Article

Combining Measurements from Mobile Monitoring and a Reference Site To Develop Models of Ambient Ultrafine Particle Number Concentration at Residences

Article

Combining Measurements from Mobile Monitoring and a Reference Site To Develop Models of Ambient Ultrafine Particle Number Concentration at Residences

If you want to read the PDF, try requesting it from the authors.

Abstract

Significant spatial and temporal variation in ultrafine particle (UFP; <100 nanometers in diameter) concentrations creates challenges in developing predictive models for epidemiological investigations. We compared the performance of land-use regression models built by combining mobile and stationary measurements (hybrid model) with a regression model built using mobile measurements only (mobile model) in Chelsea and Boston, MA (USA). In each study area, particle number concentration (PNC; a proxy for UFP) was measured at a stationary reference site and with a mobile laboratory driven along a fixed route during a ~1-year monitoring period. In comparing PNC measured at 20 residences and PNC estimates from hybrid and mobile models, the hybrid model showed higher Pearson correlations of natural log-transformed PNC (r=0.73 vs. 0.51 in Chelsea; r=0.74 vs. 0.47 in Boston) and lower root-mean-square error in Chelsea (0.61 vs. 0.72) but no benefit in Boston (0.72 vs. 0.71). All models over-predicted log-transformed PNC by 3-6% at residences, yet the hybrid model reduced the standard deviation of the residuals by 15% in Chelsea and 31% in Boston with better tracking of overnight decreases in PNC. Overall, the hybrid model considerably outperformed the mobile model and could offer reduced exposure error for UFP epidemiology.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... UFP concentrations were estimated using a recently developed highly-resolved land-use regression (LUR) model for the Boston area (Simon et al., 2018). The LUR model estimates the spatial variation of UFP throughout the study area at a 20 x 20 m resolution using spatially-resolved predictors (i.e., distance from major roadways, distance from different land uses, and distance from open spaces). ...
... The LUR model estimates the spatial variation of UFP throughout the study area at a 20 x 20 m resolution using spatially-resolved predictors (i.e., distance from major roadways, distance from different land uses, and distance from open spaces). A complete description of model development and assessment is described in Simon et al (2018). Briefly, UFP concentrations were measured via mobile monitoring along a fixed route in Boston (between 05:00 and 22:00; in all seasons; on all days of the week) and by continuously monitoring at a stationary reference site (24 hr/day, 7 days/week) near the center of the study area. ...
... This modeling framework-pairing an LUR model with measured UFP data at a reference site-was shown to perform moderately well when compared to an external dataset of 20 residential sites. Models had a Pearson correlation coefficient of 0.74 in Boston for UFP modeled at an hourly resolution [24]. The 20 x 20 m grid was superimposed over a map of block groups in Boston obtained from 2010 US Census TIGER Shapefiles using ArcMap 10.5.1. ...
Article
Full-text available
Little is known of the relationship between exposure to the smallest particles of air pollution and socio-demographic characteristics. This paper explores linkages between ultrafine particle (UFP) concentrations and indicators of both race/ethnicity and socioeconomic status in Boston, Massachusetts, USA. We used estimates of UFP based on a highly-resolved land-use regression model of concentrations. In multivariate linear regression models census block groups with high proportions of Asians were associated with higher levels of UFP in comparison to block groups with majority White or other minority groups. Lower UFP concentrations were associated with higher homeownership (indicating higher SES) and with higher female head of household (indicating lower socioeconomic status). One explanation for the results include the proximity of specific groups to traffic corridors that are the main sources of UFP in Boston. Additional studies, especially at higher geographic resolution, are needed in Boston and other major cities to better characterize UFP concentrations by sociodemographic factors.
... We were interested in determining whether a combined mobile and stationary modeling methodology (see the Supporting Information for details) 29 to predict UFP concentrations at an hourly temporal resolution and continuous spatial resolution could be applied with good accuracy both retrospectively and prospectively at the same temporal and spatial resolution. Our objectives were to (1) build longitudinal models for ambient particle number concentration (PNC; a proxy for UFP) 30 for estimating ambient UFP in urban neighborhoods and (2) assess model performance by comparing model results to measurements made at stationary sites including long-term monitoring sites both prospectively and retrospectively. ...
... Previously, we developed models to predict hourly average PNC at a continuous spatial resolution for study areas in Chelsea and Boston. 29 These models were developed by combining measurements from mobile monitoring and a stationary reference site collected over a 1 year period in each study area. We referred to these models as "hybrid" models because mobile and stationary data were used in their creation. ...
... PNC monitoring was performed for both model development and model evaluation (see Table S1 and Figure S1). For model development, monitoring was performed in the Chelsea study area continuously ( 29 we observed systematic overprediction at homes by our models due to the fact that models were developed from on-road concentrations. Therefore, to adjust models described in this manuscript, we used additional ambient PNC data collected at homes belonging to individuals in the BPRHS cohort to calculate a fixed intercept to offset the average model overprediction). ...
Article
Short-term exposure to ultrafine particles (UFP; <100 nanometers in diameter), which are present at high concentrations near busy roadways, is associated with markers of cardiovascular and respiratory disease risk. To date, few long-term studies (months to years) have been conducted due to the challenges of long-term exposure assignment. To address this we modified hybrid land-use regression models of particle number concentration (PNC; a proxy for UFP) for two study areas in Boston (MA, USA) by replacing the measured PNC term with an hourly model and adjusting for overprediction. The hourly PNC models used covariates for meteorology, traffic, and sulfur dioxide concentrations (marker of secondary particle formation). We compared model performance against long-term PNC data collected 9 years before and 3 years after the model-development period. Model predictions captured the major temporal variations in the data and model performance remained relatively stable retrospectively and prospectively. The Pearson correlation of modeled versus measured hourly log-transformed PNC at a long-term monitoring site for nine years prior was 0.74. Our results demonstrate that highly-resolved spatial-temporal models of PNC are capable of estimating ambient concentrations retrospectively and prospectively with good accuracy, giving us confidence in using these models in epidemiological studies.
... Sites (N = 69) had on average (SD) of 8090 (361) hourly readings, the equivalent of 337 [15] days of full sampling (See SI Table S3; note that this and many of the subsequent SI figures and tables also include results for NO and NO 2 ). Average (SD) hourly NOx concentrations were 16 [21] ppb (See SI Table S4). Sites had seasonal, daily, and hourly concentration patterns, with trends being more pronounced at some sites than others (See SI Figs. ...
... Annual average estimates Across the 69 monitor locations, gold standard annual average NOx concentrations had a median (IQR) of 14 [10][11][12][13][14][15][16][17][18][19][20][21] ppb and ranged from 3 to 56 ppb. Overall, the long-term and short-term sampling approach for each design had very similar distributions. ...
Article
Full-text available
Background Short-term mobile monitoring campaigns to estimate long-term air pollution levels are becoming increasingly common. Still, many campaigns have not conducted temporally-balanced sampling, and few have looked at the implications of such study designs for epidemiologic exposure assessment. Objective We carried out a simulation study using fixed-site air quality monitors to better understand how different short-term monitoring designs impact the resulting exposure surfaces. Methods We used Monte Carlo resampling to simulate three archetypal short-term monitoring sampling designs using oxides of nitrogen (NOx) monitoring data from 69 regulatory sites in California: a year-around Balanced Design that sampled during all seasons of the year, days of the week, and all or various hours of the day; a temporally reduced Rush Hours Design; and a temporally reduced Business Hours Design. We evaluated the performance of each design’s land use regression prediction model. Results The Balanced Design consistently yielded the most accurate annual averages; while the reduced Rush Hours and Business Hours Designs generally produced more biased results. Significance A temporally-balanced sampling design is crucial for short-term campaigns such as mobile monitoring aiming to assess long-term exposure in epidemiologic cohorts. Impact statement Short-term monitoring campaigns to assess long-term air pollution trends are increasingly common, though they rarely conduct temporally balanced sampling. We show that this approach produces biased annual average exposure estimates that can be improved by collecting temporally-balanced samples.
... Follow-up studies have also found significant associations between PM 0.1 and reproductive outcomes, including low birth weight and preterm birth (Laurent et al., 2016;Bergin et al., 1996). These findings have biological plausibility, since ultrafine particles may cross cell membranes and interfere with internal cell function (Sioutas et al., 2005). Ultrafine particles have greater surface area per volume due to the small particle diameter, making them more available for chemical reaction. ...
... Various methods, such as the source-resolved PMCAMx chemical transport model, the chemical mass balance (CMB) model, photochemical box models, and land use regression (LUR) models, have been used to track source contributions to primary organic matter, elemental carbon, and in some cases particle number concentration (N x ) over areas in the Eastern US and parts of Europe and Asia (Lane et al., 2007;Posner and Pandis, 2015;Wang et al., 2011;Cattani et al., 2017;Wolf et al., 2017;Simon et al., 2018;Gaydos et al., 2005;Zhong et al., 2018). However, these methods are limited in one or more aspects of their ability to predict population exposure to ultrafine particles over large analysis domains. ...
Article
Full-text available
The regional concentrations of airborne ultrafine particulate matter mass (Dp<0.1 µm; PM0.1) were predicted in 39 cities across the United States (US) during summertime air pollution episodes. Calculations were performed using a regional source-oriented chemical transport model with 4 km spatial resolution operating on the National Emissions Inventory created by the U.S. Environmental Protection Agency (EPA). Measured source profiles for particle size and composition between 0.01 and 10 µm were used to translate PM total mass to PM0.1. Predicted PM0.1 concentrations exceeded 2 µg m⁻³ during summer pollution episodes in major urban regions across the US including Los Angeles, the San Francisco Bay Area, Houston, Miami, and New York. PM0.1 spatial gradients were sharper than PM2.5 spatial gradients due to the dominance of primary aerosol in PM0.1. Artificial source tags were used to track contributions to primary PM0.1 and PM2.5 from 15 source categories. On-road gasoline and diesel vehicles made significant contributions to regional PM0.1 in all 39 cities even though peak contributions within 0.3 km of the roadway were not resolved by the 4 km grid cells. Cooking also made significant contributions to PM0.1 in all cities but biomass combustion was only important in locations impacted by summer wildfires. Aviation was a significant source of PM0.1 in cities that had airports within their urban footprints. Industrial sources, including cement manufacturing, process heating, steel foundries, and paper and pulp processing, impacted their immediate vicinity but did not significantly contribute to PM0.1 concentrations in any of the target 39 cities. Natural gas combustion made significant contributions to PM0.1 concentrations due to the widespread use of this fuel for electricity generation, industrial applications, residential use, and commercial use. The major sources of primary PM0.1 and PM2.5 were notably different in many cities. Future epidemiological studies may be able to differentiate PM0.1 and PM2.5 health effects by contrasting cities with different ratios of PM0.1∕PM2.5. In the current study, cities with higher PM0.1∕PM2.5 ratios (ratio greater than 0.10) include Houston, TX, Los Angeles, CA, Bakersfield, CA, Salt Lake City, UT, and Cleveland, OH. Cities with lower PM0.1 to PM2.5 ratios (ratio lower than 0.05) include Lake Charles, LA, Baton Rouge, LA, St. Louis, MO, Baltimore, MD, and Washington, D.C.
... The feasibility of deploying dense monitoring networks has increased with the availability of inexpensive sensors, although questions about sensor accuracy continue to be studied (e.g., Borrego et al., 2016;Castell et al., 2017;Li and Biswas, 2017;Schneider et al., 2017;Lim et al., 2019). Approaches that combine mobile monitoring with measurements made at stationary monitoring locations (Adams et al., 2012;Simon et al., 2018) or with modeling (Messier et al., 2018) are being actively researched. ...
