Oral Presentation Abstracts

Advanced measurement approaches for fenceline and fugitive monitoring applications

Combining low cost PID sensors and triggered canisters to document acute air toxics exposure episodes near oil and gas development

Jeffrey Collett, Colorado State University (Invited Talk)

Ambient concentrations of air toxics and other VOCs were measured near large, multi-well oil and gas pads in suburban Broomfield, Colorado. Monitoring began prior to drilling and continues into the production phase. Spatial networks of time-integrated VOC canisters provide clear indication of impacts of oil and gas (O&G) operations on local air quality, but cannot document acute exposure events that result from short periods of elevated emissions. A PTR-MS provides 90s time resolution measurements of BTEX, helping to identify acute exposure episodes, but lacks measurement of O&G marker species (e.g., light alkanes) needed to document the responsible source. Amobile laboratory hunts and characterizes plumes in nearby neighborhoods, but practical deployment schedules limit opportunities to capture transient events. VOC plumes from O&G operations often impact a location for just minutes. Benzene in these plumes needs to be quantified at ppbv levels to compare against acute exposure health guideline levels. To meet these needs, we have utilized APIS PID sensors with triggered canisters to continuously monitor ambient VOC levels and collect high quality, speciated VOC concentration data when VOC-rich plumes are observed. Multiple sensors are placed around an O&G operation to improve spatial coverage. Our high time resolution measurements identified a variety of short-term pollution episodes, especially during transition and maintenance activities, including pipe pulling, coiled tubing operations, and separator maintenance.. The composition of detected plumes was tied to O&G operations by elevated light alkane concentrations, a low i/n-pentane ratio and local transport conditions. Concentrations of benzene in captured plumes reached as high as 52 ppbv.

Additional Authors: Jeffrey Collett, Brent Buck, I-Ting Ku, Bryan Terry, Yong Zhou, Katie Benedict, Morgan Frazier, Emily Lachenmayer

Pairing high- and low-cost sensing technologies to understand cumulative health impacts for fenceline communities

Kirsten Koehler, Johns Hopkins University (Invited Talk)

While it is known that there are broad impacts of the petrochemical industry on the environment and public health, the current regulatory monitoring network is too sparse to fully quantify the exposures to fenceline communities and understand the cumulative impacts from multi-chemical exposures. We conducted a detailed monitoring campaign in a highly industrialized corridor between Wilmington, DE and Philadelphia, PA in September, 2021. The study included a fixed site with chemically resolved gas and particle phase measurements, a lower-cost monitoring network in fenceline communities for particulate matter (PM1,2.5,10), and a mobile monitoring laboratory for fenceline monitoring of agents known to be associated with the petrochemical industry. The approach enabled continuous monitoring at some locations with detailed characterization on a mobile platform that included repeated measures in key communities of interest to establish robust concentration surfaces. The highly spatially and temporally resolved measurements were compared with locations of known sources. Preliminary results show good agreement between the lower-cost sensors and regulatory monitors at co-located sites for PM2.5. Bipolar wind rose plots suggest that peaks in PM2.5 are associated with winds from regions with higher density of industrial sites. Mobile measurements suggest that exposures to some chemicals are higher for fenceline communities and that while exposures were highly temporally variable, high concentration plumes extended into residential areas. For other chemicals with strong contributions from traffic or gasoline filling stations, differences between fenceline communities and the urban background were smaller. This study represents a novel pairing of low-cost sensor technology, high sensitivity instruments commonly used for atmospheric science applications, and cumulative risk assessment approaches to address environmental injustice for fenceline communities.

Additional Authors: Kirsten Koehler, Ellis Robinson, Mina Tehrani, Ana Rule, Keeve Nachman, Thomas Burke, Carolyn Gigot, Andrea Chiger, Scott Van Bramer, Peter DeCarlo

Development and Evaluation of a Novel Continuous and Concurrent Sampling System for Sub-ppb Level Detection of Volatile Organic Compounds in an Industrialized Area in Los Angeles

Pami Mukherjee, South Coast Air Quality Management District (Invited Talk)

In April 2020 South Coast Air Quality Management District begun operation of 10 community air monitoring stations in communities near 7 major petroleum refineries in Los Angeles, California. In addition to refineries, these areas also contain multiple oil-wells and numerous industrial facilities that are potential sources of a wide range of Volatile Organic Compounds (VOC) and air toxics. The stations are equipped with Automated Gas Chromatography (Auto-GC) developed by Tricorntech Corp., Taiwan. Auto-GC instrument, followed by EPA PAMS VOC sampling guideline, is designed to collect ambient samples for the first 40 minutes of the hour followed by analysis of the sample in next 20 minutes. Based on our experience operating community air monitoring sites, air pollution plumes with elevated VOC are often short-lived, lasting for ~10 minutes, therefore resulting in some plumes not being captured. Moreover,the 40-minute sample concentration will report lower (averaged) value than the actual short peak concentration of VOC present in the plume. To overcome these shortcomings, we developed a sampling system (CSS) to sample the ambient air and analyze the previous sample simultaneously. In this design we will split the hour into three sampling segments, 20 minute each with >15-minute of sample collection. Therefore, we can achieve continuous and more frequent ambient samples without altering its sub ppb range detection capability. In this presentation we will present the details of CSS approach and discuss the results of testing of CSS deployment and testing the Rule 1180 sites. The CSS module will be tested for real time ambient sampling in colocation with the established Auto-GC and Optical multi-pollutant analyzers to evaluate its accuracy and efficiency.

Additional Authors: Pami Mukherjee, Robert Wimmer, Matthew Prather, Tsung-Kuan A. Chou, Jie Wang, Tirah H.F. Wu, Olga Pikelnaya, Andrea Polidori



Dynamic olfactometry (ES 137225) is the standard technique for odor intensity measurements, but not adapted for industrial sites which need continuous monitoring and fast results. Additionally, these sites need odor sources identification solutions to ensure efficient remediation. The harbour of Ventspils, (Latvia), can host up to 20 petrol tankers, and the Baltic wind can push the gas and odors towards the city. The municipality does not have the tools neither to quantify nor to identify the sources of pollution and nuisances. The port authorities deployed an array of WT1 devices on different sites close to the sources and at the fence line. During the initial training period, the devices were trained with four different types of samples at different dilution levels (black fuel oil, solvent naphta, petrol, kerosene), and a successful correlation model was established following the 13725 standard between sensory measurements with dynamic olfactometry and the WT1 outputs, allowing the quantification of odors as well as the identification with an accuracy superior to a 0,85 R2. Additionally, a PCA (Principal Component Analysis) and a LDA (Linear Discriminant Analysis) were built and the WT1 modules proved to differentiate accurately the different sources of odor and pollution. The RUBIX WT1 gas and odors sensing modules allow not only on-line monitoring of Odor Unit and various gas emissions, but also allow odor fingerprint identification. The paper will describe the methodology combining a range of smart sensors with AI statistical data processing techniques, and the experimental plan, including the training with dynamic olfactometry.


Monitoring volatiles using a mobile real-time mass spectrometer

Leslie Silva, Syft Technologies

Rapid air-quality assessment is essential to managing and reducing the impact of toxic or unpleasant volatile compounds on communities. Often deployment of instrumentation at or near the source using vehicle mobilized instrumentation is essential to an effective response. The use of a mobile, real-time volatiles analyzer will be discussed in this context, exploring how it is used to conduct continuous volatiles measurements whilst moving around the area of assessment. This could be crucial for rapid response to an alert from a sensor, where higher-order speciation or characterization of the emission is necessary. The analyzer deployed is a Voice200ultra, selected-ion flow-tube mass-spectrometer (SIFT-MS), contained within a mobile laboratory. SIFT-MS is a soft chemical ionization technique that can be used to directly quantify volatile analytes and achieves selectivity of analysis by exploiting several different chemical ionization mechanisms within a single analysis.The mobile laboratory is also equipped with GPS tracking and plotting software and a weather station mounted to the roof. Concentration data can be plotted in real time for all VOCs being monitored. The benefits of mobile SIFT-MS are highlighted in this presentation through the speciated monitoring of malodor compounds such as reduced-sulfurs and amines. This provides major advantages over conventional approaches because these species are often not amenable to chromatography and can be difficult to sample due to thermal instability, reactivity, or loss to the sample container walls. Furthermore, the processes or emissions responsible for malodors can be dynamic and short-lived necessitating continuous monitoring at the site to correctly identify the source.

Additional Authors: Leslie Silva, William Kerr

Complementary and Emerging Techniques for Fenceline Monitoring

Steven Schill, Sonoma Technology

Many industrial fenceline monitoring programs are motivated by a combination of industry proactivity, advances in monitoring technology, community concerns, and regulatory actions. There is a growing need across various commercial industries and geographic regions for ambient monitoring to determine which pollutants are crossing facility boundaries, at what time, and in what quantities. Effective fenceline monitoring approaches include point monitors and passive samplers, open-path absorption spectroscopy, and more. These approaches encompass a wide scope of technologies with varying costs, benefits, and operational considerations. In this presentation, Sonoma Technology scientists will draw on extensive experience with fenceline monitoring to discuss complementary and emerging monitoring solutions, key factors for technology decision making, and other important considerations, such as design, implementation, and operation of effective monitor networks. Techniques to evaluate source-receptor relationships in real time will also be discussed.

Additional Authors: Steven Schill, Ryan Moffet, Scott McEwan, Clinton MacDonald

Air Sensor Use in India

Supplementing air pollution data using low-cost sensor network – CSTEP studies (new addition)

Pratima Singh, Center for Study of Science, Technology and Policy (Invited Talk)

Lack of understanding on air pollution levels, sources, and the polluting activities at the city level has hindered the development of effective policies. With insufficient scientific evidence, clean air action plans have been inefficient for improving the air quality levels. To aid in generating scientific evidence, filling data gaps, and developing policy strategies at the city level, CSTEP is carrying out monitoring exercises in various cities of India. CSTEP is conducting low-cost sensor (LCS) monitoring in the Punjab state to understand the influence of stubble burning on the ambient air quality level in the state. The measurements are carried out in 5 districts in various geographical landscapes (covering urban, rural, and industrial areas) for measuring PM2.5 levels. These generated data sets will help 1) understand the air quality trend (hourly, daily, and monthly historical data), 2) prepare state-level strategies for improving the air quality, and 3) understand the indoor and outdoor air pollution levels. CSTEP is also using different kinds of LCS’s in the city of Bengaluru to generate evidence. Data from such a dense network of LCS will help run models and predict the pollution levels at high resolution. Such a data set with an understanding of the pollution levels becomes crucial in the absence of reference-grade monitors or sparsely placed monitors. High-resolution modeled data will help understand the 1) hotspots and 2) propose interventions for air quality management. CSTEP is using LCS network for the following 1) conducting inter-comparison of various LCS monitors and understanding their functioning and performance, 2) using LCS in mobile monitoring study to better understand the hyper-local air quality level, and 3) generating data and carrying out modeling for creating awareness and building capacity of citizen groups (alerting the citizens on the neighborhood air quality level, decision making at city-level and communication and outreach)

Additional Authors: Dr Pratima Singh


Edurne Ibarrola, Kunak Technologies

Governments are betting on complementary networks to the official stations, to monitor the air quality at a hyperlocal level. However, when deploying an air quality network, it is important to take into account some considerations. The utility of sensor networks continues growing. Proper QA&QC procedures allow users to better understand the quantitative capabilities of their air quality monitoring solution and be resource-efficient to keep the overall cost of the network operation low. Thus, developing, optimizing, and refining advanced techniques for sensor calibration and validation is an important area of ongoing research and is essential to obtain reliable data from sensor-based networks. The first point to consider when deploying a network is the correct installation and maintenance to ensure the proper performance of the devices and the quality of the data. The final user must consider that not only one device will need maintenance, but dozens or hundreds of them, which increases the cost of the project and the difficulties to carry it out. According to this, the air quality sensor systems must have a well-known QA&QC procedure, allowing the proper maintenance and calibration when deploying a network. Besides, it is necessary a platform that allows and efficiently facilitates this. Some real cases are shown. In Albacete (Spain) 10 devices were deployed; in Antwerp (Belgium) 20 mobile devices were installed on top of post vans; a smaller network in Bombay (India); and a green port application, with 25 devices in 5 ports in Balearic Islands (Spain).

Additional Authors: Edurne Ibarrola, Miguel Escribano, Javier Fernandez

Application of Machine Learning Regression Algorithms for Calibration of Low-Cost PM2.5 Sensor

Manoranjan Sahu, Indian Institute of Technology

Low-cost sensors (LCS) can construct ahigh spatial and temporal resolution PM2.5 network but are affected by environmental parameters such as relative humidity and temperature. The data generated by LCS are inaccurate and require calibration against a reference instrument. This study has applied nine machine learning (ML) regression algorithms for Plantower PMS 5003 LCS calibration and compared their performance with Thermal Fisher Scientific SHARP model 5030 as a reference instrument.The nine ML algorithms applied in this study are: (a) Multiple Linear Regression (MLR); (b) Lasso regression (L1); (c) Ridge regression (L2); (d) Support Vector Regression (SVR); (e) k- Nearest Neighbour (kNN); (f) Multilayer Perceptron (MLP); (g) Regression Tree (RT); (h) Random Forest (RF); (i) Gradient Boosting (GB). The comparison exhibits that kNN, RF and GB have the best performance out of all the algorithms with trainscores of 0.99 and test scores of 0.97, 0.96 and 0.95 respectively. This study validates the capability of ML algorithms for the calibration of LCS.

Additional Authors: Vikas Kumar, Manoranjan Sahu

Quantifying long-term exposures to fine particulate matter (PM2.5) using real-time low-cost sensors in the Tamil Nadu Air Pollution and Health Effects (TAPHE-II) cohort, India

Naveen Puttaswamy, Sri Ramachandra Institute of Higher Education and Research

Background: Exposures to PM2.5is typically measured for 24 or 48 hours in cohort studies involving pregnant women. Low-cost sensors (LCS) offer spatially and temporally resolved real-time PM monitoring capabilities that can capture long-term exposures to household air pollution and provide a better estimate of pregnancy period exposures to PM2.5. Methods:We used atmos™ and aerogram™real-time PM sensors developed in India to monitor indoor PM2.5in a sub-set (n=50) of the TAPHE-II cohort households (n=300) from rural (n=23) and urban (n=27) locations of the Tamil Nadu state in southern India. These LCS record PM, T, and Rh at 1-minute time intervals and transmit data in real-time to the cloud. Pump and filter (37mm PTFE, 0.2µm) set-up was collocated for 24-h alongside real-time PM sensors in 16 households to develop indoor-specific calibration equations. In addition, all sensors were collocated with a reference-grade monitor (BAM1020) and linear models were fit to derive calibration co-efficient. Results:Continuous PM data was monitored on average (s.d.) for 26 (11) and 74 (43) days in rural and urban households, respectively. The average data availability was 94.3% (10.4) across all households monitored. The NRMSE for different sensors ranged from 9.5% to 47.7%. Correlation between 24-h gravimetric and uncalibrated indoor real-time PM2.5 were 0.62 (p=0.000) and 0.36 (p=0.014) among rural and urban households, respectively. Typical daily average PM2.5 levels were high during evening cooking especially in exclusive biomass fuel using household followed by mixed-fuel and liquefied petroleum gas (LPG) users. Future work: work is underway to complete indoor PM monitoring in a total of 60 TAPHE-II households. LCS will be collocated with pump and filter set-up (i.e. gravimetric) with 24-h sampling at three time periods in a sub-set of 15 households.

Additional Authors: Naveen Puttaswamy, Sreekanth Vackacherla, Adithi Upadhya, Sudhakar Saidam, Mangalam Sundaram, Rengaraj Ramasamy, Sankar Sambandam, Santu Ghosh, Jay Dhariwal, Ajay Pillariseti, Kalpana Balakrishnan

From lab-scale research to multi city-scale implementation of low-cost sensors: A comprehensive overview of past five years works

Sachchida Nand Tripathi, Department of Civil Engineering, Indian Institute of Technology, Kanpur

As of early December 2021, India has 327 continuous air quality monitoring stations (CAAQMS) in 169 cities. There is an evident need for air quality monitoring (AQM) in India's 7000+ census cities but, investments in extending CAAQMS densities in India would be both impractical and costly. Researchers at IIT Kanpur, in collaboration with national and international institutes have conducted research works, which aimed towards establishing India's first scientifically validated and calibrated low-cost AQM network for PM and gases by comprehensively analysing them under diverse environments. Sensor calibration approach based on k-nearest neighbour distance metric learning provided R2 of up to 0.92 and 0.82 for O3 and NO2 calibration, respectively. Sensors can measure PM2.5 ~10% errors when RH corrections are done using empirical nonlinear equations. Gaussian process regression based on the fly sensor calibration method was also developed for PM2.5 sensors in Delhi. Over 22 reference nodes, the method used a leave one-out cross-validation procedure, which resulted in an overall prediction error of 30% (RMSE 33 μg.m-3) on a 24-h scale. In another work for PM2.5 sensors calibration, a domain adaptation based method that can reduce the need for co-located data is proposed and the R2 is observed as high as 0.88. A calibration approach based on model-agnostic-meta-learning showed the RMSE as low as 10.3 μg.m-3 in PM2.5 measurements. To provide real-time data across Indian cities with different characteristics, the project is now being implemented in 4 more cities namely Jaipur (arid), Guwahati (hills), Chennai (metropolitan), and Kanyakumari (sea-shore). In another undergoing work for Lucknow, 20 PM2.5 sensors co-located with the reference showed a marked improvement from pre-calibration (MAPE 37-82%) to post-calibration correlation (MAPE 12-16%). The sensor network AQM will be enhanced with microsatellite imagery to estimate PM2.5concentrations at a finer resolution.

Additional Authors: Sachchida Nand Tripathi, Ravi Sahu, Mohit Kumar, Sandeep Madhwal, Vaishali Jain, Pradhan Pranav Kumar, Ravi Prakash Maurya, Paresh Lalwani, Michael H Bergin, Prakash V. Bhave

A sensor network to map air quality across the rural-to-urban spectrum in North India

Saumya Singh, University of California, Berkeley

The Indo-Gangetic Plain in north India (IGP) experiences severe air pollution, with typical annual-average fine particulate matter (PM2.5) concentrations in the range 75-150 µg m-3, posing the largest risk for ill health. The heterogeneity of air pollution sources and adverse meteorology in this region challenges the regulatory and policy attempts made towards improving air quality. Existing policies emphasize a city-centric approach, despite air pollution being a regional challenge in the IGP. A key factor impeding effective air pollution management is the absence of monitored air quality data from smaller cities and rural areas which are home to more than 60 % of the population. Here, we present results from our ongoing project SAMOSA (Sensor-based AirMeasurement Observatory for South Asia)to characterize rural vs. urban air pollution gradients in India’s most populous/largest state Uttar Pradesh (UP) using Purple Air (PA) low-cost monitors in stratified sampling scheme across different settlement types (villages, towns, cities, etc) to cover a broad range of conditions. Our experimental approach employs a carefully maintained network of lower-cost PM2.5 sensors (Purple Air [PA] II) calibrated on the basis of hourly-average measurements made against multiple BAMs located at three sites representing in situ urban and rural regions. We deployed more than 50 PA sensors in a clustered pattern to characterize within-settlement variability along a gradient from remote to rural to highly urban conditions. PM2.5 concentrations exceeded the Indian daily standard more than 70% of the time in both rural and urban sites. Overall, we found that PM2.5 concentrations in villages and small towns were often similar to or higher than the nearest large city. These initial results suggest that additional observations in rural settings and smaller cities and towns will provide valuable new information about the sources, dynamics, and consequences of PM2.5 across this region.

Additional Authors: Saumya Singh, Mark J. Campmier, Sairam D, Adeel Khan, Sofiya Rao, Tanushree Ganguly, Adithi Upadhya, Harshraj Mishra, Ravikant Pathak, Karthik Ganesan, Sagnik Dey, Joshua S. Apte

Best practices from Breathe London: Lessons learned from integrating hundreds of air quality sensors with the largest and most advanced regulatory monitoring network in Europe

Panel Presentations:

  • Sara-Jane Millar, Greater London Authority
  • Benjamin Barratt, Imperial College London
  • Meiling Gao, Clarity Movement
  • Bloomberg Philanthropies

The Breathe London Network launched in January 2021. Following a successful 2-year pilot, the Mayor of London took over the funding of the network, appointing Imperial College London to deliver the programme. The network integrates air quality measurements from hundreds of sensors with reference-grade measurements provided by London's regulatory air monitoring network — the largest and most advanced such network in Europe. Over 350 air quality sensors will be installed in boroughs across London by the summer of 2022. Every Breathe London Node is co-located at reference sites and checked by researchers at Imperial College before deployment. Once deployed, the data is continually cross-checked against reference sites in real-time. Aside from technological innovation, one of the most remarkable components of Breathe London is that it puts air quality monitoring into the hands of London’s communities. Thanks to support from Bloomberg Philanthropies, community groups can apply for fully-funded sensors in their neighborhoods — offering a new way to engage citizens with air quality. The operators of this network — Imperial College London, the Greater London Authority, and Clarity — will share valuable information and real-world learnings about the use of air sensors at the city scale.

Topics will include:

  • Distribution - who should own the sensors, and how should they be distributed to ensure equitable air monitoring coverage?
  • Siting and installation - how to select and prioritize monitoring sites? What indemnity provisions should be made, who is responsible for maintenance, and who is liable for making sure devices are properly installed?
  • QA/QC and calibration - how to ensure data accuracy?
  • Community engagement - how to use a sensor network to build relationships between community groups and local government?
  • Research applications - how has data from the network been used for case studies and research, and what other potential applications exist?

Clean Air Monitoring and Solutions Network: getting useful, actionable data out of low cost sensors for air quality action

Olalekan Popoola, University of Cambridge (Invited Talk)

Traditionally, ambient air quality is monitored using reference/equivalent methods often at hourly resolutions. This approach is sufficient for assessing trends and compliance; however, a significant opportunity can be missed particularly in performing more robust data analysis and interpretation by adopting this limited temporal sampling approach. Low-cost sensor (LCS) networks have often been operated at higher time resolution for wide range of pollutants, and we will show in this presentation how this also offers a new data analysis approach for reference instruments and networks. In this talk, we will present data from a hyperlocal network of LCS (AQMesh) in Glasgow, Scotland and for the first time, reference air quality network also operated at high time resolution for periods in 2021. We will present results from the analysis of the high time resolution air quality network data for NOX, PM and size speciated PM measurements.

Additional Authors: Olalekan Popoola, Roderic Jones, Jessica Fleming, Amy Stidworthy, Molly Oades, Daniel Connolly, David Carruthers, Jim Mills, Felicity Sharp, Amanda Billingsley

Estimation of hourly BC from BAM tapes using image reflectance-based method

Abhishek Anand, Carnegie Mellon University

Exposure to ambient air pollution accounts for 4.2 million deaths annually. Yet, air quality information is not adequate in many densely populated lands to make informed policy decisions, especially in the Global South. A critical need is better quantification of fine particulate matter (PM2.5) composition, which can in turn be used to identify crucial pollutant sources. A major barrier is the high capital and operational costs of research-grade monitors. This study investigates a cost-effective way to leverage existing monitors to expand the limited air quality dataset in these areas. The US Department of State collects air pollutant data at US Embassies around the world to inform US personnel and citizens of air quality overseas. These measurements use Beta Attenuation Monitors (BAMs) to measure hourly ambient PM2.5concentrations. BAMs collect PM2.5 onto a filter tape and estimate particle concentrations by measuring attenuation of beta rays using the Beer-Lambert law. We utilize these BAM tapes to determine hourly ambient black carbon (BC) concentration. Each filter tape spot is individually photographed with a cell phone camera on a reference card containing a set of predefined grayscales corresponding to different BC concentrations. The image is then processed through a custom computer-vision Python script, which performs spatial correction, adjusts for lighting variations, locates the filter spot and then compares the reflectance of red light from the spot with that of the grayscales on the reference card to estimate BC concentration. The method was calibrated using filters with known BC concentrations determined from co-located aethalometer measurements and offline thermal-optical EC analysis. The limit of detection is approximately 1.1 µg.m-3for 1-hr samples. This presentation will show results from analysis of BAM tapes collected at three locations: a near-roadway site in Pittsburgh, PA; the US Embassies in Addis Ababa, Ethiopia and Abidjan, Côte d’Ivoire.

Additional Authors: Abhishek Anand, Albert Presto, Suryaprakash Kompalli, Eniola Ajiboye

Evaluation of Correction Models for a Low-Cost Fine Particulate Matter Sensor Using the Canadian AQHI+ System

Brayden Nilson, University of Northern British Columbia

We evaluated four colocation-based correction models for low-cost fine particulate matter (PM2.5) sensors from PurpleAir (PA) along with four other models from the literature. The models we developed were trained using 32 colocations sites across Canada and the United States, selected based on the proximity of PA sensors to regulatory Federal Equivalent Method (FEM) monitors. An additional 15 colocation sites were used for evaluating the performance of the models. An automated quality control process was implemented to ensure the best available data were used. Comparative evaluations were made between these correction models using the Air Quality Health Index+ (AQHI+) system, a modification of the Canadian AQHI that only considers PM2.5 concentrations. Using the AQHI+ system allowed for a direct comparison of performance across the range of observed concentrations. Emphasis was placed on improved performance in the moderate and high concentration ranges, which are the most important for health-related management decisions in North America. We provide a robust framework for the evaluation of low-cost PM2.5 sensor correction models and present an optimized correction model for North American PA sensors. We found that one of the models we developed, utilizing a hygroscopicity-based correction, performed consistently better than the other models we tested.

Additional Authors: Brayden Nilson, Peter Jackson, Corinne Schiller, Matthew Parsons

Partnerships in low-cost air quality monitoring and outreach in North Carolina

Brian Magi, UNC Charlotte (Virtual Presentation)

Low-cost air pollution sensors have emerged as a viable way to explore air quality at much finer spatial and temporal scales than data that is readily available from the existing regulatory air pollution monitoring network. Low-cost sensor data collection offers pathways to citizen science and community-based participatory research because the cost is low and the data is open-access. There are several layers of analysis that necessitate a co-researcher design to community-led inquiries, and in this presentation, I will discuss some of the successes and the roadblocks and challenges associated with low-cost air monitoring data that have happened and continue to happen in North Carolina. Successes range from peer-reviewed publications to story maps, public comment, and even motivating direct changes in regulatory monitoring. Importantly, the successes are entirely connected to engaged multi-layered partnerships among academic and government air quality and health researchers, stakeholder and community groups, and regulatory agencies. I will talk about this massive intersection, and how I use my expertise in air quality science to help facilitate long-term relationships alongside my contributions to technical data analysis. I will argue that the framework of analysis and relationship-building is broadly applicable beyond North Carolina, and that fundamentally, community climate resilience benefits from multi-layered partnerships.

Additional Authors: Brian Magi

Towards a Global Low-Cost Sensor Calibration Model via Gaussian Mixture Regression

Dan Westervelt, Columbia University

Low-cost Sensors (LCSs) for air quality monitoring have great potential to address the gap in air quality data across the globe. LCSs’ sensitivity to environment and source conditions mean they require the development of local calibration models built by collocation with reference grade monitors. Establishing sensor collocations and building local calibration models, however, is an expensive feat. Calibration methods like multiple linear regression (MLR) have been shown to effectively calibrate locally, but have failed to be applied across regions with differing climate and source conditions. Recent work in Accra, Ghana, a city with distinct climate regimes, has shown Gaussian Mixture regression (GMR) to be more successful than MLR when calibrating LCS data with heterogeneous characteristics (MLR R2= 0.81 & MAE = 2.8 µg m-3; GMR R2= 0.88 & MAE = 2.2 µg m-3).GMR is both a clustering and regression method that works by modeling the probability density of the output data conditional to the input data as a Gaussian Mixture Model (GMM).In addition to improving correlation and accuracy, GMR is shown to capture complex, physical relationships in data. GMM cluster assignments were shown to match underlying climate characteristics and seasonal trends, suggesting that a global GMM could have components for different pollutant mixes and climates and thus be applicable to multiple cities or regions. Building a GMM with data from just over 5 cities, and testing application on wide-spread global collocations via the CAMS-Net project, we have built a singular LCS correction model that can be effectively applied to any city globally. We validate the method by comparing our global correction factor to a local correction factor in several test cities. This global correction factor methodology could dramatically increase the utility of LCS data worldwide and could be incorporated into online open data repositories.

