Low-cost Air Pollution Sensors Research Articles

Data Evaluation of a Low-Cost Sensor Network for Atmospheric Particulate Matter Monitoring in 15 Municipalities in Serbia.

Conventional air quality monitoring networks typically tend to be sparse over areas of interest. Because of the high cost of establishing such monitoring systems, some areas are often completely left out of regulatory monitoring networks. Recently, a new paradigm in monitoring has emerged that utilizes low-cost air pollution sensors, thus making it possible to reduce the knowledge gap in air pollution levels for areas not covered by regulatory monitoring networks and increase the spatial resolution of monitoring in others. The benefits of such networks for the community are almost self-evident since information about the level of air pollution can be transmitted in real time and the data can be analysed immediately over the wider area. However, the accuracy and reliability of newly produced data must also be taken into account in order to be able to correctly interpret the results. In this study, we analyse particulate matter pollution data from a large network of low-cost particulate matter monitors that was deployed and placed in outdoor spaces in schools in central and western Serbia under the Schools for Better Air Quality UNICEF pilot initiative in the period from April 2022 to June 2023. The network consisted of 129 devices in 15 municipalities, with 11 of the municipalities having such extensive real-time measurements of particulate matter concentration for the first time. The analysis showed that the maximum concentrations of PM2.5 and PM10 were in the winter months (heating season), while during the summer months (non-heating season), the concentrations were several times lower. Also, in some municipalities, the maximum values and number of daily exceedances of PM10 (50 μg/m3) were much higher than in the others because of diversity and differences in the low-cost sensor sampling sites. The particulate matter mass daily concentrations obtained by low-cost sensors were analysed and also classified according to the European AQI (air quality index) applied to low-cost sensor data. This study confirmed that the large network of low-cost air pollution sensors can be useful in providing real-time information and warnings about higher pollution days and episodes, particularly in situations where there is a lack of local or national regulatory monitoring stations in the area.

Machine Learning-based Calibration Approach for Low-cost Air Pollution Sensors MQ-7 and MQ-131

Air quality is a vital concern globally, and Sri Lanka, according to WHO statistics, faces challenges in achieving optimal air quality levels. To address this, we introduced an innovative IoT-based Air Pollution Monitoring (APM) Box. This solution incorporates readily available Commercial Off-The-Shelf (COTS) sensors, specifically MQ-7 and MQ-131, for measuring concentrations of Carbon Monoxide (CO) and Ozone (O3) ,Arduino and "ThingSpeak" platform. Yet, those COTS sensors are not factory-calibrated. Therefore, we implemented machine learning algorithms, including linear regression and deep neural network models, to enhance the accuracy of CO and O3 concentration measurements from these non-calibrated sensors. Our findings indicate promising correlations when dealing with MQ-7 and MQ-131 measurements after removing outliers.

A DYNAMIC SPATIAL FILTERING APPROACH TO MITIGATE UNDERESTIMATION BIAS IN FIELD CALIBRATED LOW-COST SENSOR AIR POLLUTION DATA.

Low-cost air pollution sensors, offering hyper-local characterization of pollutant concentrations, are becoming increasingly prevalent in environmental and public health research. However, low-cost air pollution data can be noisy, biased by environmental conditions, and usually need to be field-calibrated by collocating low-cost sensors with reference-grade instruments. We show, theoretically and empirically, that the common procedure of regression-based calibration using collocated data systematically underestimates high air pollution concentrations, which are critical to diagnose from a health perspective. Current calibration practices also often fail to utilize the spatial correlation in pollutant concentrations. We propose a novel spatial filtering approach to collocation-based calibration of low-cost networks that mitigates the underestimation issue by using an inverse regression. The inverse-regression also allows for incorporating spatial correlations by a second-stage model for the true pollutant concentrations using a conditional Gaussian Process. Our approach works with one or more collocated sites in the network and is dynamic, leveraging spatial correlation with the latest available reference data. Through extensive simulations, we demonstrate how the spatial filtering substantially improves estimation of pollutant concentrations, and measures peak concentrations with greater accuracy. We apply the methodology for calibration of a low-cost PM2.5 network in Baltimore, Maryland, and diagnose air pollution peaks that are missed by the regression-calibration.