Article
Full-text available
Mobile-platform measurements provide new opportunities for characterizing spatial variations in air pollution within urban areas, identifying emission sources, and enhancing knowledge of atmospheric processes. The Aclima, Inc., mobile measurement and data acquisition platform was used to equip four Google Street View cars with research-grade instruments, two of which were available for the duration of this study. On-road measurements of air quality were made during a series of sampling campaigns between May 2016 and September 2017 at high (i.e., 1 s) temporal and spatial resolution at several California locations: Los Angeles, San Francisco, and the northern San Joaquin Valley (including nonurban roads and the cities of Tracy, Stockton, Manteca, Merced, Modesto, and Turlock). The results demonstrate that the approach is effective for quantifying spatial variations in air pollutant concentrations over measurement periods as short as 2 weeks. Measurement accuracy and precision are evaluated using results of weekly performance checks and periodic audits conducted through the sampler inlets, which show that research instruments located within stationary vehicles are capable of reliably measuring nitric oxide (NO), nitrogen dioxide (NO2), ozone (O3), methane (CH4), black carbon (BC), and particle number (PN) concentration, with bias and precision ranging from
... In contrast, measurements of PM0.1 are limited to focused field campaigns lasting for short time periods with even fewer studies attempting source apportionment calculations (Kleeman, Riddle et al. 2009). Multiple barriers have prevented the widespread deployment of PM0.1 monitoring networks including (i) 70 the low concentration of PM0.1 mass, which challenges the detection limits of analytical methods, of Europe and Asia (Gaydos, Stanier et al. 2005, Lane, Pinder et al. 2007, Wang, Hopke et al. 2011, Posner and Pandis 2015, Cattani, Gaeta et al. 2017, Wolf, Cyrys et al. 2017, Simon, Patton et al. 2018, Zhong, Nikolova et al. 2018). However, these methods are limited in one or more aspects of their ability to predict population exposure ultrafine particles over large analysis domains. ...
Article
Full-text available
The regional concentration of airborne ultrafine particulate matter mass (Dp µm; PM0.1) was predicted with 4km resolution in 39 cities across the United States during summer time air pollution episodes. Calculations were performed using a regional chemical transport model with 4km spatial resolution operating on the National Emissions Inventory created by the US EPA. Measured source profiles for particle size and composition between 0.01–10µm were used to translate PM total mass to PM0.1. PM0.1 concentrations exceeded 2µgm⁻³ during summer pollution episodes in major urban regions across the US including Los Angeles, the San Francisco Bay Area, Houston, Miami, and New York. PM0.1 spatial gradients were sharper than PM2.5 spatial gradients due to the dominance of primary aerosol in PM0.1. Artificial source tags were used to track contributions to primary PM0.1 and PM2.5 from fifteen source categories. As expected, on-road gasoline and diesel vehicles made significant contributions to regional PM0.1 in all 39 cities even though peak contributions within 0.3km of the roadway were not resolved by the 4km grid cells. Food cooking also made significant contributions to PM0.1 in all cities but biomass combustion was only important in locations impacted by summer wildfires. Aviation was a significant source of PM0.1 in cities that had airports within their urban footprints. Industrial sources including cement manufacturing, process heating, steel foundries, and paper & pulp processing impacted their immediate vicinity but did not significantly contribute to PM0.1 concentrations in any of the target 39 cities. Natural gas combustion made significant contributions to PM0.1 concentrations due to the widespread use of this fuel for electricity generation, industrial applications, residential, and commercial use. The major sources of primary PM0.1 and PM2.5 were notably different in many cities. Future epidemiological studies may be able to differentiate PM0.1 and PM2.5 health effects by contrasting cities with different ratios of PM0.1/PM2.5. In the current study, cities with higher PM0.1/PM2.5 ratios include Houston TX, Los Angeles CA, Birmingham AL, Charlotte NC, and Bakersfield CA. Cities with lower PM0.1 to PM2.5 ratios include Lake Charles LA, Baton Rouge LA, St. Louis MO, Baltimore MD, and Washington DC.
... The first lesson that I would point to is the need for ambient models of near highway pollutants that have fine grain temporal and spatial resolution. We have produced models using either mobile monitoring alone or with stationary fixed site monitoring to improve temporal prediction [10,11]. These models were at hourly resolution for a year with 20 m spatial resolution. ...
Article
Full-text available
The mainstay of air pollution health research has been fine particulate matter pollution (PM2 [...]
... Although recently developed low-cost monitors do not yet provide the accuracy and sensitivity of scientific-grade instruments, their low cost means that greater density of measurements can be obtained for longer periods of time [40,41]. Furthermore, various types of mobile vehicle-based monitoring stations have been developed for measurements of fine-scale spatial variations that can be used for LUR estimates [42][43][44]. ...
Article
Exposure to air pollution is associated with enhanced risk of developing asthma, notably in the presence of genetic risk factors. Interaction analyses have shown that both outdoor and indoor air pollution interact with genetic variability to increase the incidence of asthma. In this review, we summarize recent progress in candidate gene-based studies, as well as genome-wide gene-air pollution interaction studies. Advances in epigenetics have provided evidence for DNA methylation as a mediator in gene-air pollution interactions. Emerging strategies for study design and statistical analyses may improve power in future studies. Improved air pollution exposure assessment methods and asthma endo-typing can also be expected to increase the ability to detect biologically driven gene-air pollution interaction effects.
... 21 Mobile monitoring has also provided observations for inversion modeling using Lagrangian dispersion model footprints 13 and for building land-use regression exposure models. 8,9,24 Using multineighborhood, on-road mobile monitoring, Apte et al. 1 found that BC and NO x concentrations can vary by >5× within individual city blocks and semi-quantitatively identified hotspot locations using the median concentrations from repeated drive passes through 30 m road segments. These temporally aggregated concentration estimates are typically reported with ±10 to ±30% precision, depending on the pollutant and study region, 1,19 yet are summarized using domain-wide average uncertainties and generally do not report instrument drifts over time or between mobile vehicles. ...
Article
Diverse urban air pollution sources contribute to spatially variable atmospheric concentrations, with important public health implications. Mobile monitoring shows promise for understanding spatial pollutant patterns, yet it is unclear whether uncertainties associated with temporally sparse sampling and instrument performance limit our ability to identify locations of elevated pollution. To address this question, we analyze 9 months of repeated weekday daytime on-road mobile measurements of black carbon (BC), particle number (PN), and nitrogen oxide (NO, NO2) concentrations within 24 census tracts across Houston, Texas. We quantify persistently elevated, intermittent, and extreme concentration behaviors at 50 m road segments on surface streets and 90 m segments on highways relative to median statistics across the entire sampling domain. We find elevated concentrations above uncertainty levels (±40%) within portions of every census tract, with median concentration increases ranging from 2 to 3× for NO2, and >9× for NO. In contrast, PN exhibits elevated concentrations of 1.5-2× the domain-wide median and distinct spatial patterns relative to other pollutants. Co-located elevated concentrations of primary combustion tracers (BC and NO x ) near 30% of metal recycling and concrete batch plant facilities within our sampled census tracts are comparable to those measured within 200 m of highways. Our results demonstrate how extensive mobile monitoring across multiple census tracts can quantitatively characterize urban air pollution source patterns and are applicable to developing effective source mitigation policies.
... Over the years the partnership has grown, contributing substantially to the extant literature as it relates to near-roadway pollution, especially ultrafine particles [16][17][18][19]. Because of the CBPR framework in which community partners participate in the research decision-making, planning, implementation and dissemination, early findings have been used to inform the development of intervention studies designed to mitigate the health effects of near-roadway exposure [20]. ...
Article
Full-text available
Community-based participatory research (CBPR) aims to engage those traditionally left out of the research process. Partnering with community stakeholders to design, plan, implement and disseminate research can facilitate translation into practice. Using qualitative research methods, we set out to explore the policy and practice implications of a CBPR partnership focused on reducing exposure to near-roadway pollution. Key Informant interviews (n = 13) were conducted with individuals from various entities (municipal, state and private) for whom partners to the Community Assessment of Freeway Exposure and Health (CAFEH) provided technical assistance between 2013 and 2017. The findings indicate community research partnerships may have the power to inform local planning efforts. Developers and planners who the partnership consulted indicated a greater awareness of the implications of near-roadway exposure. They also described making changes in their practice based on study findings. The CAFEH partnership has demonstrated active attention to translating knowledge can influence local planning and practice, albeit with some challenges.
... Self-sampling corrections were not needed for the electric TAPL. Further, in the 2012-2013 campaign a butanol-based CPC was used that had a lower d 50 (4 nm) as compared to the water-based CPC (7 nm) used in the 2018-2020 campaigns; thus, an adjustment factor (based on side-by-side testing of the two CPCs) was used to allow direct comparison of the datasets (see Appendix section A2) (Simon et al., 2018). ...
Article
We investigated changes in traffic-related air pollutant concentrations in an urban area during the COVID-19 pandemic. The study area was in a mixed commercial-residential neighborhood in Somerville (MA, USA), where traffic is the dominant source of air pollution. Measurements were conducted between March 27 and May 14, 2020, coinciding with a dramatic reduction in traffic (71% drop in car and 46% drop in truck traffic) due to business shutdowns and a statewide stay-at-home advisory. Indicators of fresh vehicular emissions (ultrafine particle number concentration [PNC] and black carbon [BC]) were measured with a mobile monitoring platform on an interstate highway and major and minor roadways. Depending on road class, median PNC and BC contributions from traffic were 60–68% and 22–46% lower, respectively, during the lockdown compared to pre-pandemic conditions. A higher BC:PNC concentration ratio was observed during the lockdown period perhaps indicative of the higher fraction of diesel vehicles in the fleet during the lockdown. Overall, the scale of reductions in ultrafine particle and BC concentrations was commensurate with the reductions in traffic. This natural experiment allowed us to quantify the direct impacts of reductions in traffic emissions on neighborhood-scale air quality, which are not captured by the regional regulatory-monitoring network. Results underscore the importance of measurements of appropriate proxies for traffic emissions at relevant spatial scales. Our results are also useful for exposure analysts, and city, and regional planners evaluating traffic-related air pollution impact mitigation strategies.
... A recent study also suggests that a higher historical exposure to PM2.5 is, in particular, associated with a higher COVID-19 mortality rate [7]. Novel measurement approaches including mobile sampling (e.g., [8][9][10][11][12]) and low-cost air quality sensor networks (e.g., [13][14][15][16][17][18]) have had success in revealing urban air pollution patterns at considerably greater spatial precision than existing rather sparse regulatory air quality monitoring (AQM) stations, which have advanced our understanding of the adverse impacts of highly dynamic and heterogenous air pollutants, such as PM2.5, at higher spatiotemporal resolutions; however, these two approaches can be further complemented by a satellitebased modeling approach that requires much less manpower for sampling or instrument calibration and maintenance to potentially rapidly screen localized PM2.5 hotspots over wider spatial areas. ...
Article
Full-text available
Satellite-based rapid sweeping screening of localized PM2.5 hotspots at fine-scale local neighborhood levels is highly desirable. This motivated us to develop a random forest–convolutional neural network–local contrast normalization (RF–CNN–LCN) pipeline that detects local PM2.5 hotspots at a 300 m resolution using satellite imagery and meteorological information. The RF–CNN joint model in the pipeline uses three meteorological variables and daily 3 m/pixel resolution PlanetScope satellite imagery to generate daily 300 m ground-level PM2.5 estimates. The downstream LCN processes the estimated PM2.5 maps to reveal local PM2.5 hotspots. The RF–CNN joint model achieved a low normalized root mean square error for PM2.5 of within ~31% and normalized mean absolute error of within ~19% on the holdout samples in both Delhi and Beijing. The RF–CNN–LCN pipeline reasonably predicts urban PM2.5 local hotspots and coolspots by capturing both the main intra-urban spatial trends in PM2.5 and the local variations in PM2.5 with urban landscape, with local hotspots relating to compact urban spatial structures and coolspots being open areas and green spaces. Based on 20 sampled representative neighborhoods in Delhi, our pipeline revealed an annual average 9.2 ± 4.0 μg m−3 difference in PM2.5 between the local hotspots and coolspots within the same community. In some cases, the differences were much larger; for example, at the Indian Gandhi International Airport, the increase was 20.3 μg m−3 from the coolest spot (the residential area immediately outside the airport) to the hottest spot (airport runway). This work provides a possible means of automatically identifying local PM2.5 hotspots at 300 m in heavily polluted megacities and highlights the potential existence of substantial health inequalities in long-term outdoor PM2.5 exposures even within the same local neighborhoods between local hotspots and coolspots.