Additional Authors: Celeste McFarlane, Garima Raheja, Carl Malings, The CAMS-Net Team, Daniel Westervelt

Maximizing insights from air quality sensor networks through continuous performance evaluation

Daniel Peters, Environmental Defense Fund

As the global use of lower-cost air quality sensors (LCS) continues to grow, data quality issues including poor accuracy, drift, and environmental interference remain a critical obstacle to a user’s ability to obtain meaningful information from sensor measurements. Even if a sensor has already been tested and calibrated in a lab or field environment, a sensor’s performance and calibration parameters may vary by region (or season) depending on numerous factors including the local characteristics of air pollution and meteorology. Due to the uncertainty and variability in LCS data quality, with no universal performance standards or QA/QC protocol, the end user is responsible for ensuring the data they obtain is robust and of appropriate quality for the intended application. Here, we present results from a network of 100 electrochemical NO2 sensors as part of the Breathe London pilot project (BL) and describe the steps that were taken to develop a detailed understanding of sensor performance over the entire duration of the 26-month measurement campaign. We discuss how this project-long understanding of sensor performance informed our interpretation of the data produced by the BL network. We validate our findings by comparing BL network results to data from an extensive network of London reference monitors, including comparisons of temporal patterns (e.g., diurnal, day-of-week, and monthly averages) as well as differences in pollution profiles between near-road and urban background monitoring sites. Through these comparisons, we show how long-term collocation intercomparisons of BL sensors at two reference sites could be used as a proxy to understand overall network performance and develop corrections. We discuss how the findings could be applied to other LCS projects, including those with less reference monitoring infrastructure, provided that some source of traceable, reliable measurements is available to perform ongoing representative evaluations of sensors.

Additional Authors: Daniel Peters, Lauren Padilla, Ramón Alvarez

Closing the air pollution data gap in sub-Saharan Africa through low cost sensors, capacity building, international networking, and data science methods

Daniel Westervelt, Lamont-Doherty Earth Observatory of Columbia University

In sub-Saharan Africa, sparse air pollution monitoring imparts high uncertainty to estimates of exposure and impact.To address this problem, we have launched several internationally-supported communities of practice that spans multiple languages and cultures to unites scientists, decision-makers, city administrators, citizen groups, the private sector, and other air quality advocates and practitioners in developing new methods and best practices to hasten synthesis and implementation of clean air solutions. A cornerstone of our method includes novel collaboration accelerators such as networks of networks and communities of practice. To this end, we have launched the Clean Air Monitoring and Solutions Network (CAMS-Net), which aims to establish an international network of networks around the idea of getting useful, actionable data out of low-cost sensors to help spur air pollution mitigation strategies. We have also established two large communities of practice for air quality research and management in East and West Africa. A certificate program was held for around 100 attendees from East Africa in July 2021. Lessons learned from this and similar efforts will be presented in this talk. I will present our team's efforts at closing the air pollution data gap in sub-Saharan Africa through reference monitoring, low-cost sensors, and data science techniques. Results will include some of the first field calibrations of low-cost sensors (LCS) for PM2.5 in sub-Saharan Africa. In each city, we show that raw LCS are highly correlated with reference monitors (r2>0.8) but can be significantly biased (MAE ~ 14 µg m-3). We show that colocation with reference monitors and simple data science methods (such as random forest, linear regression, or Gaussian Mixture Regression) can reduce MAE to around 3 µg m-3 for a year of collocated data. We will also discuss emerging new high-density networks for PM2.5 in several other cities across sub-Saharan Africa.

Additional Authors: Daniel Westervelt, The CAMS-Net Team

Air quality monitoring with low-cost sensors in Pioneer Valley of Western Massachusetts: strategies for sensor deployment and calibration

Dong Gao, Yale University

Air pollution continues to be a global public health threat. With the development of sensor and internet technologies, low-cost air sensors (LCS) have emerged as an effective tool for air quality monitoring at a high time and spatial resolution. The use of LCS is of main interest for citizen science initiatives and can help answer community-driven and locally motivated questions. A collaborative air quality monitoring project has recently been conducted in Pioneer Valley of Western Massachusetts, focusing on the Springfield, Chicopee and Holyoke areas. An LCS network with 60 sensors was built up, with the goal of measuring air pollution levels and providing data to inform public health responses in the city known as having the highest rates of asthma in the country. To determine the suitable locations for sensor deployment, a weighted site selection analysis was performed by assigning varying importance levels to different factors such as traffic density, proximity to emission hotspots, and distribution of at-risk vulnerable populations. Pre- and post-deployment calibration were conducted for the deployed sensors. For pre-deployment calibration purpose, all sensors were collocated at a regulatory station for weeks. The sensor data was compared to reference methods, and calibration factors were derived for each sensor. Inter-comparison of sensor data showed good correlation between sensors (R2=0.98~0.99). Different field calibration methods (e.g., multiple linear regression, land use regression, and machine learning methods) were examined for post-deployment calibration. This study developed a formal approach for optimally locating a dense LCS network of air pollution monitoring and derived an area-specific calibration model for Pioneer Valley.

Additional Authors: Dong Gao, Jiarong Qi, Mahea Heimuli, Kayla Fennell, Anna Woodroof, Mark Chandler, David Bloniarz, Alexander Sherman, Samantha Hamilton, Yoni Glogower, Sarita Hudson, Krystal Pollitt


Driejana Driejana, Institut Teknologi Bandung

Low-cost sensor (LCS) PurpleAir PA IIfield validation was done in Jakarta, Indonesia. The study was part of Community Air Quality Monitoring (CAQM) [https://kopernik.info/id/berita-acara/news/kopernik-partners-with-islamic-development-bank-to-deliver-community-air-quality-monitoring-project],a collaboration of Kopernik, PulseLab Jakarta [https://medium.com/pulse-lab-jakarta/designing-a-community-based-air-quality-monitoring-system-to-mitigate-air-pollution-health-risks-3e527e607427] and Institut Teknologi Bandung-UDARA project. CAQM was a pilot project designed for mitigation of air pollution health risks at community level. PM10 and PM2.5concentrations were converted into Indonesian Air Quality Index (ISPU) and published in pantauudara.org [pantauudara.org]. LCSs were validated by co-location with PM10 and PM2.5BAM F-701-20FEM analyzers of DKI Jakarta Environment Agency (DLH) monitoring network. Three PA II units were deployed for 52 days from July to August 2020 at Kelapa Gading, North Jakarta. Each PA II unit has a pair of PMS5003 sensors, so that a total of six sensors were used for precision and accuracy measurements. After validation, two units were moved to Galur, (Jakarta) and Rasau Jaya (West Kalimantan) and one unit was left at the reference site until November 2020.The accuracy of the measurement was measured by comparison of 24-hr average PM10 and PM2.5with the data from BAM analyzers.The inter-model comparison showed good overall precision with average of 92 – 93% for PM1.0, PM2.5and PM10. LCS showed relatively good performance for PM10(r2=60%), butlow accuracy(r2<15%) for PM2.5. The performance was relatively similar after 127 days. The average proportion of PM2.5 in PM10 was 86.3% and PM1.0in PM2.5 was 63.3% suggesting that the majority of PM10 mass concentration was fine particles. Due to LCS PM2.5 large discrepancy to the reference method, and the fact that the large proportion of PM10 was PM2.5; ISPU was published only based on PM10.

Additional Authors: Driejana Driejana, Ahmad Daudsyah Imami, Kirana Nadhila

Field-calibrated PM2.5 Measurements, Regional Trend Assessments, and Sensor Intercomparison Results from Low-Cost Monitoring Networks in Accra, Ghana and Lomé, Togo

Garima Raheja, Columbia University

Metropolises in sub-Saharan Africa experience high levels of ambient air pollution, yet remain scarcely measured by reference-grade monitors. Combining insights from three years of data collected by newly established and field-calibrated Purple Air low-cost sensor networks, we present the first measurements of PM2.5concentrations in Lomé, Togo. For Accra, Ghana, we utilize an 18-node network of low-cost Clarity low-cost sensors to analyze regional trends, and assess the reductions in PM2.5caused by intentional interventions as well as the COVID-19 pandemic. For both low-cost sensor networks, we apply simple methods such as multiple linear regression, as well as novel more complex methods such as Gaussian mixture regression to develop correction models. We find that the annual average PM2.5in Lomé is 21.1 µg m-3and is heavily influenced by the Harmattan. The network-wide annual mean PM2.5in Accra is 26.3 µg m-3, 5.3 times higher than WHO daily guidelines and 19% higher than neighboring Lomé. Both the PurpleAir network in Lomé and the Clarity network in Accra are validated using results of co-located sensors: 18 Clarity Nodes, 2 PurpleAir, 2 Modulair, 1 Teledyne T640, and 1 Met-One Beta Attenuation Monitor at FEM instrumentation sites in Accra, Ghana. We find that Clarity devices have a mean absolute error (MAE) and R2of 3.36 µg m-3and 0.76 when compared with Teledyne (FEM) PM2.5; Purple Air devices have a MAE and R2of 2.6 µg m-3and 0.88, and Modulair devices have a MAE and R2of 1.66 µg m-3and 0.89. This sensor intercomparison study is contributing to the development of a universal low-cost sensor data correction algorithm. Using these results, we show that low-cost sensors, when combined with data science-based correction techniques, have the potential to close the air pollution data gap in resource-limited areas such as West Africa.

Additional Authors: Garima Raheja, Emmanuel Appoh, Ebenezer Appah-Sampong, Maxwell Sunu, John Nyante, Allison Hughes, Celeste McFarlane, Rob Pinder, Stefani Penn, R Subramanian, Mike Giordano, Levi Stanton, Daniel Westervelt

Public Engagement in Air Quality Management in Kenya

Godwin Opinde, Kenyatta University

Air quality management is a collective responsibility and public engagement is integral to effective management of air. The nature of public engagement influences compliance and enforcement of this aspect of the environment. Policies, laws, plans and established institutions play a significant role in ensuring its effective governance. In addition, the involvement of the public in monitoring air quality entrenches awareness creation that significantly influences the attitude and behaviour of the public on the management of air quality. This paper examines public engagement in processes and efforts geared towards ensuring clean air for improved health and mitigation of climate change in Nairobi, Kenya. It employs reviews and targeted interviews to assess public engagement in the formulation of air quality laws, policies and plans and examine the nature of public involvement in air quality monitoring. The paper identifies challenges and opportunities in the engagement process and recommends strategies for meaningful public engagement that enhances compliance and enforcement in air quality governance.

Additional Authors: Godwin Opinde

Observation of aerosol spatio-temporal variations over Ghana using MODIS-derived Aerosol Optical Depth

James Nimo, University of Ghana

Ambient particulate matter pollution in sub-Saharan Africa cities has become a major public health concern. There is limited data to assess the increasing trends in the sub-region due to the lack of monitoring stations. Remote sensing data from satellites includes aerosol optical depth (AOD) is accessible for assessment of air pollution at increasingly fine resolutions. In this study AOD data derived from the MODIS Aqua and Terra satellites were used to assess the levels of aerosols over Ghana.The results were validated with a NASA AERONET station data in the eastern region of Ghana. AOD at 550 nm was retrieved from MODIS 10 km Collection 6.1 level 2 dataset using the combined dark target and deep blue algorithm from 2013 to 2018. HYSPLIT model was used to analyse the origins of air masses and their trajectories over the 10 regions in Ghana during the peak of the dry season. The study observed an increasing trend of AOD over the northern and southern parts with highest mean 0.56 ± 0.02and lowest mean 0.15 ± 0.2over Ghana during the dry and wet seasons. About 49.05% of AODs retrieved on 1x1 kmof MODIS fall within the sensor error envelope of paired validation data against the AERONET. The Aqua and Terra sensors had slopes of 0.45 (with R = 0.78, RMSE = 0.34 and MAE = 0.29) and 0.21 (with R = 0.33, RMSE = 0.56 and MAE = 0.42). Strong outbreak of dust experienced over Ghana from 500 to 1500 m originated from continental sources in the Saharan desert, Mediterranean Sea and other biomass burning from within Ghana and neighbouring countries. Consistent cloud cover using the deep blue and dark target algorithm was observed over the southern part of Ghana.The resulted slopes indicate that MODIS algorithm underestimated the AODs.

Additional Authors: James Nimo, Allison Hughes, Nana Ama Browne-Klutse, Azoda Koffi, Abraham Amankwah, Alabi Omowumi


Josephine Kanyeria, Jomo Kenyatta University of Agriculture and Technology (Virtual Presentation)

Air pollution is a major environmental concern that affects human health worldwide. Despite recent studies indicating ambient air pollution is a growing global concern strongly linked to rapid global urbanization, little has been done to monitor the air quality levels in towns outside Nairobi, Kenya. Juja is one of the largest growing towns subjected to increased population, intense human activities and located along the busy Thika Super Highway. Thus, there is a strong need to monitor ambient Particulate Matter within the town. The purpose of this study was to assess the diurnal and seasonal variations of Ambient Particulate Matter (PM2.5) in Juja, Kenya. The data was collected November 2019 to April 2021 at JKUAT Institute of Energy and Environmental Technology (IEET) department, a residential house within Kalimoni and an additional site along the busy Thika Superhighway. The PM level was measured using the Purple Air Monitoring Sensor – PA-II-SD in μg/m3on a 24hour cycle. Data analysis was done quantitatively using Excel, SPSS and R programming. The PM2.5 level from the low-cost Purple Air Sensors were later calibrated against a reference BAM-1022 to yield corrected PM values. The results revealed that the overall PM2.5 concentration was higher during the dry season (July - August 2020) compared to March - May 2020 (wet season) where it dropped by 5-10μg/m3 on average. The highest daily PM2.5 concentration was recorded at 44μg/m3(JKUAT) and 43μg/m3 (Residential) both exceeding the WHO guidelines and the USEPA National Air Quality Standards. JKUAT had an annual mean concentration of 15μg/m3, also exceeding the WHO guidelines of 10μg/m3. In addition, comparing the month of April 2021 from the previous year, the daily mean dropped by 5-10μg/m3– the period of the new Covid -19 lockdown. Data analysis is currently ongoing and will examine the data from other additional sites.

Additional Authors: Josephine Kanyeria, Paul Njogu, Daniel Westervelt

Air quality in Togo: Monitoring status and CAMS-Net opportunities

Kokou SABI, Université de Lomé

In 2016, ambient air pollution caused more than 4.2 million premature deaths, including 300,000 deaths of children under five. The largest share of deaths among children is due to fine particulate matter (PM2.5) pollution, especially in low- and middle-income countries (WHO, 2016). As a result, air pollution has become a health emergency. PM2.5 has a greater impact on human health compared to larger particles due to its smaller size which facilitates their penetration into the human body. WHO recognizes several sources of PM2.5 in the air: combustion of biomass, road traffic, industrial and waste management sources, etc. Togo likely experiences high levels of air pollution but the exact quantities of atmospheric pollution are not well known. More specifically, the largest Togolese city of Lomé experienced up to 7.24 tonnes of PM emission in 2019 with an estimated rate of increase of about 0.82 tones per year, according to studies carried out at the Atmospheric Chemistry Laboratory (LCA). To deal with this challenge, LCA has joined forces with the CAMS-Net to find solutions. In this initiation phase, two students are enrolled in a thesis on the themes including (i) Atmospheric measurements of PM2.5 using low-cost sensors calibrated in the ambient air of the city of Lomé in Togo; and (ii) air quality modeling, characterization, and distribution of pollutant emission sources in West Africa. LCA has already deployed 5 low-cost sensors (PurpleAir) fixed in the city of Lomé for 2 years. LCA will continue with CAMS-Net's assistance to obtain equipment and technical capacity building to develop new methods and best practices for the collection and analysis of real data on air quality in the city of Lomé and eventually other parts of Togo. CAMS-Net project meeting is an opportunity for knowledge exchange, during which this initial knowledge of the state of pollution in Togo will be shared and ideas will be presented to boost research on air quality in the near future.

Additional Authors: Kokou SABI, Westervelt Dan

Evaluation of a reduced-complexity model against low-cost sensors in India and the United States

Medinat Akindele, Carnegie Mellon

It is no secret that ambient air pollution has more adverse effects in developing countries with high ambient PM2.5 concentrations. Models have an essential role in managing air pollution by estimating PM2.5 concentrations. However, the state-of-the-art air quality models, chemical transport models (CTMs), are not readily available in those regions. They are computationally expensive and restricted to regulatory agencies and specific organizations. These limitations prompted the development of reduced-complexity models (RCMs) to increase the accessibility of air quality models. RCMs have a simplified representation of atmospheric processes, but they are faster, easier to use, and have broader policy applications. Most RCMs are built for the US and commonly evaluated against CTMs and measurements from regulatory grade monitors. That presents a critical need for a more general and globally applicable RCM. Low-cost sensors have been deployed in several countries and have higher spatial coverage than traditional monitoring networks. In this work, we present an RCM that is location agnostic by applying region-specific emissions and meteorology. This RCM borrows a framework from the APEEP Gaussian plume model with chemistry. The model predicts annual-average PM2.5 concentrations from primary PM2.5, SO2, NOx, NH3, and VOC emissions. The PM2.5 predictions are evaluated against low-cost sensor measurements in India and the United States. Future work will extend this model to developing countries in Sub-Saharan Africa and others in South Asia. RCMs and low-cost sensors can democratize air quality resources and reduce barriers in regulating air pollution.

Additional Authors: Medinat Akindele, Peter Adams, Nicholas Muller

An evaluation of particulate matter (PM2.5) in the City of Nairobi, Kenya, using nephelometers

Otienoh Oguge, Center for Advanced Studies in Environmental Law & Policy (CASELAP), University of Nairobi

Particulate matter (PM) is a pollutant of high public interest globally but relatively little data exist for a time series comparison with reference data for the City of Nairobi. Given that PM is present in the atmosphere in concentrations that can vary greatly according to location, we set 10 nephelometers (E-sampler 9800) in eight planning zones of Nairobi and collocated one with a reference monitor (BAM-1022). Here we compare the data-collection capabilities of the low-cost sensor to that of collocated BAM. We obtained real-time hourly data entailing PM2.5 concentrations averages from the BAM and the collocated E-sampler for 138 days providing 3,312 valid data pairs. The E-sampler correlated well with the BAM with coefficient of determination (R2) of 0.64 over the four months. However, this was variable with the E-sampler indicating a higher PM load than the BAM during the first four weeks of September 2019 (R2=0.32, n=888) and in late November through early January 2020 (R2=0.34, n=1128), despite similarity in the vertical flux. Variation in PM2.5 between end of September and late November 2019 showed a stronger association (R2=0.76, n=1296) between the two sensors. PM2.5 pollution was found to vary in different parts of the city being highest 38.01 ug/m3 around the central business district, and lowest 12.72 ug/m3 in the low-density southern suburbs of the city. RH was the most significant atmospheric factor to affect PM2.5 values (R2=-0.32, df=137, p<0.001). An order of magnitude rises in the relative humidity (RH) led to a decrease in the level of PM2.5concentration by a factor of 0.14 ug/m3. Our findings show that low-cost sensors can improve the spatial and temporal acquisition of air quality data following calibration and standardization. It also supports existing knowledge that has demonstrated the need for observing accurate values for atmospheric factors, particularly RH in obtaining valid and reliable data from low-cost air quality monitors.

Additional Authors: Otienoh Oguge, Joshua Nyamondo, Noah Adera, Lydia Okolla, Beldine Okoth, Stephen Anyango, Augustine Afullo

Monitoring tropospheric airborne particles along a section of the busiest road in East and Central Africa (Thika road, Kenya) using low-cost monitors

Paul Njogu, Jomo Kenyatta University of Agriculture and Technology (Virtual Presentation)

Air pollution monitoring in Kenya is poorly developed due to high cost of monitoring equipment and low emphasis by the Government. Vehicular emissions are one of the greatest contributors to gaseous and particles to the troposphere. Kenyan roads are filled with vehicles mainly more than eight years after manufacture, such vehicles do not meet emission standards. Public health impacts of air pollutants are well known, this paper presents findings of air quality monitoring along one of Africa’s busiest roads, Thika Superhighway in Kenya. The road passes through residential areas with various institutions of learning. Its thus important to monitor the air quality to cushion the health of the public. Hourly data was collected from two locations along the busy road connecting Nairobi city and Thika town using low-cost Purple Air Sensors during the month of October 2021. The road has a daily traffic load of close to 150,000 vehicles. The data was corrected using colocation data from a BAM reference monitor mounted in a similar location. The levels of PM10 ranged from 5.2 – 72.5 µg/m3 whereas the mean hourly concentration was 22.4 ± 10.2 µg/m3. All recorded 24 hourly averages were within the 50 µg/m3 limit set by the WHO and the Government of Kenya. Wind speeds ranged from 3 - 12 m/s, it was observed that as winds speed increased, the PM10 levels decreased. Daily PM10 variations show two peaks, one at 0800 – 0400 hrs and 1500 – 1900 hours with a minima occurring at 0900 – 1200 hours, there were however wide temporal variations. The trends follow the traffic volumes and weather patterns. The use of LCM presents a cheap method that can be used by African government and local authorities to monitor air quality trends to enable informed appropriate actions, pollution mitigation and policy development.

Additional Authors: Paul Njogu, Daniel Westervelt, JOSEPHINE KANYERIA

Evaluation of lower-cost air quality monitors for monitoring ambient air pollution and around athletic stadiums in Qatar

R Subramanian, Qatar Environment & Energy Research Institute

Air pollution is one of the largest environmental health risks, with an estimated toll of almost 7 million deaths in 2019 (HEI State of Global Air). Fine particulate matter mass (PM2.5) is a key contributing pollutant; in dusty and rapidly growing middle-eastern cities, nitrogen dioxide (NO2) and coarse PM (PM10) are also important. Traditional "reference-grade" monitors are expensive and resource-intensive, so monitoring networks are usually sparse; QEERI operates a network of six such stations across Greater Doha. Micro-environments such as high traffic, industrial activity, and construction can result in hyperlocal pollution not captured by these reference stations. Lower-cost air quality monitors using sensors can fill this data gap, enabling high spatial and temporal resolution measurements of air pollution that can also be used to develop and improve air quality models. However, low-cost optical PM2.5/PM10 sensors, whether nephelometers or optical particle counters, need to be evaluated for conditions unique to Qatar and the middle-east generally. We are conducting extensive field testing of a variety of lower-cost devices for PM and NO2 measurements across Doha. We find that sensor design can limit product performance especially for super-micron aerosol. Changes in aerosol size distribution (e.g. during dust storms) result in significant underestimation of PM2.5 mass by nephelometric PM sensors. OPC-based sensors perform much better but they also need to be properly characterized. As seen elsewhere, aerosol hygroscopic growth can also impact sensor performance. For NO2, localized calibration models using machine learning significantly improve over as-reported sensor performance even if that uses ML models but developed elsewhere/in factory lab settings. We shall discuss these results and optimal sensing strategies for Qatar especially targeting large athletic activities.

Additional Authors: R Subramanian, Nicolas Barth, Shamjad Moosakutty, Rami Alfarra, Mohammed Ayoub

Contrasting Pattern of PM2.5 Concentrations in Urban-Rural Pair Sensors from Nepal

Rejina Maskey Byanju, Central Department of Environmental Science, Tribhuvan University

Nearly half of Nepal’s population lives in a lowland region known as the Terai, immediately north of the border with India. Air quality management in the Terai is encumbered by a lack of ambient measurements and significant uncertainty in the local vs. transboundary pollutant contributions. The Government of Nepal is expanding its air pollution monitoring network to the Terai but faces technological limitations to keep their expensive equipment calibrated and fully operational. Low-cost sensors can provide great assistance to the management of Terai air quality and quantification of transboundary pollution. This research aims to assess PM2.5 concentrations in the Terai and transboundary contributions using TSI BlueSky sensors. Our network of sampling locations across the Terai includes urban centers that are each paired with a nearby rural site. After a lengthy intercomparison, the first pair of sensors was deployed to urban and rural Birgunj in October 2021 and will run for the next 2 years. Preliminary results show daily average concentrations in urban Birgunj are only slightly higher (87.9 μg/m3) than the rural site 15 km away (84.6 μg/m3) implying a very large transboundary contribution. However, the rural measurements contain sporadic peaks due to frequent agricultural residue burning and other biomass burning that are not found in the urban time series. Additional pairs of BlueSky sensors will be deployed to the Terai over the coming months and analytical techniques employed to characterize the PM2.5 concentrations and source contributions in this populous region. Keywords: Agricultural residue, Biomass burning, Urban-rural pair, TSI Blue Sky Sensor

Additional Authors: Rejina Maskey Byanju, Ramesh Prashad Shapkota, Hasana Shrestha, Enna Mool, Prakash V Bhave, Michael H. Bergin

Insights into Urban CO2 Emissions from BEACO2N

Ronald C. Cohen, UC Berkeley

There is significant political and scientific interest in the changing landscape of urban carbon dioxide (CO2) emissions. The Bay area Environmental Air-quality & CO2Network (BEACO2N) uses low-cost sensors to continuously measure CO2and key air pollutants at high spatial resolution throughout an urban area. About 50 nodes are located at 2km spacing in the San Francisco Bay Area. We are partnering with USC in Los Angeles, and the University of Strathclyde in Glasgow, Scotland and the University of Leicester, UK. Integration of the BEACO2N dataset with other data sources (transport models, satellite data, and traffic datasets) provides insights into spatial, temporal, and sectoral changes in urban CO2emissions in these cities. Here we discuss several recent developments out of the BEACO2N project. Bayesian inverse modeling of the BEACO2N CO2measurements using a particle dispersion model and satellite-derived photosynthesis estimates provides an updated 1x1km map of emissions. Coupling the BEACO2N derived emission inventory with traffic datasets gives insight into fuel efficiency trends and the accuracy of models of the speed-dependence of fleetwide vehicle emission factors. These methods show the diversity of applications of dense ground-based measurements for improved understanding of urban emissions.

Additional Authors: Ronald C. Cohen, Helen Fitzmaurice, Jinsol Kim, Alex J. Turner, Naomi Asimow, Yishu Zhu, Catherine Newman, Paul J. Wooldridge

Air quality monitoring with TSI BlueSky sensors in the megacity Dhaka, Bangladesh

Shahid Uz Zaman, University of Dhaka

The capital city of Dhaka, Bangladesh has a population of ~21 million and is growing at an astounding rate of 4.2% annually. Accompanying the recent growth is increasingly poor air quality, although the region lacks adequate monitoring necessary to assess health impacts, and the potential success of future mitigation strategies. With this in mind, we will discuss a US Department of State (DOS) funded project that involves developing a PM2.5 sensor network in collaboration with local stakeholders aimed at improving the air quality in the region. The project involves careful placement of ~ 70 sensors across Dhaka at key locations that will be used to determine the impact of sources and source regions on PM2.5 concentrations. We will discuss site selection, sensor calibration and maintenance, as well as PM2.5 concentrations measured at locations across Dhaka and discuss the contributions of local versus regional sources on PM2.5 levels across the city.

Additional Authors: Shahid Uz Zaman, Khaled Shaifullah Joy, Shatabdi Roy, Md. Asif Iqbal Nayeem, Rabiul Islam, Farah Jeba, Prakash Bhave, Michael H. Bergin, Ross Edwards, James J. Schauer, Robert Caldow, Abdus Salam

Field calibration and performance evaluation of low-cost sensors

Sinan Yatkin, Joint Research Centre (Virtual Presentation)

The calibration of eighty-five AirSensEUR sensor systems including CO, CO2, NO, NO2 and O3 gas sensors, and two models of PM sensors were performed by 2-weeks co-location at Air Quality Monitoring Stations (AQMS) in four European cities, namely Ispra (IT), Antwerp (BE), Oslo (NO) and Zagreb (HR). Calibration was followed by the evaluation of sensor performance for several months of co-location at the same AQMS as calibration, at different AQMS and seasons. Sensor calibration models were established by an automatic process using statistical tools and laboratory experiments to determine the variables needed in the models. Measurement uncertainty, linear regression of sensor predicted data versus reference data and root-mean-squared error were utilized to evaluate the sensor performance. The calibration models for CO, NO, O3 and PM sensors used few and repeatable variables. The coefficients of calibration models were found site-dependent showing that calibration should be carried out locally rather than globally for the best performance. Nevertheless, some sensors calibrated at one AQMS predicted reasonably well at other AQMS and different seasons, meaning that calibration models derived at one AQMS/season can be applied to multiple sites/seasons. Overall, CO, O3, PM2.5 and PM1 calibration models were successful in prediction of pollutants while the performances of NO and NO2 sensors were highly dependent on similarities of pollutants, temperature and humidity conditions between prediction and calibration periods

Additional Authors: Sinan Yatkin, Michel Gerboles, Annette Borowiak, Silvije Davila, Alena Bartolova, Frank Dauge, Philipp Schneider, Martine Van Poppel, Jan Peters, Christina Matheeussen, Marco Signorini

Determination of local traffic emission and non-local background source contribution to on-road air pollution using fixed-route mobile air sensor network

Zhi Ning, Hong Kong University of Science and Technology

Traffic-related air pollutants (TRAP) are major contributors to deteriorating urban air quality and pose a serious threat to pedestrians. From both a scientific and a regulatory standpoint, it is important and challenging to understand the contributions of local and non-local sources to accurately apportion specific sources such as traffic emissions contribution to on-road and near-road microenvironment air quality. In this study, we deployed mobile sensors on-board buses to monitor NO, NO2, CO and PM2.5 along ten important routes in Hong Kong. The measurements include two seasons: April 2017 and July 2017. Two types of baseline extraction methods were evaluated and applied to separate local and background concentrations. The results show NO and NO2are locally dominated air pollutants in spring, constituting 72%-84% and 58%-71%, respectively, with large inter-road variation. PM2.5 and CO largely arise from background sources, which contribute 55%-65% and 73%-79% respectively. PM2.5 displays a homogeneous spatial pattern, and the contributions show seasonal change, decreasing during summer. Regional transport pollution is the primary contributor during high pollution episodes. Isolated vehicle plumes show highly skewed concentration distributions. There are characteristic polluted segments on routes and they are most evident at rush hours. The most polluted road segments (top 10%) cluster at tunnel entrances and congested points. Some of these polluted locations were observed in Hong Kong’s Low Emission Zones and suggest limitations to the existing control strategies, which only address larger buses. Our work gives new insights in the importance of regional cooperation to improve background air pollution combined with local control strategies to improve roadside air quality in Hong Kong.