Urban vertical air pollution gradient and dynamics investigated with low-cost sensors and large-eddy simulations

A network of five low-cost air pollution sensor (LCS) nodes was deployed vertically on the exterior of the H. C. Ørsted Institute at the University of Copenhagen, Denmark, to investigate the transport of pollution from the road below. All LCS nodes measured PM2.5, NO2, and O3 at 1-min time resolution, and one of them also measured noise. Traffic was monitored with a webcam, where traffic type and levels were derived using a machine-learning algorithm. We investigated how well traffic-related air pollution, noise, and real-time traffic counts serve as proxies for one another. The correlations between NO2, noise, and traffic count exhibited relatively low values when considering all the data. However, these correlations significantly increased under southwesterly wind direction and low wind speed, reaching R2 = 0.40 for NO2 and noise, R2 = 0.51 for NO2 and traffic volume, and R2 = 0.70 for noise and traffic volume. These results indicate a common source, namely traffic, for all three parameters. The five LCS nodes spanning 25 m vertically had extremely low intervariability with minimum R2-values of 0.98 for PM2.5, 0.89 for NO2, and 0.97 for O3. The system could not detect a vertical gradient in pollution levels. Large-eddy simulation model runs using the PALM model system generally supported the lack of gradient observed in measured observations. Under slightly unstable stratification, concentration remained relatively constant with height for southwesterly and southerly winds. Conversely, winds from the north, west, and northwest showed an increase in concentration with height. For other wind directions, the concentration decreased with height by approximately 40 % to 50 %, which is not as strong as for neutral stratification, attributed to enhanced vertical mixing under unstable stratification. Based on the measurements and modeling, we conclude that the vertical concentration profile is very sensitive to stratification, and under these conditions, the concentration outside the window of a fifth-floor office is almost the same as for an office on the ground floor.

Application of Machine Learning methods to correct the readings of low-cost air pollution sensors

The study is devoted to the analysis of the application of machine learning methods for correcting the readings of inexpensive sensors that record the concentration of suspended particles PM2.5 in the surface layer of the atmosphere, relative to readings of reference stations. The analysis was carried out on the example of coupled sensors (an inexpensive CityAir sensor and a reference E-BAM) located in Krasnoyarsk (Russia) based on observational data from January 1, 2019 to December 31, 2022. Statistical analysis of the data and comparison of parametric (Linear, Ridge, Lasso, Support vectors machine, Elastic net regressions) and nonparametric (regressions of Nearest Neighbor, Decision Tree and Random Forest) methods for establishing the relationship between two samples was carried out.

Are dense networks of low-cost nodes really useful for monitoring air pollution? A case study in Staffordshire

Abstract. Air pollution exhibits hyper-local variation, especially near emissions sources. In addition to people's time-activity patterns, this variation is the most critical element determining exposure. Pollution exposure is time-activity- and path-dependent, with specific behaviours such as mode of commuting and time spent near a roadway or in a park playing a decisive role. Compared to conventional air pollution monitoring stations, nodes containing low-cost air pollution sensors can be deployed with very high density. In this study, a network of 18 nodes using low-cost air pollution sensors was deployed in Newcastle-under-Lyme, Staffordshire, UK, in June 2020. Each node measured a range of species including nitrogen dioxide (NO2), ozone (O3), and particulate matter (PM2.5 and PM10); this study focuses on NO2 and PM2.5 over a 1-year period from 1 August 2020 to 1 October 2021. A simple and effective temperature, scale, and offset correction was able to overcome data quality issues associated with temperature bias in the NO2 readings. In its recent update, the World Health Organization (WHO) dramatically reduced annual exposure limit values from 40 to 10 µg m−3 for NO2 and from 10 to 5 µg m−3 for PM2.5. We found that the average annual mean NO2 concentration for the network was 17.5 µg m−3 and 8.1 µg m−3 for PM2.5. While in exceedance of the WHO guideline levels, these average concentrations do not exceed legally binding UK/EU standards. The network average NO2 concentration was 12.5 µg m−3 higher than values reported by a nearby regional air quality monitoring station, showing the critical importance of monitoring close to sources before pollution is diluted. We demonstrate how data from a low-cost air pollution sensor network can reveal insights into patterns of air pollution and help determine whether sources are local or non-local. With spectral analysis, we investigate the variation of the pollution levels and identify typical periodicities. Both NO2 and PM2.5 have contributions from high-frequency sources; however, the low-frequency sources are significantly different. Using spectral analysis, we determine that at least 54.3±4.3 % of NO2 is from local sources, whereas, in contrast, only 37.9±3.5 % of PM2.5 is local.