... These spikes are supported by high values of L-skewness and L-kurtosis suggesting temporal variability of deaths within our time series. Future studies must examine compound effects of additional contributors, like local air quality [76], extreme weather [77], specific calendar events (see Supplementary Table S1), an introduction of rapid over-the-counter tests [11] influencing viral transmission and thus, daily COVID-19 outcomes. Studies may also consider more exhaustive adjustment of confounding factors expected to influence case or death onset such as age, chronic health conditions, body mass index, etc. ...
Article
Full-text available
Critical temporal changes such as weekly fluctuations in surveillance systems often reflect changes in laboratory testing capacity, access to testing or healthcare facilities, or testing preferences. Many studies have noted but few have described day-of-the-week (DoW) effects in SARS-CoV-2 surveillance over the major waves of the novel coronavirus 2019 pandemic (COVID-19). We examined DoW effects by non-pharmaceutical intervention phases adjusting for wave-specific signatures using the John Hopkins University’s (JHU’s) Center for Systems Science and Engineering (CSSE) COVID-19 data repository from 2 March 2020 through 7 November 2021 in Middlesex County, Massachusetts, USA. We cross-referenced JHU’s data with Massachusetts Department of Public Health (MDPH) COVID-19 records to reconcile inconsistent reporting. We created a calendar of statewide non-pharmaceutical intervention phases and defined the critical periods and timepoints of outbreak signatures for reported tests, cases, and deaths using Kolmogorov-Zurbenko adaptive filters. We determined that daily death counts had no DoW effects; tests were twice as likely to be reported on weekdays than weekends with decreasing effect sizes across intervention phases. Cases were also twice as likely to be reported on Tuesdays-Fridays (RR = 1.90–2.69 [95%CI: 1.38–4.08]) in the most stringent phases and half as likely to be reported on Mondays and Tuesdays (RR = 0.51–0.93 [0.44, 0.97]) in less stringent phases compared to Sundays; indicating temporal changes in laboratory testing practices and use of healthcare facilities. Understanding the DoW effects in daily surveillance records is valuable to better anticipate fluctuations in SARS-CoV-2 testing and manage appropriate workflow. We encourage health authorities to establish standardized reporting protocols.
... The first lesson that I would point to is the need for ambient models of near highway pollutants that have fine grain temporal and spatial resolution. We have produced models using either mobile monitoring alone or with stationary fixed site monitoring to improve temporal prediction [10,11]. These models were at hourly resolution for a year with 20 m spatial resolution. ...
Article
There is growing concern that exposure to ultrafine particles (UFP, particles with diameter < 100 nm) may have health effects distinct from exposure to fine particulate matter mass (PM2.5). This investigative review examines spatial and temporal trends in UFP concentrations in North America. We analyze (i) multiyear (2006-2016) datasets from 11 stationary sampling sites and (ii) shorter duration, but highly spatially resolved mobile/stationary sampling data (2017-2019) in three cities (Baltimore, Oakland, and Pittsburgh). UFP concentrations have fallen by an average 30% over the past decade, similar to the reduction of PM2.5 mass concentration (35%). UFP reductions are likely a co-benefit of PM2.5 and other pollutant regulations. UFP have a factor of two to three spatial variation both within and between cities. Traffic is a major factor influencing intra-urban spatial variations. New particle formation (nucleation) is also an important source of UFP in many places. Regulations to reduce SO2 emissions from coal combustion have reduced nucleation events in the Eastern U.S., but some coastal areas with Mediterranean climates still have consistent new particle formation events. Highly spatially resolved UFP exposures in urban areas can be estimated using spatial models such as land use regression (LUR) fit to high spatial resolution data. Data collection for these models often uses mobile monitoring or other short-term sampling strategies because there is not a national-scale monitoring network for UFP. Short-term sampling produces LURs with modest (R² <0.5) performance; model performance can be improved with additional sampling. The current ability to estimate exposure at high spatial resolution over larger (e.g., national) scales is limited by a lack of data. We discuss strategies to improve UFP quantification and therefore exposure estimates.
Article
A spatiotemporal land use regression (LUR) model optimized to predict nitrogen dioxide (NO2) concentrations obtained from on-road, mobile measurements collected in 2015-16 was independently evaluated using concentrations observed at multiple sites across Toronto, Canada, obtained more than ten years earlier. This spatiotemporal LUR modelling approach improves upon estimates of historical NO2 concentrations derived from the previously used method of back-extrapolation. The optimal spatiotemporal LUR model (R²=0.71 for prediction of NO2 data in 2002 and 2004) uses daily average NO2 concentrations observed at multiple long-term monitoring sites and hourly average wind speed recorded at a single site, along with spatial predictors based on geographical information system data, to estimate NO2 levels for time periods outside of those used for model development. While the model tended to underestimate samplers located close to the roadway, it showed great accuracy when estimating samplers located beyond 100 m which are probably more relevant for exposure at residences. This study shows that spatiotemporal LUR models developed from strategic, multi-day (30 days in 3 different months) mobile measurements can enhance LUR model’s ability to estimate long-term, intra-urban NO2 patterns. Furthermore, the mobile sampling strategy enabled this new LUR model to cover a larger domain of Toronto and outlying suburban communities, thereby increasing the potential population for future epidemiological studies.
Article
Particle number concentration (PNC) is an important parameter for evaluating the environmental health and climate effects of particulate matter (PM). A good understanding of PNC is essential to control atmospheric ultrafine particles (UFP) and protect public health. In this study, we reviewed the PNC studies in the literature aimed to gain a comprehensive understanding about the levels, trends, and sources of PNC in China. The PNC levels at the urban, suburban, rural, remote, and coastal sites in China were 8500–52,200, 8600–30,300, 8600–28,400, 2100–16,100, and 5700-19,600 cm⁻³, respectively. The wide ranges of PNC indicate significant heterogeneity in the spatial distribution of PNC, but also are partly due to the different measurement techniques deployed in different studies. In general, it still can be concluded that the PNC levels at urban > suburban > rural > coastal > remote sites. Except for Mt. Waliguan (a remote site of 3816 m a.s.l.), other cities had the highest PNC in spring or winter and the lowest in summer or autumn. Long-term changes of PNCs in Beijing and Nanjing indicated that PNCs of Nucleation and Aitken modes had substantially declined following stricter emission controls in recent years, but more frequent NPF events were observed due to reduction in coagulation sink. Overall, traffic emission was the most dominant source of PNC in more than 94.4% of the selected cities around the world, while combustion2 (the energy production and industry related combustion source), background aerosol, and nucleation sources were also important contributors to PNC. This study provides insights about PNC and its sources around the world, especially in China. A few recommendations were suggested to further improve the understanding of PNC and to develop effective PNC control strategies.
Article
Limited research has been conducted in Asia on the association of maternal exposure to ambient air pollution and the increased risk of adverse pregnancy outcomes such as low birth weight and preterm birth. The aim of this study was to develop spatiotemporal land use regression (LUR) models for fine particulate matter (PM2.5) and nitrogen dioxide (NO2) in Chongqing, China, and to use the models to estimate PM2.5 and NO2 exposure for the participants in a randomized trial of complex lipid supplementation (the Complex Lipids In Mothers and Babies (CLIMB) study), before and during pregnancy. Spatiotemporal generalised additive models were developed for 2015–2016 on a daily basis incorporating measurement data from 16 sites, temporal variables on meteorology, and spatial variables produced using a geographical information system. Hold-out validation (HOV) was performed using daily and monthly averaged measurements for 2017 at 17 sites with 4 of the sites in different locations to 2015–16. The PM2.5 spatiotemporal model had good overall predictive ability (daily HOV correlation (COR)-R² = 0.75 and HOV mean-squared-error (MSE)-R² = 0.69; monthly HOV COR-R² = 0.87 and HOV MSE-R² = 0.76). The NO2 spatiotemporal model estimates had moderate-to-good correlation with measurements (daily HOV COR-R² = 0.44; monthly HOV COR-R² = 0.65), but estimates were subject to bias (daily HOV MSE-R² = 0.24; monthly HOV MSE-R² = −0.02). On this basis, we recommend that PM2.5 models are used for predicting absolute exposure and NO2 models are used for relative ranking of exposures.
Preprint
Full-text available
Mobile platform measurements provide new opportunities for characterizing spatial variations of air pollution within urban areas, identifying emission sources, and enhancing knowledge of atmospheric processes. The Aclima, Inc. mobile measurement and data acquisition platform was used to equip Google Street View cars with research-grade instruments. On-road measurements of air quality were made between May 2016 and September 2017 at high (i.e., 1-second [s]) temporal and spatial resolution at several California locations: Los Angeles, San Francisco, and the northern San Joaquin Valley (including non-urban roads and the cities of Tracy, Stockton, Manteca, Merced, Modesto, and Turlock). The results demonstrate that the approach is effective for quantifying spatial variations of air pollutant concentrations over measurement periods as short as two weeks. Measurement accuracy and precision are evaluated using results of weekly performance checks and periodic audits conducted through the sampler inlets, which show that research instruments in stationary vehicles are capable of reliably measuring nitric oxide (NO), nitrogen dioxide (NO2), ozone (O3), methane (CH4) black carbon (BC), and particle number (PN) concentration with bias and precision ranging from
Article
Background Particulate matter (PM) air pollution exposure has been linked to lung function in adolescents, but little is known about the relevance of specific PM components and ultrafine particles (UFP). Objectives To investigate the associations of long-term exposure to PM elemental composition and UFP with lung function at age 16 years. Methods For 706 participants of a prospective Dutch birth cohort, we assessed associations of forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) at age 16 with average exposure to eight elemental components (copper, iron, potassium, nickel, sulfur, silicon, vanadium and zinc) in PM2.5 and PM10, as well as UFP during the preceding years (age 13–16 years) estimated by land-use regression models. After assessing associations for each pollutant individually using linear regression models with adjustment for potential confounders, independence of associations with different pollutants was assessed in two-pollutant models with PM mass and NO2, for which associations with lung function have been reported previously. Results We observed that for most PM elemental components higher exposure was associated with lower FEV1, especially PM10 sulfur [e.g. adjusted difference −2.23% (95% confidence interval (CI) −3.70 to −0.74%) per interquartile range (IQR) increase in PM10 sulfur]. The association with PM10 sulfur remained after adjusting for PM10 mass. Negative associations of exposure to UFP with both FEV1 and FVC were observed [-1.06% (95% CI: −2.08 to −0.03%) and −0.65% (95% CI: −1.53 to 0.23%), respectively per IQR increase in UFP], but did not persist in two-pollutant models with NO2 or PM2.5. Conclusions Long-term exposure to sulfur in PM10 may result in lower FEV1 at age 16. There is no evidence for an independent effect of UFP exposure.
Article
Ultrafine particles (UFP) are quickly transformed within a few 100 m distance to the source because of aerosol dynamic processes. In order to predict the transformation of particle number size distributions (PNSD) near major streets, we developed a sectional model that represents aerosol dynamics within an air parcel that is transported from a major street along a minor street into the urban background. Simplifying assumptions on the transport and transformation due to dilution, coagulation, deposition and condensation were introduced to allow fast prediction with a limited number of model parameters. Model predictions were compared to observed PNSDs from semi-mobile measurement campaigns along 200 m long transects in Berlin, Germany. The total number concentration (TNC) along the transects declined on average by 30% for the evaluated measurement runs. The model agreed well with the observed PNSDs (coefficient of determination, R² = 0.94). The model was sensitive to the selection of a dilution parameter b (+8/-11% TNC change due to variation by factor 2) and friction velocity u∗ (up to 3% TNC change by replacing parameterized u∗ by constant values in a similar range), but almost not sensitive to the other parameters. According to the model, the dilution contributed the most to the TNC decline (approx. 73% after 200 m transport distance), followed by coagulation and deposition (13 and 14% contribution, respectively). Due to the low computational effort of the model, it may contribute to real-time forecasting of PNSDs near major streets and to determining short-term exposure to particles.