Additional Authors: Peng Wei, Zhi Ning

Assessment of NO2 and PM2.5 Variabilities in Nairobi and Evaluation of Low-Cost Sensor Performance in Long-Term Deployments

Ezekiel W. Nyaga, Universitè de Paris (Virtual Presentation)

The rapid increase in population, economic activities, and motorization have led to the deterioration of air quality in Nairobi. Exposure to particulate matter (PM2.5) is a leading risk factor responsible for 4 to 9 million deaths in the whole world each year, the majority occurring in low and middle-income countries where air quality measurements remain scarce (WHO). The existing knowledge of air quality in Nairobi is based on short-term campaigns lasting from weeks to a few months. Here, we present preliminary findings of sensor deployments and collocations with reference monitors spanning more than twenty-three months in selected sites across Nairobi. The NO2 sensor performance and interannual variabilities of PM2.5 and NO2 as reported by the Clarity nodes were investigated, though an important caveat is that the actual values and variabilities may change once corrections are applied. Preliminary analysis of uncorrected sensor data suggests that the NO2 concentrations are 43.21 µg/𝑚3 and 41.57 µg/𝑚3 at Kenyatta University (KU) and the University of Nairobi (UoN) urban sites and 30.14 µg/𝑚3 at the Innovations for Poverty Action (IPA) suburban site. The corrected PM2.5 concentrations at KU, Buru Buru (suburban), Marurui (suburban), UoN, and IPA sites are 35.87 µg/𝑚3, 35.28 µg/𝑚3, 33.15 µg/𝑚3, 28.00 µg/𝑚3 and 26.81 µg/𝑚3 respectively. The diurnal pattern of PM2.5 and NO2 follows that of the traffic cycle, therefore suggesting a strong influence of vehicle emissions on the concentrations of pollutants across the sites. The inter-comparison of one-year measurements made by all sensors at the IPA site show different rates of decline in NO2 sensors performance during the collocation period. The next steps (which will be presented at ASIC) involve development of PM2.5 and NO2 calibration models and the evaluation of performance and reliability of low-cost sensors for long-term measurements in places with minimal to no coverage by regular air monitoring networks.

Additional Authors: Ezekiel W. Nyaga, Mike Giordano, Matthias Beekmann, Savannah Ward, Daniel Westervelt, Michael James Gatari, John Mungai, Godwin Opinde, Tedy Mwendwa, Paulina Jaramillo, Albert Presto, Emilia Tjernstrom, V. Faye McNeill, R. Subramanian

Communication Strategies for Understanding, Insight, and Action

Love My Air Network- A national collaboration on messaging and education

Aubrey Burgess, City and County of Denver (Invited Talk)

Love My Air Network is an idea-sharing network with the goal of replicating Denver’s Love My Air program across the United States. The project was founded in 2018 by the City and County of Denver and started in Denver Public Schools with the goal of reducing air pollution and limiting exposure through behavior change, advocacy, and community involvement. This presentation will demonstrate how this program has grown into a flexible initiative where participating entities aim to build/ grow their air quality sensor networks and pair data with community action. Local governments and universities across the nation have begun work to adopt aspects of Denver’s framework to develop a culturally- appropriate program in their own neighborhoods. Participating entities can contribute to a repository of living documents, open-source data, and collaborative conversations. We hope to increase air quality literacy and involve the community in air quality monitoring, education, and decision-making.

Additional Authors: Aubrey Burgess

Experiences and Lessons Learned with Community Monitoring Near a Refinery

Patrick Clark, Montrose Air Quality Services, LLC (Invited Talk)

In 2021, Suncor Energy (U.S.A.) Inc. (Suncor) developed CCND Air Monitoring, a community air monitoring program in the neighborhoods surrounding its refinery in Commerce City, Colorado. The purpose of the voluntary program is to provide the community with easy-to-access air monitoring information from sensors reporting in near real-time, as well as through laboratory analysis of air samples initiated by the sensors and a mobile monitoring van. Suncor designed the program with input from stakeholders including regulators, local governments, public health professionals, and most importantly, the people who live in the community. Montrose Air Quality Services, a third-party team of engineers, scientists, analysts and technicians, runs the program. Currently eight sites are in operation utilizing a combination of low cost and mid-tier sensors. Data is fed to a public-facing website, as well a free mobile app in English and Spanish. This presentation will share the experiences and discuss lessons learned from both a technology and program management standpoint. It will discuss feedback from the community and as insights from the data. The presentation will also outline next steps to improve the program.

Additional Authors: Patrick Clark, Robert Mennillo

Testing Visual Communication Strategies of Air Quality in Pittsburgh: A behavioral science approach

Ashley Angulo, University of Oregon

Our research team developed and tested a website which displayed air quality in the greater Pittsburgh area. The research project began with the wide intention of isolating key display dimensions that affect communication accuracy and usability. We identified and tested three dimensions of our pollution map display: key guide, color overlay, and route option. The key-guide is the primary sense-making tool that allows for interpretation of the current concentration of the pollutant. Color-overlay refers to the visual use of color that is added on top of the geographic pane of the map, akin to a weather map. Lastly route option refers to the choice-set of paths testing trade-offs between distance and pollutant exposure. We tested two display patterns for the guide: a table and dial display. Both displayed three ranges of exposure that related to unhealthy, concerning, and acceptable levels of pollutants, but varied in the manner in which this information was displayed. Participants in our experiments and community surveys found the dial guide more user-friendly than the table guide. Specific details of each guide to be discussed in detail. We tested five different color overlays ranging in complexity, hue, and implicit meaning. Amongst our findings, we identified a disconnect between accuracy in responses and subjective ratings of liking. Additionally, certain colors exacerbated pollution concern more so than others. We also tested traffic-light color overlays commonly used in health domains.

Additional Authors: Ashley Angulo, Julie Downs, Albert Presto, Subu Subramanian

Community-focused Monitoring in California: Building Bridges between Community Members and Industrial Facilities

Josette Marrero, Sonoma Technology

There has been a growing national awareness about the effects of exposure to air toxics and how they may disproportionately impact certain communities. In response to these growing concerns, regulations have been passed to help provide information to those who live, attend school, and work near refineries. For example, the U.S. Environmental Protection Agency and California state regulators have passed rules requiring the continuous monitoring of various air toxics at refinery fencelines and have mandated that this information be made readily available to the public. This trend is likely to continue throughout the U.S., as both the federal government and state legislatures are seeking to support air quality monitoring in areas where environmental justice is of concern. With a thoughtful approach, these monitoring efforts can be mutually beneficial for both industries and their neighboring communities as they build trust, provide transparency, and make knowledge available to all stakeholders. In this presentation, Sonoma Technology scientists will discuss approaches to community-focused monitoring programs and highlight relevant cases in California. We will discuss key elements of these monitoring programs and how the data are presented on public websites or summarized in reports to ensure common understanding among end users, including industrial facility employees, regulators, and community members. We will discuss the importance of how a combination of high-precision instrumentation and low-cost sensors can be used to achieve monitoring goals. We will also focus on how establishing regular avenues of communication between industries and the public can lead to positive interactions when facility changes are needed or when unexpected events occur.

Additional Authors: Josette Marrero, Clinton MacDonald, Eric Winegar, Hilary Hafner

Communicating Air Sensor Data on the AirNow Fire and Smoke map

Karoline Barkjohn, U.S. Environmental Protection Agency Office of Research and Development (Virtual Presentation)

Residents and local agencies in smoke impacted areas often use air sensors to provide more localized air quality data. In 2020, air sensor measurements were added to the AirNow Fire and Smoke map. Alongside permanent and temporary smoke monitor measurements, PurpleAir sensor measurements now appear as “Low Cost Sensor” icons. Monitor and sensor icons appear on the map as distinct shapes colored by NowCast AQI category. Clicking on an icon provides information on particulate matter concentrations at various averaging intervals, trends based on the past 30 minutes of air sensor data, and recommendations for health protective actions. Before sensor data is displayed on the map, PurpleAir sensor measurements are corrected and quality assured (i.e. points removed where A and B channels disagree), resulting in higher accuracy measurements and improved category estimation for the NowCast AQI. However, the uncertainty in the sensor, temporary monitor, and stationary monitor measurements is variable and dependent on time-averaging intervals. In this talk we will discuss how insights into local air quality can be gleaned from data on the map at various averaging intervals and how these insights can be transformed into actions to reduce exposure to smoke. Based on our work with PurpleAir sensors, we will highlight how to quantify and improve the accuracy of data from other sensor networks so that it can be also be utilized to provide insights on local air quality.

Additional Authors: Karoline Barkjohn, Amara Holder, Andrea Clements, Samuel Frederick, Ron Evans, Sim Larkin

Community engagement through text-based communication with air quality sensors

Surya Venkatesh Dhulipala, University of British Columbia (Virtual Presentation)

We present a novel mode of communication for disseminating air quality data in a community setting. We fitted our low-cost air quality sensors (LCAQS) with placards containing QR codes, that when scanned, begin a friendly text message conversation with users. For this study, we partnered with Hello Lamp Post, a startup based in UK. Overall, we present 6 months’ worth of user engagement data (mostly students). To quantify the air quality on University of British Columbia (UBC) campus, we installed a network of 8 LCAQS across UBC campus in June 2021 to measure air pollutant concentrations at different traffic intersections. At the same locations, we installed placards with QR codes for community engagement. Within a two-month period (August – September 2021), a total of 624 interactions and 190 conversations were recorded. 93% of users said they would like to see air quality sensors around their home and near major traffic intersections. A further 50% of users complained about current modes of communication around air pollution exposure on UBC campus. Other feedback included – “I have no idea where to look”, “make more accessible” and “I have not heard much about air pollution exposure on campus”. Often, local communities are unable to interpret publicly available data from city-wide or province-wide air quality monitoring stations and make informed decisions. We directly asked users (mostly students) about their preferred formats for reading air quality data, perceptions about air quality on campus and their satisfaction levels about current modes of communication. Our user engagement data can also be used to make localized decisions – one community may prefer dense sensor networks for monitoring air quality whereas another may be hesitant. Other possible applications of user-engagement data will also be discussed in this presentation.

Additional Authors: Surya Venkatesh Dhulipala, James Hindson, Naomi Zimmerman

Empowering communities through data dashboards

Tara Webster, Colorado Department of Public Health and Environment

Pulling data together is just the first step in creating meaning for different audiences. At the Colorado Department of Public Health and Environment, improving data access and transparency are key parts of our efforts to work towards environmental justice. This talk will present lessons learned through our work compiling disparate datasets to inform questions from policy makers and community groups from areas that have been disproportionately impacted by pollution. Our goal is to use data dashboards to move towards frameworks of engagement that support community empowerment. We will walk through several case studies to explain the context, the questions, and the data dashboard we created. We will highlight efforts to improve on the historically fragmented nature of past data collection efforts and the challenges of bringing these datasets together to meet people where they are. We will also discuss approaches to communicating human health risks, especially in areas with great uncertainty. Throughout these case studies, we will discuss how we identify opportunities for improvement and engagement to better serve the needs of our stakeholders.

Additional Authors: Tara Webster, Kristy Richardson, Paul Romer Present, Margaret Horton, Shannon Barbare

Community Air Sensor Use

Community-driven open-data on Pakistan’s air pollution problem

Abid Omar, Pakistan Air Quality Initiative (Invited Talk)

Pakistan has an air quality data problem with little to no monitoring across the country. Community-driven air quality monitoring has been instrumental in filling the air quality data gap in Pakistan, and in creating a grassroots citizen’s movement advocating for clean air. Real-time data from this community network has put a magnifying glass on air pollution, where citizens can react to their local air pollution, evaluate impact of emission-reduction policies, and ultimately providing the impetus for a cleaner environment. This paper presents Pakistan as a case study of a typical low-income country, similar to other developing countries across South Asia and Africa, lacking reference-standard monitoring equipment, or the technical capacity to manage them. Pakistan’s community-based nationwide network of low-cost real-time air quality monitors has helped fill the data gap in Pakistan and has been instrumental in many ways, by engaging the community and corporate citizens in participatory monitoring, by creating a network of ambassadors for air quality awareness, and finally by providing the baseline data for the government to initiate reference-standard monitoring. The impact from this community monitoring network has been tremendous in kick-starting awareness and furthering monitoring in one of the most air-polluted regions of the world. Learnings from this initiative are shared to enable other communities, cities and countries to harness a similar network of low-cost monitors to kick-start positive environmental change in their region, and how to engage and develop a community of citizen scientists to deploy, maintain and manage air quality sensors in a decentralized network.

Additional Authors: Abid Omar

Environmental Justice for fence-line communities

Gertrude “Naeema” Gilyard, (Invited Talk) (Virtual Presentation)

Partnerships and coalitions, including with Municipalities and Federal Agencies, are needed to help achieve Environmental Justice for fence-line communities whose health is compromised by exposure to air pollution and harmful toxins. Residents in fence-line communities face financial barriers to obtain air monitors with filters for small particles that create data, the potential to analyze the data, and translate the chemical compounds in a report that details the health impacts of environmental stressors. Partnership and collaboration with and assistance from all government agencies could help reduce or remove these barriers. It is the responsibility of States, Counties, Municipalities and Federal Agencies, who regulate and have oversight of companies and industries that release toxins into the atmosphere, to hold them accountable for their violations. States, Counties, Municipalities and Federal Agencies have access to sensitive air monitors and the capabilities to quantify the amount of toxins released by a company that exceeds the allowable threshold. With this in mind government agencies are in a position to support the health of fence-line communities by being more collaborative and transparent, collecting air monitoring data at community fence-line and providing access to that data and its analysis. Their first priority should be to protect the health of fence-line communities by holding violators of the U.S. EPA Clean Air Act accountable for violations. Collaboration with fence-line communities to address toxic air releases and enforcing regulations are good first steps.

Additional Authors:

Community, Health and Science: Establishing the Pioneer Valley Air Quality Network

Anna Woodroof, Earthwatch Institute

As thepopularityof low-cost air quality sensor growsso does the appealand the demand fordata and anunderstandingofthemeasurements that these sensors provide. The PioneerValleyHealthyAirNetworkis one example of how three communities in Western Massachusettswith historically highrate ofasthmaand poor air pollutionare working in conjunction withnonprofits, scientists and health experts to address the challenge of air pollution in these communities.Originally commissioned by theMassachusetts Attorney General’s Office’s Environmental Protection Division and the Massachusetts Municipal Vulnerability Program,the Pioneer Valley Air Quality Networkdeployslow-cost air quality sensors with city and school partnerstoprovide real world datasets to helpassess hyper-local air quality.By engaging withcommunity advisory groupstoco-builda school-based dashboard,the Network works toeducate andinform residence of local air quality and actions to mitigate exposure.This partnership has combined both a bottom-up andtop-downdirection and buy-in.The presenters from YaleSchool of Public Health, Earthwatch Institute,and the Public Health Institute of Western Massachusettswill share challenges and successes from the experience of this program.Earthwatch Institute and Yale School of Public Health have partnered on additional projects to provide education and tools to monitor and understand data coming from low-cost air quality sensors.

Additional Authors: Anna Woodroof, Dong Gao

Air Quality Investigation and Research for Equity (AIRE) in Commerce City, CO

Aracely Navarro, Cultivando

Commerce City, CO is one of the most polluted cities in the entire United States. It is also a community that it predominantly Latinx and low-income, making it a community suffering from environmental racism and injustice. They are surrounded by oil and gas refineries which contribute to some of the worst air quality in the nation and contribute to poor health outcomes for the community. Cultivando is a community-based non-profit organization that utilizes the Promotora model to promote health and racial justice in Commerce City. Due to the great need to address environmental racism, Cultivando, in partnership with Boulder A.I.R., implemented the community-based Air Quality Investigation and Research for Equity (AIRE) program. This is a year-long research study that utilizes a stationary air monitor which measures 50 different air toxics in real-time, a mobile air lab and simple home-based air monitors, which will measure various pollutants including VOC's, particulates and radioactivity. This is a unique project because no similar project has ever been done in this community before. This research is also unique because it is community led project that uses state of the art air monitoring technology. The aim of this project is to quantify the amounts of air pollution in Commerce City, CO and also investigate the health impact on surrounding communities to address environmental racism and injustice.

Additional Authors: Aracely Navarro, Olga Gonzalez, Detlev Helmig

Community led air monitoring informs land use policies in Kansas City

Beto Lugo, CleanAirNowKC

Regulatory ambient air monitoring under the Clean Air Act is concerned with compliance with federal and state standards for criteria pollutants, not what community members are breathing at the fenceline to industrial facilities. The EPA cannot detect major contaminants to which residents are exposed daily. Even when regulatory monitors are available, regulators prioritize industry over public health. The placement of regulatory monitors lack inclusion of community representatives in decisions, these networks fail to capture air pollution hot spots. As a result, CleanAirNowKC (CANKC) and community members in Kansas City living in areas with high concentrations of diesel emissions and industrial polluters are forced to live with elevated levels of pollution and the associated health effects without the resources necessary to advocate for improved air pollution standards. (CANKC) created a community-driven open source air monitoring network to measure PM2.5, NO2, and O3 and black carbon concentrations in different locations about which the community has expressed concern. These monitors, which measure hyperlocal pollution atintervals shorterthan federal monitors, have measured levels that exceed EPA standards. These resources can serve an important role in civic engagement, advocating for community-driven and just policies to improve public health; this project has enabled participation in decision making policies and land use plans in Kansas City. This data will also be integrated into an interactive map with an overlay that includes toxic release facilities in close proximity to residential homes, this will help the community members identify major air pollution contributors in their neighborhoods.

Additional Authors: Beto Lugo, Elizabeth Friedman, Casey Kalman, Atenas Mena

Establishing an Air Quality Monitoring Network to Inform Local Strategies in Franklin County, Ohio

Brandi Whetstone, Mid-Ohio Regional Planning Commission

The Mid-Ohio Regional Planning Commission, a regional council of governments, and Franklin County Public Health, located in Columbus, Ohio, are partnering to investigate differences in air quality across zip codes in Franklin County using low-cost particulate matter sensors manufactured by PurpleAir. Placement of these monitors is prioritized using sociodemographic and health data including poverty rates, percent minority population, numbers of children under age 5 and people over age 65, as well as asthma and COVID-19 infection rates. Team members conducted extensive outreach in the community to site the monitors with residents, organizations, and businesses in the identified zip codes. Prior to deployment, the PurpleAir monitors were collocated with a regulatory monitor for data quality evaluation. Here, we will present on the process, lessons learned to date, and highlight how data will be compared across zip codes to identify potential inequities in air pollution exposure and opportunities to improve public health and quality of life through program and policy implementation.

Additional Authors: Brooke White, Jennie McAdams, Layla Abraham, Anna Oestreich, Carl Malings, Michelle Bailey

Aires Nuevos: Driving Meaningful Air Quality Action in Latin America

Christi Chester Schroeder, IQAir North America

Aires Nuevos is a Citizen Air Quality Network created in 2020 to promote community-based air quality data generation and communication to reduce early childhood air pollution exposure in Latin America. Based on the IQAir air quality monitoring and communication platform, Aires Nuevos has installed and published 95 low-cost air quality monitors in 28 cities in Mexico, Uruguay, Peru, Brazil, Argentina, Ecuador, Chile, and Colombia building a sensor network capable of monitoring air quality for nearly 1.5 million children under the age of 4. Aires Nuevos uses a grass-roots approach to combatting air pollution by engaging the local community and offering air pollution education using real-time data accessible for free on IQAir’s website and mobile app. In each city, working groups consisting of local public officials, university researchers, and local community members collaborate using sensor data to identify local policies that will result in limiting air pollution exposure to young children. In this presentation, the authors will discuss some of the lessons learned in building up an air quality monitoring network across multiple countries and how the air quality data generated is starting to affect local policies to better protect citizens and improve air quality in the region.

Additional Authors: Christi Chester Schroeder, Marcela Otto, Loreto Stambuk

Community-engaged air sensor analysis: Visualizing PM2.5 data from PurpleAir sensors in Southeast Los Angeles

Claire Bai, University of Southern California

The advent of using low cost sensors to monitor air quality in environmental justice (EJ) communities has prompted a need to understand and interpret data among members and activists. PurpleAir PM2.5 sensors have gained popularity due to their accessibility and increased work of communities to characterize neighborhood air quality. However, less attention is paid to community-oriented data analyses. Our initial work focused on outdoor PurpleAir sensors placed along a major freeway in an international trade corridor in southeast Los Angeles County connecting the Ports complex to a major railyard. Sensors were included if they had coverage for at least 30% of the year and at least 80% reliability between channels. Data was identified for 26 sensors in 2019 and 41 in 2020, and combined to average daily PM2.5 levels for all sensors, and by 3 regions: south of the railyard, the freeway corridor and the Port area. Using R, we created time variation plots compared to the closest NAAQS reference monitor, calendar plots, bar plots for percentage of time above NAAQS thresholds by region, and seasonal daily variation plots. The visualizations were shared with multiple EJ organizations for feedback, who emphasized the importance of clear messaging and health-related information associated with PM2.5 levels. We later developed a Shiny app, hosted on the University of Southern California’s Environmental Health Centerswebsite [https://envhealthcenters.usc.edu/pm-2-5-shiny], to provide an interactive platform that compares visualizations by region and calendar year. We hope to expand this initial work to develop tools for PM2.5 data visualization across Los Angeles.

Additional Authors: Claire Bai, Wendy Gutschow, Jill Johnston

Revolutionising air quality monitoring using DIY and IoT approaches to beat air pollution in Africa.

Collins Gameli Hodoli, Clean Air One Atmosphere

The emergence and use of low-cost environmental sensing tools have shifted the paradigm on how air quality monitoring is carried out. This includes community engagement integrated with the democratisation of air quality monitoring. This consequently gives power back to the people to understand baseline air quality levels to influence behavioural changes and appreciate air pollution and climate change mitigation policies. Specifically, for communities in the global south, utility of low-cost sensors (LCS) has provided citizens with relatively cheaper and simpler mechanisms for understanding the impacts of our activities on local air quality including vehicular emissions as well as other anthropogenic activities and natural sources. Of a particular importance is how the utility of LCS can help expand community’s understanding of local air quality levels and corresponding health effects tied to health advice using “dot it yourself” (DIY) and “internet of things” (IoT) approaches. Clean Air One Atmosphere (CAOA) has collaborated with local and international organisations and research institutions including regulatory and local governmental agencies to deploy LCS in West Africa as well as establish links between air quality experts from the global north and academic institutions in West Africa. CAOA also initiated the first ever mobile application that meaningfully communicate air quality data with corresponding health effects and provide health tips for subscribers as well as a general advice on how individuals can reduce exposure and contribute to reducing atmospheric emissions in the global south. The application "Yakokoe" populates open-source data from all over Africa to sensitise citizens.

Additional Authors: Collins Gameli Hodoli

Low-Cost Air Pollution Sensor Characterizes Excessive Smoke from a Neighborhood Restaurant and Highlights Gaps in Environmental Health Laws: An Observational, Citizen Science Study

Nicholas Newman, University of Cincinnati, College of Medicine, Dept of Pediatrics (Virtual Presentation)

Air pollution monitoring is done using a sparce monitoring network and is designed to provide background concentrations of pollutants. Due to this design, small area variations due to local emission sources are missed. The advent of low-cost air pollution sensors provides an opportunity to fill this gap, but air quality & public health agencies are not staffed or funded for this type of monitoring. Using trained citizen scientists to measure local level air pollution is one method to address this gap. We describe the development and implementation of an air pollution monitoring and community engagement plan in response to resident concerns regarding excessive smoke production from a neighborhood restaurant. Neighborhood-level particulate matter <2.5 micron in diameter (PM2.5) was measured using a portable sensor. In 82% (14/17) of the air monitoring sessions, the peak PM2.5levels were at or above 35.5µg/m3, considered “Unhealthy for Sensitive Groups” by the US EPA. The highest PM2.5 readings were in the immediate vicinity of the source. Results were shared with the community at a health fair and with local government officials. Neither of these resulted in any action to address the source of the smoke emissions. Agencies cited a lack of jurisdiction as the primary reason. Current environmental public health laws may not adequately address non-industrial or non-vehicular sources of PM2.5. Community members using modern lower cost air pollution sensors may demonstrate potentially important local sources of air pollution, but environmental public health laws will need updating to address these.

Additional Authors: Nicholas Newman, Deborah Conradi, Rachael Shepler, Alexander Mayer, Patrick Ryan, Erin Haynes

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

Ryan Brown, US EPA Region 4

The collocation of air quality sensors with regulatory monitors at existing 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. This presentation discusses a project during which 20 collocation shelters, which provide weather shielding and security for sensors, were built and recently deployed at state/local/tribal air monitoring locations across the U.S. to provide the infrastructure for monitor and sensor collocation opportunities. This work is a collaboration between the U.S. Environmental Protection Agency (EPA) and multiple state, local, and tribal air monitoring agencies. Participating agencies are working to make these shelters accessible to the public and to encourage communities to collocate sensors with regulatory monitors and incorporate sensor data quality assurance and correction into their sensor projects. These efforts lead to increased community engagement, leveraged expertise of the regulatory air monitoring staff, lowered barriers for collocation, and more quality-assured sensor data. As part of the project, EPA Office of Research and Development commissioned and published a design document which will allow shelters to be built and installed at regulatory air monitoring sites across the U.S. The presentation will also focus on specific challenges related to site access, security, and infrastructure that state, local, and tribal air monitoring agencies addressed in implementing this project.

Additional Authors: Ryan Brown, Daniel Garver, Karoline Barkjohn, Andrea Clements, Amanda Kaufman, Suzy Apodaca, Mark Berry

Air Quality Chicago: Mobile Monitoring and Capacity-building with Chicago's Environmental Justice Communities.

Tiffany Werner, Environmental Law & Policy Center (Virtual Presentation)

As technology advances and low-cost sensors are becoming more accessible, understanding one’s exposure to air quality pollution is becoming easier to do each year. While arming communities with handheld sensors to collect real time air quality data is useful for public education and awareness can it also be used as an advocacy tool to push for clean air policy? Understanding the quantity of data needed to properly advocate for policy change is difficult, especially when using low-cost, mobile sensors, but not impossible. In Chicago, the Environmental Law & Policy Center (ELPC) has spent the last five years exploring how armingcommunities with handheld air quality monitors to assess street by street PM2.5 levels can empower community members to become vocal advocates for clean air policy and enforcement. Using AirBeams, ELPC has been partnering with residentsto collect data around their neighborhoods and near pollution sources. With targeteddata collection andempowered community scientistswe have been able to gain the attention of policy makers on air quality injustice and are now exploring how solutions such as vegetative buffers and fleet electrification can mitigate neighborhood PM2.5 levels.

Additional Authors: Kiana Courtney, Susan Mudd, Tiffany Werner

Filling in the air quality data gap and enabling air quality management in LMICs using low-cost sensors

Redspira: Sharing information to transform communities.

Alberto Mexia, Redspira (Invited Talk)

The collaborative environmental monitoring network REDSPIRA emerged in 2018 as a social responsibility initiative, with the purpose of offering a solution to the deficit of information on air quality. This technological platform is integrated by hardware and software. The hardware is made up of low-cost sensors installed at different points in a given geographic area, designed to measure air pollution; while the software includes both mobile applications and a web portal that displays in a georeferenced way the location of the sensors and their measurements. One advantage that REDSPIRA offers, thanks to its collaborative nature, is its ability to integrate monitoring infrastructure and air quality data collected by both environmental authorities and citizens into a common platform. In 2020, the NGO "Foundation for Air Quality Research" was created. with the aim of operating and ensuring the growth and consolidation of the network, adding socially responsible companies to donate sensors, and dozens of volunteers who host them, providing them with electricity, internet and adequate protection. In addition, funds have been obtained for the expansion, research and development of the technologies, obtaining in 2019 a grant of $ 50,000 dollars (United States Environmental Protection Agency, 2020) in alliance with the Imperial County Air Pollution District, to expand the binational air monitoring network using 30 particle sensors and low-cost sensors in the Mexicali-Imperial atmospheric basin. The project also considered the incorporation of 30 schools in the Mexicali valley to the Air Quality Flags Program, through the Border 2020 program.