Data reconstruction applications for IoT air pollution sensor networks using graph signal processing

The analysis of sensor networks for air pollution monitoring is challenging. Recent studies have demonstrated the ability to reconstruct the network measurements with graphs derived from the acquired data, thus describing the complex relationships between the sensors that compose the network. In this work, we propose a graph-based data reconstruction framework that can be used to carry out different post-processing applications that arise in real-world low-cost sensor deployments for air pollution monitoring. This data reconstruction framework first describes the relationships between the different network sensors by means of a graph learned from the measured data, and then a signal reconstruction model is superimposed to reconstruct sensor data. This methodology allows reconstructing sensor data to carry out missing value imputation, signal reconstruction at points where there are no physical sensors (virtual sensing), or data fusion. The results, using real data taken with an air pollution monitoring network including reference stations and low-cost sensors, show how this framework performs well for these applications compared to estimates obtained from individual sensors and task-specific state-of-the-art methods that are not able to deal with this necessary range of applications. In short, the results show the potential of using graphs, whose topology is based on the measurements taken, to calibrate and reconstruct signals in heterogeneous networks of low-cost air pollution sensors for a wide variety of applications.

Field Evaluation and Calibration of Low-Cost Air Pollution Sensors for Environmental Exposure Research

This paper seeks to evaluate and calibrate data collected by low-cost particulate matter (PM) sensors in different environments and using different aggregated temporal units (i.e., 5-s, 1-min, 10-min, 30 min intervals). We first collected PM concentrations (i.e., PM1, PM2.5, and PM10) data in five different environments (i.e., indoor and outdoor of an office building, a train platform and lobby of a subway station, and a seaside location) in Hong Kong, using five AirBeam2 sensors as the low-cost sensors and a TSI DustTrak DRX Aerosol Monitor 8533 as the reference sensor. By comparing the collected PM concentrations, we found high linearity and correlation between the data reported by the AirBeam2 sensors in different environments. Furthermore, the results suggest that the accuracy and bias of the PM data reported by the AirBeam2 sensors are affected by rainy weather and environments with high humidity and a high level of hygroscopic salts (i.e., a seaside location). In addition, increasing the aggregation level of the temporal units (i.e., from 5-s to 30 min intervals) increases the correlation between the PM concentrations obtained by the AirBeam2 sensors, while it does not significantly improve the accuracy and bias of the data. Lastly, our results indicate that using a machine learning model (i.e., random forest) for the calibration of PM concentrations collected on sunny days generates better results than those obtained with multiple linear models. These findings have important implications for researchers when designing environmental exposure studies based on low-cost PM sensors.

Community-Engaged Use of Low-Cost Sensors to Assess the Spatial Distribution of PM2.5 Concentrations across Disadvantaged Communities: Results from a Pilot Study in Santa Ana, CA.

PM2.5 is an air pollutant that is widely associated with adverse health effects, and which tends to be disproportionately located near low-income communities and communities of color. We applied a community-engaged research approach to assess the distribution of PM2.5 concentrations in the context of community concerns and urban features within and around the city of Santa Ana, CA. Approximately 183 h of one-minute average PM2.5 measurements, along with high-resolution geographic coordinate measurements, were collected by volunteer community participants using roughly two dozen low-cost AtmoTube Pro air pollution sensors paired with real-time GPS tracking devices. PM2.5 varied by region, time of day, and month. In general, concentrations were higher near the city’s industrial corridor, which is an area of concern to local community members. While the freeway systems were shown to correlate with some degree of elevated air pollution, two of four sampling days demonstrated little to no visible association with freeway traffic. Concentrations tended to be higher within socioeconomically disadvantaged communities compared to other areas. This pilot study demonstrates the utility of using low-cost air pollution sensors for the application of community-engaged study designs that leverage community knowledge, enable high-density air monitoring, and facilitate greater health-related awareness, education, and empowerment among communities. The mobile air-monitoring approach used in this study, and its application to characterize the ambient air quality within a defined geographic region, is in contrast to other community-engaged studies, which employ fixed-site monitoring and/or focus on personal exposure. The findings from this study underscore the existence of environmental health inequities that persist in urban areas today, which can help to inform policy decisions related to health equity, future urban planning, and community access to resources.