Article
Sampling strategies for collecting ultrafine particle (UFP) data to develop land use regression (LUR) models can strongly influence the resulting exposure estimates. Here we systematically examine how much sampling is needed to develop robust and stable UFP LUR models. To address this question, we collected 3-6 weeks continuous measurements of UFP concentrations at 32 sites in Pittsburgh, Pennsylvania covering a wide range of urban land-use attributes. Through systematic subsampling of this dataset, we evaluate the performance of hundreds of LUR models with varying numbers of sampling days and daily sampling durations. Our base LUR model derived from wintertime average concentrations explained about 80% of the spatial variability in the data (adjusted R2 ~ 0.8). The performance of the LUR models degrades with decreasing number of sampling days and sampling duration per day. For our dataset, 1-3 hours of sampling per day for 10-15 days provided UFP concentration estimates comparable to models derived from the entire dataset. Small numbers of repeated sampling per site (1-3 days) at short duration (~15-60 minutes per day) result in poor performance (R2 < 0.5), similar to previous UFP LUR models. This study provides guidelines for the design of future measurement campaigns and monitoring networks to generate robust UFP LUR models for exposure assessments. Further study in other locations with more sites is needed to evaluate these guidelines over a broader range of conditions.
Article
Most empirical air quality models (e.g., land use regression) focus on urban areas. Mobile monitoring for model development offers the opportunity to explore smaller, rural communities – an understudied population. We use mobile monitoring to systematically sample all daylight hours (7am-7pm) to develop empirical models capable of estimating hourly concentrations in Blacksburg, VA – a small town in rural Appalachia (population: 182,635). We collected ~120 hours of mobile monitoring data for Particle Number (PN) and Black Carbon (BC). We developed (1) daytime (12-hour average) models that approximate long-term concentrations and (2) spatiotemporal models for estimating hourly concentrations. Model performance for the daytime models is consistent with previous fixed-site and short-term sampling studies; adjusted R² (10-fold CV R²⁾ was 0.80 (0.69) for the PN model and 0.67 (0.58) for the BC model. The spatiotemporal models had comparable performance (10-fold CV R² for the PN [BC] models: 0.42 [0.25]) to previous mobile monitoring studies that isolate specific time periods. Temporal and spatial model coefficients had similar magnitudes in the spatiotemporal models suggesting both factors are important for exposure. We observed similar spatial patterns in Blacksburg (e.g., roadway gradients) as in other studies in urban areas suggesting similar exposure disparities exist in small, rural communities.
Article
Convolutional Neural Networks (CNNs) are a promising technique to predict highly localized fine particulate matter (i.e., PM2.5 levels) based on high-resolution satellite imagery. Unfortunately, CNNs typically require large amounts of supervised data to perform well, whereas this application generally has lots of unsupervised data (all satellite imagery) and relatively sparse supervised data (measurements from ground sensors). Previous work used transfer learning from another visual task to initialize the CNN weights; however, we hypothesize that standard transfer learning strategies would bias the CNN to focus on irrelevant details of the image for our applications. Instead, we develop a novel framework called Spatiotemporal Contrastive Learning (SCL) to pre-train the CNN. We test both regular contrastive learning and SCL on predicting PM2.5 levels from satellite images in two different cities, Delhi and Beijing, and compare to CNNs with parameters initialized randomly and by transfer learning. Our results show that regular contrastive learning and our SCL frameworks both manage to better capture spatial variation of ground-level PM2.5 concentrations compared to traditional initialization schemes, and that this performance gap increases as the number of ground sensors decreases, implying that the approach will be even more valuable in cities with fewer ground sensors. Our work demonstrates that contrastive learning is a powerful pre-training technique to build better spatial maps of PM2.5, and can be broadly applied in related situations.
Article
Full-text available
Purpose of review: Epidemiological studies of health effects of long-term exposure to outdoor air pollution rely on different exposure assessment methods. This review discusses widely used methods with a special focus on new developments. Recent findings: New data and study designs have been applied, including satellite measurements of fine particles and nitrogen dioxide (NO2). The methods to apply satellite data for epidemiological studies are improving rapidly and have already contributed significantly to national-, continental- and global-scale models. Spatiotemporal models have been developed allowing more detailed temporal resolution compared to spatial models. The development of hybrid models combining dispersion models, satellite observations, land use and surface monitoring has improved models substantially. Mobile monitoring designs to develop models for long-term UFP exposure have been conducted. Methods to assess long-term exposure to outdoor air pollution have improved significantly over the past decade. Application of satellite data and mobile monitoring designs is promising new methods.
Article
Full-text available
The epidemiological research benefits from an accurate characterization of both spatial and temporal variability of exposure to air pollution. This work aims at proposing a method to combine the high spatial resolution of Land Use Regression (LUR) models with the high temporal resolution of fixed site monitoring data, to model spatiotemporal variability of NO2 over a wide geographical area in Northern Italy. We developed seasonal LUR models to reconstruct the spatial distribution of a scaling factor that relates local concentrations to those measured at two reference central sites, one for the northern flat area and one for the southern mountain area. We calculated the daily average concentrations at 19 locations spread over the study areas as the product of the local scaling factor and the reference central site concentrations. We evaluated model performance comparing modeled and measured NO2 data. LUR model's R(2) ranges from 0.76 to 0.92. The main predictors refers substantially to traffic, industrial land use, buildings volume and altitude a.s.l. The model's performance in reproducing measured concentrations was satisfactory. The temporal variability of concentrations was well captured: Spearman correlation between model and measures was >0.7 for almost all sites. Model's average absolute errors were in the order of 10?gm(-3). The model for the southern area tends to overestimate measured concentrations. Our modeling framework was able to reproduce spatiotemporal differences in NO2 concentrations. This kind of model is less data-intensive than usual regional atmospheric models and it may be very helpful to assess population exposure within studies in which individual relevant exposure occurs along periods of days or months.
Article
Full-text available
Prior research on ultrafine particles (UFP) emphasizes that concentrations are especially high on-highway, and that time on highways contribute disproportionately to total daily exposures. This study estimates individual and population exposure to ultra-fine particles in the Minneapolis ? St. Paul (Twin Cities) metropolitan area, Minnesota. Our approach combines a real-time model of on-highway size-resolved UFP concentrations (32 bins, 5.5?600 nm); individual travel patterns, derived from GPS travel trajectories collected in 144 individual vehicles (123 h at locations with UFP estimates among 624 vehicle-hours of travel); and, loop-detector data, indicating real-time traffic conditions throughout the study area. The results provide size-resolved spatial and temporal patterns of exposure to UFP among freeway users. On-highway exposures demonstrate significant variability among users, with highest concentrations during commuting peaks and near highway interchanges. Findings from this paper could inform future epidemiological studies in on-road exposure to UFP by linking personal exposures to traffic conditions.
Article
Full-text available
Quantification of exposure to traffic-related air pollutants near highways is hampered by incomplete knowledge of the scales of temporal variation of pollutant gradients. The goal of this study was to characterize short-term temporal variation of vehicular pollutant gradients within 200–400 m of a major highway (>150 000 vehicles/d). Monitoring was done near Interstate 93 in Somerville (Massachusetts) from 06:00 to 11:00 on 16 January 2008 using a mobile monitoring platform equipped with instruments that measured ultrafine and fine particles (6–1000 nm, particle number concentration (PNC)); particle-phase (>30 nm) NO<sub>3</sub><sup>−</sup>, SO<sub>4</sub><sup>2−</sup>, and organic compounds; volatile organic compounds (VOCs); and CO<sub>2</sub>, NO, NO<sub>2</sub>, and O<sub>3</sub>. We observed rapid changes in pollutant gradients due to variations in highway traffic flow rate, wind speed, and surface boundary layer height. Before sunrise and peak traffic flow rates, downwind concentrations of particles, CO<sub>2</sub>, NO, and NO<sub>2</sub> were highest within 100–250 m of the highway. After sunrise pollutant levels declined sharply (e.g., PNC and NO were more than halved) and the gradients became less pronounced as wind speed increased and the surface boundary layer rose allowing mixing with cleaner air aloft. The levels of aromatic VOCs and NO<sub>3</sub><sup>−</sup>, SO<sub>4</sub><sup>2−</sup> and organic aerosols were generally low throughout the morning, and their spatial and temporal variations were less pronounced compared to PNC and NO. O<sub>3</sub> levels increased throughout the morning due to mixing with O<sub>3</sub>-enriched air aloft and were generally lowest near the highway reflecting reaction with NO. There was little if any evolution in the size distribution of 6–225 nm particles with distance from the highway. These results suggest that to improve the accuracy of exposure estimates to near-highway pollutants, short-term (e.g., hourly) temporal variations in pollutant gradients must be measured to reflect changes in traffic patterns and local meteorology.
Article
Full-text available
Ultrafine particles are emitted at high rates by jet aircraft. To determine the possible impacts of aviation activities on ambient ultrafine particle number concentrations (PNCs), we analyzed PNCs measured from 3 months to 3.67 years at three sites within 7.3 km of Logan International Airport (Boston, MA). At sites 4.0 and 7.3 km from the airport, average PNCs were 2- and 1.33-fold higher, respectively, when winds were from the direction of the airport compared to other directions, indicating that aviation impacts on PNC extend many kilometers downwind of Logan airport. Furthermore, PNCs were positively correlated with flight activity after taking meteorology, time of day and week, and traffic volume into account. Also, when winds were from the direction of the airport, PNCs increased with increasing wind speed, suggesting that buoyant aircraft exhaust plumes were the likely source. Concentrations of other pollutants [CO, black carbon (BC), NO, NO2, NOx, SO2, and fine particulate matter (PM2.5)] decreased with increasing wind speed when winds were from the direction of the airport, indicating a different dominant source (likely roadway traffic emissions). Except for oxides of nitrogen, other pollutants were not correlated with flight activity. Our findings point to the need for PNC exposure assessment studies to take aircraft emissions into consideration, particularly in populated areas near airports.
Article
Full-text available
Background: Long-term exposure to fine particulate matter has been linked to cardiovascular disease and systemic inflammatory responses; however, evidence is limited regarding the effects of long-term exposure to ultrafine particulate matter (UFP, <100nm). We used a cross-sectional study design to examine the association of long-term exposure to near-highway UFP with measures of systemic inflammation and coagulation. Methods: We analyzed blood samples from 408 individuals aged 40-91years living in three near-highway and three urban background areas in and near Boston, Massachusetts. We conducted mobile monitoring of particle number concentration (PNC) in each area, and used the data to develop and validate highly resolved spatiotemporal (hourly, 20m) PNC regression models. These models were linked with participant time-activity data to determine individual time-activity adjusted (TAA) annual average PNC exposures. Multivariable regression modeling and stratification were used to assess the association between TAA-PNC and single peripheral blood measures of high-sensitivity C-reactive protein (hsCRP), interleukin-6 (IL-6), tumor-necrosis factor alpha receptor II (TNFRII) and fibrinogen. Results: After adjusting for age, sex, education, body mass index, smoking and race/ethnicity, an interquartile-range (10,000particles/cm(3)) increase in TAA-PNC had a positive non-significant association with a 14.0% (95% CI: -4.6%, 36.2%) positive difference in hsCRP, an 8.9% (95% CI: -0.4%, 10.9%) positive difference in IL-6, and a 5.1% (95% CI: -0.4%, 10.9%) positive difference in TNFRII. Stratification by race/ethnicity revealed that TAA-PNC had larger effect estimates for all three inflammatory markers and was significantly associated with hsCRP and TNFRII in white non-Hispanic, but not East Asian participants. Fibrinogen had a negative non-significant association with TAA-PNC. Conclusions: Our findings suggest an association between annual average near-highway TAA-PNC and subclinical inflammatory markers of CVD risk.