Additional Authors: Alberto Mexia

Overview of the LCS-SA Campaign: Opportunities for the application of low-cost air quality sensors in South Africa

Brigitte Language, North-West University (Invited Talk)

Air pollution is a leading environmental health risk in low- and middle-income countries. Ambient and household air pollution have an impact on the health of vulnerable population groups. Internationally, there is appropriate evidence showing links between impacts and patterns of exposure to air pollution. However, there is less evidence in African countries. Policy- and decision-makers, and people advocating for clean air, including communities themselves, need evidence to understand the severity of Africa's air pollution problem and the associated burden. This highlights the importance of hard evidence and data to understand, explain and call for action, and interventions to prevent air pollution and its harmful health impacts. South Africa has a network of 183 ground-based air quality monitoring stations (AQMS). AQMSs are sparse and do not capture HAP as the AQMS network is biased in location towards industrial areas. The AQMSs typically monitor criteria pollutants such as PM10, PM2.5, SO2, NO2, O3, and CO. The AQMS network is plagued with high levels of required maintenance and lower levels of data recovery, which is linked to the cost of maintaining such a network. A solution to the lack of data in ambient and household settings is the concept of low-cost air quality sensors (LCS), which are relatively inexpensive and easy to deploy in an area of interest. Most of the LCSs used in South Africa are developed and manufactured outside the country. We questioned the precision, accuracy, reliability, rigidity and usability within a South African setting. The Low-Cost Sensor South Africa (LCS-SA) campaign was launched to understand the advantages and limitations of using LCS. The LCS and reference instrument co-location campaign was conducted during 2021 at the SAWS AQMS, situated on the Vanderbijlpark Campus of the NWU. This site is ideal as it is centrally located in a region experiencing a wide range of pollutant concentrations resulting from multiple sources.

Additional Authors: Brigitte Language, Bianca Wernecke, Roelof Burger, Joshua Vande Hey, Rikesh Panchal, Caradee Wright, Stuart Piketh

Intercomparison of Low-Cost PM2.5 Sensors with Federal Regulatory Monitor in Sub-Saharan Africa

Emmanuel Appoh, (Invited Talk)

Urbanization, migration and urban population growth has increased in Sub-Saharan African (SSA) cities over the last decades, affecting air quality in cities. Low-cost sensors are becoming popular in use especially in cities where regulatory grade monitors are lacking. Air quality monitoring in SSA is limited as a result of the high cost associated with Federal Reference Monitors (FRM). Clarity Node and PurpleAair sensors are currently being used in Ghana on a lower scale for measuring PM mass concentration as an indicator of air quality. Evaluation of sensors with FRM in SSA is still a work-in-progress. Additional efforts documenting data quality are necessary. In this study, 2 PurpleAir, 2 Modulair and 18 Clarity node sensors were evaluated by collocating them at University of Ghana air quality monitoring station equipped with Teledyne T640. PM2.5 measurements were compared with data generated from the FRM from April-July 2021. The study showed that (i) the R2 values between the T640, Clarity Node, PurpleAir and ModulAir were 0.77, 0.88 and 0.89 respectively. (ii) With respect to the FRM the 3 sensors have mean absolute error of 3.36, 2.60 and 1.66 (µg m-3) respectively. The research period covers relatively short time and it is necessary to carry out longer-term studies covering both dry and wet seasons, including meteorological conditions. From the findings, it is clear that these sensors have the potential to supplement, but not replace high-quality and reliable air quality monitoring systems. The implications of the results for future urban air quality assessment using low-cost sensors in SSA are discussed.

Additional Authors: Appoh

Air sensing to action in the African context: design and deployment of a community-driven digital air quality sensing network for African cities.

Engineer Bainomugisha, AirQo/Makerere University (Invited Talk) (Virtual Presentation)

African cities are characterised by unique contexts for air quality sensing including deployment environments, the technology infrastructure, varying degrees of regulatory frameworks, capacity, public education and awareness. This talk will cover the lessons learned and success stories from the AirQo initiative based in Kampala, Uganda. Founded at Makerere University, AirQo designs and develops of a community-driven low-cost air sensing network to fill the air quality data gaps in African cities. The AirQo system includes: (i) a network of over 100 custom air sensing devices hosted by communities (ii) digital air quality platforms to facilitate timely access to air quality information by decision makers and the public (iii) machine learning-based calibration models for data quality (iv) Translation of air quality data into actions such as citizen and community engagement to raise awareness, and providing evidence to inform regulations and air quality management plans in African cities.

Additional Authors: Engineer Bainomugisha, Deo Okure, Joel Ssematimba

Improving low-cost PM2.5 sensor networks through retrospective analyses and satellite observations

Michael R Giordano, AfriqAir (Invited Talk)

Over much of the Global South, air quality monitoring, especially for PM2.5, is lacking. Low-cost PM sensors (LCS) offer one way to help close that data gap. Unfortunately, the various sensitivities of LCS require that local calibrations (usually done with reference or research-grade sensors) are applied before low-cost data quality can be assured. This requirement can be a problem in many parts of the world where expensive reference or research-grade equipment is simply unavailable. Two ways to potentially alleviate this requirement are to use either satellite observations or data assimilation tools such as retrospective analyses to calibrate LCS. The two approaches differ in both spatial and temporal resolution but both can provide useful information. Here we will discuss both these approaches using various satellites and the Modern Era Retrospective analysis for Research and Applications (MERRA) tool from NASA. Satellite AOD and MERRA aerosol speciation are used to calibrate LCS. This work will demonstrate how this process works and how it performs for 3 areas in sub-Saharan Africa: Ghana, Kenya, and South Africa. Results suggest that simple hygroscopicity corrections from the MERRA reanalysis datasets and AOD measurements from satellites do not generally improve the correlation between LCS and reference monitors but can both reduce the mean-normalized bias by up to 20% and can increase the accuracy by up to 25% (reductions in MAE, CvMAE) for otherwise uncalibrated LCS, depending on area.

Additional Authors:

Prospects of emerging low-cost air quality sensors for bridging air pollution epidemiologic evidence gaps in Africa

A. Kofi Amegah, University of Cape Coast

The burden of disease (BOD) attributable to ambient air pollution exposure in Africa is growing and yet estimates of its magnitude and impact are underestimated due to lack of air quality monitoring capacity as well as limited air pollution epidemiological studies. Integrated exposure response (IER) functions used to estimate the BOD attributable to air pollution are based on data from North America and Western Europe, locations with lower ambient PM2.5 exposures, and hence, have limited application to Africa. There is therefore the urgent need to expand air quality monitoring networks in Africa to enable the conduct of high quality air pollution epidemiologic studies to better quantify the BOD attributable to air pollution exposure in Africa and provide the evidence base to trigger investments in air pollution control for public health protection. The growing influx of low-cost sensors represents an excellent opportunity for bridging these data gaps and in this presentation, I would outline how the Ghana Urban Air Quality which seeks to deploy a dense network of low cost sensors interspersed with reference grade monitors in urban settlements of Ghana has been leveraged for investigating adverse respiratory and cardiovascular outcomes among street traders, a major occupational group in cities of Africa and other developing regions. The presentation would also highlight the use of low-cost sensor networks in other LMIC countries in bridging the air pollution epidemiologic evidence gaps in these countries.

Additional Authors: A. Kofi Amegah

First measurements of PM2.5 and NO2 in Mombasa, Kenya

Daniel Westervelt, Lamont-Doherty Earth Observatory of Columbia University

Air pollution caused an estimated 1.2 million premature deaths in Africa in 2019. Many cities have limited or non-existent air quality monitoring networks which makes air quality management difficult or even impossible, including Mombasa, a major port city in Kenya with a population of at least 1.2 million people and growing. To address this lack of air pollution data, we deploy a reference Beta Attenuation Monitor 1022 (BAM-1022) and five Clarity Node-S low cost sensors in diverse environments within Mombasa. Locations include 2 urban sites on the main Mombasa island, one residential site, one site near the largest port in East Africa and the international airport, and one site in proximity to Bamburi Cement Factory, a major point source slightly north of the city. Low-cost PM2.5 from the Clarity Node-S are calibrated against the reference BAM-1022 and yield high correlation on daily and hourly timescales but moderate bias. Initial results show relatively good air quality for the latter half of 2021 (~10 µg m-3 average, 58 µg m-3 maximum) at the urban and residential sites, perhaps modulated by pristine air mixing in from the Indian Ocean. These levels are lower than similar time frame mean levels in Nairobi which are about 23 µg m-3 in urban environments. Concentrations at the industrial cement site in Mombasa are elevated, with a 2021 average of ~13 µg m-3 and peak hourly concentrations reaching 70 µg m-3. Concentrations at the port site are significantly higher than any other site in Mombasa, with mean concentrations of about 23 µg m-3and maximum hourly values up to 75 µg m-3. NO2 concentrations in Mombasa are relatively low, at only about 13 ppbv compared to 16 ppbv in Nairobi, though better understanding and characterization of low-cost NO2 sensors are needed before placing high confidence in this data. Ongoing work will examine additional data as it comes in and attempt to understand sources and explanations for the air quality in Mombasa.

Additional Authors: Daniel Westervelt, Josephine Kanyeria Ndiangui, Paul Njogu, Michael Gatari, Ezekiel Waiguru Nyaga, R Subramanian, Michael Giordano, Savannah Ward

Assessment of Traffic-derived Air Pollutants by Smart Sensors: Comparison of Pollutants at Street Levels

Mahesh Senarathna, Postgraduate Institute of Science, University of Peradeniya (Virtual Presentation)

Traffic-Related Air Pollution (TRAP) is a primary source of urban atmospheric pollution worldwide. Changes in city traffic flow is sensitive to air pollution levels. When city’s streets are congested with traffic due to improper planning, the levels of air pollution at street levels may vary and this variation sometimes could not be captured by stationary air pollution monitors. This study aimed to evaluate air pollution before and during a new traffic plan established in March 2019 in the city of Kandy, Sri Lanka, using smart sensor technology. Street-level air pollution data (Particulate Matter < 2.5 µm in diameter (PM2.5) and Nitrogen dioxide (NO2)) was obtained using a mobile air quality sensor unit before and during the new traffic plan. This sensor unit was mounted to a Police traffic motorcycle which travelled in selected streets around the city four times per day. Air pollution in road segments was compared before and during the new traffic plan, and air pollution trends at different times of the day were compared using data from a stationary smart sensor. During the monitoring period, both PM2.5and NO2 levels were well above the World Health Organization (WHO) 24-hour standards, regardless of the traffic plan period. Most of the road segments had comparatively higher air pollution levels during compared to before the new traffic plan. For any given time (morning, midday, afternoon, evening), day of the week, and period (before or during the new traffic plan), the highest PM2.5and NO2 concentrations were observed at the road segment from Girls High School to Kandy Railway Station. The mobile air pollution monitoring data provided evidence that the mean concentration of PM2.5during the new traffic plan (116.71µgm-3) was significantly higher than before the new traffic plan (92.32µgm-3) (p <0.007). Careful planning and improving the road infrastructure before implementing a new traffic plan could reduce air pollution in urban areas.

Additional Authors: Mahesh Senarathna, Sajith Priyankara, Rohan Weerasooriya, Gayan Bowatte

The role of philanthropy in filling air quality data gaps

Tom Grylls, Clean Air Fund

Only half of the world’s national governments produced air quality data in some capacity in 2020, with data gaps particularly acute in low-income countries. Recent innovations (low-cost sensors, improved satellite data, machine learning etc.) have the potential to increase data coverage at potentially reduced costs, but widespread and sustained adoption of these emerging approaches requires increased funding for deployment and capacity building.

Official development funders (ODF) and philanthropic foundations spend less than 1% and 0.1% of their total budgets on air quality initiatives respectively. Of that funding, just 10% goes to countries in Latin America and 5% to Africa, resulting in a corresponding air quality funding gap.

Philanthropic funding to air quality initiatives is growing (albeit from a small base), with an estimated $44.7m in funding provided in 2020 compared to only $12.9m in 2015. Foundations have a key role to play. They can be agile, have a higher tolerance to risk, and catalyse other ODF and governmental funding to help projects go to scale.

The Clean Air Fund (CAF) is a philanthropic initiative with a mission to tackle air pollution globally. A key part of CAF’s work is to improve the accessibility, usability and actionability of air quality data – delivered by funding a range of projects that test emerging technologies and build capacity towards effective air quality management. CAF has developed an air quality data strategy, and through its implementation has supported hybrid and low-cost sensor networks across continents, as well as work on open data platforms, knowledge hubs and technical assistance.

This talk will provide a funder perspective on filling air quality data gaps. CAF’s strategic approach to air quality data, findings from CAF’s annual State of Global Air Quality Funding report, and learnings from CAF-funded projects will be drawn together to reflect on future needs and priorities for both project implementers and other funders.

Additional Authors: Matt Whitney, Tom Grylls

Low-cost PM2.5 measurements in a binational metropolitan area along the U.S.-Mexico border

Mayra Chavez, University of Texas at El Paso

The Paso del Norte (PdN) encompasses a binational metropolitan area along the U.S./Mexico border that includes El Paso, Texas in the U.S and Ciudad Juárez, Chihuahua in Mexico. Air quality in the area is a major concern for the 2.5 million residents due to social, economic, geological, and political factors. Resource disparity between the two international jurisdictions has hindered the progress of developing a systemic air quality management plan for the airshed. A binational air quality study using low-cost sensors was therefore conducted to generate PM2.5 data to supplement the limited PM monitoring network in the airshed. The study attempts to 1) evaluate the efficacy of low-cost sensors when compared to reference monitors, and 2) advance scientific measurement and analysis of air quality for exposure assessment in the PdN using low-cost air sensors. We conducted a 2-month air monitoring campaign at 17 elementary schools and 14 industrial sites throughout the PdN with a control site collocated at a state-operated air monitoring station equipped with a FRM instrument. Quality control for the PM2.5 data included removal of anomalies based on internal inconsistencies between different channels; removal of outliers; and application of a calibration equation considering concurrent humidity, temperature, and FRM readings. Confidence in the sensors’ performance is evaluated through the agreement between internal channels and inter-sensor consistency. Multivariate linear correlation with FRM-quality data provided a unique calibration equation and a performance evaluation for each sensor. The adjusted PM2.5 data were used to generate a concentration map for identifying potential hotspots near industrial facilities and schools affected by traffic. The monitoring campaign is extended for a 12-month period in the PdN. Collocated monitoring at two reference stations will be continued for quality control and performance evaluation of the low-cost sensors.

Additional Authors: Mayra Chavez, Leonardo Vazquez, Yazmin Hernandez Garcia, Frida Toquinto Manjarrez, Evan Williams, Adrian Vazquez Galvez, Wen-Whai Li

Using low-cost PM2.5 and GPS sensors with surveys to understand exposure in informal settlements in Nairobi, Kenya

Michael Johnson, Berkeley Air Monitoring Group

We monitored personal PM2.5(Purple Air PA-II-SD) and GPS location (Columbus P-1) for 71 mothers in two informal settlement communities outside Nairobi, Kenya (Dagoretti and Starehe). Participants were outfitted with backpacks to carry the instruments for 24-hour periods, and ambient PM2.5monitors were installed in each community. Time-activity surveys were administered to contextualize the PM2.5and location data with the sources and activities contributing to exposure. Mean daily exposures were relatively high (43.9 and 44.5 µg/m3, in Dagoretti and Starehe, respectively), exceeding the WHO annual interim 1 target (35µg/m3), and all participants had exposures above the WHO annual guideline (10 µg/m3). Diurnal ambient PM2.5 patterns typically tracked well with the personal exposures, although the daily median personal exposures were higher than ambient, which was generally quite high as well (27.2 and 35.8 µg/m3, respectively). Exposures during cooking with wood or charcoal were higher than during other activities, and participants who used these solid fuels for cooking had a daily mean PM2.5 exposure of 59.7 µg/m3 (n=16), compared to 40.3 µg/m3 for those who did not (n=57; t-test p = 0.003). The results suggest the most promising and practical intervention to reduce exposures in the target population would be to transition households using wood and/or charcoal for cooking to clean fuels such as LPG, ethanol, or electricity.

Additional Authors: Michael Johnson, Timothy Abuya, Ricardo Piedrahita, Deborah Sambu, Daniel Mwanga, George Odwe, Charity Ndwiga, Heather Miller, Madeleine Rossanese, Sathy Rajasekharan

Spatial variation of fine particulate matter levels in Nairobi before and during the COVID-19 curfew: implications for environmental justice

Priyanka deSouza, University of Colorado Denver

The temporary decrease of PM2.5concentrations in many parts of the world due to the COVID-19 lockdown spurred discussions on urban air pollution and health. However there has been little focus on sub-Saharan Africa, as few African cities have air quality monitors and if they do, these data are often not publicly available. Spatial differentials of changes in PM2.5concentrations as a result of COVID also remain largely unstudied. To address this gap, we use a serendipitous mobile air quality monitoring deployment of eight Sensirion SPS 30 sensors on motorbikes in the city of Nairobi starting on 16 March 2020, before a COVID-19 curfew was imposed on 25 March and continuing until 5 May 2020. We developed a random-forest model to estimate PM2.5surfaces for the entire city of Nairobi before and during the COVID-19 curfew. The highest PM2.5concentrations during both periods were observed in the poor neighborhoods located to the east of the city center. Changes in PM2.5were heterogeneous over space. PM2.5concentrationsincreasedduring the curfew in rapidly urbanizing, lower-middle-class neighborhoods likely because residents switched from LPG to biomass fuels due to loss of income. Our results indicate that COVID-19 and policies to address it may have exacerbated existing air pollution inequalities in the city of Nairobi. The quantitative results are preliminary, due to sampling limitations and measurement uncertainties, as the available data came exclusively from low-cost sensors. This research serves to highlight that spatial data that is essential for understanding structural inequalities reflected in uneven air pollution burdens and differential impacts of events like the COVID pandemic. With the help of carefully deployed low-cost sensors with improved spatial sampling and at least one reference-quality monitor for calibration, we can collect data that is critical for developing targeted interventions that address environmental injustice in the African context.

Additional Authors: Priyanka deSouza

Improving Air Quality in 133 'Non Attainment' cities of India with Low-Cost Sensors & National Clean Air Policies

Ronak Sutaria, Respirer Living Sciences Pvt. Ltd.

As of November 2021, India has around 300 Continuous Ambient Air Quality Monitoring Stations (CAAQMS - reporting air quality data every 15 minutes) and 800 "manual" monitoring stations (which report 2 air quality data points every week). India's Rs. 4000 Crore (USD $650 Million) flagship national policy for clean air called "National Clean Air Programme (NCAP)" is based on Particulate Matter (PM10) data collected from the 'manual' monitors. Advances in Low-Cost Air Quality Sensors in India have been supported by several Indian government agencies (such as the Department of Science and Technology, Ministry of Environment, Forests and Climate Change and few state Pollution Control Boards). Based on field evaluation and validation results, the central government in India has now started exploring applications which use low-cost air quality sensor data. The "Atmos - Realtime Air Quality" monitoring network has been part of several state and central government funded initiatives around low-cost air quality sensors. In this paper, we will present the performance metrics (including calibration factors, correlation R-square, precision, bias and error percentages) of the low-cost sensors from 20 colocated sites (with FEM/FRM grade BAM monitors) in India. The paper will also discuss government directives related to low-cost air quality sensors and the 'citizen and urban planning' applications that are expected to be built using the LCS data. Brief overview of the air quality monitoring initiatives across India and the role of LCS in the 133 'non attainment' cities of India will also be discussed.

Additional Authors: Ronak Sutaria

Annual observations of Air Quality using Cost-efficient Sensors in Cabo Verde

Sandra M. S. Freire, University of Cabo Verde

In Cabo Verde, Saharan dust is the primary source of air pollution in the particulate matter form. Several studies carried out particularly in Praia and on the São Vicente Island have shown the impact of dust from the Sahara Desert on the local air quality. In this study, Size-resolved (PM10, and PM2.5) aerosol concentrations were measured in Praia city and Sal Island between January 2020 and January 2021 using two low-cost sensors in Sal and Praia, AERONET and satellite observation. Results show a seasonal trend of PM concentration on both sites with the largest concentrations in Praia. The presented PM data for a year-long record are addressed for the potential impact of local air pollution on public health during the COVID-19 pandemic and provide input for the long-term air quality management policies in Cabo Verde.

Additional Authors: Sandra M. S. Freire, Gregory S. Jenkins, Mateus Andrade, Nicolau Araújo

Indoor Sensing for Air Quality Control and Ventilation Applications

Wildfire smoke and ash: particle size, chemistry, and measurement needs

Jeff Wagner, California Dept. of Public Health (Invited Talk)

Wildfire smoke is a mixture of particle- and gas-phase chemicals produced by incomplete combustion of various fuels, including biomass, building materials, and vehicles. Indoor exposures to wildfire smoke particulate matter (PM) in populated areas depend upon on smoke and ash PM size, chemistry, transport distance, duration, and meteorology, as well as indoor penetration and filtration. This presentation discusses how these variables are addressed with different study designs and method choices. Wildfire smoke exposures have been associated with increased hospital admissions and symptoms for respiratory, cardiovascular, and cerebrovascular events in the general population and wildland firefighters. However, the specific impacts of excursions on the order of hours to weeks of wildfire smoke (~200-500 nm tar balls) and ash (coarse, inorganic, metal-rich PM) possess several unknowns. Current guidelines for wildfire smoke exposure are based on guidelines established for non-wildfire PM2.5 and PM10 over specific durations. However, several studies suggest that PM enriched in organic carbon or metals can have differential health effects, as can size distribution subcomponents within PM2.5 (PM1) or PM10 (PMcoarse). Similarly, varying PM exposure and measurement durations may correspond to different chronic or acute health endpoints. Improved characterization of PM size, composition, and exposure durations can help elucidate their relevant inhalation and filtration parameters, as well as their potential for health effects. The identification of wildfire-specific tracers may enable better classification of smoke episodes. For these reasons, the use of multiple, complementary measurement methods is sometimes desirable. Recent and current wildfire PM projects are discussed that combine different continuous/low-cost sensors and integrated samples, including measurements of UVPM, PM1, PM2.5, PMcoarse, and individual particle chemistry and morphology.

Additional Authors: Jeff Wagner

Standardized test instructions and test gases for VOC detectors for indoor air quality measurement

Christian Bur, Saarland University, Lab for Measurement Technology

Many sensor systems are offered for monitoring indoor air quality (IAQ), often based on low-cost gas sensor elements for quantification of the total concentration of volatile organic compounds (TVOC). This field has seen a tremendous effort from many industrial players to provide novel “digital” gas sensors allowing ubiquitous AQ monitoring using IoT devices or even mobile phones. However, existing test standards, e.g., as defined in ISO 16000-29 (test methods for VOC detectors), are highly inadequate for providing a suitable frame of reference for manufacturers and end users alike. Based on the workshop “Setting standards for low-cost Air Quality sensors” held in Berlin in 2019, VDI/VDE-GMA, the society for measurement and automation technology (Gesellschaft Mess- und Automatisierungstechnik) of the two leading German engineering associations VDI and VDE, has started developing a guideline for performance targets of indoor air quality sensor targeting VOCs. The contribution will describe the technology-agnostic concept of the proposed test standard which is based on simulating the complex indoor environment by selecting the most abundant VOC(s) of the main VOC classes and testing sensor systems with complex mixtures of these VOCs as well as further permanent background gases like CO2, CO, H2, NOx plus humidity. Tests are based on random mixtures in the relevant concentration ranges for each gas component to simulate real-world environments in a well-controlled laboratory test. In the framework of the national project VOC4IAQ, we are currently performing extensive lab and field experiments, accompanied by analytical reference measurements, to ensure that the test standard has high relevance for end users while being affordable for sensor element and system manufacturers. One further goal is to define a simplified Indoor Air Quality Index (IAQI) based on VOC concentrations similar to the six level AQI of the EPA for outdoor pollutants.

Additional Authors: Christian Bur, Tobias Baur, Johannes Amann, Christian Meyer, Andreas Schütze

Testing of a Low-Cost Sensor and Sampling Platform Alongside Reference Instruments in a Home Kitchen

Jessica Tryner, Colorado State University

People in the United States spend most of their time indoors at home, but comprehensive characterization of in-home air pollution is limited by the cost and size of reference-quality monitors. We assembled small “Home Health Boxes” (HHBs) to measure indoor PM2.5, PM10, CO2, CO, NO2, and O3 concentrations using filter samplers and low-cost sensors. Nine HHBs were collocated with reference monitors in the kitchen of an occupied home in Fort Collins, Colorado, USA for 168 h while wildfire smoke impacted local air quality. When HHB data were interpreted using gas sensor manufacturers’ calibrations, HHBs and reference monitors (a) categorized the level of each gaseous pollutant similarly (as either low, elevated, or high relative to published air quality guidelines) and (b) both indicated that gas cooking burners were the dominant source of CO and NO2 pollution (with 1-h average NO2 concentrations exceeding 100 ppb during normal cooking); however, HHB and reference O3 data were not correlated. When HHB gas sensor data were interpreted using linear mixed calibration models derived via collocation with reference monitors, root-mean-square error decreased for CO2 (from 408 to 58 ppm), CO (from 645 to 572 ppb), NO2 (from 22 to 14 ppb), and O3 (from 21 to 7 ppb); additionally, correlation between HHB and reference O3 data improved (Pearson’s r increased from 0.02 to 0.75). Mean 168-h PM2.5 and PM10 concentrations derived from nine filter samples were 19.4 µg m-3 (6.1% relative standard deviation) and 40.1 µg m-3 (7.6% relative standard deviation). The 168-h PM2.5 concentration was overestimated by PMS5003 sensors (median sensor/filter ratio = 1.7) and underestimated slightly by SPS30 sensors (median sensor/filter ratio = 0.91).

Additional Authors: Jessica Tryner, Mollie Phillips, Casey Quinn, Gabe Neymark, Ander Wilson, Shantanu Jathar, Ellison Carter, John Volckens

Low-cost high-performance VOC sensor systems: comparison with analytical measurements and long-term stability

Johannes Amann, Saarland University, Lab for Measurement Technology

With air quality being one target in the sustainable development goals set by the United Nations for 2030, the accurate monitoring of Indoor Air Quality (IAQ) is more important than ever. Volatile organic compounds (VOCs) from a wide range of sources (building materials, furniture, household items, cooking, cleaning, human metabolism) are the main pollutant of concern indoors. One inexpensive and promising solution for VOC monitoring for IAQ are metal oxide semiconductor (MOS) gas sensors. To fully exploit the potential of MOS-based sensor systems for accurate quantification of the total VOC concentration as well as specific VOCs of interest, advanced operating modes, calibration, and data evaluation methods are required. This presentation provides an overview of the potential and limits of MOS gas sensor systems for IAQ monitoring using temperature cycled operation (TCO), calibration with complex randomized gas mixtures and data-based models trained with advanced machine learning. We studied the performance of several sensors in a typical office environment over a period of more than one year. In addition to monitoring normal ambient air, release tests were performed with compounds that were included in the lab calibration, but also with additional VOCs. The tests were accompanied by different analytical systems (GC-MS with Tenax sampling, mobile GC-PID, and GC-RCP). The results show excellent quantitative agreement between analytical systems and MOS gas sensor systems under real-world conditions. The study shows that MOS sensors are highly suitable for determining the overall VOC concentrations with high temporal resolution and, with some restrictions, also for selective measurements of individual components. Based on these results, MOS sensor systems are well suited for demand-controlled, and thus energy efficient ventilation in typical indoor environments.

Additional Authors: Johannes Amann, Tobias Baur, Caroline Schultealbert, Christian Bur, Andreas Schütze

Changing the Indoor Air Quality (IAQ) Landscape: New and Emerging Tools and Technologies Can Improve Traditional IAQ Best Practices.

Randy Chapman, US Environmental Protection Agency

Poor indoor air quality (IAQ) is ranked by the United States Environmental Protection Agency(EPA) among the top environmental risks facing the public. It is well established that maintaining good IAQ is an important part of creating healthy and resilient buildings and communities. However, often there is a gap between knowing that IAQ is important and the ability of building operators and occupants to adequately assess the quality of the air in their indoor environments or take appropriate actions to improve IAQ. As part of its mission, EPA works to bridge this gap by developing and providing clear and credible guidance on strategies for improving indoor air quality (IAQ) and health. Traditionally, the strategies for improving IAQ focus on the integration of three IAQ best practices - source control, ventilation and supplemental air cleaning and/or filtration. However, advances in IAQ-related science as well as the increasing development and availability of lower-cost tools and technologies could provide the general public with methods to interpret and assess the benefits of their IAQ-related actions and are potential important complements to traditional IAQ best practices. EPA is therefore working to not only continue improving traditional IAQ strategies -- source control, ventilation and supplemental filtration/air cleaning --but also to advance scientific knowledge about the use of tools such as indoor contaminant and building metrics, indoor building indices and indoor sensor technology as complements to these IAQ best practices. This presentation will elaborate on some of the types of tools that could be used to complement traditional IAQ best practices; how these tools can help improve healthy buildings and communities; and, the knowledge gaps and research needs necessary to fully utilize these newer tools and technologies to improve traditional IAQ best practices.

