science programs, and research.  This session/poster will cover the interaction with partners, development and implementation of the loan program, lessons learned, and potential future uses for the program.

Kids Making Sense®: Educating Youth about Air Quality and How to Effect Change using Small Sensor Technology

By: Hilary Hafner, Sonoma Technology, Inc.

Summary: Kids Making Sense ® (KMS) is an air quality education program that utilizes sensor technology and science, technology, engineering, and math (STEM)-aligned curriculum. Sonoma Technology, Inc. (STI) has worked in partnership with several air quality agencies and community-based organizations to successfully deploy the KMS program in more than 100 schools over the past several years. This presentation will provide an overview of the KMS program and discuss the successes, challenges, and lessons learned from the many KMS deployments to date. We will highlight the experiences of teachers and government representatives working with students using sensor technology to empower youth to engage with their communities on air quality issues. Advances in air quality sensor technology are creating new educational opportunities. 
KMS engages students with real-world, hands-on science. The curriculum begins with in-classroom experiments that teach students about the properties of air, sources and characteristics of particle pollution, and health effects of air pollution. Working in teams, students use the scientific method to develop a testable research question (hypothesis) and design a study to monitor air pollution around their school and communities. Using low-cost sensors, such as AirBeams, students collect credible air quality data. Back in the classroom they analyze the collected data and discuss their findings with their classmates and professional scientists. The KMS website,, allows students to view their routes on a sharable map, identify key areas of pollution, and share their data with the global air quality community. At the conclusion of the program, students develop an air quality awareness campaign or an emissions reduction action plan to share what they’ve learned with members of their community and drive positive change for clean air.

Opening up the “Black Box”: Engaging Youth with Low-Cost Air Quality Sensors

By: Kerry Kelly, University of Utah

Summary: Poor air quality ranks as the number one detractor to quality of life in Northern Utah, a region that periodically experiences the highest fine particulate matter levels in the United States. Consequently, local community members are highly interested and engaged in improving air quality. This region is also home to one of the densest particulate matter (PM) sensing networks in the United States, with more than 500 low-cost air quality sensors (including PurpleAir, state, and AQ&U, the University of Utah’s sensor network and infrastructure). These sensor networks have been operating since 2017 and offer a rich framework for engaging youth in understanding the region’s air-quality challenges and potential solutions. Through outreach efforts to local schools and the community, we have developed hands-on, engaging activities that help citizens: (1) understand how meteorology and topography lead to pollution episodes, (2) build their own light-scattering-based PM sensor from LegosÔ and simple electronics, thereby opening up the black box of how a PM sensor functions; (3) learn how to be good sensor hosts; (4) make sense of real-world data; and (5) understand how dots on a map translate into visualizations, thereby helping them to understand how location can affect air quality during different pollution episodes and how the geospatial density of measurements affects pollution estimates. These activities have engaged students and community members in hundreds of classrooms and community events. Our surveys suggest that these activities are highly engaging and effective for improving awareness of air quality, the principles behind light-scattering based sensors, and the geospatial and temporal variations in PM2.5 levels during a variety of pollution episodes. Working with youth and citizens in general also pose a number of challenges, and we describe some of these and strategies to address them.

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

By: Natalie Sampson, University of Michigan-Dearborn

Summary: 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.  

Heater and Substrate Profile Optimization for Low Power Portable Breathalyzer to Diagnose Diabetes Mellitus

By: Ramji Kalidoss, Chennai Institute of Technology

Summary: Chemi-resistive sensors used in breathalyzers have become a hotspot between the international breath research communities. These sensors exhibit a significant change in its resistance depending on the temperature it gets heated thus demanding high power leading to non-portable instrumentation. In this work, numerical simulation to identify the suitable combination of substrate and heater profile using COMSOL multiphysics was studied. Ni-Cr and Pt-100 joule resistive heater with various profiles were studied beneath the square and circular alumina substrates.  The temperature distribution was uniform throughout the square substrate with the meander shaped pt100 heater with 48 mW power consumption for 200 oC. Moreover, this heater profile induced minimal stress on the substrate with 0.5 mm thick. A novel Graphene based ternary metal oxide nanocomposite (GO/SnO2/TiO2) was coated on the optimized substrate and heater to elucidate the response of diabetes biomarker (acetone). The sensor exhibited superior gas sensing performance towards acetone in the exhaled breath concentration range for diabetes (0.25 – 3 ppm). These results indicated the importance of substrate and heater properties along with sensing material for low power portable breathalyzers. 

Indoor Air Quality

Use of low-cost air quality sensors for quantifying indoor pollution exposures

By: Jiayu Li, Carnegie Melon University


According to the National Human Activity Pattern Survey, people spend 87% of their time in buildings. However, epidemiological studies usually interpret the harmful effects of air pollutants based on outdoor concentrations measured by EPA regulatory sites. This compromise is mainly due to the lack of affordable and accurate measurement techniques to sample air pollutants with a high spatiotemporal resolution and in indoor environments. Current advances in low-cost air quality sensors may address this problem. This study focuses on characterizing indoor and outdoor air pollution profiles utilizing multiple real-time multi-pollutant sensors (RAMPs). 

In this study, we sampled over 40 indoor environments with RAMPs. Each location was sampled for 14 consecutive days. The gaseous and particulate pollutants we measured include CO, CO2, SO2, NO, NO2, VOC, and PM2.5. The 40 indoor samples can be characterized as six major categories (residential housing, school and public building, office and factory, bar and restaurant, mall and store, and public transportation space). The diurnal patterns of various pollutants from each category are significantly different. In residential buildings, PM2.5 and CO emissions related to cooking dominate during the mealtimes, especially for families with gas stoves. These indoor emissions may increase personal exposure considerably compared to that estimated based on ambient measurements. Therefore, we coupled the human activity pattern (HAP) with the diurnal profile measured for each building environment to estimate the HAP-based personal exposure. In general, the activity pattern-based CO, NO2, and PM2.5 exposures were higher than those based on outdoor measurements alone. The affordability and accuracy of RAMPS offer an opportunity to reveal the diversity of various indoor environments, which is usually prohibited due to the high cost and maintenance requirement of conventional sampling techniques. 

IoT Sensor Package Monitoring Indoor Environment for Optimal Cognitive Function & Employee Productivity  

By: Ajith Kaduwela, Air Quality Research Center, University of California, Davis


People spend a large fraction of their time (up to 90%) indoors and the indoor environmental conditions need to be maximized for our comfort, cognitive ability, and productivity. Some of the critical indoor environmental factors worth monitoring include Air Pollution, Temperature, Humidity, Sound, Vibrations, Lighting, colors and shapes appealing to humans. 