Article
Full-text available
Background Land Use Regression (LUR) is a popular method to explain and predict spatial contrasts in air pollution concentrations, but LUR models for ultrafine particles, such as particle number concentration (PNC) are especially scarce. Moreover, no models have been previously presented for the lung deposited surface area (LDSA) of ultrafine particles. The additional value of ultrafine particle metrics has not been well investigated due to lack of exposure measurements and models. Methods Air pollution measurements were performed in 2011 and 2012 in the eight areas of the Swiss SAPALDIA study at up to 40 sites per area for NO2 and at 20 sites in four areas for markers of particulate air pollution. We developed multi-area LUR models for biannual average concentrations of PM2.5, PM2.5 absorbance, PM10, PMcoarse, PNC and LDSA, as well as alpine, non-alpine and study area specific models for NO2, using predictor variables which were available at a national level. Models were validated using leave-one-out cross-validation, as well as independent external validation with routine monitoring data. ResultsModel explained variance (R2) was moderate for the various PM mass fractions PM2.5 (0.57), PM10 (0.63) and PMcoarse (0.45), and was high for PM2.5 absorbance (0.81), PNC (0.87) and LDSA (0.91). Study-area specific LUR models for NO2 (R2 range 0.52–0.89) outperformed combined-area alpine (R2 = 0.53) and non-alpine (R2 = 0.65) models in terms of both cross-validation and independent external validation, and were better able to account for between-area variability. Predictor variables related to traffic and national dispersion model estimates were important predictors. ConclusionsLUR models for all pollutants captured spatial variability of long-term average concentrations, performed adequately in validation, and could be successfully applied to the SAPALDIA cohort. Dispersion model predictions or area indicators served well to capture the between area variance. For NO2, applying study-area specific models was preferable over applying combined-area alpine/non-alpine models. Correlations between pollutants were higher in the model predictions than in the measurements, so it will remain challenging to disentangle their health effects.
Article
Full-text available
Existing evidence suggests that ambient ultrafine particles (UFPs) (<0.1µm) may contribute to acute cardiorespiratory morbidity. However, few studies have examined the long-term health effects of these pollutants owing in part to a need for exposure surfaces that can be applied in large population-based studies. To address this need, we developed a land use regression model for UFPs in Montreal, Canada using mobile monitoring data collected from 414 road segments during the summer and winter months between 2011 and 2012. Two different approaches were examined for model development including standard multivariable linear regression and a machine learning approach (kernel-based regularized least squares (KRLS)) that learns the functional form of covariate impacts on ambient UFP concentrations from the data. The final models included parameters for population density, ambient temperature and wind speed, land use parameters (park space and open space), length of local roads and rail, and estimated annual average NOx emissions from traffic. The final multivariable linear regression model explained 62% of the spatial variation in ambient UFP concentrations whereas the KRLS model explained 79% of the variance. The KRLS model performed slightly better than the linear regression model when evaluated using an external dataset (R(2)=0.58 vs. 0.55) or a cross-validation procedure (R(2)=0.67 vs. 0.60). In general, our findings suggest that the KRLS approach may offer modest improvements in predictive performance compared to standard multivariable linear regression models used to estimate spatial variations in ambient UFPs. However, differences in predictive performance were not statistically significant when evaluated using the cross-validation procedure.
Article
Full-text available
Urban air pollution represents one of the greatest environmental challenges facing mankind in the 21st century. Noticeably, many developing countries, such as China and India, have experienced severe air pollution because of their fast-developing economy and urbanization. Globally, the urbanization trend is projected to continue: 70% of the world population will reside in urban centers by 2050, and there will exist 41 megacities (with more than 10 million inhabitants) by 2030. Air pollutants consist of a complex combination of gases and particulate matter (PM). In particular, fine PM (particles with the aerodynamic diameter smaller than 2.5 μm or PM_(2.5)) profoundly impacts human health, visibility, the ecosystem, the weather, and the climate, and these PM effects are largely dependent on the aerosol properties, including the number concentration, size, and chemical composition. PM is emitted directly into the atmosphere (primary) or formed in the atmosphere through gas-to-particle conversion (secondary) (Figure 1). Also, primary and secondary PM undergoes chemical and physical transformations and is subjected to transport, cloud processing, and removal from the atmosphere.
Article
Full-text available
We measured the spatial pattern of particle number (PN) concentrations downwind from the Los Angeles International Airport (LAX) with an instrumented vehicle that enabled us to cover larger areas than allowed by traditional stationary measurements. LAX emissions adversely impacted air quality much farther than reported in previous airport studies. We measured at least a 2-fold increase in PN concentrations over unimpacted baseline PN concentrations during most hours of the day in an area of about 60 km(2) that extended to 16 km (10 miles) downwind and a 4- to 5-fold increase to 8-10 km (5-6 miles) downwind. Locations of maximum PN concentrations were aligned to eastern, downwind jet trajectories during prevailing westerly winds and to 8 km downwind concentrations exceeded 75 000 particles/cm(3), more than the average freeway PN concentration in Los Angeles. During infrequent northerly winds, the impact area remained large but shifted to south of the airport. The freeway length that would cause an impact equivalent to that measured in this study (i.e., PN concentration increases weighted by the area impacted) was estimated to be 280-790 km. The total freeway length in Los Angeles is 1500 km. These results suggest that airport emissions are a major source of PN in Los Angeles that are of the same general magnitude as the entire urban freeway network. They also indicate that the air quality impact areas of major airports may have been seriously underestimated.
Article
Full-text available
Estimating ultrafine particle number concentrations (PNC) near highways for exposure assessment in chronic health studies requires models capable of capturing PNC spatial and temporal variations over the course of a full year. The objectives of this work were to describe the relationship between near-highway PNC and potential predictors, and to build and validate hourly log-linear regression models. PNC was measured near Interstate 93 (I-93) in Somerville, MA (USA) using a mobile monitoring platform driven for 234 hours on 43 days between August 2009 and September 2010. Compared to urban background, PNC levels were consistently elevated within 100-200 m of I-93, with gradients impacted by meteorological and traffic conditions. Temporal and spatial variables including wind speed and direction, temperature, highway traffic, and distance to I-93 and major roads contributed significantly to the full regression model. Cross-validated model R(2) values ranged from 0.38-0.47, with higher values achieved (0.43-0.53) when short-duration PNC spikes were removed. The model predicts highest PNC near major roads and on cold days with low wind speeds. The model allows estimation of hourly ambient PNC at 20-m resolution in a near-highway neighborhood.
Article
Full-text available
Epidemiological and clinical studies have linked exposure to particulate matter (PM) to adverse health effects, which may be registered as increased mortality and morbidity from various cardiopulmonary diseases. Despite the evidence relating PM to health effects, the physiological, cellular, and molecular mechanisms causing such effects are still not fully characterized. Two main approaches are used to elucidate the mechanisms of toxicity. One is the use of in vivo experimental models, where various effects of PM on respiratory, cardiovascular, and nervous systems can be evaluated. To more closely examine the molecular and cellular mechanisms behind the different physiological effects, the use of various in vitro models has proven to be valuable. In the present review, we discuss the current advances on the toxicology of particulate matter and nanoparticles based on these techniques.
Article
Full-text available
Ultrafine particles (UFP; aerodynamic diameter < 0.1 micrometers) are a ubiquitous exposure in the urban environment and are elevated near highways. Most epidemiological studies of UFP health effects use central site monitoring data, which may misclassify exposure. Our aims were to: (1) examine the relationship between distant and proximate monitoring sites and their ability to predict hourly UFP concentration measured at residences in an urban community with a major interstate highway and; (2) determine if meteorology and proximity to traffic improve explanatory power. Short-term (1 - 3 weeks) residential monitoring of UFP concentration was conducted at 18 homes. Long-term monitoring was conducted at two near-highway monitoring sites and a central site. We created models of outdoor residential UFP concentration based on concentrations at the near-highway site, at the central site, at both sites together and without fixed sites. UFP concentration at residential sites was more highly correlated with those at a near-highway site than a central site. In regression models of each site alone, a 10% increase in UFP concentration at a near-highway site was associated with a 6% (95% CI: 6%, 7%) increase at residences while a 10% increase in UFP concentration at the central site was associated with a 3% (95% CI: 2%, 3%) increase at residences. A model including both sites showed minimal change in the magnitude of the association between the near-highway site and the residences, but the estimated association with UFP concentration at the central site was substantially attenuated. These associations remained after adjustment for other significant predictors of residential UFP concentration, including distance from highway, wind speed, wind direction, highway traffic volume and precipitation. The use of a central site as an estimate of personal exposure for populations near local emissions of traffic-related air pollutants may result in exposure misclassification.
Article
Full-text available
openair is an R package primarily developed for the analysis of air pollution measurement data but which is also of more general use in the atmospheric sciences. The package consists of many tools for importing and manipulating data, and undertaking a wide range of analyses to enhance understanding of air pollution data. In this paper we consider the development of the package with the purpose of showing how air pollution data can be analysed in more insightful ways. Examples are provided of importing data from UK air pollution networks, source identification and characterisation using bivariate polar plots, quantitative trend estimates and the use of functions for model evaluation purposes. We demonstrate how air pollution data can be analysed quickly and efficiently and in an interactive way, freeing time to consider the problem at hand. One of the central themes of openair is the use of conditioning plots and analyses, which greatly enhance inference possibilities. Finally, some consideration is given to future developments.
Article
Full-text available
Accurate quantification of exposures to traffic-related air pollution in near-highway neighborhoods is challenging due to the high degree of spatial and temporal variation of pollutant levels. The objective of this study was to measure air pollutant levels in a near-highway urban area over a wide range of traffic and meteorological conditions using a mobile monitoring platform. The study was performed in a 2.3-km(2) area in Somerville, Massachusetts (USA), near Interstate I-93, a highway that carries 150,000 vehicles per day. The mobile platform was equipped with rapid-response instruments and was driven repeatedly along a 15.4-km route on 55 days between September 2009 and August 2010. Monitoring was performed in 4-6-hour shifts in the morning, afternoon and evening on both weekdays and weekends in winter, spring, summer and fall. Measurements were made of particle number concentration (PNC; 4-3,000 nm), particle size distribution, fine particle mass (PM(2.5)), particle-bound polycyclic aromatic hydrocarbons (pPAH), black carbon (BC), carbon monoxide (CO), and nitrogen oxides (NO and NO(x)). The highest pollutant concentrations were measured within 0-50 m of I-93 with distance-decay gradients varying depending on traffic and meteorology. The most pronounced variations were observed for PNC. Annual median PNC 0-50 m from I-93 was two-fold higher compared to the background area (>1 km from I-93). In general, PNC levels were highest in winter and lowest in summer and fall, higher on weekdays and Saturdays compared to Sundays, and higher during morning rush hour compared to later in the day. Similar spatial and temporal trends were observed for NO, CO and BC, but not for PM(2.5). Spatial variations in PNC distance-decay gradients were non-uniform largely due to contributions from local street traffic. Hour-to-hour, day-to-day and season-to-season variations in PNC were of the same magnitude as spatial variations. Datasets containing fine-scale temporal and spatial variation of air pollution levels near highways may help to inform exposure assessment efforts.