Machine learning calibration of low-cost NO<sub>2</sub> and PM<sub>10</sub> sensors: non-linear algorithms and their impact on site transferability

Abstract. Low-cost air pollution sensors often fail to attain sufficient performance compared with state-of-the-art measurement stations, and they typically require expensive laboratory-based calibration procedures. A repeatedly proposed strategy to overcome these limitations is calibration through co-location with public measurement stations. Here we test the idea of using machine learning algorithms for such calibration tasks using hourly-averaged co-location data for nitrogen dioxide (NO2) and particulate matter of particle sizes smaller than 10 µm (PM10) at three different locations in the urban area of London, UK. We compare the performance of ridge regression, a linear statistical learning algorithm, to two non-linear algorithms in the form of random forest regression (RFR) and Gaussian process regression (GPR). We further benchmark the performance of all three machine learning methods relative to the more common multiple linear regression (MLR). We obtain very good out-of-sample R2 scores (coefficient of determination) >0.7, frequently exceeding 0.8, for the machine learning calibrated low-cost sensors. In contrast, the performance of MLR is more dependent on random variations in the sensor hardware and co-located signals, and it is also more sensitive to the length of the co-location period. We find that, subject to certain conditions, GPR is typically the best-performing method in our calibration setting, followed by ridge regression and RFR. We also highlight several key limitations of the machine learning methods, which will be crucial to consider in any co-location calibration. In particular, all methods are fundamentally limited in how well they can reproduce pollution levels that lie outside those encountered at training stage. We find, however, that the linear ridge regression outperforms the non-linear methods in extrapolation settings. GPR can allow for a small degree of extrapolation, whereas RFR can only predict values within the training range. This algorithm-dependent ability to extrapolate is one of the key limiting factors when the calibrated sensors are deployed away from the co-location site itself. Consequently, we find that ridge regression is often performing as good as or even better than GPR after sensor relocation. Our results highlight the potential of co-location approaches paired with machine learning calibration techniques to reduce costs of air pollution measurements, subject to careful consideration of the co-location training conditions, the choice of calibration variables and the features of the calibration algorithm.

The effectiveness of drones in measuring particulate matter

The use of low-cost air pollution sensors mounted on drones is an exciting new approach to conduct affordable and highly resolved air quality measurements. However, the air flow around a drone consists of complex, unsteady turbulent structures which interact directly with ambient particulate matter and gases. These interactions influence the transport of fine particulate matter (e.g. PM2.5) and, subsequently, its measurement using on-board PM sensors. In this work, we aim to quantify this effect by conducting wind tunnel and open-air experiments using a quadrotor drone and an array of low-cost sensors and research grade instruments. Sensors and instruments were placed around the drone to assess the effect of mixing, dilution, and particle segregation in the wake generated by the drone. Our results indicate little impact on the shape of the particle size distribution but significant differences (p<0.05) in PM2.5 concentration ranging from -70% to greater than 400% when comparing measurements before and after the drone is turned on. These effects varied greatly with sensor location and across experiments. In the wind tunnel, the sensor least affected was located far upstream from the drone with concentration differences less than 7%. The sensor far downstream from the drone as well as one of the sensors to the side of the drone also showed relatively small concentration differences (<50%) compared to the other sensors. In the open-air tests, the sensor directly under the drone and one of the sensors to the side drone showed the smallest changes after the drone was turned on with concentration differences less than 20%. This study highlights many important factors to consider when working with low-cost sensors mounted on drones. It also lays the groundwork for more in-depth studies into how to effectively take air-quality measurements using drones.