Air Pollution is the leading cause of death, and one of the key factors when evaluating indoor environmental factors. Indoor Air Pollution is mostly caused by elevated concentrations of Carbon Dioxide (CO2), particulate matter (PM that also include disease causing bacteria and viruses), volatile organic compounds (VOC), carbon monoxide (CO). It is well established that high CO2 concentrations lead to reductions in cognitive functions.  VOCs and CO2 are independently associated with cognitive scores.  There are several effective steps that can be taken to reduce the indoor CO2 concentrations, if made aware. 

A low cost IoT sensor package has been created to measure CO2, particulate matter (PM) number and mass, Temperature, and Relative Humidity. Additional sensors (e.g., VOC, CO, light, sound, radiation) can be easily added. Alerts can be automated when factors exceed optimal thresholds, so immediate preventative action can be taken (e.g. via text messaging). Results can also be monitored realtime via a mobile app. It is recommended that all offices, schools, and buildings that have high occupancy install similar IoT sensors to monitor various environmental factors to keep conditions optimal for improved cognitive function and productivity.

Development of a Time-Dependent Mass-Balance Equation for Carbon Dioxide in a Classroom: Determination of Outdoor Air Mixing Rates using an Optimization Method

By: Ajith Kaduwela, Air Quality Research Center, University of California, Davis

Summary: We have developed a time-dependent mass-balance equation for Carbon Dioxide (CO2) concentrations in a classroom. The time dependence is important as the occupancy of the room varies significantly depending on the time of the day. CO2 concentration in the classroom is measured every 30 seconds using a low-cost sensor (MH Z-19) interfaced with a Raspberry Pi computer. The occupancy of the classroom is known every class period. We have also developed low-cost ultrasonic instruments to measure the on/off timing of the inlet and outlet HVAC registers. These instruments also measures the inlet CO2 concentrations using the same type of a low-cost sensor. The only two unknowns are the inlet/outlet air flow rates. These rates are generally difficult to measure without specialized and expensive equipment. Using an optimization technique, together with our measurements, we obtain these rates and compare them with the recommended rates for a classroom environment. This method allows us to infer the efficiency of the HVAC system on a daily time scale. This method is also very transferable and can be used in any enclosed indoor area with some form of air circulation.

Evaluation of a low cost particle sensor for applicability to indoor air quality measurements

By: Aloka Khanna, 3M Company

Summary: A low-cost, laser-based particle sensor has been evaluated through multiple tests with lab reference (monodisperse Polystyrene Latex Beads) aerosols, in the size range 0.2 to 3 microns, and real-world challenge aerosols typically encountered in indoor environments.  Measurements were compared against lab-grade reference instruments, i.e. a TSI Scanning Mobility Particle Sizer (SMPS) and TSI Aerodynamic Particle Sizer (APS). Based upon the afore-mentioned tests, a detailed investigation has been carried out on the performance of the particle sensor as it relates to response time, zero error/bias, correlation/linearity, minimum and maximum detectable particle size, and mass concentration accuracy for myriad particle size distributions, different challenge aerosols, and minimum and maximum detectable mass concentrations in various particle size bins. The potential reasons for the particle sensor performance are examined and, through this analysis, it has been determined that the response time was majorly influenced by the regulation of airflow through the sensor, using either a fan or a pump, while the optical components were the deciding factor for minimum and maximum detectable particle size. The zero error/bias and correlation/linearity as well as the minimum and maximum detectable mass concentrations in various particle size bins were limited by optical components as well the optical design of the sensor. The mass concentration accuracy depended heavily upon the optical components and optical design along with the calibration conditions. Finally, recommendations on methods to utilize the particle sensor data for effectively monitoring the dynamic indoor environment will be presented.

Clean Air Tourism: Assessment of Individual Exposure to Indoor and Outdoor Air Pollution on the Las Vegas Strip Using Portable Sensors, Las Vegas, Nevada, USA

By: Carlos Arambula-Quintero, Nevada State College

Summary: Poor air quality has been shown to increase the rates of mortality and morbidity in urban environments, where most of the world’s population lives. Urban dwellers increasingly tend to spend more time indoors. Indoor air quality is assumed to be better than outdoors’ due to extensive sources of contaminants in the later. Exposure to outdoor pollutants varies due to local meteorological conditions. In contrast, indoor air quality is determined by indoor pollution sources and the ventilation systems. Non-smoking indoor areas should have better air quality than smoking areas if ventilation systems are able to sufficiently contain drifting tobacco smoke to avoid the risk to secondhand smoke exposure for non-smokers. 
This study presents the measurement of outdoor air quality using portable sensors along the Las Vegas Strip, a year-long touristic destination in Nevada, USA with ~70,000 daily traffic counts at very low speed and more than 55,000 pedestrians. Additionally, risk to secondhand smoke at smoking- and non-smoking areas inside six casinos was assessed. Monitoring air quality will help to assess the health risks at these environments. Measurements were taken twice each season for one year to evaluate air quality seasonal variability. Measured parameters included particulate matter (PM2.5 and PM10 in µg/m3) by two Nova SDL 607 wearable air quality monitors and carbon dioxide/carbon monoxide (in ppm) by a pSense portable CO2 meter. Air temperature (°C) and relative humidity (%) were also recorded simultaneously. 
Smokers were identified as the major source of particulate matter in both, outdoor and indoor environments. Proximity to smokers resulted in PM2.5 concentration increases up to 343 and 854 µg/m3 in the outdoor and the indoor smoking-areas microenvironments, respectively. We will further discuss individual exposure, air quality seasonal changes and the relationships between PM2.5, PM10, CO2 concentrations and environmental conditions at each environment.

Air quality where we live, work and play – what the air sensors tell us

By: Gedi Mainelis, Rutgers University


Our recent study showed that AirVisual air quality sensors (IQAir, North America) are sufficiently reliable devices and that they can be used in various indoor studies. They measure PM2.5 concentrations, air temperature, relative humidity, and CO2. In this project, we deployed AirVisuals: in homes of Rutgers faculty/staff, their offices, as well as their outdoor environments – all in central NJ, one of the most densely populated areas in the US. The deployments lasted from a few months to more than a year. In select homes, we also deployed personal air pollution monitors Flow 1 and Flow 2 (Plume Labs, France) to test their accuracy. 