Article
Full-text available
Quantification of exposure to traffic-related air pollutants near highways is hampered by incomplete knowledge of the scales of temporal variation of pollutant gradients. The goal of this study was to characterize short-term temporal variation of vehicular pollutant gradients within 200-400 m of a major highway (> 150 000 vehicles/d). Monitoring was done near Interstate 93 in Somerville (Massachusetts) from 06:00 to 11:00 on 16 January 2008 using a mobile monitoring platform equipped with instruments that measured ultrafine and fine particles (6-1000 nm, particle number concentration (PNC)); particle-phase (> 30 nm) NO(3)(-), SO(4)(2-), and organic compounds; volatile organic compounds (VOCs); and CO(2), NO, NO(2), and O(3). We observed rapid changes in pollutant gradients due to variations in highway traffic flow rate, wind speed, and surface boundary layer height. Before sunrise and peak traffic flow rates, downwind concentrations of particles, CO(2), NO, and NO(2) were highest within 100-250 m of the highway. After sunrise pollutant levels declined sharply (e.g., PNC and NO were more than halved) and the gradients became less pronounced as wind speed increased and the surface boundary layer rose allowing mixing with cleaner air aloft. The levels of aromatic VOCs and NO(3)(-), SO(4)(2-) and organic aerosols were generally low throughout the morning, and their spatial and temporal variations were less pronounced compared to PNC and NO. O(3) levels increased throughout the morning due to mixing with O(3)-enriched air aloft and were generally lowest near the highway reflecting reaction with NO. There was little if any evolution in the size distribution of 6-225 nm particles with distance from the highway. These results suggest that to improve the accuracy of exposure estimates to near-highway pollutants, short-term (e.g., hourly) temporal variations in pollutant gradients must be measured to reflect changes in traffic patterns and local meteorology.
Article
Full-text available
The relationship between traffic emissions and mobile-source air pollutant concentrations is highly variable over space and time and therefore difficult to model accurately, especially in urban settings with complex terrain. Regression-based approaches using continuous real-time mobile measurements may be able to characterize spatiotemporal variability in traffic-related pollutant concentrations but require methods to incorporate temporally varying meteorology and source strength in a physically interpretable fashion. We developed a statistical model to assess the joint impact of both meteorology and traffic on measured concentrations of mobile-source air pollutants over space and time. In this study, traffic-related air pollutants were continuously measured in the Williamsburg neighborhood of Brooklyn, New York (USA), which is affected by traffic on a large bridge and major highway. One-minute average concentrations of ultrafine particulate matter (UFP), fine particulate matter [≤ 2.5 μm in aerodynamic diameter (PM2.5)], and particle-bound polycyclic aromatic hydrocarbons were measured using a mobile-monitoring protocol. Regression modeling approaches to quantify the influence of meteorology, traffic volume, and proximity to major roadways on pollutant concentrations were used. These models incorporated techniques to capture spatial variability, long- and short-term temporal trends, and multiple sources. We observed spatial heterogeneity of both UFP and PM2.5 concentrations. A variety of statistical methods consistently found a 15-20% decrease in UFP concentrations within the first 100 m from each of the two major roadways. For PM2.5, temporal variability dominated spatial variability, but we observed a consistent linear decrease in concentrations from the roadways. The combination of mobile monitoring and regression analysis was able to quantify local source contributions relative to background while accounting for physically interpretable parameters. Our results provide insight into urban exposure gradients.
Article
Full-text available
The Boston Puerto Rican Health Study is an ongoing longitudinal cohort study designed to examine the role of psychosocial stress on presence and development of allostatic load and health outcomes in Puerto Ricans, and potential modification by nutritional status, genetic variation, and social support. Self-identified Puerto Ricans, aged 45-75 years and residing in the Boston, MA metro area, were recruited through door-to-door enumeration and community approaches. Participants completed a comprehensive set of questionnaires and tests. Blood, urine and salivary samples were extracted for biomarker and genetic analysis. Measurements are repeated at a two-year follow-up. A total of 1500 eligible participants completed baseline measurements, with nearly 80% two-year follow-up retention. The majority of the cohort is female (70%), and many have less than 8th grade education (48%), and fall below the poverty level (59%). Baseline prevalence of health conditions is high for this age range: considerable physical (26%) and cognitive (7%) impairment, obesity (57%), type 2 diabetes (40%), hypertension (69%), arthritis (50%) and depressive symptomatology (60%). The enrollment of minority groups presents unique challenges. This report highlights approaches to working with difficult to reach populations, and describes some of the health issues and needs of Puerto Rican older adults. These results may inform future studies and interventions aiming to improve the health of this and similar communities.
Article
Full-text available
The caret package, short for classification and regression training, contains numerous tools for developing predictive models using the rich set of models available in R. The package focuses on simplifying model training and tuning across a wide variety of modeling techniques. It also includes methods for pre-processing training data, calculating variable importance, and model visualizations. An example from computational chemistry is used to illustrate the functionality on a real data set and to benchmark the benefits of parallel processing with several types of models.
Article
Full-text available
Although humans have been exposed to airborne nanosized particles (NSPs; < 100 nm) throughout their evolutionary stages, such exposure has increased dramatically over the last century due to anthropogenic sources. The rapidly developing field of nanotechnology is likely to become yet another source through inhalation, ingestion, skin uptake, and injection of engineered nanomaterials. Information about safety and potential hazards is urgently needed. Results of older biokinetic studies with NSPs and newer epidemiologic and toxicologic studies with airborne ultrafine particles can be viewed as the basis for the expanding field of nanotoxicology, which can be defined as safety evaluation of engineered nanostructures and nanodevices. Collectively, some emerging concepts of nanotoxicology can be identified from the results of these studies. When inhaled, specific sizes of NSPs are efficiently deposited by diffusional mechanisms in all regions of the respiratory tract. The small size facilitates uptake into cells and transcytosis across epithelial and endothelial cells into the blood and lymph circulation to reach potentially sensitive target sites such as bone marrow, lymph nodes, spleen, and heart. Access to the central nervous system and ganglia via translocation along axons and dendrites of neurons has also been observed. NSPs penetrating the skin distribute via uptake into lymphatic channels. Endocytosis and biokinetics are largely dependent on NSP surface chemistry (coating) and in vivo surface modifications. The greater surface area per mass compared with larger-sized particles of the same chemistry renders NSPs more active biologically. This activity includes a potential for inflammatory and pro-oxidant, but also antioxidant, activity, which can explain early findings showing mixed results in terms of toxicity of NSPs to environmentally relevant species. Evidence of mitochondrial distribution and oxidative stress response after NSP endocytosis points to a need for basic research on their interactions with subcellular structures. Additional considerations for assessing safety of engineered NSPs include careful selections of appropriate and relevant doses/concentrations, the likelihood of increased effects in a compromised organism, and also the benefits of possible desirable effects. An interdisciplinary team approach (e.g., toxicology, materials science, medicine, molecular biology, and bioinformatics, to name a few) is mandatory for nanotoxicology research to arrive at an appropriate risk assessment.
Article
Background Exposure to airborne ultrafine particle (UFP; <100 nm in aerodynamic diameter) is an emerging public health problem. Nevertheless, the benefit of using high efficiency particulate arrestance (HEPA) filtration to reduce UFP concentrations in homes is not yet clear. Methods We conducted a randomized crossover study of HEPA filtration without a washout period in 23 homes of low-income Puerto Ricans in Boston and Chelsea, MA (USA). Most participants were female, older adults who were overweight or obese. Particle number concentrations (PNC, a proxy for UFP) were measured indoors and outdoors at each home continuously for six weeks. Homes received both HEPA filtration and sham filtration for three weeks each in random order. Results Median PNC under HEPA filtration was 50–85% lower compared to sham filtration in most homes, but we found no benefit in terms of reduced inflammation; associations between hsCRP, IL-6, or TNFRII in blood samples and indoor PNC were inverse and not statistically significant. Conclusions Limitations to our study design likely contributed to our findings. Limitations included carry-over effects, a population that may have been relatively unresponsive to UFP, reduction in PNC even during sham filtration that limited differences between HEPA and sham filtration, window opening by participants, and lack of fine-grained (room-specific) participant time-activity information. Our approach was similar to other recent HEPA intervention studies of particulate matter exposure and cardiovascular risk, suggesting that there is a need for better study designs.
Article
Land-use regression (LUR) models for ultrafine particles (UFP) and Black Carbon (BC) in urban areas have been developed using short-term stationary monitoring or mobile platforms in order to capture the high variability of these pollutants. However, little is known about the comparability of predictions of mobile and short-term stationary models and especially the validity of these models for assessing residential exposures and the robustness of model predictions developed in different campaigns. We used an electric car to collect mobile measurements (n = 5236 unique road segments) and short-term stationary measurements (3 × 30 min, n = 240) of UFP and BC in three Dutch cities (Amsterdam, Utrecht, Maastricht) in 2014–2015. Predictions of LUR models based on mobile measurements were compared to (i) measured concentrations at the short-term stationary sites, (ii) LUR model predictions based on short-term stationary measurements at 1500 random addresses in the three cities, (iii) externally obtained home outdoor measurements (3 × 24 h samples; n = 42) and (iv) predictions of a LUR model developed based upon a 2013 mobile campaign in two cities (Amsterdam, Rotterdam). Despite the poor model R² of 15%, the ability of mobile UFP models to predict measurements with longer averaging time increased substantially from 36% for short-term stationary measurements to 57% for home outdoor measurements. In contrast, the mobile BC model only predicted 14% of the variation in the short-term stationary sites and also 14% of the home outdoor sites. Models based upon mobile and short-term stationary monitoring provided fairly high correlated predictions of UFP concentrations at 1500 randomly selected addresses in the three Dutch cities (R² = 0.64). We found higher UFP predictions (of about 30%) based on mobile models opposed to short-term model predictions and home outdoor measurements with no clear geospatial patterns. The mobile model for UFP was stable over different settings as the model predicted concentration levels highly correlated to predictions made by a previously developed LUR model with another spatial extent and in a different year at the 1500 random addresses (R² = 0.80). In conclusion, mobile monitoring provided robust LUR models for UFP, valid to use in epidemiological studies.
Article
Traffic-related ultrafine particles (UFP; <100 nm diameter) are ubiquitous in urban air. While studies have shown that UFP are toxic, epidemiological evidence of health effects, which is needed to inform risk assessment at the population scale, is limited due to challenges of accurately estimating UFP exposures. Epidemiologic studies often use empirical models to estimate UFP exposures; however, the monitoring strategies upon which the models are based have varied between studies. Our study compares particle number concentrations (PNC; a proxy for UFP) measured by three different monitoring approaches (central-site, short-term residential-site, and mobile on-road monitoring) in two study areas in metropolitan Boston (MA, USA). Our objectives were to quantify ambient PNC differences between the three monitoring platforms, compare the temporal patterns and the spatial heterogeneity of PNC between the monitoring platforms, and identify factors that affect correlations across the platforms. We collected >12,000 h of measurements at the central sites, 1000 h of measurements at each of 20 residential sites in the two study areas, and >120 h of mobile measurements over the course of ∼1 year in each study area. Our results show differences between the monitoring strategies: mean 1 min PNC on-roads were higher (64,000 and 32,000 particles/cm³ in Boston and Chelsea, respectively) compared to central-site measurements (23,000 and 19,000 particles/cm³) and both were higher than at residences (14,000 and 15,000 particles/cm³). Temporal correlations and spatial heterogeneity also differed between the platforms. Temporal correlations were generally highest between central and residential sites, and lowest between central-site and on-road measurements. We observed the greatest spatial heterogeneity across monitoring platforms during the morning rush hours (06:00-09:00) and the lowest during the overnight hours (18:00-06:00). Longer averaging times (days and hours vs. minutes) increased temporal correlations (Pearson correlations were 0.69 and 0.60 vs. 0.39 in Boston; 0.71 and 0.61 vs. 0.45 in Chelsea) and reduced spatial heterogeneity (coefficients of divergence were 0.24 and 0.29 vs. 0.33 in Boston; 0.20 and 0.27 vs. 0.31 in Chelsea). Our results suggest that combining stationary and mobile monitoring may lead to improved characterization of UFP in urban areas.