Additional Authors: Randolph Chapman, Laureen Burton, Jordan Zambrana, Sheila Batka

Investigating Indoor Air Quality in On-Campus Residences Using Low Cost Air Quality Sensors

Ran Zhao, University of Alberta

University and college students spend the majority of their life studying and living on campus. Indoor air quality (IAQ) significantly affects their health, wellbeing, and productivity. Investigations of IAQ in student residences are sparse, with a few previous studies measuring IAQ with research-grade instruments. A major challenge lies in the fact that IAQ is dependent on multiple factors and can vary drastically between buildings and even between individual rooms. The use of a low-cost air sensor network is expected to provide IAQ data not accessible with a traditional measurement approach. My research group has initiated a pilot project to deploy a network of air sensors in the University of Alberta (U of A) campus residences to monitor key IAQ parameters, including particular matter, temperature, RH, and CO2 concentration. The overall objective of this project is to provide insights into IAQs in student residences, which serve as a guide for building ventilation, energy management, and student health. In particular, we aim to compare residences with and without in-suite kitchens, a new air handling system (an HRV), and the presence of potential external sources (e.g., food court). Community engagement and education are also essential missions of this project. Participants were recruited from the campus community, and a home-built sensor is being deployed in their individual residences. The sensor records and send in-situ IAQ data to ThingSpeak, an online data repository. The monitoring is currently underway, and preliminary results and conclusions will be presented.

Additional Authors: Ran Zhao, Mst Rowshon Afroz, Xinyang Guo, Ariel Delorme, Chu-wen Cheng, Ryan Duruisseau-Kuntz

Development of ASTM Standard Test Methods for PM2.5 and CO2 Sensors Used for Indoor Air Quality Measurements

Wilton Mui, South Coast Air Quality Management District

Real-time indoor air quality sensing devices, which are used to provide occupants with information on indoor air pollutant concentrations and concurrently provide data to ventilation systems for improved system operation, have become increasingly popular with consumers and manufacturers. However, the effectiveness of these devices in accurately assessing pollutant concentrations is limited. A variety of pollutants can be measured using sensors, with several being the subjects of future standardization activity. The focus of the two ASTM International standards covered in this presentation is to provide a mechanism for evaluating the performance of PM2.5/CO2 sensors or sensor systems for indoor applications through laboratory-based testing. Under a pass-through grant from the DOE to South Coast AQMD, the Air Quality Sensor Performance Evaluation Center (AQ-SPEC) drafted the test standards, conducted the engineering and scientific work, including laboratory setup and testing, data processing and analysis. Using the established AQ-SPEC laboratory protocol as a starting framework, new comprehensive protocols were developed 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. The performance of indoor air quality sensors was evaluated based on data recovery, intra-sensor variability, accuracy, precision, and correlation to Federal Reference/Equivalent Method instruments. We summarize the requirements and procedures of these test methods.

Additional Authors: Wilton Mui, Xiaobi (Michelle) Kuang, Hang Zhang, Sahil Bhandari, Joe Nebbia, Mike Moore, Raul Dominguez, Jr., Vasileios Papapostolou, Andrea Polidori

Air pollution exposures in rural and urban solid fuel-using households in sub-Saharan Africa

Stephanie Parsons, North Carolina State University (Invited Talk)

Use of solid fuels, including fuel wood and charcoal, for cooking is ubiquitous in both rural and urban settings in sub-Saharan Africa and is associated with extreme air pollutant exposures. Introduction of alternative fuels or cooking devices have often targeted these populations, but few studies have quantified individual cook exposures in urban settings, where other sources of air pollution exposure are more prevalent. Here, we present a synthesis of multiple field studies of personal and household CO and PM2.5exposures collected in urban and rural settings in Malawi and Rwanda and in urban Lusaka, Zambia. Measurements collected include both baseline exposures (in households using traditional fuels/combustion devices) and in households who have adopted alternative stoves (pellet + gasifier stove, improved charcoal stoves). Observations were collected with personal and kitchen monitoring and via surveys describing fuels and stoves in use and household (e.g. cooking location and building materials) and personal characteristics. Fuel selection and seasonality are dominant drivers of exposure variability. For example, in the Malawi study, 99% of rural cooks primarily used biomass and had 2.5 times higher 24-hour average PM2.5exposure (433 ± 484 μg m-3) than urban cooks (172 ± 164 μg m-3), while 82% of urban cooks used charcoal primarily and had CO exposures (7.1 ± 7.0 ppm) almost three times greater than rural cooks (2.7 ± 2.0 ppm). In Lusaka, CO exposures were consistent across both traditional and ‘improved’ charcoal stove users, but substantially reduced for users of pellet+gasifier stoves. CO exposures were driven by evening peaks, presumably from charcoal use for indoor space heating. Lusaka PM exposures showed no consistent variation across stove/fuel types, and were mainly driven by the regional/urban characteristics of PM2.5 concentration. We conclude by summarizing lessons learned from field studies in diverse settings.

Additional Authors: Stephanie Parsons, Cheryl Weyant, Wesley Hayes, Joe Pedit, Pamela Jagger, Andrew Grieshop

Innovative Sensor Technologies

Expanding stationary and mobile PM2.5 measurement capabilities near fires

Ashley Bittner, North Carolina State University (Virtual Presentation)

This study used two easily deployable lower-cost air sensor systems to collect high-frequency measurements of fine particulate matter (PM2.5) near fires. The data collection systems include (1) portable, solar-powered backpack monitors equipped with a PurpleAir (PA-II-SD) sensor and (2) a mobile monitoring system equipped with a pDR-1500 (Thermo Scientific). During a fire event, the backpack monitors are deployed as temporary fixed-site monitors to capture longer-term temporal variation, while the mobile monitoring system is driven along routes to characterize spatial variation upwind and downwind, including higher-concentration smoke pulses near the burn. This monitoring approach was tested during prescribed burns in Tall Timbers, FL and Konza Prairie, KS and during the Monument wildfire in Humboldt County, CA. We present these examples to demonstrate the ability of this monitoring approach to improve the spatiotemporal resolution of PM2.5measurements compared to the nearest regulatory air quality monitoring network site.The results show that both prescribed fire and wildfire plumes can have localized impacts while also contributing to widespread elevated PM2.5 concentrations. Additionally, we explore alternative applications of this technology, including evaluating predictions from commonly used smoke dispersion models, assessing real-time local air quality impacts from the burning of spilled oil, and mobile deployments on all-terrain vehicles and helicopters. 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. The mention of trade names does not constitute endorsement.

Additional Authors: Ashley Bittner, Maiko Arashiro, Amara Holder, Andrew Grieshop, William Mitchell, Brian Gullett

Unmanned Aerial Air Quality measurements: the potential for industrial fire plumes characterization with onboard low-cost sensor measurements.

Brice Berthelot, Ineris

During an industrial fire, potentially hazardous substances may be released outside the site by atmospheric transfer. In order to characterize their potential impact, it is essential to gather data as quickly and reliably as possible on the consequences of the event on the environment and on the population. However, recent events related to the Lubrizol 2019 factory fire in Rouen have highlighted the difficulty in certain circumstances of listing the composition of fumes and knowing the mechanisms that contributed to their emission. These shortcomings taint the numerical models used to prioritize the fallout zones and thus distort the assessment of the impact of the fire on the environment and the health of the populations. The DESIHR project (Drones en Essaims pour la Surveillance de Sites Industriels à Hauts Risques), winner of an French National Research Agency grant, aims to develop innovative tools and methods based on the use of a fleet of autonomous drones to characterize in real-life situations the substances present in a fire plume as well as their emission and propagation conditions. The definition of the control laws governing the behaviour of each drone must in this context be based on real-time air quality information from sensors onboard each mobile vector. In order to limit the mass and the energy cost, low-cost sensors are among others considered in the DESIHR project. The objective of the work presented here is to address the potential for industrial fire plumes characterization with onboard low-cost sensor measurements in complex environments. On the basis of experiments carried out before the start of the project, it will address issues ranging from the selection of pollutants monitored and subsequent sensors to the use of data by drones, through the definition of innovative protocols for qualifying measurement performance based on real air matrices.

Additional Authors: Brice Berthelot, François Guerin, Jason Bardou, Laurent Spinelle, Frederic Guinand, Antoine Level, Etienne Petitprez, Pierre Avanzini, Sofiane Ahmed Ali, Jérôme Cortinovis, Sébastien Le Meur, Samuel Cutullic, Queron Jessica, Marc Durif

A Low-Cost Industrial-Grade Carbon Sensor

David A. Gobeli, Ph.D., Met One Instruments, Inc.

Met One Instruments, Inc. has developed an industrial-grade black carbon “BC” sensor-based upon near-infrared (880 nm) attenuation across filter media upon which ambient PM is being deposited in real-time. This device, known as the “C-12 Carbon Monitor”, can operate for up to two months without user intervention and report “BC” with a sensitivity of less than 10 ng/m3. The base version of the C-12 Carbon Monitor, costing less than $3,000 and using a roll of filter tape allowing more than 1,000 spots, can be augmented with an additional illumination wavelength at 370 nm, and a PM2.5fractionator. The C-12 Carbon Monitor comes standard with a Cloud modem and GPS, and can be set up and report data within 10 minutes. The instrument is lightweight and self-contained in a weatherproof enclosure, meaning that no environmental shelter is required for its operation. In this presentation, we will report on field tests conducted at a near-roadside location with the base version of the C-12 monitor. In it, we demonstrate near-equivalence to our BC-1054 MultiSpectrum carbon monitor since there is no reference standard for BC. We also provide data on a version of the C-12 monitor with an optional UV (370 nm) illumination channel, demonstrating its ability to perform source apportionment computations.

Additional Authors: David A. Gobeli, Ph.D., Jennifer Brown, Leah Miller

A Compact High-Precision Microfluidic Platform for Wearable Sensing of Particulate Matter

Ehsan Ashoori, Michigan State University

Particulate matter (PM) is one the main air pollutants which has been linked with severe health conditions. In a recent study, 10.2 million premature deaths have been attributed to only fossil-fuel component of PM2.5 (PM with diameter smaller than 2.5 µm). Many existing commercial PM monitoring devices are bulky and expensive. They are typically used for stationary or portable weather monitoring stations and hence, they are not appropriate for monitoring PM with high spatiotemporal resolution for everyday use of individuals. In this work, toward implementing a wearable PM monitoring system, we present a microfluidic platform for separating PM into different size bins as well detecting particles with high precision. For separation, we implemented deterministic lateral displacement (DLD) technique where I-shape pillars were built into the microfluidic channel to achieve high separation efficiency. This is a necessary step for the sensing stage where the accuracy of the particle detection is sensitive to the size of particles. The implemented DLD separator showed nearly 100% separation efficiency for a critical separation size of 2.5 µm. For the sensing stage, we examined two different methods to achieve high sensitivity. We first utilized differential capacitive measurement technique to detect the presence of particles in the sample. We designed and fabricated the sensor as well as the readout circuit and were able to detect particle concentration levels that are equivalent to air quality index of 151 which is considered to be “unhealthy” according to the US environmental protection agency. To better detect the PM particles with dielectric constants similar to the microfluidic medium in use, we also developed electrochemical sensor for PM detection. We successfully implemented cyclic voltammetry as well as differential pulse voltammetry techniques. The promising results that these methods produced enable detection of a vast majority of particle pollutants.

Additional Authors: Ehsan Ashoori, Heyu Yin, Sina Parsnejad, Andrew Mason

Detecting toxic metals in ambient particulate matter using a low-cost and near real-time analyzer

Hanyang Li, Air Quality Research Center, University of California Davis

Urban populations are exposed to hazardous air pollutants (HAPs), including the gaseous and particulate matter that are known or suspected to cause cancer or other serious health effects. Airborne trace metals are part of the HAPs, which originate from both natural processes and anthropogenic emissions. XRF and ICP/MS have been traditionally used to measure ambient metal concentrations, but their high cost and lab-intensive analysis make them unsuitable to deploy in vulnerable neighborhoods. To fill this gap, we have developed an inexpensive instrument for measuring the presence of toxic metals in ambient PM (named as Toxic-metal Aerosol Real Time Analyzer, TARTA). TARTA was developed based on the theory of spark-induced breakdown spectroscopy. It consists of an aerosol sampling system, electrical components for spark generation, a delay generator to suppress the initial continuum radiation, and an optical spectrometer for spectra collection. Air pollution is first drawn through an inlet nozzle that deposits particles onto the flat surface of a tungsten electrode. After sampling for a period of time, a spark discharge is generated between the electrodes to form a plasma which atomizes the deposited particles. The emitted light from the plasma is then collected by a spectrometer for the determination of aerosol composition and concentration via machine learning methods. Our laboratory and field testing of TARTA show that the instrument is capable of detecting multiple metal particles in near real-time. Specifically, TARTA has LODs ranging from 0.04 µg m-3 ­at a sampling duration of 30 minutes and a flow rate of 15 l min-1. The measurement uncertainty is estimated to be 19% based on a multivariate calibration model. During the field experiments at Caldecott Tunnel and a rural environment, the relatively high temporal resolution of TARTA enables us to examine the dynamics of metal particle emissions and compare the results against real-time total PM concentration.

Additional Authors: Hanyang Li, Leonardo Mazzei, Christopher Wallis, Anthony Wexler

RADICAL: Developing an electronic sensor for detecting short-lived atmospheric radicals and other gases

Justin Holmes, University College Cork

Atmospheric radicals, particularly hydroxyl and nitrate, are the drivers of chemical processes that determine atmospheric composition and thus influence local and global air quality and climate. However, the detection of these short-lived atmospheric radicals is far from routine, and only a few labs worldwide can accurately measure their concentrations in air. Current techniques for measuring radicals are based on spectroscopic and mass spectrometric methods, which although sensitive and robust, are technically complex, cumbersome and expensive. This presentation will provide an overview, and a discussion of the latest results, from the EU-funded project ‘RADICAL’ which is developing a small, low-cost sensor to electrically detect short-lived atmospheric radicals in real-time. This will be the first gas sensor built from an array of junctionless nanowire transistors, which has proven popular for liquid-based sensors. Although challenging, RADICAL sensors not only have the potential to be rolled out on a global scale but can also be adapted to detect other important atmospheric gases, particularly on short-timescales. The project team welcome ideas and future collaborations for how these sensors might be best applied in real-life environmental monitoring situations. Website:https://radical-air.eu/ [https://radical-air.eu/]

Additional Authors: Justin Holmes, Subhajit Biswas, Vaishali Vardhan, Stig Hellebust, John Wenger, Tamela Maciel

IoT VOC Monitoring with a Fully Autonomous MEMS-based Analyzer


IoT chemical sensing for environmental health & safety is definitely the future. In today’s refineries, petrochemical plants and cities operations, conventional methodologies are costly and do not provide real-time actionable data with compounds specificity for critical intervention. The Omniscent platform offers autonomous real-time molecular level data insight that enables EH&S officers to swiftly intervene and prevent costly disasters. Using advanced MEMS & Cloud technologies, we have developed an autonomous and portable VOC analyzer capable of detecting and speciating more than 20 VOCs in near real-time. The technology is using ambient air as a carrier gas, therefore reducing the consumables requirements to an absolute minimum. Users can view and compare the analysis results with historical data anytime through an internet or cellular connection from anywhere in the world. The MEMS-based micro Gas Chromatograph technology has a small footprint design that lends itself to portability and can be powered from solar panels or other 24V outlets. The data is exhibited in a tabulated format or a chromatogram format and can be correlated in the portal's user interface with Wind Speed and directions measurements for leak source attribution purposes. The complete solution has either LTE and WiFi communication module embedded in the microGC analyzer, which is integrated with an anemometer and the solar panels on a mobile platform for ease of deployment and relocation at the fenceline or for area VOC monitoring.

Additional Authors: Nabil Saad

USEPA Alternative Method 082, Next Generation Air Quality Monitoring, Forget the school and use the tool

Shawn Dolan, Virtual TEchnology LLC

Opacity of Visible emissions has been used to tune and automate combustion sources for hundreds of years, in the late 1800's Ringleman developed the Ringleman scale which has been used to determine the efficiency of burn ever since. Enter the world of Particulate sensors alerting the public of the dangers of particulate, and the concentration they breath at a specific point on the globe. Both technologies very fitting for the task but neither good for communicating health risks.Ringleman can tell us the fuel is being burned efficiently, but not the particulate exposure to the population. Particle monitors are only good for that specific place on earth and 10 feet away is an entirely different exposure profile. The Digital Opacity Compliance System Third Generation (DOCS III), is the bridge to this dilemma, DOCS III uses digital camera technology to record visible emission events from various sources, Deisel generators, Trucks, Ships, Stationary Sources, and fugitive Emissions from farms to process vents. The DOCS III uses patent pending algorithms to determine opacity of the scene, viewed through the digital camera. With a valid opacity DOCS III converts the Opacity value to concentrations of Particulate sizes, e.g. PM 0-3m, PM 3-10 M and PM>10M, additionally the scene is evaluated for flow and flux to speciate the Volatile concentrations in the Opacity. Re applying the original Air Pollution monitoring to todays high technology and we now have the ability to measure Particulate for Wild fire from 5 miles away, the PM from a HD Deisel truck running 65 MPH down the hwy. The Particulate from a dust storm in the open desert, or a mining blast in a quarry. The volatiles from a quenched flare, or the exhaust stacks of a CruiseLine.

Additional Authors: Shawn Dolan

Merging sensor data with other air pollution data sources: methods and benefits

Data fusion for air quality mapping using low-cost sensor observations

Alicia Gressent, INERIS (Invited Talk) (Virtual Presentation)

The recent technological developments and the strong increased interest for public information lead to a fast-growing use of low-cost sensors (LCS) for air quality monitoring. In Nantes (a French city), LCS have been installed in the city center and deployed on driving school cars, ambulances and service vehicles to measure PM concentrations. We used the large amount of PM10 observations provided by the sensors for air quality mapping to show the potential added-value with respect to the dispersion model (ADMS-Urban) calculations (Gressent et al., 2020). Data fusion was performed by combining the preprocessed fixed and mobile LCS observations in November 2018 and the 2016 annual average of the ADMS-Urban outputs. The measurement uncertainty related to the LCS and the dispersion of the data are considered in data fusion that is achieved at hourly resolution. Results show that considering the model alone implies 8% bias whereas including the LCS observations reduces the bias to 2.5%. However, the fusion tends to smooth the PM10 peaks. In addition, the effect of the measurement uncertainty has been investigated by doubling it or reducing it to the reference station measurement uncertainty. The sensitivity study demonstrates that the performance is increasing by reducing the uncertainty. This highlights the importance to estimate accurately the measurement uncertainty of the devices to ensure relevant air quality mapping. The method efficiency is also quite limited by the low correlation between the sensor observations and the model used as external drift in the kriging that may be explained by the remaining bias on LCS data. This has been addressed in Rollin et al., in prep, for PM2.5 by developing a new methodology to identify “rendez-vous” between sensors that relies on a graph approach linking measurements of sensors close to each other’s that allows characterizing and correcting the measurement bias.

Additional Authors: Alicia Gressent, Hugo Rollin, Laure Malherbe, Augustin Colette

Closing the gap between air pollution data sources, tools and end users in LMIC

Beatriz Cardenas, WRI Mexico (Invited Talk)

Air pollution in cities at LMIC is one of the health and environmental most urgent problems to attend. Citizens and government officers are increasingly demanding access to a wide variety of air pollution data sources to attend data and knowledge needs. These range from identification of sources of pollution, ambient concentrations, microenvironment, and personal exposure to information ahead of time of high pollution events to protect the most vulnerable ones and less reactive actions. Using a few case studies that are part of the World Resources Institute (WRI)'s air quality data strategy, we will discuss how local insights can be generated by aggregating global data sources and integrating them with local data. This strategy focuses on translating existing science into decision-relevant, action-provoking, and locally relevant tools. Examples of how different air pollution data sources may be used in combination for specific demands in Latin American cities will include: 1) Development and piloting data and modeling tools that aggregate diverse datasets (satellites, ground monitoring data (from both reference and LCS networks) to provide actionable information in cities. 2) Participatory science processes using portable and small sensors to evaluate the effectiveness of intervening at neighborhood or city scale. 3) Development of bottom-up participatory emissions inventories in which local authorities are key data providers and users. 4) Visualization and auralization tools to see and listen to air pollutants dynamics. Based on these experiences, a reflection on the benefits and barriers will inspire and promote collaborative efforts to close the gap between existing and available air pollution sources, scientific and technical tools and end users (citizens and officers) in cities from LMIC to motivate actions.

Additional Authors:

Air quality forecasting at sub-city-scale by combining models, satellites, and surface measures

Carl Malings, Postdoctoral Program Fellow, NASA GSFC (Invited Talk)

While there are a variety of sources for air quality information, no one source simultaneously allows for high accuracy, low bias, fine spatial resolution, wide spatial coverage, high temporal frequency, and the capability for near-term forecasting of air quality. Global models, like the NASA’s Goddard Earth Observing System - Composition Forecasting (GEOS-CF) model, provide global coverage and forecasting capabilities, but operate at relatively coarse spatial resolution and require ground-truthing with in-situ data. Polar-orbiting satellite data products, like those of the ESA TROPOspheric Monitoring Instrument (TROPOMI), provide higher-spatial-resolution remote sensing of atmospheric composition, but are limited by cloud cover and overpass times and report column-integrated quantities. Surface measurements, both from regulatory-grade monitors and low-cost networks, measure “nose-level” air quality, but may not represent concentration variability across large spatial domains, and (in the case of low-cost sensors) are subject to interference and biases. There exists a great potential to combine these diverse data sources together, using the strengths of some to offset the weaknesses of others to build a more comprehensive picture of air quality. This presentation will summarize results from ongoing efforts to produce such a combined forecast, with application case studies for surface-level Nitrogen Dioxide forecasting in several major US cities. Furthermore, we will examine the relative impacts and benefits of different data sources on the forecasting accuracy at different spatial and temporal scales. Finally, we will examine the potential for integrating low-cost sensors into such a system, both in terms of using these integrated air quality estimates as a baseline from which to calibrate networks of low-cost sensors in the field, and in terms of using dense networks of low-cost sensors to refine the spatial resolution of integrated air quality forecasts.

Additional Authors: Carl Malings, K. Emma Knowland, Christoph Keller, Stephen Cohn

Integrating multi-modal transportation data with low-cost air quality sensor data to improve understanding of traffic-related air pollution

James Hindson, University of British Columbia

In order to quantify air quality on the University of British Columbia (UBC) campus, a network of 8 low-cost air quality sensors were installed in June 2021 at several traffic intersections. The “Remote Air Quality Monitoring Platform” (RAMP) sensors measure PM2.5­, Ozone (O3), Carbon Monoxide (CO), Carbon Dioxide (CO2) and Nitrogen Oxides (NOX­) every 15 seconds. The RAMP sensors are solar-powered, battery-operated and can measure wind speed and direction (with attachments). Each sensor is connected to the Rogers Communications 5G network testbed at UBC, allowing for real-time calibration and reporting of air quality data. Multi-modal transportation sensors have also been installed at some of these intersections that collect anonymized data through the use of video technology, tracking and quantifying cars, trucks, buses, pedestrians and bicyclists (partnership with Numina and Blue City). Large scale, anonymized mobile data has also been analyzed to understand and map local population movements including pedestrians, cars and public transport (partnership with downtown.ai). Here we present our findings on linking transportation sensor traffic as well as mobile location data to pollution concentrations measured by the RAMP sensors. Merging these data stream enabled a greater understanding of the relationship between roadside air pollution levels and traffic count by mode. Exposure levels experienced by pedestrians and cyclists around the chosen intersections can also be explored. Data will be presented for the period June 2021 to October 2021, covering a number of interesting pollution events such as wildfires. At most sensor locations, preliminary results show a clear correlation between vehicular count and CO and NOX levels, with significant contribution from buses and trucks. Results indicate that pedestrian and bicyclist peak counts coincide with peak pollutant levels, suggesting increased exposure levels.

Additional Authors: James Hindson, Surya Dhulipala, Naomi Zimmerman

Supporting timely, high-resolution air quality data availability in Africa by fusing satellite observations of aerosol optical depths, PM2.5 model data, and PM2.5 surface-based measurements

Nathan Pavlovic, Sonoma Technology, Inc

Globally, over 4 million premature deaths per year can be attributed to fine particulate matter (PM2.5) pollution, the majority of which occur in lower- and middle-income countries. The availability of timely, accurate, and high-spatial resolution PM2.5 data can increase awareness of global air quality conditions and help protect public health. However, in many parts of the world, there are a limited number of ground-based air quality monitors. Satellite observations offer global coverage and have the potential to provide a comprehensive view of PM2.5 in near-real time when combined with data from ground-level monitors and air quality models. In this work, we assess the viability of an uncertainty-weighted data fusion approach to provide high-resolution PM2.5 concentration data in Africa. We estimate ground-level PM2.5 concentrations using aerosol optical depth (AOD) retrievals from the Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm applied to NASA Moderate Resolution Imaging Spectroradiometer (MODIS) observations in combination with Copernicus Atmosphere Monitoring Service (CAMS) near-real time air quality model data. Using ground-level PM2.5 data, we fused satellite-based estimates with observations and assessed the performance of the application. Fusion of surface data with satellite-derived PM2.5 shows improvements in agreement with independent observations compared with satellite-derived PM2.5 alone, where tested. Based on the results presented here, we plan to use Google Earth Engine to implement the data fusion approach and develop daily real-time PM2.5 concentrations over Africa. These results can support further efforts to evaluate and enhance the use of satellite data to estimate ground-level PM2.5. Using the system we have developed as a foundation, data fusion harnessing growing monitoring networks across Africa, including low-cost sensors, presents the opportunity to fill the air quality monitoring gap.

Additional Authors: Nathan Pavlovic, Sean Khan, Brian Sullivan

Integration of Air Quality Sensor Data into the South Coast AQMD Real-Time Air Quality Index Map

Nico Schulte, South Coast Air Quality Management District

Despite significant progress, air quality in the South Coast Air Quality Management District (South Coast AQMD) remains among the worst in the nation. South Coast AQMD provides a real-time 5 km resolution air quality index (AQI) map for 17 million residents in a region encompassing major portions of Los Angeles, Orange, Riverside, and San Bernardino counties in Southern California. This AQI map, deployed in late 2020, combines regulatory-grade monitor data, consumer-grade sensor data (i.e., PurpleAir PA-II), and data from the NOAA National Air Quality Forecast Capability, a chemical transport model that uses a bias correction algorithm to improve accuracy. The model and sensor data supplement the regulatory data by filling in the gaps between regulatory monitors in the region, where South Coast AQMD operates a network of 42 air monitoring stations. However, data gaps still exist in areas where ozone and PM2.5concentrations routinely reach unhealthy levels. To improve the accuracy of the AQI map in between the monitors and existing sensors, we worked with volunteers (members of the public) to host additional PurpleAir PA-II and also Aeroqual AQY sensors at strategic locations throughout the South Coast AQMD jurisdiction and integrated data from these additional sensor networks into a revised version of the AQI map. Sensor data is corrected using remote calibration techniques (i.e., MOment MAtching (MOMA) calibration and network management framework) which are used in real-time to correct the sensor measurement data before integration into the AQI map. Collocated sensor measurements at regulatory monitoring stations are used to estimate measurement uncertainty, which is used to weigh sensor data through the residual Kriging method. We will demonstrate the improved spatial coverage in the AQI map as we aim to provide accurate AQI data at a neighborhood scale.

Additional Authors: Nico Schulte, Ashley Collier-Oxandale, Kyrstin Fornace, Brandon Feenstra, Vasileios Papapostolou, Scott Epstein

The AirHeritage Hierarchical Network: Sensing, Calibration, Deployment strategies for fixed, mobile air quality monitoring and modeling in urban scapes.

Saverio De Vito, ENEA

The AirHeritage project aims to improve citizens and administrators AQ knowledge in small and medium cities in Italy, empowering awareness, behavioral change and adherence to participated remediaton policies. Funded by EU Urban Innovative Action framework, the project involved the city of Portici, located 7km south of Naples and partners from academia, research agencies, environmental protection, associations and SMEs. It deployed a hierarchical network relying of fixed and mobile stations, involving regulatory grade analyzers and low cost sensors to map citizens exposure through crowdsensing and long term fixed monitoring. Concentration of NO2, O3, CO, PM1 – 2.5 – 10. Monitored data are shared and integrated in ultra high resolution (20m) exposure maps. Data from fixed stations, both low cost and regulatory grade, are assimilated in 3D chemical transport model using weather data and 3D urban scape data for now- and forecasting purposes. 7 Fixed and 30 Mobile MONICA™ platforms, relying on electrochemical sensors, are calibrated twice a year and used in 4 opportunistic measurement campaigns. Platforms are co-located with local regulatory grade analyzer for at least 3 weeks each. Colocation data driven calibration algorithms are derived, validated and executed at the edge on citizens smartphone through an Android App resulting in real time exposure estimations. In each session, routes are color coded using European Air Quality Index allowing to assess exposure on the move and selecting low exposure paths. The same result is available, using modeled air quality maps, to citizens which do not received MONICA analyzers. Shared data are fused through IDW and by computing measurements median on a predefined grid discarding low populated cells. This contribution will show the results of 6 colocation driven calibrations and 4 mobile crowdsensing campaigns in the project, exploring architectural (Sensors and IoT management), accuracy, engagement and communication issues.