We found that hourly PM2.5 levels in suburban NJ areas were typically below 25 µg/m^3 with very instances where hourly PM2.5 concentrations exceeded 35ug/m^3. Overall, the PM2.5 concentrations measured by AirVisuals stationed outdoors were in reasonable agreement with the PM2.5 levels measured by an EPA station located within a 10-mile radius. The coefficient of correlation was ~ 0.4 and the linear trendline closely followed the 1:1 line with a slope coefficient of 0.98. The hourly PM2.5 levels indoors typically stayed below 20 µg/m^3 except during cooking activities, when it could spike to several hundreds of µg/m^3 depending on an individual home and the cooking activity. The PM2.5 levels in offices were typically very low: in single µg/m^3. The carbon dioxide concentrations in offices were typically well under control, usually under 600 pm. However, the relative humidity levels in offices were poorly controlled. In summer, they exceeded 70%, while in winter they fell below 20%. On a subjective level, participants indicated that having the monitors nearby made them more cognizant of activities that increase PM2.5 levels, such as cooking and cleaning; many also adjusted their outdoor activities based on outdoor pollution levels, which are conveniently displayed by AirVisuals based on data from the nearest EPA measurement station. 

Development of Standard Test Methods for PM2.5 and CO2 Sensor Units Intended for Indoor Air Quality Measurements

By: Michelle Kuang, South Coast Air Quality Management District

Summary: Air quality sensors are able to provide feedback data necessary to improve indoor air quality through ventilation control, filtration, or other air quality control treatments. In order to provide better sensor information for users to make informed environmental control decisions, Standard Test Methods for CO2/PM2.5 Sensor Units Intended for Indoor Air Application were developed in coordination with ASTM International to establish comprehensive protocols to evaluate commercially available indoor air quality sensors for their ability to measure a wide range of pollutant concentrations, recover from loss of power, and perform under various climate conditions and in the presence of interferents, using South Coast AQMD’s Air Quality Sensor Performance Evaluation Center (AQ-SPEC) environmental test chamber. The performance of indoor air quality sensors were evaluated based on data recovery, intra sensor variability, accuracy, precision, hysteresis and correlation to Federal Reference/Equivalent Method instruments. We focus on the development of these methods, and exemplary results from testing air sensors following these methods are presented.


A study of Energy Transition, Health and Indoor Environmental Quality

By: Shelly Miller, The University of Colorado Boulder


The change in global climate will increasingly change the way we work and live in our communities. Individual choices, for
example in transportation and home ventilation, can affect health and well-being. Previous studies have shown that beneficial improvements in health and well-being are achieved with an improvement of indoor air quality relating to energy efficiency and energy transition in homes.

The objective of our project is to document physical evidence that the move towards a clean energy economy will improve environmental quality, health and well-being of residents in the City of Boulder and Boulder County. We will explore the following hypothesis: Households shifting towards netzero energy use by installing electrical heat pumps to replace natural gas furnaces will result in improvements in residents' indoor environmental quality, well-being and health. 

The City of Boulder has partnered with NREL and the University of Colorado Boulder to recruit participants who utilize heat pumps which runs on solar electricity, or as a control typically using natural gas for heating. Since heat pumps and solar PV are expensive, initial adoption will be difficult. During this process, the University of Colorado Boulder’s air quality team will distribute seasonal surveys to document the health and well-being of participants as well as monitor individual household air quality using low cost sensors. 

A pilot study conducted in 2018 showed low-cost IAQ sensors were were effectively integrated into everyday life. Participant survey response was low. Data were consistent across homes showing generally low pollution levels with episodic peaks mainly due to traffic and cooking. 

Evaluation of the Impact of Indoor Air Filtration on Particulate Matter Exposures and Cardiovascular Health: A Pilot Study

By: Shirley Huang, Department of Environmental and Occupational Health Sciences, University of Washington

Summary: This study assessed the impacts of using auto-mode air cleaners with high-efficiency particulate air (HEPA) filters designed to automatically turn on in response to elevated indoor particle concentrations on cardiovascular responses among healthy adults. Six non-smoking healthy adults, aged 24 – 37 years were recruited in Seattle, Washington for a randomized crossover study, in which each participant used an air cleaner under three intervention scenarios: (1) sham-mode filtration with the air cleaner operated with the filter removed, (2) auto-mode filtration with the air cleaner set to auto-mode, and (3) adjustable-mode filtration with the participant adjusting the air cleaner manually. Each intervention session was one-week long and separated by a two-week washout period. Repeat measures of blood pressure were collected as an indicator of participants’ cardiovascular health. Two sets of self-administered home blood pressure measurements were taken at 8 am and 8 pm each day by the study participant followed a standardized protocol during the air filtration intervention using a validated wrist-worn oscillometric blood pressure monitor. Results indicated that compared to the sham-mode and adjustable-mode filtration, use of auto-mode filtration resulted in lower mean indoor PM2.5 levels by 5.8 (95% CI [4.6, 7.0]) μg/m3 and 1.6 (95% CI [0.4, 2.8]) μg/m3, respectively. The use of auto-mode filtration was also associated with a non-significant reduction in both systolic and diastolic blood pressure as compared to the sham-mode and adjustable-mode filtration. HEPA air cleaners running in auto-mode effectively reduced indoor PM2.5 levels. We observed reductions in blood pressure as expected for sham-mode filtration, but these did not reach statistical significance. A larger study is needed to determine health benefits of the PM reduction afforded by auto-mode filtration air cleaners.

Assessment of PM2.5 concentration and transport in indoor environments using low cost sensors 

By: Sumit Sankhyan, University of Colorado Boulder

Summary: Fine particulate matter (PM2.5) is an important constituent of air pollution that has been linked to a variety of health effects. Consumer-grade low-cost PM sensors are gaining popularity as a convenient tool for consumers to monitor indoor air quality inside their homes. We tested five commercially available PM monitors (IQAir AirVisual Pro, Foobot Home, PurpleAir PA-II-SD, and PurpleAir PA-I-Indoor) and compared them to a research- and industry-grade optical particle monitor (TSI Optical Particle Sizer, OPS 3330) by deploying them in three homes of different sizes over a period of 9-12 weeks each. Two identical units of each monitor were deployed in the kitchen and bedroom of each home to evaluate PM2.5 transport between the kitchen and bedroom. Indoor monitors were collocated for 3 days at the beginning and end of each deployment period to assess their accuracy. Two PurpleAir PA-II-SD sensors were also deployed outdoors to compare the indoor-to-outdoor ratios of PM2.5 concentration during periods of no activity inside the house. A second component of the project was to investigate the effects of deploying a consumer-grade portable air cleaner in the kitchen and in the bedroom on PM2.5 levels. A range of sensors (Samsung SmartThings) were also deployed to detect opening of doors and windows and operation of the air purifier. Results are expected to report the efficacy of low-cost monitors to inform the concentrations and transport of PM2.5 between the kitchen and bedroom of homes.