Disparities in Growth of Low-Cost Air Pollution Sensor Deployment

Background: Low-cost air quality sensors offer the capability to measure and generate a large volume of real-time air pollution data. Two of the most popular low-cost sensors are PurpleAir, a stationary outdoor/indoor monitor; and AirBeam, a portable monitor paired with a smartphone app. Despite the low cost of entry, the popularity and usage of low-cost air quality sensor usage are not yet quantified, and it is unclear where they are being deployed. Methods: Geocoded data from PurpleAirs and AirBeams were downloaded and processed through their respective application programming interfaces. We examined the growth of the numbers of outdoor PurpleAir units and AirBeam mobile sampling sessions over time around the globe. We also evaluated whether there exist disparities in the number of low-cost sensors in the U.S across geographic, racial, and socioeconomic factors at the census tract-level. Results: In 2017 there were approximately 800 outdoor PurpleAir units, and by the end of 2019 this number grew to 5700 units. The total number of AirBeam sampling sessions grew from 1300 in 2013 to more than 80000 by 2019. Most of the sensors were deployed in North America and Europe, although there was significant growth in Africa and Asia in 2019. In the U.S., PurpleAirs were mostly located in the Western states, and within census tracts with high socioeconomic status and majority white populations. AirBeam sessions were mostly carried out in densely populated census tracts and in areas with lower median incomes. Conclusions: The rapid growth of the low-cost sensor deployment across the globe highlights the interest in high-resolution air pollution data. In the U.S., there were noticeable disparities across race, socioeconomic status, and urbanicity for low-cost sensor usage. The observed differences between PurpleAir and AirBeam suggest that their user-profiles and usage purpose are likely to differ.

Behind the scenes of planning for public participation: planning for air-quality monitoring with low-cost sensors

We report from an environmental monitoring project planning for public participation to evaluate low-cost air pollution sensors. With an ethnographic approach, we studied how challenges were expressed and negotiated in discussions among project members when planning to involve the public in monitoring with low-cost sensors. Data was collected through participant observation of project meetings. Our analysis shows that perceived challenges involve data quality (i.e. reliability and validity), support, creating a sense of ownership and trust, as well as how to handle a possible rearrangement of power between authorities and the public. In order for the project to have control over different parts of the process when planning for public participation, they cannot stay true to all of the goals. This study contributes to the understanding of factors that foster the use of community-based data, and on the barriers for engaging the public in policy issues.

Monitoring Excess Exposure to Air Pollution for Professional Drivers in London Using Low-Cost Sensors

In this pilot study, low-cost air pollution sensor nodes were fitted in waste removal trucks, hospital vans and taxis to record drivers’ exposure to air pollution in Central London. Particulate matter (PM 2.5 and PM 10 ), CO 2 , NO 2 , temperature and humidity were recorded in real-time with nodes containing low-cost sensors, an electrochemical gas sensor for NO 2 , an optical particle counter for PM 2.5 and PM 10 and a non-dispersive infrared (NDIR) sensor for CO 2 , temperature and relative humidity. An intervention using a pollution filter to trap PM and NO 2 was also evaluated. The measurements were compared with urban background and roadside monitoring stations at Honor Oak Park and Marylebone Road, respectively. The vehicle records show PM and NO 2 concentrations similar to Marylebone Road and a higher NO 2 -to-PM ratio than at Honor Oak Park. Drivers are exposed to elevated pollution levels relative to Honor Oak Park: 1.72 μ g m − 3 , 1.92 μ g m − 3 and 58.38 ppb for PM 2.5 , PM 10 , and NO 2 , respectively. The CO 2 levels ranged from 410 to over 4000 ppm. There is a significant difference in average concentrations of PM 2.5 and PM 10 between the vehicle types and a non-significant difference in the average concentrations measured with and without the pollution filter within the sectors. In conclusion, drivers face elevated air pollution exposure as part of their jobs.

Developing of Low-Cost Air Pollution Sensor-Measurements with the Unmanned Aerial Vehicles in Poland.