Additional Authors: Saverio De Vito, Grazia Fattoruso, Sergio Ferlito, Girolamo Di Francia, Paolo D' Auria, Roberta Gianfreda, Fabrizio Carteni

Using Crowd-Sourced Low-Cost Sensors in a Land Use Regression of PM2.5 in 6 US Cities

Tianjun Lu, California State University, Dominguez Hills

Assessing exposure to ambient Fine Particulate Matter (PM2.5) is important for improving human health. Low-cost sensor networks, located by a variety of users (i.e., crowd sourcing), are expanding rapidly on a global scale thereby increasing measurement density and coverage for land use regression (LUR) models. Few studies have developed methods (based on hygroscopic growth factors) to integrate low-cost sensors into LUR models across multiple cities, limiting the ability of modelers to fully utilize growing low-cost sensor networks worldwide. We developed five LUR models to predict annual average PM2.5concentrations using combinations of regulatory (six-city: n = 68; national: n = 757) and low-cost monitors (n = 149) from six US cities. We found that developing hybrid LURs that include the low-cost (i.e., PurpleAir) network may better capture within-city variation. LURs with the 6-city PurpleAir data only (10-fold CV R2 = 0.66, MAE = 2.01 µg/m3) performed slightly worse than a conventional LUR based on the 6-city regulatory data (10-fold CV R2 = 0.67, MAE = 0.99 µg/m3). Hybrid models that included both 6-city low-cost and national regulatory data performed similarly to existing national models that rely on regulatory data (Hybrid models: 10-fold CV R2 = 0.85, MAE = 1.02 µg/m3; regulatory monitor models: R2 = 0.83, MAE = 0.72 µg/m3). Integrating crowd-sourced low-cost sensor networks in LUR models has promising applications to help identify intra-city exposure patterns especially for rural areas and low and middle income countries where high-grade regulatory monitor networks are sparse or do not exist.

Additional Authors: Tianjun Lu, Matthew Bechle, Albert Presto, Steve Hankey

Mobile Monitoring/Monitoring Mobile Sources

Opportunistic mobile air quality mapping using service fleet vehicles: from point clouds to actionable insights

Jelle Hofman, Flemish Institute for Technological Research (VITO)

Within the framework of the VLAIO City of Things program, IMEC The Netherlands deployed mobile Kunak® sensor systems on a postal service fleet, collecting high spatiotemporal air quality data (NO2, O3 and PM1, PM2.5and PM10) “at every doorstep” in the city of Antwerp, Belgium. Before deployment, all 20 sensor systems were co-located next to a regulatory monitoring station for validation and calibration purposes. After deployment, they collected opportunistic data, at 10 second resolution during daytime and 5 minute resolution (when parked/static) during nighttime, from March till September, 2021. Data validation and monitoring was foreseen via the Kunak Cloud. Three devices remained next to the air quality monitoring station to evaluate the sensor performance (precision, accuracy, correlation) over time. Calibration resulted in good sensor performance (R²=0.85-0.93, MAE = 3.35-4.35) up to supplementary (<25% expanded uncertainty) data quality levels. When mounted in their mobile housing but not moving, their correlation slightly lowered (R²=0.77-0.88, MAE = 3.29-5.21) due to the lower ventilation rate. Mobile measurements (e.g. n=2080067 for June 2021) were aggregated at street segment level and evaluated in terms of spatial coverage, range, temporal dynamics and representativity with regard to reference measurements. After ~6 months of operation, the sensor performance had lowered but remained acceptable for 2 out of 3 devices (R²=0.5-0.71, MAE=7.78-13.64), nevertheless indicating the need for recalibration. This mobile sensor testbed enabled the collection of real-time fine-grained air quality data to gain a better understanding of spatiotemporal pollutant dynamics and exhibited range in urban air pollution exposure. Proper data aggregation methods, performance validation and coverage thresholds are needed to draw representative pollutant maps from the collected mobile point measurements.

Additional Authors: Jelle Hofman, Valerio Panzica La Manna, Edurne Ibarrola-Ulzurrun, Miguel Escribano Hierro, Martine Van Poppel

Mobile air sensing to detect PM2.5 hot spots in Houston, Texas

Tim Dye, TD Environmental Services

Mobile air monitoring is a new and innovative way to collect air quality data over a large area. Like stationary air monitoring, the size, sophistication, and cost of mobile monitoring technology has been decreasing over the last five years. The City of Houston and the Environmental Defense Fund (EDF) have been at the forefront of this data collection method. In 2017, EDF used two Google Street View cars with reference instruments to map hyperlocal air quality [https://www.edf.org/airqualitymaps/houston] across 35 neighborhoods. Then in 2018, EDF partnered with the City of Houston to deploy roof-mounted, mid-cost instruments on two city vehicles that measure PM2.5and Black Carbon, testing operational viability of this type of system. In 2021, building on the successful 2018 pilot, the City of Houston deployed ten low-cost mobile sensor units on city vehicles that measure PM2.5, NO2, CO, and CO2. To maximize spatial coverage with minimal use of additional staff time, 10 mobile units were deployed on municipal vehicles to collect PM2.5data continuously as staff drive predominantly “business as usual” routes to collect baseline concentrations with broad geographic coverage, supplemented by occasional targeted driving around sources of interest to improve accuracy of baseline calculations in those locations. This project is currently collecting data to determine PM2.5 hot spots within the city and improve spatial coverage for air quality data. Data will be used with a control chart to determine where additional effort (monitoring, investigation, enforcement) should take place. This talk will present the technology used, present lessons learned, and discuss several case studies from this project.

Additional Authors: Tim Dye, Dave Bush, Loren Hopkins, Maia Draper, David Yoho, Story Schwantes, Randy Baxter

Air Quality Sensors Deployed on Mobile Platforms: A Performance Evaluation Protocol and Recent Advances

Wilton Mui, South Coast Air Quality Management District

Performance evaluation studies for air quality sensors collecting stationary measurements have been conducted by various academic and government bodies, and current efforts by international standards organizations may lead to convergence of these methods. In contrast, such performance studies and methods are nonexistent for sensors used in mobile deployments, even though sensors are being used in this manner by academics, government agencies, non-governmental organizations, and citizen scientists. Nonstationary measurements with air quality sensors are a relatively nascent but growing use-case, and questions of appropriateness and data quality will become increasingly important. The South Coast Air Quality Management District’s Air Quality Sensor Performance Evaluation Center (AQ-SPEC) has developed a novel evaluation protocol in which air sensors are compared to reference- or research-grade instruments while deployed on a mobile platform. Air sensors are assessed in testing phases of various degrees of environmental control, ranging from placement in a controlled-flow sampling duct to unsheltered mounting on a vehicle rooftop. These evaluations probe the performance of air sensors in mobile monitoring setups that may be appropriate for community members to carry out at the neighborhood level. The testing procedures aim to quantify the performance of air sensors and the effects of sensor siting, orientation, and vehicle velocity, which can provide guidance to users on appropriate sensors and configurations for their use-case. Recent advances in the design of a new mobile platform are also discussed.

Additional Authors: Wilton Mui, Berj Der Boghossian, Ashley Collier-Oxandale, Steve Boddeker, Jason Low, Vasileios Papapostolou, Andrea Polidori

High resolution mapping of on-road air pollution using a large taxi-based mobile sensor network in Shanghai

Yuxi Sun, Hong Kong University of Science and Technology (Virtual Presentation)

Severe urban air pollution caused by rapid urbanization has become a major environmental and social problem around the world. Traffic emissions are major sources of urban air pollution in most of megacities. To explore heterogeneity in concentrations of traffic-related air pollution (TRAP) in city area, we established a mobile sensor network on 150 taxis in Shanghai, China. Each taxi was equipped with an air monitoring device to capture spatial and temporal patterns of CO, NO2, and PM2.5concentrations from November 2019 to December 2020. Through our regular quality assurance and quality control campaign of devices, 15,045,137 intermittent data points during 3647.7 hours were collected and used in this study period. Road coverage and pollutant concentration spatial distribution were better identified and quantified. Since the sampling stage spanned the Coronavirus Disease of 2019 (COVID-19) outbreak, the average concentrations of CO, NO2, and PM2.5in Shanghai decreased by 33.59%, 33.83% and 45.55% during the lockdown period compared to the before period, respectively. Meanwhile, the local contribution related to traffic emissions changed slightly before and after the epidemic period, and the background contribution changes may relate to the seasonal variation. Such real-time air monitoring systems can be a valuable tool for policymakers and environmental protection agencies to implement effective policies for the future development of cities. Besides, our analysis provides new insights into changes in urban air quality, particularly during the COVID-19 epidemic period. In the following research, we will combine real-time pollutant concentration data measured by mobile devices with traffic emission inventories, urban geographical features, meteorological information, and other factors to study the horizontal relationship between them and establish relevant models, which can be applied to air pollution prediction on other fields.

Additional Authors: Yuxi Sun, Zhi Ning, Peng Wei, Peter Brimblecombe

Performance targets for air quality sensors

ASTM Standards for the Performance Evaluation of Outdoor Air Quality Sensors

Geoff Henshaw, Aeroqual Ltd (Invited Talk)

Over the past ten years, the availability of lower priced air quality sensors has encouraged innovative measurement of air quality by a wide range of users. There has also been a significant effort to understand how to use the technology and this has identified that data quality from these devices can be variable. Over the past few years several entities have undertaken the development of methods to objectively evaluate air sensors and create criteria for performance-based grades. This presentation describes the work undertaken by ASTM International, a consensus based standards organisation, to develop standards that will support users of these devices. ASTM established a working group in 2018 to develop a standard for the evaluation of outdoor air sensors focused on the criteria pollutants: CO, SO2, NO2, O3, PM2.5 and PM10. The standard has evolved to include both laboratory and field tests that provide information on instrument repeatability, sensitivity, linearity, cross- interferences, drift, and comparability with Federal Equivalent Methods. The metrics selected, issues resolved to achieve consensus and the compatibility with the recent indoor PM2.5 standard ASTM D8405 will be discussed. Test examples will be presented to illustrate how different sensors perform. ASTM also began work in 2020 on a specification for outdoor air sensors that would create grades of performance based on the standard test practice. The key criteria and their rationale will be presented and compared and contrasted with approaches by other organisations.

Additional Authors: Geoff Henshaw

Is PM sensor testing really testing the sensors? Experiences from 400 days of field tests in the Life VAQUUMS project.

Jordy Vercauteren, Flemish Environment Agencey (Invited Talk) (Virtual Presentation)

We often get the question what the best low-cost PM-sensor is, or whether a certain PM-sensor is 'good enough' to use for certain applications. But what does 'the best' or 'good enough' really mean?During the Life VAQUUMS project we tried to answers these questions. Six different sensor types (5 units per type) were compared with the EU gravimetric reference and an automated equivalent method (Palas Fidas 200). The sensors were: + Honeywell HPMA 115S0 + Dylos DC1700 + Nova Fitness SDS011 + Plantower PMS7003 + Winsen ZH03B + Shinyei PPD60PV In general sensors showed acceptable to good correlation for PM2.5(R² between 0.62 and 0.84). Due to the high proportion of PM2.5 in PM10 some sensors did show some correlation for PM10, but this could be considered artificial since there was poor to non-existing correlation for the PMcoarse fraction (=PM10- PM2.5). Apart from correlation we also looked at: + the importance of (human expert) data validation; + the effect of additional calibration; + the between sensor-uncertainty; + the effect of relative humidity on the sensor/reference ratio + the uncertainty at the limit value when compared to the reference methods Additional analysis of SDS011 sensors co-located with Fidas Palas 200 monitors at 8 different sites in Flanders showed that other locations can give less favourable results than the urban test site in Borgerhout, Antwerp. These differences could be linked to more frequent episodes of high relative humidity at other locations. The presence of vegetation close to the monitoring sites appears to play a role which indicates that test results will often bedependent on time and location of the test and that a thorough understanding of aerosols in the atmosphere and the limitations of the sensors is vital for the applications of PM-sensors. Life VAQUUMS website:https://vaquums.eu/ [https://vaquums.eu/]

Additional Authors: Jordy Vercauteren

A French certification scheme for the evaluation of sensor systems dedicated to the ambient air quality monitoring.

Laurent Spinelle, Ineris (Invited Talk) (Virtual Presentation)

The continuous interest for sensor systems dedicated to the air quality monitoring led the French national agencies (Ministry in charge of the environment, the French National Reference Laboratory for monitoring air quality (LCSQA) members and the regional monitoring networks (AASQA)) to study the reliability of these new devices. However, there is currently no national or European normative framework regulating their uses or giving the guidelines to evaluate their performances against reference measurement systems. Two members of the LCSQA (Ineris & LNE) joined forces to establish a voluntary certification scheme called Air Quality Sensor that focused on the evaluation of the metrological performances of sensor systems for both gaseous and particulate matter. This evaluation, based on the European standard drafts of the working group on sensor systems for air quality monitoring (CEN TC264/WG42), is divided in two steps: This evaluation focuses on fixed measurement performed with a stationary sensor system. In a first stage, this voluntary evaluation is aimed at measuring PM2.5 and NO2 and will be further extended to other pollutants such as O3 and PM10. At the end of the two parts of the metrological assessment and after an audit of the devices production, a performance division is assigned to each pollutant measured by the sensor system, based on specific metrological criteria described in the protocol for evaluating sensor systems for ambient air quality monitoring at fixed point (MO-1347, Evaluation Protocol for Sensor Systems for Ambient Air Quality Monitoring at Fixed Site, https://prestations.ineris.fr/en/certification/certification-sensors-system-air-quality-monitoring).

Additional Authors: Laurent Spinelle, Caroline Marchand, Amandine Fievet, Tatiana Macé, Marguerita El Boustani, Marc Durif, Dominique Charpentier

Performance evaluation of sensors for gaseous pollutants and particulate matter in ambient air: status of European standardization

Martine Van Poppel, VITO (Invited Talk) (Virtual Presentation)

European standardization of test protocols to evaluate performance characteristics of sensor systems is needed and is an important step to include sensor system measurements into the monitoring of air quality for regulatory and non-regulatory purposes. TC264 WG42 (Working Group 42 of Technical Committee 264 on Air Quality of the European Committee for Standardization) has been working on a protocol for the evaluation of sensor systems for air quality monitoring. The protocol defines procedures and requirements for the evaluation of sensor systems. It applies to sensor systems as individual measurement devices for outdoor measurements at fixed sites. The procedure evaluates if the measurement uncertainty defined in Directive 2008/50/EC as Data Quality Objective (DQO) for indicative measurements and for objective estimation is met. The protocol includes also a less demanding performance for non-regulatory sensor measurements. This results in three sensor performance classes. The protocol includes pollutants regulated under Directive 2008/50/EC (PM10, PM2.5, O3, NO, NO2, CO, SO2 and benzene), and also guidance is given for CO2. The protocol for gases includes lab and field tests. Two routes are possible: extended lab tests and short field test, or only extended field tests. The protocol defines concentration levels and interferences to be tested. It defines also the number and types of field sites, seasons and concentration levels. Classification of sensor systems is based on performance requirements and DQO. DQO is calculated as expanded uncertainty at Limit Value. Performance requirements include response time, lack of fit, repeatability, Limit of Detection, between sensor uncertainty and data capture. The protocol for PM sensors is currently being developed. A lab test has been introduced to evaluate if sensor systems can measure also coarse PM. This presentation will discuss the current procedure and open issues and compare to other standardization initiatives.

Additional Authors: Martine Van Poppel

Highlights on U.S. EPA Efforts on Developing Performance Testing Protocols and Targets for Air Sensors

Rachelle Duvall, U.S. EPA (Invited Talk)

As the development and use of air sensors continues to expand rapidly, understanding the performance of sensors remains a critical need given the variability in sensor data quality. To help support consumers and developers of air sensor technologies, the U.S. Environmental Protection Agency (U.S. EPA) published reports in 2021 outlining recommended testing protocols, metrics, and target values to evaluate the performance of ozone (O3) and fine particulate matter (PM2.5) air sensors used in non-regulatory supplemental and informational monitoring applications. The U.S. EPA is expanding these efforts to cover additional pollutants including particles with diameters of 10 microns or less (PM10), nitrogen dioxide (NO2), carbon monoxide (CO), and sulfur dioxide (SO2). The goal of this work is to provide a consistent approach for evaluating sensor performance while also helping provide confidence in sensor data quality, encouraging technology improvements and development in the marketplace, and assisting users select appropriate sensors for their application of interest. This presentation will provide highlights on the U.S. EPA’s completed and future work in developing guidance for evaluating the performance of air sensors. Disclaimer: Although this abstract was reviewed by EPA and approved for presentation, it may not necessarily reflect official Agency policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

Additional Authors: Rachelle Duvall, Andrea Clements, Karoline Barkjohn, Samuel Frederick

AQ-SPEC: Our Transition to the Latest Sensor Testing Protocols and Standards

Vasileios Papapostolou, South Coast Air Quality Management District (Invited Talk)

Performance standards and certifications for air quality sensors (AQS) will be increasingly important to manufacturers for product marketability, as well as to consumers for making purchasing decisions. Governmental entities can play important roles as testing and certification organizations that are authoritative, independent, and objective. The South Coast AQMD Air Quality Sensor Performance Evaluation Center (AQ-SPEC) is a governmental program with a global reputation for AQS performance evaluation due to its established field and laboratory testing protocols that have been used since 2015 to evaluate over 200 AQS models. AQ-SPEC is poised to transition into a new phase of sensor performance standards and protocols with its sophisticated field and laboratory testing capabilities that include South Coast AQMD operated air monitoring stations and two state-of-the-art test chamber systems integrated with an array of reference instruments. In this presentation, we will discuss select testing procedures, protocols, and standards for evaluating AQS and how AQ-SPEC has started evaluating sensors based on these new standards. This talk will focus on the recently approved ASTM International test standards (PM2.5 and CO2 sensors – used for indoor air quality applications) and the EPA performance testing protocols and target values (PM2.5­­ and O3 sensors – used for ambient applications). AQ-SPEC has the experience and capabilities of executing the ASTM test standards along with both the “base” field and “enhanced” laboratory evaluation components of the EPA performance testing protocols. Hear what is next for AQ-SPEC as we transition towards implementing the latest standard testing procedures and protocols and how this will impact AQS performance evaluation.

Additional Authors: Wilton Mui, Michelle Kuang, David Herman, Berj Der Boghossian, Randy Lam, Ashley Collier-Oxandale, Brandon Feenstra, Jason Low, Vasileios Papapostolou, Andrea Polidori

Using International Standards to prove the performance of low-cost sensors - the regulatory perspective

Richard Gould, Environment Agency

Reliable data is critical for compliance assessment and regulation. Under European legislation, data quality is specified through Data Quality Objectives (DQOs) and measurement uncertainty at the limit value, expressed as a 95% confidence interval. European legislation for air quality specifies maximum, allowable uncertainties for both reference monitoring instruments and for indicative instruments (i.e. low-cost sensors). Reference instruments have to meet the requirements for performance specified in European Standards, which specify performance criteria and test procedures. These Standards ensure that reference instruments meet the required uncertainty of measurement. Additionally, legislation states that such reference instruments must be type-approved. Until now, there has been an absence of equivalent performance standards and test procedures for indicative instruments. However, the European Standards Body, CEN, has been developing European Standards for indicative instruments measuring gases and particulate matter. These two developing standards have been produced to fill the gaps to test, validate and certify indicative sensors. This presentation describes how regulatory bodies will be able to use the two, new European Standards for low-cost sensors to assure the performance of such indicative-monitors, and how this will benefit manufacturers of low-cost sensors, the organisations which use them, and importantly, how certified low-cost sensors will be vital in both measuring and improving ambient air-quality, and how these sensors can be used in synergy with reference instruments and modelling.

Additional Authors: Richard Gould

What is the Impact of Common Sources of Error on Air Quality LCS Measurements Performance? A Practical Guide

Sebastian Diez, University of York

Accurately measuring atmospheric pollutants is critical for decision-making and designing policies aiming to improve air quality and reduce human health exposure. The advent of inexpensive sensor-based technologies means that there are now a growing number of measuring devices available that could be useful for this purpose. However, not all this range available today will necessarily be adequate to the problem that the user wants to tackle. It is therefore key to ask whether the information provided by instrument "X" is appropriate for the intended purpose. Clearly focusing on the question to be answered, and defining the required data quality accordingly, is key to identifying those instruments/techniques capable of meeting this requirement. Since the measurement uncertainty ultimately determines the information content, it is therefore critical to estimate this parameter in a robust and transparent way, thus allowing the potential application of the instrument in question to be defined. In this work, we explore the nature of common air pollution measurement sources of errors in the real world and the implications they have for traditional uncertainty metrics and other potentially more insightful approaches to assessing measurement uncertainty. For this, we employed first simulated datasets combining different sources of error/interferences from (i) a non-target chemical, (ii) physical parameters and (iii) electromagnetic fields. Then we study real-world data from the QUANT project which involves a range of LCS technologies and multiple reference instruments in 3 urban sites in the UK. We then use this information to explore the performance of these technologies and develop methods that will enable their integration into the air quality monitoring infrastructure and use in atmospheric chemistry research. This work will ultimately make it possible to optimize the applications of these technologies based on the quality of the LCS data.

Additional Authors: Sebastian Diez, Pete Edwards, Stuart Lacy

Sensor Networks: From nuts and bolts to real-world impacts

Hyper-Local Air Quality Sensor Network in the Town of Cheverly, MD

Karen Moe, Green Infrastructure Committee (Invited Talk)

Concerned about air quality in its immediate vicinity, the Town of Cheverly partnered with the Community Engagement, Environmental Justice, and Health Laboratory, Maryland Institute for Applied Environmental Health at the University of Maryland to create a hyper-local air monitoring network of sensors. Results are needed to provide a baseline for gauging the immediate and cumulative effects of additional industrial development and road traffic. Equally important, is the need to address community concerns regarding pubic health impacts of local air pollution in industrial areas with heavy truck traffic, and provide guidance regarding exposure risks. By April 2021, a network of 21 commercial PurpleAir PA-II-SD air quality sensors were installed in the town (an area of 1.32 sq. miles). In July, the Maryland Dept. of Environment joined the partnership to pilot a local air monitoring project to conduct the Cheverly Targeted Inspection Initiative. The steps taken to implement the network working with host volunteers, and engaging the community to address growing concerns about cumulative impacts of air pollution on public health are discussed. How to present meaningful data is a key goal of the partnership moving forward from data collection to providing actionable air quality information for Town managers and for residents.

Additional Authors: Karen Moe

Community Monitoring: Using Citizen Science, Technical Expertise, & Lived Experiences for Real World Impacts

Luis Olmedo, Comite Civico del Valle (Invited Talk)

Good air quality is a fundamental right that everyone should have. But that is not the case, especially throughout communities of color & low-income neighborhoods where industry pollutes because it’s cheaper for them to do so, the environment is not well taken care of, and regulators have struggled to fulfill their role to improve air quality. Because of this inaction, many community groups such as ours, Comite Civico del Valle, have initiated local community air-quality monitoring networks (CAMNs) to fill data gaps the government has & provide a public service to disadvantaged communities. Our IVAN Air CAMN was established using the principles of citizen science, participatory community input, technical expertise, and lived experiences from community advisors. In the last half-decade, the network has collected data and pushed for action, either through legislation (California’s AB617) or community programs (Air Notification Program). As the principles are what make the IVAN network a valuable community resource, our technological approach is open to innovation, with methane monitors now in the network and more innovation on the horizon. The network will soon deploy mobile monitoring & maintenance operations with a Zero Emissions Vehicle, pushing towards reducing emissions as we manage our more than 60 stationary particulate matter monitoring sites throughout the Imperial and Eastern Coachella Valleys. The approach has led us to create one of the largest and sustainable community monitoring networks in California.

Additional Authors: Luis Olmedo

Air pollution monitoring in Vietnam with low-cost sensor network

An Le, Vietnam Germany University

Air quality is declining in Vietnam due to human activities such as transportation, industry, agriculture. The air quality information collection has been addressed as a crucial task as it can trace the pollution sources. In this work, we propose an air quality monitoring system for tracing transportation activities. The device placed on a bus measures PM2.5 and associate the GPS data of the bus locations. The data can show us the insight of the road pollution. Combine with the fixed air quality monitoring station, sources of pollution can be pointed out. The tracking of the transportation activities can be implemented with a vehicle tracking system. In this work, the background subtraction is first used as a detector for vehicle detection, and Kalman Filter and Hungarian methods are used for tracking and ID assignment for detected objects. In addition, beside the traditional method, background subtraction, deep learning methods for object detection are also used as the vehicle detector for the system, and compared with background subtraction method as the baseline model.

Additional Authors: An Le, Nhat Le, Hien Vo, Huy-Dung Han

Using a Remote Calibration Technique to Improve Data Quality for Large Networks of Particulate Matter Sensors

Ashley Collier-Oxandale, South Coast Air Quality Management District

A continual challenge in the long-term deployment and management of large sensor networks lie with data quality and calibration. For example, calibration procedures may require physical co-locations – limiting the scale of deployments. Alternatively, global correction equations may not account for changes driven by PM source type, seasonal influences, or drift. Using remote calibration techniques may offer an alternative approach that supports the deployment and maintenance of large sensor networks. Here we applied a technique, entitled MOMA (MOment MAtching), to PM2.5data from PurpleAir PA-II sensors. MOMA was developed by Aeroqual Ltd and is currently being explored under a collaboration between the South Coast Air Quality Management District's Air Quality Sensor Performance Evaluation Center (AQ-SPEC) Program and Aeroqual Ltd. Under this collaboration, the MOMA approach is being piloted with different sensor networks to determine its suitability as a sensor agnostic tool for improving data quality and enhancing the benefits of large-scale networks. This approach involves identifying suitable proxy sites from regulatory monitoring networks. Then, at regular intervals, or when drift is detected via a drift detection algorithm, appropriate calibration periods are determined and used to calculate gains and offsets for each sensor, allowing sensors to remain in the field through calibration. In this presentation, we will examine the efficacy of this approach for sensors co-located at regulatory air monitoring stations and deployed over multiple years (2017-2020), in addition to sharing interesting observations from corrected data. Different failure modes and how well this approach addresses these cases will also be discussed. In terms of implementation, we will illustrate how an open-source R package, AirSensor, can be leveraged to streamline data access and Quality Control (QC) processing, thus enabling the use of this calibration technique with sensor networks.

Additional Authors: Ashley Collier-Oxandale, Brandon Feenstra, Randy Lam, Lena Weissert, Geoff Henshaw, Vasileios Papapostolou, Andrea Polidori

Calibration of citizen sensor networks using a mobile air monitoring platform

Brian LaFranchi, Aclima Inc.

Citizen air sensors are typically low cost, easy to install, and can deliver valuable data at high temporal resolution from many locations at once. However, generating high quality data from these air sensors can be challenging. Sensor accuracy and precision is often unclear out-of-the-box and can also degrade over time. Even with collocation alongside reference instrumentation prior to deployment, sensor performance is strongly dependent on the conditions during collocation and after deployment. This presentation will focus on results from a field study conducted by scientists and operational staff at Aclima and the California Air Resources Board in Sacramento, CA between October 2020 and February 2021 involving a heterogeneous network of regulatory stationary sites, mobile monitoring, and a dense network of various types of air sensors. The study demonstrates the feasibility of using a mobile air monitoring platform as a calibration transfer method between a single stationary reference site and a network of air sensors measuring particulate matter (PM2.5), nitrogen dioxide (NO2), and/or ozone (O3). The presentation will incorporate the results and lessons learned from the study towards development of a methodology that is broadly applicable to a wide variety of stationary sensor networks.

Additional Authors: Brian LaFranchi, David Ridley, Cassandra Trickett, Marek Kwasnica, Matthew Chow, Melissa Lunden, Kathryn George, Jeremy Smith, Matthew Vona

Operationalizing air sensor data for EH&S at the nation’s second-largest school district

Carlos Torres, Los Angeles Unified School District

The increased frequency and intensity of wildfires pose risks for sensitive groups such as children at schools. Exposure to outdoor air pollution can inhibit cognitive development and negatively impact academic achievement — studies have found air pollution levels to be a statistically significant predictor of student performance. Access to real-time air quality data is critical in order for schools and school district staff to make informed decisions that protect the health and safety of our students and school communities. In 2020, the Los Angeles Unified School District (LAUSD) completed a pilot study with 22 air quality sensors placed at schools most impacted by episodes of poor air quality. As of December 2021, LAUSD has procured and installed more than 200 air quality sensors to provide a uniform network across their entire school district. The district was divided into grid squares of approximately 1.6 miles, with a sensor placed in each grid — resulting in one of the largest and densest air sensor networks in the world. This first-of-its-kind network provides real-time monitoring across more than 200 school campuses, yielding reliable, local air quality data for over 1,000 schools and offices and 700,000 students. The sensors will be used during emergency events such as wildfires and poor air quality episodes — which occur frequently in Los Angeles — to determine what actions need to be taken at the school level. Potential actions include keeping students indoors, restricting outdoor activities, attending to students with asthma or other respiratory/cardiovascular illness — and even school closures in the most extreme conditions. Presenters from LAUSD, the Coalition for Clean Air, and Clarity Movement will discuss how this air sensor network has allowed LAUSD to operationalize air quality in their school operations, as well as the intriguing possibilities this network holds for research related to air quality and health in Los Angeles.