Characterization of fine particulate matter (PM2.5) concentrations in single-family homes in the southeastern U.S. using low-cost air quality monitors 

By: Tanvir Khan, Florida Solar Energy Center

Summary: In this study, continuous indoor measurements of PM2.5 concentrations were conducted in 30 single-family, non-smoking occupied homes for a duration of 1 to 2 weeks in the southeastern U.S. states (Florida, Georgia, and South Carolina) as a part of a large on-going study aimed at investigating indoor air quality in ~60 homes in this region. To capture the spatial variability in measured indoor concentrations, low-cost optical monitors with high temporal resolution (1-min) were placed in multiple indoor locations (living area near the kitchen, master bedroom, and second common area). To monitor cooking activities (a major source of indoor PM2.5), temperature sensors were used on cooking burners, toasters, ovens, and other indoor hot cooking surfaces. In addition, the use of spot mechanical ventilation such as kitchen range hood and exhaust fans were monitored. Additional data on factors affecting PM2.5 such as candle/incense use and bad outdoor air quality were collected through participants’ self-reported daily activity logs. The primary goal of this presentation is to investigate the sources to PM events as measured by the low-cost monitors by utilizing the concentration measurements, and relevant data from temperature sensors, and occupants’ daily activity logs. Preliminary data analysis from 30 homes reveals that mean PM2.5 concentrations measured in the main living areas (adjacent to the kitchens) varied from 0.5 to 13.5 µg m^-3. However, large variability in measured (maximum) indoor concentrations (4-1105 µg m^-3) was found, suggesting strongly varying indoor source strengths and occupant behavior across these homes. In addition, it was found that the maximum concentrations measured simultaneously at multiple locations by the monitors varied by a factor of 1 to 8. We will present our approach to examine this large variability and utility of using low-cost particle monitors to track occupant activities linked to PM events. 

Deploying Air Quality Sensors by Communities

Developing Innovative Solutions for Networked Air Quality Sensors Near Wildfire Prone Communities 

By: Jacob Holle, Intellisense Systems Inc


Intellisense Systems Inc. (ISI) is advancing the development of a new Remote Air Quality Reporting (RAQR) device to measure and track wildland fire pollutants, specifically for communities located near forested areas.  In partnership with the Environmental Protection Agency with their Small Business Innovative Research (SBIR) team, ISI is enabling detection and reporting of contaminants including carbon monoxide, carbon dioxide, nitrogen dioxide, and particulate matter in a small, easy-to-use unit capable of data storage and dissemination. By working alongside the EPA with an SBIR Phase I, ISI is excited to bring a critical capability to cities impacted by wildfire air quality. 

The RAQR system leverages ISI’s existing low size, weight, and power technology used in both the Micro Weather Sensor (MWS) and AWARE Flood. This particular Air Quality reporting system in development can harvest solar power and communicate through radio, cellular, or Iridium, coupled with a new innovative, compact air-quality sensor suite. 

As RAQR is small, simple to set up, and does not require hardline power, the sensor will enable end-users to deploy a robust network of air-quality sensors rapidly and at low cost. These new data sets will then be used to track air quality around dense population centers at significantly higher resolution than currently possible and also serve as an early warning system for fire detection. Each sensors information can then be shared with researchers, government agencies, and commercial modelling developers to not only predict near term impacts, but also forecast mid and long-term impacts and expected changes in air quality. 

ISI has developed a unique ability to support user initiatives of building dense sensor networks without committing to long-term sustainment and maintenance requirements.The presentation will highlight ISI’s development path of RAQR, share details participating in EPA’s SBIR program, and creating a networked ‘IoT’ sensor. 

User-friendly mobile air sensing devices: the key to the success of community-based participatory research on air pollution monitoring

By: Yoo Min Park, East Carolina University

Summary: This study develops a new personal air monitoring device that considers an end-user’s perspective/experience and discusses how the user-friendly device can contribute to the success of community-engaged projects that aim to enhance public awareness of air quality and mitigate environmental health disparities. Although low-cost portable air sensors have offered tremendous opportunities for accurate personal real-time pollution monitoring and exposure assessments, community-based participatory research using these technologies still poses several challenges. Because low-cost air sensing devices have been developed focusing primarily on improving their accuracy to meet the needs of researchers, the usability of the devices among the general public has not been fully taken into account in the design. Therefore, low-income communities and socially marginalized groups tend to be underrepresented in such technology-driven studies due to the limited technological literacy, poor user-friendliness of air sensing devices, limited awareness of environmental health risk factors, and an unwillingness to participate in scientific research. This can lead to the most vulnerable groups being excluded from policy-making processes and, ultimately, may exacerbate environmental health disparities. We propose a personal mobile air monitoring device that is user-friendly for everyone regardless of their technological literacy to improve their awareness of air quality in their community and their daily activity places (e.g., homes, workplaces, and social/recreational venues) while ensuring that it can produce high-quality data so that researchers can perform data analysis and visualization. We also discuss how the user-friendly air monitoring devices can help increase participation of underrepresented communities in the citizen science projects and how they can produce research outcomes that are beneficial to both the communities and researchers.

The challenges of deploying air quality sensors by communities

By: Chrystal Gaither, IQAir


IQAir AirVisual has worked with hundreds of communities and contributors around the world to deploy low cost air quality sensors. Deploying sensors is a challenge, but managing them over long periods of time and making the air quality data actually reach the people is even more challenging. This presentation will cover the following topics illustrated with concrete use cases met in projects:

- Technical and human challenges in selecting location and installing sensors
- Sensor data validation and calibration
- Sensor maintenance and management
- Getting the data out to make a significant impact
- Case studies of successful communities that have created significant impacts in their society

Mapping hyperlocal air pollution to drive clean air policies

By: Harold Rickenbacker, Environmental Defense Fund


Lower cost air quality sensors are redefining the power of comprehensive spatial-temporal data. But while technology is advancing and creating new hyperlocal insights, cities are struggling to turn that data into local solutions that clean the air and improve local health.

Figuring out how to design and deploy an air pollution monitoring system, and then developing clean air policies based on the data, can be daunting. Environmental Defense Fund (EDF) will help guide local leaders to scientifically rigorous, meaningful clean air decisions, by giving a behind-the-scenes look at our monitoring efforts in pilot cities across the globe.  