Article
Secondary organic aerosol (SOA) is formed from the atmospheric oxidation of gas-phase organic compounds leading to the formation of particle mass. Gasoline- and diesel-powered motor vehicles, both on/off-road, are important sources of SOA precursors. They emit complex mixtures of gas-phase organic compounds that vary in volatility and molecular structure-factors that influence their contributions to urban SOA. However, the relative importance of each vehicle type with respect to SOA formation remains unclear due to conflicting evidence from recent laboratory, field, and modeling studies. Both are likely important, with evolving contributions that vary with location and over short timescales. This review summarizes evidence, research needs, and discrepancies between top-down and bottom-up approaches used to estimate SOA from motor vehicles; focusing on inconsistencies between molecular-level understanding and regional observations. The effect of emission controls (e.g. exhaust aftertreatment technologies, fuel formulation) on SOA precursor emissions needs comprehensive evaluation, especially with international perspective given heterogeneity in regulations and technology penetration. Novel studies are needed to identify and quantify "missing" emissions that appear to contribute substantially to SOA production, especially in gasoline vehicles with the most advanced aftertreatment. Initial evidence suggests catalyzed diesel particulate filters greatly reduce emissions of SOA precursors along with primary aerosol.
Article
Background: Epidemiologic evidence on the association between short-term exposure to ultrafine particles and mortality is weak, due to the lack of routine measurements of these particles and standardized multi-center studies. We investigated the relationship between ultrafine particles and particulate matter (PM) and daily mortality in eight European urban areas. Methods: We collected daily data on non-accidental and cardio-respiratory mortality, particle number concentrations (as proxy for ultrafine particle number concentration), fine and coarse PM, gases and meteorologic parameters in eight urban areas of Finland, Sweden, Denmark, Germany, Italy, Spain, and Greece, between 1999 and 2013. We applied city-specific time-series Poisson regression models and pooled them with random-effects meta-analysis. Results: We estimated a weak, delayed association between particle number concentration and non-accidental mortality, with mortality increasing by approximately 0.35% per 10,000 particles/cm increases in particle number concentration occurring 5 to 7 days before death. A similar pattern was found for cause-specific mortality. Estimates decreased after adjustment for fine particles (PM2.5) or nitrogen dioxide (NO2). The stronger association found between particle number concentration and mortality in the warmer season (1.14% increase) became null after adjustment for other pollutants. Conclusions: We found weak evidence of an association between daily ultrafine particles and mortality. Further studies are required with standardized protocols for ultrafine particle data collection in multiple European cities over extended study periods.
Article
Important health relevance has been suggested for ultrafine particles (UFP) and ozone, but studies on long-term effects are scarce, mainly due to the lack of appropriate spatial exposure models. We designed a measurement campaign to develop land use regression (LUR) models to predict the spatial variability focusing on particle number concentration (PNC) as indicator for UFP, ozone and several other air pollutants in the Augsburg region, Southern Germany. Three bi-weekly measurements of PNC, ozone, particulate matter (PM10, PM2.5), soot (PM2.5abs) and nitrogen oxides (NOx, NO2) were performed at 20 sites in 2014/15. Annual average concentration were calculated and temporally adjusted by measurements from a continuous background station. As geographic predictors we offered several traffic and land use variables, altitude, population and building density. Models were validated using leave-one-out cross-validation. Adjusted model explained variance (R²) was high for PNC and ozone (0.89 and 0.88). Cross-validation adjusted R² was slightly lower (0.82 and 0.81) but still indicated a very good fit. LUR models for other pollutants performed well with adjusted R² between 0.68 (PMcoarse) and 0.94 (NO2). Contrary to previous studies, ozone showed a moderate correlation with NO2 (Pearson's r = − 0.26). PNC was moderately correlated with ozone and PM2.5, but highly correlated with NOx (r = 0.91). For PNC and NOx, LUR models comprised similar predictors and future epidemiological analyses evaluating health effects need to consider these similarities.
Article
Mobile and short-term monitoring campaigns are increasingly used to develop land use regression (LUR) models for ultrafine particles (UFP) and black carbon (BC). It is not yet established whether LUR models based on mobile or short-term stationary measurements result in comparable models and concentration predictions. The goal of this paper is to compare LUR models based on stationary (30 minutes) and mobile UFP and BC measurements from a single campaign. An electric car collected both repeated stationary and mobile measurements in Amsterdam and Rotterdam, The Netherlands. A total of 2,964 road segments and 161 stationary sites were sampled over two seasons. Our main comparison was based on predicted concentrations of the mobile and stationary monitoring LUR models at 12,682 residential addresses in Amsterdam. Predictor variables in the mobile and stationary LUR model were comparable resulting in highly correlated predictions at external residential addresses (R2 of 0.89 for UFP and 0.88 for BC). Mobile model predictions were on average 1.41 and 1.91 times higher than stationary model predictions for UFP and BC respectively. LUR models based upon mobile and stationary monitoring predicted highly correlated UFP and BC concentration surfaces, but predicted concentrations based on mobile measurements were systematically higher.
Article
We measured size–resolved PNCs in the 5–560 nm range at two different types (4– and 3–way) of TIs in Guildford (Surrey, UK) at fixed sites (~1.5 m above the road level), sequentially at 4 different heights (1, 1.5, 2.5 and 4.7 m), and along the road at five different distances (10, 20, 30, 45 and 60 m). The aims were to: (i) assess the differences in PNCs measured at studied TIs, (ii) identify the best fit probability distribution curves for the PNCs, (iii) determine vertical and horizontal decay profiles of PNCs, (iv) estimate particle number emission factors (PNEFs) under congested and free–flow traffic conditions, and (v) quantify the pedestrian exposure in terms of respiratory deposition dose (RDD) rates at the TIs. Daily averaged particle number distributions at TIs reflected the effect of fresh emissions with peaks at 5.6, 10 and 56nm. Despite the relatively high traffic volume at 3–way TI, average PNCs at 4–way TI were about twice as high as at 3–way TI, indicating less favourable dispersion conditions. Generalised extreme value distribution fitted well to PNC data at both TIs. Vertical PNC profiles followed an exponential decay, which was much sharper at 4–way TI than at 3–way TI, suggesting ~40% less exposure for people at first floor (4.7 m) to those at ground floor around 4-way TI. Vertical profiles indicated much sharper (~132–times larger) decay than in horizontal direction, due to close vicinity of road vehicles during the along-road measurements. Over an order of magnitude higher PNEFs were found during congested, compared with free–flow, conditions due to frequent changes in traffic speed. Average RDD rate at 4–way TI during congested conditions were up to 14–times higher than those at 3–way TI (0.40×1011 h˗1). Findings of this study are a step forward to understand exposure at and around the TIs.
Article
In this study, NOx and particle number concentration (PNC) at an urban background and a traffic location were measured in the city of Amsterdam (the Netherlands). Modelled and measured contributions to NOx and PNC at the traffic location were used to derive real-world PN emission factors for average urban road traffic. The results for NOx were applied to validate our approach. The real-world PN emission factors (#.km-1) were 2.9E+14 (urban road) and 3E+14 (motorway). These values were at least a factor eight higher than dynamometer-based PN emission factors from COPERT 4 and HBEFA databases. The real-world PN emission factors were used to model the contribution to PNC near road traffic in 2014. This was two to three times higher than the PNC urban background along urban roads over 20,000 vehicles per day and near motorways. The discrepancy between dynamometer-based and real-world emission factors demonstrates the need for more PNC observations to assess actual PN emissions from road traffic.
Article
Atmospheric ultrafine particles (UFP; diameter < 0.1 μm) represent a growing global health concern in urban environments and has a strong link to traffic related emissions. UFP is usually the dominating fraction of atmospheric particle number concentrations (PNC) despite being a minor part of total particle mass. The aim of this study was to empirically investigate the relationship between PNC and other air pollutants (NOX, NO2 and PM10) in the urban environment and their dependence on meteorology and weather type, using the Lamb Weather Type (LWT) classification scheme. The study was carried out in Gothenburg, Sweden, at an urban background site during April 2007–May 2008. It was found that daily average [PNC] correlated very well with [NOx] (R² = 0.73) during inversion days, to a lesser extent with [NO2] (R² = 0.58) and poorly with [PM10] (R² = 0.07). Both PNC and NOx had similar response patterns to wind speed and to the strength of temperature inversions. PNC displayed two regimes, one strongly correlated to NOx and a second poorly correlated to NOx which was characterised by high wind speed. For concentration averages based on LWTs, the PNC-[NOx] relationship remained strong (R² = 0.70) where the windy LWT W deviated noticeably. Exclusion of observations with wind speed >5 ms⁻¹ or ΔT < 0 °C from LWTs produced more uniform and stronger relationships (R² = 0.90; R² = 0.93). Low wind speeds and positive vertical temperature gradients were most common during LWTs A, NW, N and NE. These weather types were also associated with the highest daily means of NOx (∼30 ppb) and PNC (∼10 000 # cm⁻³). A conclusion from this study is that NOx (but not PM10) is a good proxy for PNC especially during calm and stable conditions and that LWTs A, NW, N and NE are high risk weather types for elevated NOx and PNC.
Article
In several studies, exposure to fine particulate matter (PM) has been associated with inflammation, with inconsistent results. We used repeated measurements to examine the association of long-term fine and ultrafine particle exposure with several blood markers of inflammation and coagulation. We used baseline (2000-2003) and follow-up (2006-2008) data from the Heinz Nixdorf Recall Study, a German population-based prospective cohort of 4814 participants. A chemistry transport model was applied to model daily surface concentrations of PM air pollutants (PM10, PM2.5) and particle number on a grid of 1 km(2). Applying mixed regression models, we analysed associations of long-term (mean of 365 days prior to blood draw) particle exposure at each participant's residence with the level of high-sensitivity C reactive protein (hs-CRP), fibrinogen, platelet and white cell count (WCC), adjusting for short-term PM exposure (moving averages of 1-7 days), personal characteristics, season, ambient temperature (1-5 days), ozone and time trend. We analysed 6488 observations: 3275 participants with baseline data and 3213 with follow-up data. An increase of 2.4 µg/m(3) in long-term PM2.5 was associated with an adjusted increase of 5.4% (95% CI 0.6% to 10.5%) in hs-CRP and of 2.3% (95% CI 1.4% to 3.3%) in the platelet count. Fibrinogen and WCC were not associated with long-term particle exposure. In this population-based cohort, we found associations of long-term exposure to PM with markers of inflammation (hs-CRP) and coagulation (platelets). This finding supports the hypothesis that inflammatory processes might contribute to chronic effects of air pollution on cardiovascular disease. Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://group.bmj.com/group/rights-licensing/permissions.
Article
Land Use Regression (LUR) models typically use fixed-site monitoring; here, we employ mobile monitoring as a cost-effective alternative for LUR-development. We use bicycle-based, mobile measurements (~85 hours) during rush-hour in Minneapolis, MN to build LUR models for particulate concentrations (particle number [PN], black carbon [BC], fine particulate matter [PM2.5], particle size). We developed and examined 1,224 separate LUR models by varying pollutant, time-of-day, and method of spatial and temporal smoothing of the time-series data. Our base-case LUR models had modest goodness-of-fit (adjusted R2: ~0.5 [PN], ~0.4 [PM2.5], 0.35 [BC], ~0.25 [particle size]) and included predictor variables which captured proximity to and density of emission sources. The spatial density of our measurements resulted in a large model-building dataset (n=1,101 concentration estimates); ~25% of buffer variables were selected at spatial scales of <100m, suggesting that on-road particle concentrations change on small spatial scales. LUR model-R2 improved as sampling runs were completed, with diminishing benefits after ~40 hours of data collection. Spatial autocorrelation of model residuals indicated that models performed poorly where spatiotemporal resolution of emission sources (i.e., traffic congestion) was poor. Our findings suggest that LUR modeling from mobile measurements is possible, but that more work could usefully inform best practices.