Additional Authors: Carlos Torres, Dr. Jennifer Lentz, Katie Moore

Investigating Use of Low-Cost Sensors to Increase Accuracy and Equity of Real-Time Air Quality Information

Ellen Considine, PhD Student, Department of Biostatistics, Harvard T.H. Chan School of Public Health

Reference monitoring networks are sparse. Low cost sensors (LCS) are increasingly being used by the public to fill the gaps in air quality monitoring. In this project, we investigate different factors that influence the accuracy of information when individuals interpret measurements from the nearest LCS or monitor as their air pollution exposure. This has implications for the integration of LCS data into real-time air quality reporting platforms. Three main components of error (between what is reported and experienced) in real-time air quality reporting are distance to the nearest monitor or sensor, local variability in air quality, and sensor (device) measurement error. We examine these components under various hypothetical LCS placement strategies using simulations based on data from California. For these simulations, we assume that the “true” PM2.5 in each 1x1km grid is obtained from an existing PM2.5 prediction model and that PM2.5 is measured without error at the fixed locations of EPA reference monitors. Then, we select locations to have LCS and induce sensor measurement error at those locations. By varying the amount and type of sensor measurement error, we can observe changes in accuracy of exposure information when LCS are integrated. We evaluate the accuracy of daily air quality information available from individuals’ nearest monitor or sensor with respect to absolute errors and misclassifications of the Air Quality Index, stratified by Census tract-level socioeconomic and demographic characteristics. Comparing air quality information accuracy under different hypothetical LCS placement strategies (at schools, near major roads, and in environmentally and socioeconomically marginalized census tracts) and different numbers of LCS also illustrates how real-time air quality reporting could be improved (or, in some cases, worsened) by using LCS, both for the population overall and for marginalized communities specifically.

Additional Authors: Ellen Considine, Rachel C. Nethery, Danielle Braun, Priyanka deSouza

High density sensor network for air quality monitoring and source identification in Shanghai Ports

Han Mei, The Hong Kong University of Science and Technology

The sensor network is increasingly known as an effective way to explore the characteristics of local emissions with greater spatiotemporal details. Here we present how high-density air and meteorological sensor network provides unprecedented insights into the patterns of pollutant emissions at two ports of Shanghai during the China International Import Expo (CIIE). The diurnal variations of pollutants reasonably indicate the non-local emissions have little impact on the air quality in the port area. The nitrogen oxides (NOx) is identified as the major local pollutant of the ports, while there is no obvious sources of CO and particulate pollution have been found in this area. The ports area is divided into three types of areas for NOx pollution, including the source area, susceptible area and less affected operations and berth area. The roads in the source area are identified as the main local source because the trucks in this area emit a large amount of NOx, which impact on the surrounding susceptible area and the operation and berth area under different wind conditions. In addition, the emission controls during the CIIE have sufficiently improved the air quality of ports area and reduced its negative impact on the surrounding environment. This study reveals the sensor network can accurately locate the key pollution sources and their variations in a complex emission area by analyzing the heterogeneity of the monitoring data, with strong implications for designing effective control measures.

Additional Authors: Yang Xing, Han Mei, Zhi Ning

The Smart and Trustworthy AIR quality network (STAIR): practical considerations in network design and community outreach

Haofei Yu,

The Smart and Trustworthy AIR quality network (STAIR) is a low-cost sensor network designed to measure ambient concentrations of particulate matter (PM) and carbon dioxide (CO2) at up to 100 locations across the Orlando, FL area. Some of primary purposes of this highly interdisciplinary project are to design a cyber-secured sensor network that can provide reliable and trustworthy data, and to utilize this network as a tool to facilitate the communications between citizens and the government and to enhance trust in governance. Monitor deployment is still underway and we anticipate the network to be operational in early 2022. Here, we will briefly discuss some of the valuable lessons we learned in community outreach, and in designing, deploying, and maintaining the STAIR network. Our low-cost sensor network has been shown to be a powerful tool in engaging citizens in ambient air quality concerns.

Additional Authors: Elchin Kazimov, Lan Luo, Clayton White, Haofei Yu, Xinwen Fu, Deliang Fan, Kelly Stevens, Thomas Bryer

Air quality use cases: assessing the impact of different events using air quality data from model and sensor network

Jill Chevalier, eLichens

Road traffic is a main target source of action to improve air quality (AQ). The monitoring and evaluation of the impact of emission abatement actions on AQ is essential to better inform decision-makers. However, short-term, or localised actions remain very complex to evaluate without the implementation of significant means of modelling and measurements allowing to consider, the impact of meteorological variability on the concentration levels. High-resolution, continuous data on pollutants concentrations are needed to observe possible impacts of changes in mobility policies. However, model-only cannot reflect the reality accurately enough, as the local and short-term changes due to peculiar events might not be caught. A dense network of station is of paramount importance to catch such events. eLichens developed a mapping method to provide hourly, street-level data on urban areas. The model uses a neural network pollution model improved by fusing measurements from a network of eLos (eLichens outdoor station). eLos is a real-time recalibrated low-cost AQ station, providing hourly measurements of NO2, O3, PMs and CO2, temperature and humidity. Using this data, we assessed the impact of different events on AQ: Covid-19 lockdowns, summer fires or changes in local mobility policies. For long-lasting events, we compare the present data to the data from the previous year to get rid of the meteorological variability. We detected a drop in nitrogen dioxide concentrations in several cities during lockdowns. In Grenoble, France, the creation of a bike lane changed the traffic, creating jams. However, we evidenced a decrease in pollution levels.

Additional Authors: Jill Chevalier, Franck Lascaux, Benjamin Lebegue, Julie Allard, Pierre Jallon

Using sensors to measure the impact of air pollution on early childhood. Lima Air Quality Network for Children Project.

Kyara Diaz Carrasco, Municipalidad de Lima

Until 2019, the province of Lima only had 19 reference stations of air quality, this number of stations is insufficient to cover the extension of the city (the biggest city in Perú with a third of the entire country population). This lack of monitoring systems, open and systematized information conceals the problem and makes it difficult for the authorities and citizenship to take action to protect the environmental health of the most vulnerable groups. In response to this, the Metropolitan Municipality of Lima is implementing an Air Quality Monitoring Network focused on children to design and implement actions that reduce air pollution and the exposure of infants. To this end, the project will be implemented in 8 phases; the first, involves the installation of the Air Quality Network for children; the second, consists of capacity building for teachers and caregivers in the use of the air quality data; the third, seeks to bring air quality knowledge to children, while the fifth and sixth phases focus on generating research on children's exposure to air pollution with information from the network. An air quality normative focused on children will be drawn up in the seventh stage. Finally, based on the previous results, in the eighth phase, the aim is to coordinate with the institutions involved to design and implement actions for the benefit of children. Currently, the first and second phases of the project have been completed and the third phase is being implemented for affiliated institutions, training more than 200 caregivers and teachers, and around 1,400 children on air quality. These three phases are dynamic, restarting each time new institutions join the Network. This project seeks to generate sufficient data for studies of children's exposure to pollution to contribute to the improvement of air quality management, encourage greater commitment on behalf of stakeholders and guide actions to reduce exposure to adequate environmental levels.

Additional Authors: Kyara Diaz Carrasco, Caroline Henriquez Anaya

A tale of our hometown: how the low-cost sensor network helped in changing the air pollution legislation and started the fight with smog in Poland

Marcin Szwagrzyk, Airly

Poland has one of the worst air pollutions in Europe, mostly due to individual, coal-based combustion. Cracow - a former capital city - has been known to be the most polluted city in Poland due to its unfavorable topographic location and high density of pollution sources. However, due to a lack of public awareness, this issue has not been addressed for decades. In the winter of 2017/18 there happened to be one of the biggest pollution episodes in Europe in the XXI century. In Cracow, day after day the enormous exceedances of PM’s daily norms occurred, reaching 1000% of daily WHO PM10 norms. Citizens became vividly interested in tracking the pollution levels, with whom help Airly started its mission and set up a pro bono low-cost sensor network of almost 100 sensors. The screenshots from the Airly mobile application, showing terrible air quality norms exceedances, flooded social media all over the country. Therefore, the Cracow authorities were forced to act and the ban on combustion with solid fuel was introduced within the city borders. In the following years, the Airly network in Cracow and its vicinity started to show an interesting pattern of pollution. The air quality in the city itself rapidly improved, yet in the neighbouring towns and villages, pollutants concentrations remained high. The following led to the emergence of the so-called ‘Cracow bagel’ effect. Moreover, the Airly sensor network began to expand and many other Polish cities, which were lacking proper air quality monitoring before, began to realize the scale of the problem. Some of them were found to have pollution levels much higher than Cracow itself which wasn’t known before. Data from thousands of Airly sensors, working from 2017 are now forming a fascinating image of the air pollution in recent years, which has been shaped by many factors such as topography, climate change, urbanization and legal actions. This was not possible with only a few costly official stations.

Additional Authors: Marcin Szwagrzyk, Piotr Kowalski, Aleksander Konior

From CO and CO2 Measurements to Emissions Maps

Naomi Asimow, UC Berkeley

High-density, low-cost air quality sensor networks promise insight into urban emissions and neighborhood-scale concentration gradients of pollutants. Such concentration maps can show intra-community heterogeneity of air pollutants and often reveal environmental injustice (i.e. disproportionately high concentrations of harmful pollutants in low-income communities of color). However, specific quantitative information on the emission sources of the air pollutants that drive these concentration gradients are essential to understanding and managing them. Here we use hourly carbon monoxide (CO) and carbon dioxide (CO2) measurements from The Bay area Environmental Air-quality & CO2Network (BEACO2N) to develop updated maps of urban CO and CO2 emissions in the San Francisco Bay Area and CO2emissions in Glasgow, Scotland. By coupling dense measurements with a prior emissions inventory, meteorological data, and a particle dispersion model we develop improved maps of urban emissions and an understanding of underlying processes.

Additional Authors: Naomi Asimow, Alexander Turner, Helen Fitzmaurice, Yishu Zhu, Jinsol Kim, Paul Wooldridge, Catherine Newman, Ronald Cohen

Evaluating the Spatial and Temporal Sensitivity of Sensor Networks to the Calibration Algorithm Applied

Priyanka deSouza, University of Colorado Denver

Low-cost monitoring networks are increasingly being used to supplement regulatory networks and to aid in a neighborhood-scale understanding of air pollution levels. However, the variability and uncertainty inherent in the measurements from such devices presents many challenges. In order to correct the data from low-cost sensor networks, one or multiple sensor devices are usually collocated with a reference-grade monitor in an environment that is representative of the sampling conditions. This co-location time frame serves as the training period for which a data-adjustment method is developed that incorporates the sensor raw data and corrects the data to most closely match the reference-grade data. The calibration algorithm developed is then applied to measurements from the entire network of low-cost sensors. Although there has been a lot of work evaluating the efficacy of different correction algorithms, less work has been done on evaluating the sensitivity of spatial and temporal trends in the low-cost air quality monitoring network to the calibration algorithm applied. Such an evaluation is especially important as several groups are now routinely using non-linear algorithms and machine learning techniques to calibrate their low-cost sensor networks. Accordingly, this research evaluates the spatial and temporal trends of the Love My Air network in Denver comprising of ~ 40 low-cot sensors in public schools to the correction method applied. Our work allows for the development of a robust air quality management plan using the data from the Love My Air network

Additional Authors: Priyanka deSouza

Increasing Community Participation in Air Pollution Mitigation in Indore City, India

Tim Dye, TD Environmental Services

India’s National Clean Air Plan has recently called for the expansion of air quality monitoring using both reference instruments and air sensors, marking a change in the country’s monitoring trajectory. As a pilot project, the U.S. Agency for International Development-funded Building Healthy Cities (BHC) project has deployed a network of 20 low-cost air quality sensors in slum and non-slum areas in Indore, India. A city with a population of 2 million, Indore has only one continuous PM2.5 reference station. This mixed-method, longitudinal study is led by BHC and was co-created with the City of Indore and Indore School of Social Work to fill in air quality monitoring gaps across the city and mobilize communities on the issue of air quality. The sensor air quality data is shared to the city’s Integrated Command and Control Center, allowing the public and decision-makers to also access the data. A unique element of this study is the Clean Air Guides – local community members that were trained to install and operate the air sensors, collect qualitative data on sources of air pollution in their neighborhoods, and lead community advocacy efforts to improve air quality. These Clean Air Guides learned about the effects of air quality on health and began interacting with residents in their communities to explain air quality concepts, identify neighborhood-specific sources of air pollution, and help advocate for addressing local air quality. This presentation will briefly review the technical logistics of setting up the network and will focus on the successes and challenges of the Clean Air Guides model. We will discuss how we are working to continuously engage the Guides, share examples of how the Guides used local air quality data to advocate for change, and show journey maps created by the Guides to document changes over time.

Additional Authors: Tim Dye, Amanda Pomeroy-Stevens, Damodar Bachani

Citizen science monitoring of air pollution from residential wood burning using low-cost sensors

Nuria Castell, NILU – Norwegian Institute for Air Research

Conventional monitoring systems such as reference stations can provide accurate and reliable pollution data in the urban environment, but only in single points. In this study, we investigated how low-cost sensor technologies mounted at citizens’ houses can contribute to fill existing gaps in pollution monitoring. During winter 2021 we engaged with 20 residents in Kristiansand, Norway to monitor air pollution using low-cost sensors. For this study we employed Airly sensor units monitoring PM10, PM2.5, and PM1. The selection of the sensor systems was based on their usability and reliability. The Airly sensor systems integrate a Plantower PMS5003 and showed a correlation of 0.6 against FIDAS optical reference-equivalent for PM2.5 and of 0.8-0.9 for PM1 hourly observations. Additionally, a Kleinfiltergerät providing measurement of fine particle mass concentration (PM2.5) was installed in the garden of one of the houses showing R2=0.8 for PM2.5 daily averages. The diurnal pattern of the data collected with the sensors clearly showed two peaks, one in the morning, around 7-8 and one in the evening around 17-20, both likely associated with residential wood burning. Those peaks were not picked up by the reference station in Kristiansand, that is located close to a road. However, they were very clear, particularly in the afternoon and in the cooler months in the residential areas where wood burning is used for residential heating. Pollution levels from PM2.5 where especially high in one of the neighbourhoods that is located is a small valley in the northern part of Kristiansand. The study shows that citizen science data, when data quality routines are in place, can contribute to in-situ environmental monitoring in urban environments, providing measurements than can help science and authorities in locations where we do not have reference stations. This work was partly supported by NordForsk (http://nordicpath.nilu.no).

Additional Authors: Nuria Castell, Matthias Vogt, Philipp Schneider, Sonja Grossberndt

Standard, Supplemental and Informational Monitoring

Aggregating and Harmonizing Air Quality Data on a Global Scale

Chris Hagerbaumer, OpenAQ (Invited Talk)

Despite the urgency of confronting air pollution, only 50% of governments worldwide produce air quality data, leaving 1.4 billion citizens without access to fundamental information that could protect them from the harmful effects of air pollution. Where air quality data does exist, data are often in inconsistent and temporary data sharing formats or buried in hard-to-find websites, making it difficult for the public to readily access and make use of the data.

The OpenAQ platform aggregates and harmonizes real-time and historical air quality data from 155 countries from a variety of sources including reference-grade government monitors to community-led low-cost sensors to research-grade data. The open source platform hosts over 18 billion data points and the open API serves an average of 18 million requests per month. The data has been used for a wide variety of applications, from air quality forecasting systems produced by NASA scientists, to public awareness raising apps, to data-driven media communications. By filling a basic data-access gap and building foundational open source tools, OpenAQ has empowered diverse stakeholders to use data toward the end goal of clean air.

What methods have been used to aggregate and harmonize air quality data on a global scale? What kinds of partnerships have helped reshape the air quality data ecosystem? Looking into the future, how do we make data more accessible to catalyze more effective and efficient action? Drawing from data sharing partnerships that the organization has built over the past few years, OpenAQ will share key lessons learned on how to work on a global scale to make air quality data accessible and usable to all.

Additional Authors: Chisato Calvert, Russ Biggs

tVOC sensor use in Colorado oil & gas Regulation 7

Michael Ogletree, Colorado Department of Public Health & Environment, Denver, Colorado, USA (Invited Talk)

The state of Colorado Air Quality Control Commission passed a regulation in late fall of 2020 that went into effect May 1 2021 requiring real time fenceline monitoring of pre-production O&G development. It required that all new operations would monitor for "Methane, Benzene, or tVOC" pre, during, and post construction for a period of time. tVOC sensors have been the prefered measurement of choice by operators due to their lower cost and high resolution. This presentation will highlight key components of the regulation, how monitoring plans use tVOC sensors to meet the requirements, and a preliminary look at some data from year 1 of implementation.

A real-time calibration and device management system for air quality sensors deployed in hierarchical networks

Lena Weissert, Aeroqual Ltd (Invited Talk)

Regular calibration and maintenance are critical to provide reliable real-time data from sensor-based air monitoring networks. Given the costs related to individual site visits and calibrations, new approaches are required for large scale sensor networks to be viable. We report on the performance of the MOment MAtching (MOMA) network management framework that allows sensors to be managed remotely ensuring reliable data can be accessed in real-time without having to physically move any sensors. The framework has been operating for 13 months and has been applied to a mixed asset network of 50+ air quality sensors measuring O3, NO2, PM2.5and PM10and 13 regulatory air monitoring stations operated by South Coast AQMD in Southern California. The sensors were calibrated at monthly intervals using previously developed algorithms that match the sensor distribution to the distribution of a regulatory measurement which is referred to as a ‘proxy’. In addition, sensor diagnostics were developed to monitor sensor health and trigger maintenance requirements which minimises data loss and maximises sensor life. The performance of the network management framework was assessed by sensor co-location and proxy comparisons using RMSE and other statistical metrics. Regular calibration resulted in a significant reduction in the RMSE compared to uncalibrated data across all pollutants. We observed higher month to month variability in the performance of the calibration framework for PM2.5 which was associated with changes in wind direction affecting pollution sources and proxy suitability. The combination of regular calibration and maintenance, facilitated by the MOMA framework, enables the collection of reliable real-time data at neighbourhood scales. This offers an opportunity to identify local effects on air pollution and determine the probability of exceedances at different times and locations providing valuable information to local communities.

Additional Authors: Lena Weissert, Geoff Henshaw, David Williams, Brandon Feenstra, Randy Lam, Ashley Collier-Oxandale, Vasileios Papapostolou, Andrea Polidori

Evaluation of high-spatial-resolution air pollutant concentration and AQI estimates across the U.S. by fusing low-cost and reference monitor observations with chemical transport model forecasts

Jennifer DeWinter, Sonoma Technology

Near-real-time community-scale air quality information is essential to provide actionable insights for government agencies, industry, and the public. Existing products that predict current air quality conditions range from simple interpolation of reference monitor observations (e.g., the U.S. Environmental Protection Agency AirNow system) to complex machine learning approaches. The fusion of observations and chemical transport models via kriging interpolation provides a straightforward, computationally feasible, interpretable method to accurately predict continuous air quality index (AQI) nowcast surfaces in near-real time. Based on recent methods reported by Schulte et al. (Environmental Research Letters, 2020) focusing on the Los Angeles basin in California, we demonstrate that high accuracy nowcasts can be achieved across the entire U.S. by kriging the differences between gridded mean observations and chemical transport model simulations. We use gridded median AirNow reference monitor and PurpleAir low-cost sensor PM2.5 observations and the National Oceanic and Atmospheric Administration (NOAA) NAQFC forecast model. Hourly fine particulate matter (PM2.5) and ozone concentrations are predicted at 1x1 km resolution, along with grid-scale measurement and sampling uncertainties. These results are used to report the AQI nowcast for each pollutant at high-spatial resolution. We apply this method across diverse urban and rural geographies with variable topographies and spatial coverage of PurpleAir and reference monitors. We also describe computationally efficient kriging within subdomains and post-kriging merging of overlapping subdomains. We evaluate the accuracy, bias, and precision of air pollutant concentration predictions via 10-fold cross validation across each subdomain and nearest neighborhood, including in sparsely monitored regions and during extreme regional wildfire events. New automation techniques for reporting the AQI nowcast will also be presented.

Additional Authors: David Miller, Nathan Pavlovic, Anondo Mukherjee, Lisa Churchman, Anthony Cavallaro, Steven Brown, Fred Lurmann, Jennifer DeWinter

Data Quality Assessment Methods to Support Community-Level Air Quality Monitoring

Cheryl Winfield, CARB

Community-level air quality monitoring presents new opportunities to promote citizen science, better understand local air quality, and address its impacts. This also poses new challenges for quality assurance systems in centralized data repositories. With the introduction of commercially accessible low-cost sensors (LCS), the challenge of evaluating data quality at the centralized database level requires an expanded set of quality control (QC) methods. Large disparities in the grade of monitoring platforms and site-level quality assurance (QA) conducted demand a wider scope of QC methods and tools at the central database level. AQview is CARB’s centralized database repository and public data portal for storing, providing, and visualizing community-level air quality data collected to support California’s AB 617 legislation and other communities across the state. In contrast to regulatory air quality monitoring data, the monitoring data collected to support communities originate from a variety of instruments ranging from regulatory and research-grade monitors to networks consisting of many LCS devices. This presentation will discuss the challenges of community air quality data and describe current and future elements for assigning and messaging data quality in AQview.

Additional Authors: Emily Gorrie, Raiford Hann, Annemarie Flores, Taylor Helgestad, Yanju Chen


Edurne Ibarrola, Kunak Technologies SL

Summary The aim was to evaluate a new methodology for managing air quality throughout ABB Formula-E events. Data from reference stations, satellites and sensors were combined at Berlin E-Prix 2021 to evaluate hyper-local exposure of sensitive receptors (visitors, teams, staff) and the potential wider ecological impact of the event. The study concluded that it had no significant impact on local conditions, which remained within international standards throughout. Introduction Formula E is a leader in sustainability at its events and beyond. In enhancing its net-zero, e-mobility and ISO20121 credentials, it was identified that availability of detailed, high-value data for source identification and exposure characterisation was limited at events. Methodology, Results Data for conventional (PMs, NOx, CO, O3), emerging pollutants (VOCs, BC) and meteo (wind, temp., RH, WBGT) was recorded. The nearest 5 stations (BLUME) and satellite (SENTINEL) were included, and 15 medium to low-cost monitors with 131 sensors were deployed at the E-Prix (Kunak AIR Pro, Kunak AIR Mobile, Airly, AE51). The temporal patterns of pollutants were analysed independently according to EUAQI/WHO guidelines and evaluated against exposure for sensitive receptors in a tentative causal analysis. The main results were: (i) Strong positive correlation for O3, Temp, WGBT and reference data. (ii) Very low PM and NOx values, following local traffic patterns. (iii) Discreet NOx and VOCs emissions at specific locations, with winds cleaning the air. Conclusions A dense, distributed fixed & mobile network helped to (i) provide evidence of an event within international standards (ii) assess the impact on receptors as dynamic and low (iii) identify the sources outside and within the events remit of control (iv) provide integrated digital tools to educate fans (vi) fill the data gap; enabling air quality control strategies to be devised for the future design of healthier motorsports competitions.

Additional Authors: Tom Verity, Iona Neilson, Miguel Escribano, Edurne Ibarrola

AirNow Fire and Smoke Map

Ron Evans, USEPA

In recent years, data from particulate matter (PM) sensors have become a valuable tool for broadening our understanding of air quality impacts from wildfire smoke and demonstrating the localized nature of smoke plumes. In 2020, EPA, in conjunction with state, tribal and local air monitoring agencies and the US Forest Service, initiated the Sensor Data Pilot on the AirNow Fire and Smoke map. This pilot displayed publicly available PurpleAir sensor data side-by-side with data from the permanent and temporary ambient air quality monitors. Because PM sensor data are often biased compared to federal reference method monitors, a correction equation was developed from a long-term sensor collocation across geographically diverse areas, spanning a wide range of PM concentrations. This correction was improved increase accuracy at extremely high concentrations and validated with smoke-impacted collocations during the 2020 fire season. Data cleaning and quality control steps were developed to quality assure the sensor data that is displayed on the map. This presentation will provide an overview of the AirNow Fire and Smoke map and describe the research behind the sensor data processing that has enabled the addition of thousands of new air quality observations helping communities monitor smoke impacting their area. The presentation will focus on updates Version 2 released in July 2021and planned updates for Version 3 to be released in Summer 2022. 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.

Additional Authors: Ron Evans, Andrea Clements, Phil Dickerson, Karoline Barkjohn, Amara Holder, Sim Larkin, Stuart Illson

An Exploration of Gas Chromatograph and tVOC Sensor Data Collected During Two Different Releases from Oil and Natural Gas Well Pads in Colorado

Presented By: Alicia Frasier, Colorado Department of Public Health and Environment, Denver, CO

Colorado currently uses both tVOC sensors and gas chromatographs for monitoring around oil and gas development sites.  The sensors have primarily been used as air quality indicators in health and odor-based air quality complaints.  With the passing of recent oil and gas monitoring regulations, however, further study was needed to see how well the sensors performed in real-world settings, and how effective they could be in a regulatory capacity.  To date, the state has recorded two instances of releases at well pads near semi-rural communities which resulted in hourly spikes of benzene concentrations to levels greater than 9 ppb. At these times, both a tVOC sensor and a GC were on site and operational. This case study will show how the sensor compared to the GC in a practical field setting, as well as it's capability to help identify these types of incidents, and the potential impact in the regulatory world.

Additional Authors: Alicia Frasier

Swimming in Data: The current and future state of data management platforms

Lessons Learned in designing, developing, and implementing the South Coast AQMD AQPortal environmental data management solution

Vasileios Papapostolou, South Coast Air Quality Management District (Invited Talk)

Technological advances along with regulatory advances mandating increased air quality monitoring have provided the opportunity for air districts to expand their monitoring capabilities and build hierarchical networks that include regulatory-grade instruments, research-grade equipment, and consumer-grade air quality sensors. These expanding monitoring networks collect data at increased spatial and temporal resolution to support different purposes: community, fence-line, supplemental, and incident response monitoring. These expanding networks create data management challenges to ingest, store, process, analyze and visualize the collected data. The South Coast Air Quality Management District has developed the AQPortal, a single-point of access cloud-based data management platform to meet these new data management challenges. AQPortal supports over 10 District-level and several external data streams, removes data and analytical silos, and establishes new data science workflows for data analysts and scientists to develop comprehensive data visualizations for public consumption. This presentation will discuss lessons learned during the design, development, and implementation of the AQPortal for environmental data management. We will review a use case scenario in which AQPortal was utilized to develop a publicly available data dashboard for air quality sensors deployed in support of community air monitoring networks under the California Assembly Bill 617 Community Air Monitoring and Emission Reduction plans efforts.

Additional Authors: Brandon Feenstra, Ashley Collier-Oxandale, Aneeta Dev, Brian Roche, Xin Chen, Ron Moskowitz, Jason Low, Vasileios Papapostolou, Andrea Polidori

Air sensor data management, visualization, and analysis: understanding and meeting the needs of government air quality organizations in the United States

Gayle Hagler, US EPA Office of Research and Development

Air sensor use has grown in multiple sectors in the United States, including use by air agencies (federal, state, local, tribal) for a variety of non-regulatory supplemental and informational monitoring purposes. Realizing the full benefit of this new technology, however, is limited by the extent to which the data can be attained, processed, and analyzed by the user. To understand this particular end user sector, the United States Environmental Protection Agency (EPA) Office of Research and Development led unstructured and open-ended interviews in late 2019 with 19 government air organizations to understand their current practices, unmet needs, and future outlook for using air sensor data. Based on the dialogues, the organizations were grouped by type (e.g., state vs. local air agency) and three identified levels of use: Level 1) limited use (e.g., educational demonstrations); Level 2) Growing use in temporary monitoring, data quality evaluation; and Level 3) Sensors are routinely integrated into meeting the organization’s goal. We found that organizational type was not a good predictor of their level. After this dialogue stage, a cross-EPA team evaluated the landscape of existing and in-development solutions to the identified unmet needs, focusing on Level 2 users who faced the greatest barriers to progression in their use of air sensor data. The unmet needs we targeted for this study include data hosting, data quality, code sharing, and data analytics tools. This presentation will cover the interview findings, assessment of solutions, and gaps that remain.