Through partnerships with technology firms, scientists, grassroots organizations and city leaders, EDF has garnered best practices for using both mobile and stationery monitoring networks to inform land use zoning and permitting, implement emergency public health interventions, and advise the design of traffic management measures and transportation projects. Learn key takeaways from our work in:
•    London, UK, to measure pollution levels before and after the introduction of a new Ultra Low Emissions Zone;
•    Houston, TX, to identify elevated levels of benzene (~300 ppb) near petrochemical facilities after Hurricane Harvey; and
•    West Oakland, CA, to develop city-wide exposure reduction strategies, such as truck management and electrification, to benefit nearby port communities. 

EDF aims to create a resource center for city leaders and academics interested in using air pollution data to design new solutions, build political support for action, increase compliance, and hold polluters accountable. For example, our newly released how-to guide for hyperlocal air pollution monitoring (available at demonstrates how a new wave of transdisciplinary research bridging air pollution science, grassroots advocacy, and policy making and governance. 

Performance assessment of low-cost stationary PM2.5 sensor networks, deployed in Brooklyn and the Bronx, New York

By: Holger Eisl, Barry Commoner Center for Health and the Environment

Summary: Traditional approaches to air quality monitoring typically involve regulatory agencies that utilize expensive and complex stationary equipment, maintained by trained staff, to provide the type of highly accurate data needed to demonstrate attainment with federal air quality standards.  While this type of monitoring is a vital component to air quality management, in urban areas these monitors are often deployed at, only, a limited number of rooftop locations.  Though intended to track urban scale trends in pollution levels, the placement of these monitors is not spatially dense enough to characterize intra-urban spatial variation in air quality, due to local emissions sources such as traffic.  To address this limitation, this project explored the feasibility of using stationary low-cost monitoring networks for spatial and temporal estimation of ambient fine particulate concentrations in two environmental justice communities in New York City – El Puente (Brooklyn) and Youth Ministries for Peace and Justice (Bronx). The data from the community-based low-cost stationary monitoring networks were compared to FEM/FRM data and the findings land use regression (LUR) analysis of the New York City Community Air Survey (NYCCAS). The stationary networks in both neighborhoods consisted of a total of 22 monitoring locations. The data collection started in January 2019 and lasted until November 2019. In collaboration with the New York State Department of Environment Conservation (DEC), the low-cost air quality monitors (AirBeam2) were surveyed and assessed through field colocation and integrated into a cellular data acquisition system. QC/QA data were collected both, before and after the deployment for a duration of 3 weeks. Based on the r2-value a strong agreement was observed between FEM and AirBeam2 low-cost sensors.

BAAQMD’s Portable Air Quality Monitor (PAQMon) design and potential applications

By: Charity Garland, Bay Area Air Quality Management District


The Bay Area Air Quality Management District (BAAQMD) is currently developing a highly portable air quality monitor to meet growing data needs faced by BAAQMD and Bay Area residents. The lightweight (~75 lb.) Portable Air Quality Monitor (PAQMon) will contain relatively low-powered Federal Equivalent Method (FEM) and other high-quality instrumentation resulting in measurements of particulate matter mass and size, NO, NO2, NOx, and O3, among other potential pollutants which may be selected depending on study objectives. Metadata will also be collected to assess system health and security, qualify data, and account for differences in performance of collocated instruments. Data will be logged with an onboard computer, which will allow for synchronized timestamps across all metrics. Remote data download and control of the PAQMon will be made possible by a cellular modem connection. Currently, the system will require shore power, however, eventually solar power will allow for an “off-the-grid” option.

The PAQMon will prove valuable for use as a back-up monitoring platform when additional or stand-in air quality measurements are required, such as during wild fires or power outages, both increasingly common occurrences in the Bay Area. The PAQMon can also be used as a “transfer standard”, using NIST traceability to ensure the highest quality data possible. The “transfer standard” capability combined with the PAQMon’s portability will make it especially valuable as a quality control tool for sensor networks.

This poster will discuss in detail the design and potential applications of the PAQMon, as well as limitations and challenges faced during development.

A Board Game to Identify Local Air Pollution Sources

By: Landon Bassett, The University of Connecticut

Summary: Air pollution research has come a long way since the 20th century. With the advent of cheaper personal air quality sensors, air pollution research and awareness has greatly increased, not only in academia but also in local communities and cities. Fully mapping a city with air quality sensors can seem like a daunting task, therefore, it is important to develop relationships with the people of local communities in order to receive assistance with monitoring and maintenance. Working with local communities is also beneficial in promoting air pollution awareness. A healthy relationship with local neighborhoods allows both researchers and residents the chance to work together in setting up air quality sensor networks. We will present an air pollution board game that could provide a more engaging way to bridge the gap between researchers and residents. The game itself works towards helping players identify local sources of particulate matter and provides open ended opportunities for them to suggest local areas suspected of having high pollution. This game is designed to be helpful to both researchers and residents. Playing the game cultivates awareness in residents about local particulate matter sources while providing researchers with insight of potential sources not otherwise captured by the data. This added information can inform the future siting of low-cost air pollution monitoring networks to gather data on local air pollution concerns. We are testing whether or not this game approach provides better information about these “missing” local sources than surveys. We will present results from a preliminary test of this game at the University of Connecticut - Storrs Campus though our ultimate goal is to expand it for use in community-engaged research in Hartford, CT.

Application of low-cost portable air quality sensor to evaluate the association between personal exposure of fine particulate matter and heart rate variability in a community-based study

By: Ming-Chien Tsou,  Research Center for Environmental Changes, Academia Sinica, Taiwan

Summary: Personal fine particulate matter (PM2.5) exposure may vary from person to person due to different lifestyles. Previous studies had showed associations between altered heart rate variability (HRV) and PM2.5 exposure. In this study, we evaluate the effect of personal PM2.5 on HRV using the low-cost portable air quality sensor (AS-LUNG) in a community.
We recruited 28 adult residents living in a community in Taiwan. Every subject was asked to wear a AS-LUNG and a wearable HRV sensor (RootiRx) to measure PM2.5 concentrations and HRV for consecutive 48 hours, respectively. Meanwhile, subjects were also asked to record time-activity diary. The effects of 5-minute average PM2.5 on HRV, including standard deviation of NN intervals (SDNN) and ratio of low frequency to high frequency powers (LF/HF ratio) during awake time were evaluated by generalized additive mixed model.
PM2.5 concentrations derived from AS-LUNG were adjusted to the research-grade results by comparing the results of AS-LUNG to Grimm portable aerosol spectrometer. A total of 8395 5-minute observations were obtained. A significant association between LF/HF ratio and PM2.5 (p=0.004) were found after adjusting age, gender, BMI, activity, location. We also found that a borderline significant association between SDNN and PM2.5 (p=0.065). We further focus the time when subjects stayed indoors. SDNN and LF/HF ratio were both significantly associated with PM2.5 (p=0.015 and 0.001, respectively) after adjusting age, gender, BMI, activity, ventilation and PM2.5 exposure events when subjects stayed indoors.
We demonstrated that the low-cost, more accurate portable air quality sensor can be used to investigate the 24-hour consecutive PM2.5 concentrations, i.e. the real-life PM2.5 exposures, for a large population. The HRV indices were associated with PM2.5 exposures. Future studies should consider seasonal differences in association between PM2.5 exposures and HRV.