This article presents the capabilities and selected measurement results from the newly developed low-cost air pollution measurement system mounted on an unmanned aerial vehicle (UAV). The system is designed and manufactured by the authors and is intended to facilitate, accelerate, and ensure the safety of operators when measuring air pollutants. It allows the creation of three-dimensional models and measurement visualizations, thanks to which it is possible to observe the location of leakage of substances and the direction of air pollution spread by various types of substances. Based on these models, it is possible to create area audits and strategies for the elimination of pollution sources. Thanks to the usage of a multi-socket microprocessor system, the combination of nine different air quality sensors can be installed in a very small device. The possibility of simultaneously measuring several different substances has been achieved at a very low cost for building the sensor unit: 70 EUR. The very small size of this device makes it easy and safe to mount it on a small drone (UAV). Because of this device, many harmful chemical compounds such as ammonia, hexane, benzene, carbon monoxide, and carbon dioxide, as well as flammable substances such as hydrogen and methane, can be detected. Additionally, a very important function is the ability to perform measurements of PM2.5 and PM10 suspended particulates. Thanks to the use of UAV, the measurement is carried out remotely by the operator, which allows us to avoid the direct exposure of humans to harmful factors. A big advantage is the quick measurement of large spaces, at different heights above the ground, in different weather conditions. Because of the three-dimensional positioning from GPS receiver, users can plot points and use colors reflecting a concentration of measured features to better visualize the air pollution. A human-friendly data output can be used to determine the mostly hazardous regions of the sampled area.

The performance assessment of low-cost air pollution sensor in city and the prospect of the autonomous vehicle for air pollution reduction

Urban air pollution is clearly a constantly growing problem and the high level of urban air pollution has been shown to pose a significant risk to city dwellers. It is necessary to have low-cost sensors for data collection, ready data source allowing normal citizen to access to and gain information, and have prospect solutions (e.g. autonomous vehicle) for air pollution reduction in the city. The aim of the study is to illustrate the drivers behind the use of low-cost sensors and to review the performance of sensors. In addition, autonomous vehicle is expected to reduce pollution; therefore, the paper analyzes the benefits and adoption of autonomous vehicles in the future. The challenges and outlook for both low-cost sensor deployment and the adoption of self-driving vehicle will also be discussed. A literature review is used to obtain these aims. The study indicates that the main driver of low-cost sensor is to provide high-density spatiotemporal pollution data, assisting in creating emission inventories of pollutants and detecting pollution hotspots without capital investments. A number of performance aspects considered include the coefficients of determination (R2), variance (CV), repeatability, reproducibly and stability of the sensor. Self-driving vehicle is promising in the change of travel patterns and having impacts on the health of society. Technology and economic challenges, the willingness to use the sensor and the autonomous policy approach are the major challenges. In terms of the outlook, standard guidelines and the calibration methods for low-cost sensors, energy consumption savings and policy supporting for autonomous vehicle should be further investigated.

Novel Approaches to Air Pollution Exposure and Clinical Outcomes Assessment in Environmental Health Studies

An accurate assessment of pollutants’ exposure and precise evaluation of the clinical outcomes pose two major challenges to the contemporary environmental health research. The common methods for exposure assessment are based on residential addresses and are prone to many biases. Pollution levels are defined based on monitoring stations that are sparsely distributed and frequently distanced far from residential addresses. In addition, the degree of an association between outdoor and indoor air pollution levels is not fully elucidated, making the exposure assessment all the more inaccurate. Clinical outcomes’ assessment, on the other hand, mostly relies on the access to medical records from hospital admissions and outpatients’ visits in clinics. This method differentiates by health care seeking behavior and is therefore, problematic in evaluation of an onset, duration, and severity of an outcome. In the current paper, we review a number of novel solutions aimed to mitigate the aforementioned biases. First, a hybrid satellite-based modeling approach provides daily continuous spatiotemporal estimations with improved spatial resolution of 1 × 1 km2 and 200 × 200 m2 grid, and thus allows a more accurate exposure assessment. Utilizing low-cost air pollution sensors allowing a direct measurement of indoor air pollution levels can further validate these models. Furthermore, the real temporal-spatial activity can be assessed by GPS tracking devices within the individuals’ smartphones. A widespread use of smart devices can help with obtaining objective measurements of some of the clinical outcomes such as vital signs and glucose levels. Finally, human biomonitoring can be efficiently done at a population level, providing accurate estimates of in-vivo absorbed pollutants and allowing for the evaluation of body responses, by biomarkers examination. We suggest that the adoption of these novel methods will change the research paradigm heavily relying on ecological methodology and support development of the new clinical practices preventing adverse environmental effects on human health.