Additional Authors: Gayle Hagler, Andrea Clements, Ryan Brown, Daniel Garver, Ron Evans, Colin Barrette, Ethan McMahon, Dena Vallano, Robert Judge, Sarah Waldo, William Wallace, Anna Mebust, Corey Mocka, David Smith, Amanda Kaufman

Unlocking the Value in Sensor Data

Graeme Carvlin, Puget Sound Clean Air Agency

How do we take the mountains of publicly available sensor data and create meaningful data displays and reports that answer people’s air quality questions? Graeme will discuss the Sensor Dashboard app suite he has developed (http://apps.pscleanair.gov [http://apps.pscleanair.gov]). This toolkit includes ways to view, download, and analyze QC’d and calibrated sensor data. Two new modules, Surface and Compare, allow the creation of 2D air pollution surfaces and comparison between Purple Air sensors respectively. Graeme will show application of these tools to examples of air pollution concerns in the Puget Sound region.

Additional Authors: Graeme Carvlin

Universal Data Structures for Air Quality Data

Jonathan Callahan, Desert Research Institute

Air quality data comes in a bewildering array of formats from an ever increasing list of regulatory monitors and low-cost sensors. Data providers make data available in csv, json and binary formats at everything from raw, “engineering level” to highly processed final versions. Ingesting, harmonizing, aggregating and storing this data for use by downstream applications is a formidable challenge that requires an understanding of the fundamental nature of environmental monitoring time series data. This presentation describes a “universal” data structure that supports both raw engineering data and national collections of aggregated data associated with stationary air quality sensors. A suite of R packages has been created that helps with data ingest, harmonization, restructuring and spatial assignment of data into this format. Additional R packages provide analysis and visualization of air quality data that has been converted into this format. In an era of hybrid monitoring networks, increasing data volumes and intense public interest in air quality data, it is incumbent upon the air quality community to work toward compact, uniform data standards that allow the ready exchange of air quality data. The data format described in this presentation is maximally compact, allowing data from thousands of devices to be stored in easily downloadable archive files. Archive files can be read in by software written in all common languages. Example scripts written in R, python and javascript will demonstrate the interoperability of this format.

Additional Authors: Jonathan Callahan

Integrating an in-house developed sensor platform with the existing AQM network and its off-the-shelf DAS solution

Matt Shrensel, Oregon Department of Environmental Quality (Virtual Presentation)

Oregon Department of Environmental Quality (ODEQ) Air Quality Monitoring recently doubled its particulate monitoring network using its own in-house sensor platform (SensOR™). The sensor platform leverages ODEQ’s existing systems, low-cost Plantower particulate sensors, open source computer code, and existing staff time. We adapted our sensor platform and data acquisition system (EnviDAS) to provide data seamlessly into our existing informational-level particulate monitoring network. This includes support for ODEQ’s near-realtime Air Quality Index, USEPA’s AirNow, and wildfire smoke monitoring. This approach enabled ODEQ to get granular control over data, customize data collection, and use existing review and validation procedures. This approach further allowed ODEQ to make measurements that were consistent with other continuous informational-level monitors maintained by ODEQ, providing consistency across ODEQ’s network. We will discuss advantages and disadvantages of this approach.

Additional Authors:

Using spatiotemporal infrastructure to manage and process air quality data for a rapid response to COVID-19 impact on air quality

Phil Yang, George Mason University

Air quality is such a critical public health factor, that worsened air takes the lives of approximately 7M annually. It also intertwines with other disasters and provides an indicator for both environmental and human response to these disasters. We’ve utilized the spatiotemporal infrastructure built by the NSF Spatiotemporal I/UCRC, including cloud computing, big data analytics, and GPU computing to process various sources of air quality observations to investigate how the global pandemic has impacted the air quality in China, California, the U.S. and globally: a) a spatiotemporal data discovery portal was utilized to identify the data of interest in a smart fashion; b) spatiotemporal analytical tools are built to conduct statistical analyses. Machine learning tools are used to discover correlations among air quality and socioeconomic factors; c) the results are visualized with the use of various tools, d) cloud computing (AWS and OpenStack) was utilized to provide a rapid response and GPU computing was used to speed up the learning & mining process. The spatiotemporal infrastructure was able to provide efficient support to the on-demand and complex requirements of analyzing the pandemic and air quality.

Additional Authors:

Distributed network of sensors for real time environmental monitoring

Radu Motisan, Magnasci SRL

Introduction Technology has reached the point where we can build sensor networks that monitor the environment in real time. We present the results of a large-scale implemented project consisting of a network of interconnected sensors to generate a real-time environmental data stream. Methods Distributed systems are composed of detection elements (sensors) and connectivity elements (for data transmission). The sensors address a large number of parameters of interest, both for air and water: Temperature, Barometric pressure, Humidity, Volatile organic compounds, Suspended particles (PM1, PM2.5, PM10), Carbon dioxide, Formaldehyde, Other Specific gases (Ozone, Nitrogen dioxide, Sulfur dioxide, Carbon monoxide, Benzene, Hydrogen sulfide, Ammonia), PH, Conductivity, Turbidity, REDOX, Dissolved oxygen and more. Results All this data is available for real-time analysis in the form of a structured flow, easily interconnectable with specific analysis and statistical programs. Discussion The current challenges refer to the reliability of the distributed systems in time (lifespan), to the need to adopt a constantly changing technology, to the maintenance effort but also to the processing part of a very large volume of data. From what we have at the moment, it is obvious that the system offers an increased capacity to analyze the environment over large areas, at a much lower cost.

Additional Authors: Radu Motisan

The Potential of Air Sensors for Personalizing and Advancing Human Health Research

Personalized environmental sensing for health research and disease management - lessons learnt and future challenges

Benjamin Barratt, Imperial College London (Invited Talk)

The overarching aim of most air quality measurement campaigns is to better understand where air pollution comes from, how it behaves and how it impacts human health. While air sensor campaigns now regularly accomplish the first two aims, relatively little attention is given to the third. Over the past decade, the Environmental Research Group has conducted a series of large personal monitoring campaigns directly linking acute and chronic health responses to exposure to environmental stressors. These range from tracking the cardiovascular health of elderly residents in Beijing, to the lung function of asthmatic children in Sub Saharan Africa, to the respiratory health of COPD patients in London. Each campaign has provided insight into the strengths and limitations of this research method, along with valuable practical lessons in how to, and how not to, carry out personal air sensor campaigns for human health research. While campaign results have provided valuable evidence for disease prevention and management, there are many outstanding challenges, including extrapolation of results over longer time periods, participant burden, data interpretation and transferrability of results between cohorts. These challenges must be overcome if personalized sensing is to realize its full potential.

Additional Authors: Benjamin Barratt

Ecologically-Valid, Multimodal Data Collection Platforms to Measure the Effects of Indoor Air Quality on Sleep Quality

Zoltan Nagy, The University of Texas at Austin (Invited Talk)

An adequate amount of good-quality sleep is essential for a person's physiological and psychological well-being, but can be affected by a wide variety of factors. The Indoor Air Quality (IAQ) of one's bedroom environment has recently garnered attention since many of the health effects caused by indoor air also directly affect sleep. A few studies have assessed the relationship between IAQ and sleep quality in controlled environments which can affect both self-report and measured sleep quality outcomes. Examining the relationship within a participant's home environment is provides a more ecologically valid measurement which helps to reduce participant bias. Provided the wide variety and availability of consumer-grade IAQ sensors, a study of this type is now more feasible. Here, we present results from a field study conducted with 20 participants over 2.5 months in Austin, TX. We monitored five components of IAQ using a purpose-built environmental monitor, the BEVO Beacon, and measured sleep quality with wearable activity trackers and short sleep surveys sent four times a week to participants. When compared to prior related studies, our study was conducted over the longest period and includes measurements of the most IAQ parameters. We found significant degradation in self-reported and device-measured sleep quality during nights with elevated CO, CO2, and temperature. Evenings with elevated PM2.5 and TVOC concentrations were found to be associated with general improvements in sleep quality which indicates a need to study relationships between these aggregate IAQ measures and sleep quality more closely. Our study highlights the application of consumer-grade, affordable IAQ and sleep quality monitoring devices and the complicated relationship that exist between these two domains.

Additional Authors: Zoltan Nagy, Hagen Fritz, Kerry Kinney, David Schnyer

Indoor Air Quality Data Captured from Consumer-Grade Devices and Its Effect on Occupant Mood

Hagen Fritz, The University of Texas

Poor mental health can hinder a person's productivity, ability to learn, or participate fully in society. Researchers have found particular effects of outdoor air pollution on mood, primarily focusing on depression, but less is known about the effect of Indoor Air Quality (IAQ) on mood. To address this question, we use information fusion across disparate datasets collected during a 2.5 month study including 20 participants to understand how parameters of a participant's IAQ affect four components of mood: discontent, sadness, loneliness, and stress. We collect data on four components of IAQ -- carbon dioxide, particulate matter (PM), total volatile organic compounds, and temperature -- through the use of our own consumer-grade device and mood data through Ecological Momentary Assessments (EMAs) completed four times a week by participants on their smartphones. Using GPS data provided by participants' smartphones, we determine instances when EMAs were submitted while at their residence and the amount of time the participant was home prior to submitting. We summarize IAQ parameters during these periods to understand (1) how IAQ conditions at home influence EMA outcomes, (2) if the length of time participants are at home before submitting influences their mood, and (3) if mood responses are more pronounced following more extreme measurements of IAQ. Initial results indicate associations between increased CO2 and elevated responses on all four components of mood. Elevated temperatures were associated with increased discontentment and stress while elevated PM concentrations seemed to produce positive responses, showing associations with decreases in sadness and loneliness. Our study is one a few that directly compares IAQ measurements and components of mood and helps to emphasize the importance of proper IAQ control for the benefit of the psychological health of building occupants.

Additional Authors: Hagen Fritz, Congyu Wu, David Schnyer, Kerry Kinney, Zoltan Nagy

Daily Associations of Air Pollution and Pediatric Asthma Risk using the Biomedical REAI-Time Health Evaluation (BREATHE) Kit

Hua Hao, Department of Population and Public Health Sciences, University of Southern California

Background: Exposure to air pollution is associated with pediatric asthma exacerbation, yet acute effects are poorly understood. Methods: We conducted a panel study of 40 children aged 8-16 years with moderate to severe asthma. We used the BREATHE kit to monitor personal exposure to particulate matter of aerodynamic diameter <2.5 µm (PM2.5), lung function, medication use, and asthma symptoms for 14 days. Hourly ambient (PM2.5, nitrogen dioxide (NO2), ozone (O3)) and traffic-related (NOx and PM2.5) air pollution exposures were modeled based on location. We used mixed effects models to examine the association of same day and lagged (up to 2 days) exposures with daily changes in % predicted forced expiratory volume in 1 sec (FEV1) and % predicted peak expiratory flow (PEF), rescue inhaler puffs, and symptoms. Results: Participants were on average 12.0 years old with mean (SD) morning %predicted FEV167.9% (17.3%), PEF 69.1% (18.4%), and rescue inhaler use of 1.4 (3.5) puffs/day. Participants reported chest tightness, wheeze, trouble breathing, and cough symptoms on 36.4%, 17.5%, 32.3% and 42.9%, respectively (N=217 person-days). One SD increase in previous day O3 was associated with reduced morning (beta [95% CI]: -4.11 [-6.86, -1.36]), evening (-2.65 [-5.19, -0.10]) and daily %predicted FEV1(-3.45 [-6.42, -0.47]). Daily (lag 0) traffic-related PM2.5exposure was associated with reduced morning %predicted PEF (-3.97 [-7.69, -0.26]). Exposure (lag 0) to ambient O3, NOx, and NO was associated with rescue inhaler use (rate ratio [95% CI]: O3 1.52 [1.02, 2.27], NOx1.61 [1.23, 2.11], NO 1.80 [1.37, 2.35]). Traffic-related PM2.5(lag 0) was associated with “feeling scared of trouble breathing” symptom (odds ratio [95% CI]: 1.83 [1.03, 3.24]). Conclusions: Our study demonstrates the potential of informatics, air sensors, and wearables at enabling the collection of highly space and time resolved, contextual and personal data for understanding acute pediatric asthma triggers.

Additional Authors: Hua Hao, Sandrah P. Eckel, Anahita Hosseini, Eldin Dzubur, Genevieve Dunton, Shih Ying Chang, Kenneth Craig, Rose Rocchio, Theresa Bastain, Frank Gilliland, Sande Okelo, Mindy K. Ross, Majid Sarrafzadeh, Alex A. Bui, Rima Habre

Integration of Tools for Real-time Assessment of Residential Air Quality and Asthma Symptoms: Challenges and Lessons Learned

Luz Huntington-Moskos, University of Louisville School of Nursing

Introduction: The COVID-19 crisis has altered cleaning practices and time spent at home. Research assessing residential exposure to cleaning/disinfecting products among adults with asthma and the impact on asthma symptoms are lacking. This presentation will discuss the challenges and lessons learned during integration of an air quality monitor, spirometer, and ecological momentary assessment (EMA) to assess residential air quality and asthma symptoms among adults with asthma. Methods: This study is being conducted with 50 adults with uncontrolled asthma. We are assessing the feasibility and usability of: (1) providing participants with a commercially available indoor air quality monitor (Awair Omni) to continuously capture total volatile organic compound (VOCs) and particulates (PM2.5); (2) EMA notifications via a smartphone app to capture context of real-time indoor air quality, alert participants to high residential VOCs and PM2.5, and assess real-time asthma symptoms. We measure lung function with a home spirometer (ZEPHYRx). Results: Data collection using 3 software platforms (Awair Omni, ZEPHYRx, PiLR EMA) with diverse dashboards and technological requirements required creativity and diligence. Additional time for novel integration between the Awair Omni and the PiLR EMA platforms and testing of the integration was required. Developing study instructional materials for each platform in an iterative process with the study team was critical to minimizing burden on study participants. Feedback from an asthma Community Advisory Board led to modifications in instructional materials, reduced EMA notifications, limited data collection to 14 days, and added a 3-day run-in period. Conclusions: This presentation will provide insight into the integration of diverse technologies, including indoor air quality monitors while assessing real-time events along with real-time symptoms. The lessons presented will guide development of future studies using personal air monitors.

Additional Authors: Luz Huntington-Moskos, Barbara Polivka, Sharmilee Nyenhuis, Emily Cramer, Kathryn Krueger, Matthew Grande, Kamal Eldeirawi



The aim of this work was to estimate the quantitative exposure reductions achievable by monitoring personal exposure to fine particles during recreational runs. We used portable monitors to characterise personal exposure of runners across the city, and quantified the reductions achievable by modifying the routes and times of day. The exposures were then correlated with data on running habits as a function of age and gender, to identify the most exposed population groups. Portable, personal particle monitors (AirBeam2, HabitatMap.org) were provided to volunteer runners to map personal exposures to particles (PM10, PM2.5, PM1) across a 2x2 km area in central Barcelona (Spain). The time resolution of the monitors was 5 seconds, and the position and route of the runners were traced by GPS from their mobile phones, to which the sensors were connected via Bluetooth. Meteorological variability was accounted for by comparison with reference PM2.5 data from a local air quality monitoring station. When using the portable monitors to select the cleanest routes, runners were able to reduce their average exposures by 44%. When comparing exposures along the major roads (most polluted environments) and the cleanest routes (e.g., crossing park areas), average reductions of up to 71% in personal exposure to PM2.5 were achieved. We conclude that it is possible to achieve significant reductions in personal exposure (up to 71%) without significantly modifying running habits. Due to running habits, female runners tend to suffer higher exposers, mostly during morning runs.

Additional Authors: Mar Viana, Cristina Reche, Miguel Escribano, Paolo Adami, Stéphane Bermon

Feasibility study on the application of low-cost sensors for epidemiological investigations

Miriam Chacón-Mateos, University of Stuttgart

The World Health Organization has updated on its global air quality guidelines on September 2021. The new standards resulted from the findings based on the latest epidemiological studies. This update was only possible thanks to the improvement on the technologies for air monitoring and personal exposure. In this context, low-cost sensors appear as a useful tool to increase the number of participants, ensuring adequate statistical power in epidemiological studies. In order to prove the feasibility of low-cost sensors for health studies, a pilot study with patients suffering from COPD (chronic obstructive pulmonary disease) or Asthma was carried out in Stuttgart (Germany) in cooperation with University Hospital Charité in Berlin and an outpatient pulmonary practice in Stuttgart. For this purpose, two different low-cost sensor boxes were designed to measure particulate matter, nitrogen dioxide, temperature, and relative humidity inside the houses as well as outside in the vicinity of the buildings where the patients live and spend most of the time during a 30 days time period. To avoid the effect of the hygroscopic growth, a low-cost dryer was included in the outdoor boxes. Participants completed spirometry, a questionnaire assessing respiratory symptoms and a protocol of activities at a daily basis. In this talk we will focus on technical details of the methodology used in the pilot project considering the calibration of the sensors, the validation of the measurement results and the data post-processing needed to get a useful data set for carrying out an epidemiological study. Furthermore, some results will be presented and an evaluation approach for quality assurance and the possibilities and limits of low-cost sensors for epidemiological studies will be discussed.

Additional Authors: Miriam Chacón-Mateos, Ulrich Vogt, Bernd Laquai, Ioannis Chourdakis, Grecia Carolina Solís-Castillo, Héctor García-Salamero, Frank Heimann, Uta Liebers, Christian Witt

Advancing personal air pollution exposure for pregnancy studies using air sensors

Yisi Liu, University of Southern California

Background: Epidemiological studies typically assign ambient exposures to air pollution at participants’ residential location as the surrogate of personal exposures. Ignoring time-activity and mobility patterns introduces exposure measurement error, especially for pregnant woman, whose mobility may change during pregnancy and early postpartum. Methods: We conducted 4-day continuous, personal PM2.5 and geolocation monitoring in 62 pregnant, Hispanic women enrolled in the MADRES cohort during the 1st and 3rd trimester and 4-6 months postpartum using the RTI MicroPEM v3.2A and our madresGPS app, respectively. Ambient PM2.5 exposure was estimated at each participant’s residential location using inverse distance squared weighted interpolation from regulatory air monitors. We compared sub-daily level of personal and ambient PM2.5 exposure and examined variation of exposure measurement error (i.e., the absolute difference between personal and ambient PM2.5 exposure) over time using mixed effect models. Results:A total of 478 person-days were measured with a mean (SD) daily personal PM2.5 exposure of 15.5 (18.3) µg/m3. The exposure measurement error decreased 0.53 (95% CI: 0.02, 1.04) µg/m3in the 3rd trimester, but increased 1.94 (95% CI: 1.27, 2.61) µg/m34-6 months postpartum as compared to the 1st trimester. Exposure measurement error was larger in winter (2.69 µg/m3, 95% CI: 2.03, 3.35) and on weekends (2.06 µg/m3, 95% CI: 1.68, 2.44). In addition, exposure error was larger when women visited commercial and services locations (2.13 µg/m3, 95% CI: 1.35, 2.91), but lower when they took pedestrian trips as compared to home residence (-3.61 µg/m3, 95% CI: -5.20, -2.01). Conclusion:Our findings highlight the discrepancy in personal exposure measurement as compared to the typically used ambient exposure at residential location. The results suggest the importance of considering time-activity and mobility in exposure assessment in pregnancy studies.

Additional Authors:

Youth-Focused Education and Youth-Lead Initiatives

Tribal Air Quality Education and Outreach

Mansel Nelson, Northern Arizona University (Invited Talk)

In partnership with the USEPA, the Institute for Tribal Environmental Professionals (ITEP) established Air Sensor Loan and Education Programs to enable tribal communities to learn about air quality. ITEP staff will share about ITEP’s work providing air quality sensors to tribal communities and tribal schools. Tribal staff and school staff can use the sensors to monitor air quality both inside and outside, for a variety of situations, including during wildfire seasons. The program also includes Air Sensor Toolbox Educational Resources, such as lesson plans and resource guides. ITEP interns have helped expand sensor networks in Alaska, as well as in Arizona, Oklahoma and other states. The ITEP staff are also encouraging the use of sensors to help monitor air quality for school indoor environments. Concerns about COVID exposure has highlighted the importance of adequate ventilation and filtration.

Additional Authors: Mansel Nelson, Josephine Kamkoff

Using personal air monitors with fast response sensors to enhance understanding of air quality for college level students

Austin Moon, University of Wyoming, Department of Atmospheric Science (Invited Talk)

Since 2018 the University of Wyoming’s School of Energy Resources Air Quality Management class and 2B Technology Inc. have collaborated with the use of the AQ Treks Personal Air Monitor (PAM). The PAM includes multiple sensors that enable the rapid measurement of air pollutants including carbon monoxide, particulates (PM1, PM2.5, and PM10), and carbon dioxide. Given these pollutants are present within emission sources with different contributions, basic source apportionment is possible. Together with measurements of meteorological parameters the PAM allows individuals to differentiate between different micro-environments that people experience in their lives. We explore the importance of the measurement quality to reveal genuine understanding of pollution exposure. We highlight the benefits of using state of the art technology that allows the student to self-investigate real world air quality problems to enhance learning outcomes in the field of measurement science. We note the importance of hands-on experiential learning to not only better understand measurements themselves but also the steps related to sharing data. We share examples of a pivotal class assignment that challenges students to create their own experimental designs. Students are able to explore and share results of a variety of indoor and outdoor pollution problems. One exploration of indoor air by a student revealed extreme levels of carbon dioxide within an apartment that was related to harmful health impacts. One outdoor example revealed the impact of a wood burning stove affecting another adjacent residence. Personal air monitors enable individuals to deepen their understanding of air quality. In a wider context we note the significance of the use of high-quality sensors to enhance community understanding about the air they breathe beyond the information provided through regulatory approaches.

Additional Authors: Robert Field, Austin Moon

Utilizing Indoor Air Quality Measurements as a Youth-Action Project During COVID-19

Sarah Peterson, Denver Public Schools (Invited Talk)

The COVID-19 pandemic has underscored the importance of understanding indoor air quality and the role that ventilation plays in protection from airborne illness. This session will focus on the role that students can play in both leading initiatives around air quality measurements in school sites as well as student advocacy at the school and district level for leaders to make science-based decisions to best support the safety of students and staff in the building. The session will feature sample unit and lesson plans, artifacts from student learning, and a discussion on how partnerships with local public health entities can best support in class learning.

Additional Authors: Sarah Peterson

Engaging Youth and the Community in Citizen Science with Air Sensors

Christina Yoka, Cleveland Department of Public Health - Division of Air Quality

Cleveland Division of Air Quality (CDAQ) is the Ohio EPA designated local air agency serving Cuyahoga County. CDAQ frequently receives inquiries from residents and community organizations about utilizing low cost air quality sensors to expand on the EPA ambient air monitoring network. In many instances, CDAQ collaborates with these local organizations on air sensor deployment. These partnerships are a valuable tool used to inform the community about air quality issues and an opportunity for CDAQ to learn more about specific community concerns. In addition to working with community organizations, CDAQ uses low cost air quality sensors to deliver project based learning experiences for youth. CDAQ has a limited number of particulate matter sensors that are available for schools, summer camps and youth groups. CDAQ developed a curriculum that is designed to engage students in the full citizen science experience. Youth are empowered to develop project ideas that address local issues that interest the student. The EPA Handbook for Citizen Science Quality Assurance has been a tool used by CDAQ to develop this curriculum. Major topics of the program include developing citizen science projects, quality assurance, air sensor siting, and data analysis. CDAQ has also incorporated the University of Utah AirU Teaching Module on Building Lego Air Quality Sensors into overall programming. This presentation will provide an overview of the key areas of CDAQ’s air sensor programs with youth and the community; strategies used for program implementation; evaluation results from youth and educators; and lessons learned from initial implementation.

Additional Authors: Christina Yoka

Air Quality InQuiry: Adapting air quality sensors for use in high school settings in the United States and universities in Mongolia

Helena Pliszka, University of Colorado Boulder

Air quality sensors have been a growing tool employed in the field of citizen science to enhance public understanding and engage local stakeholders in issues of air pollution. We adapted low cost air quality sensor packages, known as pods, to educational settings in suburban and rural high schools in Colorado and universities in Mongolia to increase students’ exposure to environmental and STEM fields. Our science engineering outreach program, known as Air Quality InQuiry (AQIQ), trains university students to mentor high school students to build and execute their own projects using pods. The program started in 2014 in the US and 2020 in Mongolia, and currently reaches 12 schools and 686 students in Colorado and 6 schools and 130 students in 5 rural provinces and the capital in Mongolia.The program follows a project-based curriculum housed on TeachEngineering.org that covers five major educational modules: air quality concepts, pod use, project brainstorming, data analysis and scientific poster presentation. We conducted a comparative analysis of high school students’ pre-and-post program surveys and end-project posters among 2018-19, 2019-20, and 2020-21 academic years. Key results suggest that despite the fluctuating learning modes as a result of the COVID-19 pandemic, science, engineering and air quality-related learning outcomes remained relatively stable. 2019-20 survey results from a high school classroom suggest an increase in student agreement that “science is personally relevant,” and that “engineering is helpful to [students’] future careers.” Further analysis will investigate results from the current academic year and will strive to evaluate the long-term impact that this program may have on participating students’ college and career decision-making. Our vision for the AQIQ program and expectations for future results center on empowering high school students to investigate and solve air quality issues within their own communities.

Additional Authors: Helena Pliszka, Kristen Okorn, Michael Hannigan, Joseph Polman, Trang Tran, Daniel Knight, Evan Coffey

Building an aerosol sensing sensor network and inspiring citizen scientists

Kerry Kelly, University of Utah

Since 2018, the AQ&U network has been collecting aerosol measurements from a growing number of low-cost sensors (more than 200) in the Salt Lake Valley. AQ&U integrates low-cost, research-grade, and reference measurements with robust data screening, event-specific calibrations, and a Gaussian Process model to understand neighborhood-scale PM2.5 concentrations as well as uncertainty estimates. It provides near-real time visualizations of PM2.5 concentration along with uncertainty estimates through a public-facing website. AQ&U also provides a rich framework for citizen science. Here, we discuss AQ&U’s community engagement strategies, particularly those to encourage participation of under-served communities, and highlight two of our most successful efforts to engage students as citizen scientists. The first effort centers around a hands-on activity to build and test an aerosol sensor from LegosTM and simple microelectronics. It emphasizes the underlying principles of aerosol light scattering. This has been our college’s most requested outreach activity and has been successfully demonstrated at hundreds of high-school classrooms, numerous STEM tabling events, and the National Science and Engineering Fair. The second effort focuses on an interactive, team-based teaching module using local real-world measurements. This activity's goal is to engage students in generating and testing hypotheses while also encouraging citizen scientists to use real-world air quality data for their own interests, such as exploration, science fair projects, or environmental oversight. This second activity has been a successful strategy for virtual learning. Finally, we discuss lessons learned. COI Declaration: Drs. Kelly, Gaillardon, and Whitaker have a financial interest in Tetrad Sensor Network Solutions, LLC, which commercializes environmental measurement technologies.

Additional Authors: Kerry Kelly, Anthony Butterfield, Katrina Le, James Moore, Wei Xing, Tom Becnel, Pierre-Emmanuel Gaillardon, Miriah Meyer, Ross Whitaker

Championing Environmental Awareness in Communities Through Air Sensor Loan Programs

Ryder Freed, U.S. EPA R9

Air sensors are exploratory and educational tools for communities. In recent years, the public has increasingly turned to these technologies to measure air quality. In response, U.S. EPA established pilot air sensor loan programs to better facilitate access to sensors and provide resources necessary for operating them. The loan programs include a living museum [The Morton Arboretum (IL)], Tribal Partners [Heritage University on the Yakama Reservation (WA), the Nez Perce Tribe (ID), and the Institute for Tribal Environmental Professionals (ITEP)], and national, urban, and rural libraries [Los Angeles Public Library System (CA), Evansville-Vanderburgh Library (IN), L’Anse Public Library (MI), Bayliss Library/Superior District (MI), and the Nez Perce Tribal Community libraries (ID)]. These programs offer community members an opportunity to borrow particulate matter (PM) sensors (AirBeam2 and/or PurpleAir) to collect PM measurements to learn about local air quality and pollution sources or to collect data to support community science projects. As part of the loan programs, five lesson plans were developed to guide participants through investigations of (1) outdoor air, (2) indoor air, (3) personal exposure, (4) vegetative barriers, and (5) smoke. These interactive lesson plans were shared with librarians and tribal air quality and educational professionals who plan to use the materials in educational programming with communities. The materials will also be posted to EPA’s Air Sensor Toolbox webpage [https://www.epa.gov/air-sensor-toolbox/educational-resources-related-air-sensor-technology] for anyone to use. An overview of these innovative, EPA-supported loan programs will be discussed, along with lessons learned. 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.

Additional Authors: Ryder Freed, Sheila Batka, India Young, Christi Duboiski, Sarah Waldo, Megan Gavin, Andrea Clements, Rachelle Duvall

Using Air Monitoring Projects to Plant Social, Academic, and Economic Seeds in African American Youth

Gwen Smith, Community Health Aligning Revitalization Resilience & Sustainability (CHARRS

To maximize the future impact of the air monitoring field, projects must be more intentional about depositing resources and skills in youth and focus on greater integration of challenges impacted communities directly face. This presentation will discuss specific challenges African American youth face in gaining access to the field and how citizen science and air monitoring projects are used as tools to improve math and science performance and exposure to long-term career opportunities.

Additional Authors: N/A