Design of an Air Sensor Loan Program with Public Libraries

By: Rachelle Duvall, US EPA


Air sensors are more frequently being used by the public to learn about air quality in their communities. To facilitate access to sensors and provide resources for operating them, a pilot air sensor loan program will be established at select branches of the Los Angeles Public Library System. Community members will have an opportunity to check out sensors, similar to checking out a book, and use them to collect air quality measurements. The air sensors will measure fine particulate matter (PM2.5) which is a common air pollutant that contributes to respiratory and cardiovascular health problems. As part of the loan program, curriculum will be developed and shared with librarians who will then host workshops and classes for library patrons on the air sensors and air quality. This loan program uses a “train the trainer” approach where librarians will receive training and train other librarians to implement the program in their library branches. An overview of this first-of-its-kind loan program will be discussed along with lessons learned in the development of the program. 

Disclaimer: Although this work was reviewed by EPA and approved for presentation, it may not necessarily reflect official Agency policy.

Report on a Low-Cost Particulate Matter Local Sensor Network

By: Rebecca E. Skinner, Manylabs- San Francisco AQ


We report on progress and lessons learned with Manylabs' low-cost PM 2.5 sensor network for San Francisco's Eastern Neighborhoods, emphasizing both technical aspects of PM sensor deployment and the procedures for such projects. This neighborhood has high levels of criteria pollutants, as is widely documented.

Our Cal ARB Supplemental Environmental Projects grant was awarded and funded October 2018. We obtained permission from CalARB Enforcement to adjust our plan from DIY sensors to calibrated off the shelf Purple Air IIs. We arranged co-location and calibration using a borrowed eBAM, colocating two dozen PA-II sensors with the eBAM and a MetOne ES 642 dust monitor. The Grafana data display portal is currently displaying PM 2.5 data from several deployed sensors. We've installed eight calibrated sensors, and are arranging several other hosts.

Initial results vindicate hyperlocal data collection, with distinct diurnal differences between neighborhoods, and characteristic spikes near-roadways, although the calibrated sensors generally follow the trend of San Francisco's FRM station, northwest of most of the sensors. 

Our plans include a second CalARB grant application, further technical initiatives, ongoing community activities, and work on presentation of sensor data. The latter includes a) local data display, using LIFX bulbs colored to correspond with the EPA’s AQI color code; b) ongoing civic and community outreach; c) an improved website with a live data map. The former include operation of a “community reference site” including PM reference monitor purchase (TS640 or similar); more sophisticated calibration investigations, and further evaluation of the Purple Air II dual sensor channels.

We'll also discuss ongoing development of best practices for the emerging field of community AQ monitoring projects. These practices will include installation preparation, host recruitment, timing of projects, and expectations regarding funding and personnel requirements.

Long-term Machine-learning Calibration Model to Predict Ozone Concentrations Using a Metal Oxide Sensor

By: Tofigh Sayahi, University of Utah

Summary: Ozone (O3) is a potent oxidant that is linked to adverse health effects.  Low-cost ozone sensors, such as metal oxide (MO) sensors, can complement regulatory O3 monitors and enhance the spatial resolution of measurements. However, the data quality of MO sensors remains a challenge. The University of Utah has a network of low-cost air quality sensors (called AirU) that primarily measures PM2.5 concentrations around the Salt Lake City area. The AirU package also has a low-cost MO sensor ($8) that measures oxidizing/reducing species. These MO sensors exhibited excellent laboratory response to O3 although they exhibited intra-sensor variability.  Field performance was evaluated by placing eight AirUs at two Division of Air Quality (DAQ) monitoring stations with O3 federal equivalent methods for one year to develop long-term ordinary least squares (OLS) and artificial neural network (ANN) calibration models to predict O3 concentrations. Six sensors served as train/test sets. The remaining two sensors served as a hold-outs to evaluate the applicability of the new calibration models in predicting O3 concentrations for other sensors of the same type. The independent variable selection was also performed by OLS and least absolute shrinkage and selection operator (LASSO) regressions. The variable selection results indicated that the AirU’s MO oxidizing species and temperature as well as solar radiation from DAQ were the most important variables for the OLS and ANN calibration models. The OLS calibration model exhibited moderate performance (R2 = 0.58 for test and R2 = 0.62 for hold-out set), and the ANN exhibited good performance (R2 = 0.82 for the test set and 0.81 = the hold-out set). The ANN was able to help address the intra-sensor variability challenge. These low-cost MO sensors combined with ANNs can lower the cost of air sensor packages, making them more accessible to communities.

SmartAIR ‘20: Community-driven Smart Street-Level Air Quality Monitoring, Mapping System with Automatic Calibration using Regulatory Sensor Networks

By: Vikram Rao, University of California, Davis

Summary: Real-time street-level Air Quality (AQ) monitoring is an important application of wireless sensor networks (WSN). Presenting accurate AQ data to users in real-time is still a major challenge and actively under research. Current models predict ambient AQ using data captured from city-wide mobile monitoring campaigns (like Google Street Cars) using approaches such as feedforward artificial neural networks (ANN), hybrid ANN or statistical model. Such campaigns are expensive and done only infrequently. Nowadays low-cost low-power portable AQ sensors can be easily mounted on vehicles (bike, car, public transports). Using these devices, we introduce a new community-driven participatory sensing approach to collect AQ information efficiently from participants, while 1) increasing security, 2) calibrating the data to nearby regulatory monitors and 3) handling the rapid growth in community-derived data. We devise a secure protocol to mitigate HTTP/NoSQL injection attacks and subsequently eliminate malicious nodes from the system. We introduce automated calibration sampling protocol that leverages AQ data from well-calibrated Regulatory Sensor Networks (RSN). Calibration occurs whenever a low-cost sensor comes close to an RSN. As the portable sensor community increases, data requests increase. We introduce a novel Zoned Distributed Hash Table (Z-DHT) – a NoSQL resilient database to populate AQ data across various zones within a zone.  We define the zone based on area zip code, latitude and longitude coordinates. Each zone has its own DHT. By this, we decentralize data-fetch and data-push operations, reduce cloud usage costs, reduced network delay and importantly increases scalability. We have implemented a prototype of our framework at UC Davis. Results and analysis of our model show robust performance in terms of data quality, network latency, scalability and data security.