Examining spatiotemporal variability of urban particulate matter and application of high-time resolution data from a network of low-cost air pollution sensors

Traditional air monitoring approaches using regulatory monitors have historically been used to assess regional-scale trends in air pollutants across large geographical areas. Recent advances in air pollution sensor technologies could provide additional information about nearby sources, support the siting of regulatory monitoring stations, and improve our knowledge of finer-scale spatiotemporal variation of ambient air pollutants and their associated health effects. Sensors are now being developed that are much smaller and lower cost than traditional ambient air monitoring systems and are capable of being deployed as a network to provide greater coverage of a given area. The CitySpace project conducted by the US EPA and the Shelby County Health Department included the deployment of a network of 17 sensor pods using Alphasense OPC-N2 particulate matter (PM) sensors integrated with meteorological sensors in Memphis, TN for six months. Sensor pods were collocated with a federal equivalent method (FEM) tapered element oscillating microbalance (TEOM) monitor both before and after the primary study period. Six of the sensor pods were found to meet the data quality objective (DQO) of coefficient of determination (R2) greater than 0.5 when collocated with the TEOM. Seven pods were decommissioned before the end of the study due to mechanical failure. The six pods meeting the DQO were used to examine the spatiotemporal variability of fine PM (PM2.5) across the Memphis area. One site was found to have higher relative PM2.5 concentrations when compared to the other sites in the network. The 1-min data from this sensor pod were evaluated to quantify the regional urban background and local-scale contributions to PM2.5 at that monitoring location. This method found that approximately 20% of the PM2.5 was attributed to local sources at this location, compared to 9% at a local regulatory monitoring site. Additionally, the 1-min data were combined with 1-min wind speed and wind direction data to examine potential sources in the area using the nonparametric trajectory analysis (NTA) technique. This method geographically identified local source areas that contributed to the measured concentrations at the high reading sensor location throughout the course of the study.

Use of Low-Cost Air Pollution Sensors in the Adult Changes in Thought Air Pollution Study

Incorporating modern air pollution sensor technology into epidemiologic cohort studies is appealing because of its low cost, allowing much better spatial representation of pollution exposure. Exposure sampling design can also leverage emerging understanding from measurement error correction methods that the monitoring network design should be spatially compatible with the locations where the health study participants live. However, research has not yet been conducted to determine whether these low-cost sensor data are reliable, accurate, and consistent enough to use for epidemiological study exposure assessment. Our scientific goal is to estimate individual-level long-term average exposures to PM2.5 and oxides of nitrogen for inference about the effects of air pollution on brain health. To accomplish this we are developing new spatio-temporal pollution predictions based on recently collected low-cost sensor data combined with existing ambient monitoring data. I will discuss the necessary quality control criteria for using the low-cost sensor data in a spatio-temporal prediction model, including calibration of these measurements to federal reference measurements. I will present the improvement in the performance of the spatio-temporal predictions from adding the low-cost sensor data and discuss the implications of these results for improving inference from epidemiologic cohort studies such as the Adult Changes in Thought Air Pollution study.

A Survey on Sensor Calibration in Air Pollution Monitoring Deployments

Air pollution is a major concern for public health and urban environments. Conventional air pollution monitoring systems install a few highly accurate, expensive stations at representative locations. Their sparse coverage and low spatial resolution are insufficient to quantify urban air pollution and its impacts on human health and environment. Advances in low-cost portable air pollution sensors have enabled air pollution monitoring deployments at scale to measure air pollution at high spatiotemporal resolution. However, it is challenging to ensure the accuracy of these low-cost sensor deployments because the sensors are more error-prone than high-end sensing infrastructures and they are often deployed in harsh environments. Sensor calibration has proven to be effective to improve the data quality of low-cost sensors and maintain the reliability of long-term, distributed sensor deployments. In this paper, we review the state-of-the-art low-cost air pollution sensors, identify their major error sources, and comprehensively survey calibration models as well as network recalibration strategies suited for different sensor deployments. We also discuss limitations of exiting methods and conclude with open issues for future sensor calibration research.