Improving Tribal and Citizen Science with Low-Cost Air Sensor Collocation Shelters

By: Ryan Brown, US EPA Region 4


Collocation of air quality sensors at existing regulatory air monitoring sites, where sensor data can be compared against quality-assured reference data, is a cost-effective method for understanding the performance of lower-cost air sensors. In some cases, collocations can allow for location-specific correction factors to be developed to improve sensor accuracy. For previous air sensor research and projects, EPA designed and deployed a few air sensor collocation shelters to provide weather shielding and security to support collocation efforts. In addition, Mecklenburg County Air Quality (MCAQ) built and deployed a similar air sensor collocation shelter, called the Community Science Station, at their regulatory multi-pollutant air monitoring site in Charlotte, North Carolina.  The Community Science Station supports community groups seeking air sensor collocation space and guidance from MCAQ staff.

Based on these previous designs as well as the experiences of EPA and other agencies, EPA/ORD is currently working to develop and publish a reusable, affordable design for an air sensor collocation shelter and a best practice guide for its use in supporting community science.  This will allow similar shelters to be built and installed at regulatory air monitoring sites across the U.S.  During this project, EPA will pilot the installation and utilization of air sensor collocation shelters at multiple tribal and state monitoring locations in the Southeastern U.S. Having air sensor collocation shelters accessible to the public and encouraging community, and outside groups to work with nearby regulatory air monitoring staff, will leverage existing air quality infrastructure and lower barriers for collecting quality-assured sensor data. This data can be used by a variety of stakeholders with increased confidence.

Optical PM2.5 Low-cost Sensor Quality Control and Verification Analysis

By: Chen Fan Lun, Industrial Technology Research Institute

Summary: Optical PM2.5 low-cost sensors are widely applied in Taiwan with the purpose of improving environmental quality. A quality control mechanism for a set of sensing data is designed for data quality management. All sensors installed in industrial parks, transportation, community and auxiliary sensing areas are designed to go through five stages, which include quality control in manufacturing, sensor-performance testing and verification platform, field evaluation prior to deployment, preventative maintenance and inspection and third-party checking to ensure the data received is accurate and effective. The result shows, take the example of 500 sensors installed in Taichung, that 27% of sensors are covered by obstacles, 96% of the sensor data has an integrity rate of 90% or more, and 80% of sensors compared with a calibration instrument have a bias error lower than (or equal to) 50%. In the verification through on-site checks and quality control mechanisms through the lower layers of the stack, using analysis of population variability, abnormal operation of the sensor can be clearly identified. As such, the manpower and budget needed for maintenance of Air Quality Internet of Things would be reduced substantially, and in the meantime, the abnormal data can be removed from the big data analysis, so as to assist the environmental inspection unit to conduct the hot spot analysis and to play a role in assisting law enforcement on air pollution.

Communication and Interpretation of AQ Data

T640 Optical PMx Inter-comparability With Beta Attenuation And Gravimetric Filter Samplers In Auckland, New Zealand

By: Woodrow Pattinson, Mote Ltd

Summary: Data from USA and Europe have shown that the T640x typically over-estimates PM2.5-10 concentrations by up to ~20%, compared to traditional regulatory (FRM/FEM) monitoring methods (beta attenuation, oscillating microbalance and filter samplers). Similar observations have been made from co-located units in New Zealand, but inter-comparability at individual stations as well as between sites can vary daily with changes in climatic conditions and particulate sources (composition, shape, size). Preliminary data from South Island towns suggests the T640 tends to read approximately 4 µg/m3 higher for PM10 while the relationship between PM2.5 is far closer to 1:1. This work presents instrument inter-comparison data from three distinct areas of Auckland city; the North Shore, Central City and Central-South Auckland. To capture the transition into wintertime, continuous sampling took place from 1st Dec 2018 to 31st June 2019, with the gravimetric samples taken in December and January. For two separate sites, long-term 24-hour T640 optical PM10 was highly correlated with the FH62 BAM PM10 (R2=0.90), but not for PM2.5 (5014i BAM, R2=0.50), presumably due to 5014i instrument noise. For 10 days’ PM10 gravimetric sampling, the coefficient of determination at the North Shore was 0.99 but was only 0.74 at the Central City site likely due to the instruments’ differing responses to dynamic source variation and losses of volatile organic species. On average, the North Shore T640 reported PM10 concentrations 15% and 20% higher than the BAM and gravimetric samplers, respectively (~13 – 15 µg/m3). Despite the poor correlation for PM2.5, long-term averages across co-located samplers were the same (~ 5 – 8 µg/m3 depending on site). These findings for Auckland city generally align with previous work during South Island winters and reflect observations around the world.  

Closing the loop between laboratory and field-based sensor calibrations 

By: Eben Cross, QuantAQ, Inc.

Summary: Integrated sensor systems that measure and report ambient air quality on locally relevant scales provide an exciting new data layer for real-time decision making.  Unfortunately, the ‘truth’ of a given sensor measurement when deployed across different microenvironments can vary dramatically, ranging from ‘misleading’ to ‘indicative’ to ‘quantitative’.  Distributed AQ sensor networks can provide valuable information to community members and decision makers if the sensor calibration works sufficiently well for the target domain.  The vast majority of sensor calibrations to-date have been generated from ambient co-location with reference monitors.  While inherently authentic to the co-location conditions, such calibration datasets often require lengthy sampling periods (weeks-to-months) and limited transferability to domains outside of the co-location environment.  In this presentation we demonstrate the effectiveness of a laboratory-based sensor calibration approach which can reduce overall calibration time and expand the application domains where local AQ data is critically needed. 

Data QA/QC provides opportunities and challenges for a school-based air quality sensor network program

By: Ian Ozeroff, Earthwatch Institute

Summary: A/QC and correction of low-cost sensor data can increase result accuracy and utility. However, incorporating these control processes into a workflow that delivers data products to communities raises unique challenges and opportunities. We present the outcomes of developing this workflow for a collaborative project that involves placing particulate matter sensors in schools and libraries at four global locations: Boston, Southern California, India, and Sri Lanka. We have designed a tripartite workflow, composed of ingestion, QA/QC and correction, and analysis. This workflow both leverages existing tools, like the AirSensor R-package, and necessitates the development of new tools. Each section is designed to balance the competing priorities of scientific rigor and public accessibility. Our workflow begins with ingestion, cataloguing, and reading data from over 70 sensors deployed in diverse socio-environmental contexts. Automated data pipelines assume standardized inputs and precise records, and lapses, in either, lead to critical errors or false outputs. As our data is collected from varying communities, inputs and records may vary. Our team has implemented metadata conventions to preclude input error and allow for scalable project scope. The next step, QA/QC and correction increases confidence in the measurements. However, there are challenges not only in developing these methods, but also in communicating how these methods work and their impacts on the data. Transparency and discussion with our partners, adapts rigorous data processing into a learning opportunity on the science of measurement and the blossoming field of low-cost particulate matter sensor correction.  The final goal of our workflow is to generate reliable, accessible, and useful outputs, such as datasets and visualizations, to share with the public. We expect our deliverables to inform our partners’ actions and provide real-world, localized data as material for data-literacy curricula.  

Performance of Low-Cost PM Sensors Before and After Simulated Aging

By: Jessica Tryner, Colorado State University

Summary: Studies that characterize the performance of low-cost PM sensors are needed to provide practitioners with a better understanding of the accuracy of the mass concentrations and particle number size distributions reported by different sensor models. In this study, we conducted a set of laboratory experiments to evaluate the performance of three PM sensors: the Plantower PMS5003, the Sensirion SPS30, and the Amphenol SM-UART-04L. The sensors were exposed to various polydisperse aerosols, monodisperse aerosols, and environmental conditions to answer the following questions: (1) do the sensors respond linearly to PM concentrations ranging from 0 to 1000 μg m-3, (2) how does the relationship between the PM mass concentration reported by each sensor and the true PM mass concentration vary with aerosol type, (3) how are sensor readings affected by high relative humidity, (4) how accurate are the particle number size distributions reported by the sensors, and (5) do the sensor responses drift over time in a high-concentration environment? We found that the correction factor relating the sensor-reported and gravimetrically-determined PM2.5 concentrations varied with aerosol type for both the PMS5003 and SPS30 sensors. For all four aerosols tested (ammonium sulfate, Arizona road dust, NIST Urban PM, and wood smoke), the PMS5003 consistently reported higher PM2.5 concentrations than the SPS30. When exposed to monodisperse aerosols with particle sizes between 0.1 and 2.0 μm, the size distribution reported by the SPS30 shifted in accordance with the known particle size, whereas the size distribution reported by the PMS5003 remained roughly constant. After aging the sensors by exposing them to an average Arizona road dust concentration of 34000 μg m-3 (total PM) for 18 hours, the PM2.5 concentrations reported by the SPS30 sensors remained consistent, whereas concentrations reported by several of the PMS5003 and SM-UART-04L sensors became nonsensical. 

New Vision for EPA’s AirNow Program

By: John White, US EPA


The EPA’s AirNow program resides in the Office of Air Quality Planning and Standards in RTP. Air quality reporting to the public has rapidly changed as well, where government agencies are not the only voice anymore in providing Air Quality Index (AQI) information.  

As such the AirNow team has taken a fresh look at the current state of the program and developed a new working strategic vision that will address some of these questions: 

•    As a science-based agency, how can EPA provide the public with the best information to protect their health? 
•    Can (or should) the Agency use sensor data and/or other datasets (e.g., numerical models, satellite estimates, etc.) to provide a robust Air Quality Index surface/value for everywhere in the U.S.? If so how do we properly message this information? 
•    How do we work with our partners, including the private sector, to provide a consistent AQI value to the public?
•    Does AirNow intend to incorporate sensor data and if so, how?

AQview: Community Air Quality Viewer

By: Mena Shaw, California Air Resources Board


The California Air Resources Board (CARB) is developing a new online data portal, AQview, to respond to the growing interest in community air monitoring and the need for a common platform to store and disseminate California’s air quality data. AQview seeks to improve access to and interpretation of air quality data from traditional monitoring networks, communities (in support of community air monitoring programs implemented through Assembly Bill 617), and other air quality monitoring networks by serving as a comprehensive online air quality data portal. 

Significant effort has gone into creating web-based tools for accessing and visualizing air quality monitoring data in the past at both the statewide and community levels. However, existing tools have traditionally focused on only a portion of the air quality data, requiring users to visit multiple websites to achieve a comprehensive picture of air quality from all data sources. 

The fragmented state of current air quality tools can be attributed to the disparate nature of air monitoring, including differences in air monitoring objectives, pollutant categories, equipment techniques and maintenance, data quality, etc. While it will be challenging to combine these datasets into a single tool and simultaneously provide simple interpretation of measurement data and data quality for users, we believe that existing web portals can serve as models. CARB will build on these existing programs to enhance accessibility to all types of air quality data. 


Community Air Monitoring in Puget Sound (CAMPS): Communicating Air Monitoring Results

By: Nancy Carmona, University of Washington

Summary: Communicating results to community partners is an essential component of community engaged research. However, in environmental exposure studies it is not clear how to communicate these results without confusion from the public. The Community Air Monitoring in Puget Sound (CAMPS) study recruited a variety of groups concerned with air quality in their neighborhoods including day care centers, community centers, and senior centers. Community partners hosted low-cost air monitors measuring fine particulate matter, nitrogen dioxide, carbon monoxide, and ozone for two seasons from Summer 2018 – Winter 2019. Report-back included a letter with information about the pollutants measured, average concentrations compared to regulatory standards, average concentrations by hour of day, and actionable behavior modifications to reduce personal exposures. In person meetings were held to review the report, address any uncertainties, and discuss modifiable actions. Communicating results with a variety of communities with different socioeconomic status, educational attainment and racial/ethnic composition found that reporting information required tailored methods. Feedback from the community steering committee was crucial in tailoring the report-back for each neighborhood. Tailored methods included reducing the quantity of information provided, simplifying language, and  straightforward figures. Meeting with community members to learn about their concerns of sources such as airplanes, traffic, and diesel trucks provided a richer story than only comparing results to national standards. Comparing results to standards is not informative when communities perceive sources of concern. Given the confusion around air quality, air pollution sources, and health effects it is vital that air quality information is presented in a manner that addresses the concern of the public and improves knowledge of air quality.

Comparing Air Quality Worldwide 

By: Chrystal Gaither, IQAir North America, Inc.

Summary: This presentation will elaborate on why the US AQI needs to be the standard. AQI is computed in different ways around the world. China and America have the two most widely used systems. Both are calculated weighting the six key pollutants. The results of these two functions differ only in AQI scores of 200 and below.
Since the American index system yields higher scores for AQI’s under 200, it is thought to be more strenuous. For this reason, the American index has become the general world standard. Finally, this presentation will dive into how to deploy air quality sensors for schools and teach air quality.