Article
Health effects of long-term exposure to ultrafine particles (UFP) have not been investigated in epidemiological studies because of the lack of spatially resolved UFP exposure data. Short-term monitoring campaigns used to develop land use regression (LUR) models for UFP typically had moderate performance. The aim of this study was to develop and evaluate spatial and spatiotemporal LUR models for UFP and Black Carbon (BC), including their ability to predict past spatial contrasts. We measured 30 minutes at each of 81 sites in Amsterdam and 80 in Rotterdam, the Netherlands in three different seasons. Models were developed using traffic, land use, reference site measurements, routinely measured pollutants and weather data. The percentage explained variation (R2) was 0.35-0.40 for BC and 0.33 - 0.42 for UFP spatial models. Traffic variables were present in every model. The coefficients for the spatial predictors were similar in spatial and spatiotemporal models. The BC LUR model explained 61% of the spatial variation in a previous campaign with longer sampling duration, better than the model R2. The UFP LUR model explained 36% of UFP spatial variation measured 10 years earlier, similar to the model R2. Short-term monitoring campaigns may be an efficient tool to develop LUR models.
Article
Land use regression (LUR) models have been used to assess air pollutant exposure, but limited evidence exists on whether location-specific LUR models are applicable to other locations (transferability) or general models are applicable to smaller areas (generalizability). We tested transferability and generalizability of spatial-temporal particle number concentration (PNC) LUR models for Boston-area (MA, USA) urban neighborhoods near Interstate 93. Four neighborhood-specific regression models and one Boston-area model were developed from mobile monitoring measurements (34-46 days/neighborhood over one year each). Transferability was tested by applying each neighborhood-specific model to all four neighborhoods; generalizability was tested by applying the Boston-area model to each neighborhood. Both were tested with and without neighborhood-specific calibration. Important PNC predictors (adjusted-R(2) = 0.24-0.43) included wind speed and direction, temperature, highway traffic volume, and distance from the highway edge. Direct model transferability was poor (R(2) <0.17). Locally-calibrated transferred models (R(2) = 0.19-0.40) and the Boston-area model (adjusted-R(2) = 0.26, range: 0.13-0.30) performed similarly to neighborhood-specific models. However, the coefficients of locally-calibrated transferred models were sometimes uninterpretable. Our results show that transferability of neighborhood-specific PNC LUR models is limited, but that general models perform acceptably in multiple areas, especially when calibrated with local data.
Article
Although there is evidence that ultrafine particles (UFP) do affect human health there are currently no legal ambient standards. The main reasons are the absence of spatially resolved exposure data to investigate long-term health effects and the challenge of defining representative reference sites for monitoring given the high dependence of UFP on proximity to sources. The objectives of this study were to evaluate the spatial distribution of UFP in four areas of the Swiss Study on Air Pollution and Lung and Heart Diseases in Adults (SAPALDIA) and to investigate the representativeness of routine air monitoring stations for residential sites in these areas. Repeated UFP measurements during three seasons have been conducted at a total of 80 residential sites and four area specific reference sites over a median duration of seven days. Arithmetic mean residential PNC scattered around the median of 10,800 particles/cm3 (interquartile range [IQR] = 7,800 particles/cm3). Spatial within area contrasts (90th/10th percentile ratios) were around two; increased contrasts were observed during weekday rush-hours. Temporal UFP patterns were comparable at reference and residential sites in all areas. Our data show that central monitoring sites can represent residential conditions when locations are well chosen with respect to the local sources - namely traffic. For epidemiological research, locally resolved spatial models are needed to estimate individuals' long-term exposures to UFP of outdoor origin at home, during commute and at work.
Article
Air pollution in New Delhi, India is a significant environmental and health concern. To assess determinants of variability in air pollutant concentrations we develop land use regression (LUR) models for fine particulate matter (PM2.5), black carbon (BC) and ultrafine particle number concentrations (UFPN). We used 136 hours (39 sites), 112 hours (26 sites), 147 hours (39 sites) of PM2.5, BC and UFPN data respectively, to develop separate morning (0800-1200) and afternoon (1200- 1800) models. Continuous measurements of PM2.5 and BC were also made at a single fixed rooftop site located in a high-income residential neighborhood. No continuous measurements of UFPN were available. In addition to spatial variables, measurements from the fixed continuous monitoring site were used as independent variables in the PM2.5 and BC models. The median concentrations (and interquartile range) of PM2.5, BC and UFPN at LUR sites were 133 (96-232) μg m-3, 11 (6-21) µg m-3 and 40 (27-72) × 103 cm-3 respectively. In addition a) for PM2.5 and BC, the temporal variability was higher than the spatial variability; b) the magnitude and spatial-variability in pollutant concentrations was higher during morning than during afternoon hours. Further, model R2 values were higher for morning (for PM2.5, BC and UFPN, respectively: 0.85, 0.86, and 0.28) than for afternoon models (0.73, 0.69 and 0.23); (c) the PM2.5 and BC concentrations measured at LUR sites all over the city were strongly correlated with measured concentrations at a fixed rooftop site; d) spatial patterns were similar for PM2.5 and BC, but different for UFPN; (e) population density and road variables were statistically significant predictors of pollutant concentrations; and, (f) available geographic predictors explained a much lower proportion of variability in measured PM2.5, BC and UFPN than observed in other LUR studies, indicating the importance of temporal variability and suggesting the existence of uncharacterized sources.
Article
High concentrations of air pollutants on roadways, relative to ambient concentrations, contribute significantly to total personal exposure. Estimation of these exposures requires measurements or prediction of roadway concentrations. Our study develops, compares and evaluates linear regression and non-linear generalized additive models (GAMs) to estimate on-road concentrations of four key air pollutants, particle-bound polycyclic aromatic hydrocarbons (PB-PAH), particle number count (PNC), nitrogen oxides (NOx), and particulate matter with diameter <2.5 µm (PM2.5) using traffic, meteorology, and elevation variables. Critical predictors included wind speed and direction for all the pollutants, traffic-related variables for PB-PAH, PNC, and NOx, and air temperatures and relative humidity for PM2.5. GAMs explained 50%, 55%, 46%, and 71% of the variance for log or square-root transformed concentrations of PB-PAH, PNC, NOx, and PM2.5 respectively, an improvement of 5 to over 15% over the linear models. Accounting for temporal autocorrelation in the GAMs further improved the prediction, explaining 57-89% of the variance. We concluded that traffic and meteorological data are good predictors in estimating on-road traffic-related air pollutant concentrations and GAMs perform better for non-linear variables, such as meteorological parameters.
Article
Methods to characterize chronic exposure to ultrafine particles (UFP) can help to clarify potential health effects. Since UFP are not routinely monitored in North America, spatiotemporal models are one potential exposure assessment methodology. Portable condensation particle counters were used to measure particle number concentrations (PNC) to develop a land use regression (LUR) model. PNC, wind speed and direction were measured for sixty minutes at eighty locations during a two-week sampling campaign. We conducted continuous monitoring at four additional locations to assess temporal variation. LUR modeling utilized 135 potential geographic predictors including: road length, vehicle density, restaurant density, population density, land use and others. A novel approach incorporated meteorological data through wind roses as alternates to traditional circular buffers. The range of measured (sixty-minute median) PNC across locations varied seventy-fold (1500 - 105000 particles/cm3, mean [SD] = 18200 [15900] particles/cm3). Correlations between PNC and concurrently measured two-week average NOX concentrations were 0.6 - 0.7. A PNC LUR model (R2 = 0.48, leave-one-out cross validation R2 = 0.32) including truck route length within 50m, restaurant density within 200m and ln-distance to the port represents the first UFP LUR model in North America. Models incorporating wind roses did not explain more variability in measured PNC.
Article
Working in the context of the linear model y = Xβ + ε, we generalize the concept of variance inflation as a measure of collinearity to a subset of parameters in β (denoted by β1, with the associated columns of X given by X1). The essential idea underlying this generalization is to examine the impact on the precision of estimation—in particular, the size of an ellipsoidal joint confidence region for β1—of less-than-optimal selection of other columns of the design matrix (X2), treating still other columns (X0) as unalterable, even hypothetically. In typical applications, X1 contains a set of dummy regressors coding categories of a qualitative variable or a set of polynomial regressors in a quantitative variable; X2 contains all other regressors in the model, save the constant, which is in X0. If σV denotes the realized variance of , and σU is the variance associated with an optimal selection of X2, then the corresponding scaled dispersion ellipsoids to be compared are ℰv = {x : x′Vx ≤ 1} and ℰU = {x : x′Ux ≤ 1}, where ℰU is contained in ℰv. The two ellipsoids can be compared by considering the radii of ℰv relative to ℰU, obtained through the spectral decomposition of V relative to U. We proceed to explore the geometry of generalized variance inflation, to show the relationship of these measures to correlation-matrix determinants and canonical correlations, to consider X matrices structured by relations of marginality among regressor subspaces, to develop the relationship of generalized variance inflation to hypothesis tests in the multivariate normal linear model, and to present several examples.
Article
The introduction of condensation particle counters (CPCs) utilizing water as the condensing fluid provides an alternative to traditional butanol based CPCs. Previous evaluations, using atmospheric and laboratory test aerosols, have verified performance. This study compares the performance of multiple water and butanol based CPC models using a diesel engine exhaust challenge aerosol. A total of 5 CPCs used in a scanning mobility particle sizer (SMPS) configuration were compared. TSI models 3786, and 3782 use water as the condensing fluid while models 3010, 3025A, and 3775 use butanol. The test aerosol was generated by a turbocharged, direct injection diesel engine running at constant speed and load, with two fuels, a low sulfur diesel and 99% soy methyl ester biodiesel fuel. Tests were conducted using a single SMPS platform and switching CPCs for each set of tests. In addition, the tests were repeated with long and nano differential mobility analyzer (DMA) columns. Four of the five CPCs agreed well, giving a standard deviation of the overall average geometric mean diameter of less than 1 nm between the 4 CPCs. The fifth CPC, TSI model 3782 did not agree well with the others. The cause of this disagreement is thought stem in part from the use of water as a condensing fluid, but primarily from a lack of sheath air in the 3782 design. The performance of the TSI 3786, an ultrafine water-based CPC with sheath flow showed far better agreement with the butanol CPCs throughout most mobility diameters.
Article
a b s t r a c t Continuous measurements of number size distributions of ultrafine particles (UFPs) and other pollut-ants (PM 2.5 , SO 2 , CO and O 3) have been performed in Rochester, New York since late November 2001. The 2002e2009 average number concentrations of particles in three size ranges (10e50 nm, 50e100 nm and 100e500 nm) were 4730 cm À3 , 1838 cm À3 , and 1073 cm À3 , respectively. The lowest annual average number concentrations of particles in 10e50 nm and 50e100 nm were observed during 2008e2009. The lowest monthly average number concentration of 10e50 nm particles was observed in July and the highest in February. The daily patterns of 10e50 nm particles had two peaks at early morning (7e8 AM) and early afternoon (2 PM). There was a distinct declining trend in the peak number concentrations from 2002e2005 to 2008e2009. Large reductions in SO 2 concentrations associated with northerly winds between 2007 and 2009 were observed. The most significant annual decrease in the frequency of morning particle nucleation was observed from 2005 to 2007. The monthly variation in the morning nucleation events showed a close correlation with number concentrations of 10e50 nm particles (r ¼ 0.89). The frequency of the local SO 2 -related nucleation events was much higher before 2006. All of these results suggest significant impacts of highway traffic and industrial sources. The decrease in particle number concentrations and particle nucleation events likely resulted from a combination of the U.S. EPA 2007 Heavy-Duty Highway Rule implemented on October 1, 2006, the closure of a large coal-fired power plant in May 2008, and the reduction of Eastman Kodak emissions.
Article
New particle formation in the polluted continental boundary layer was studied, based on 1.5-year observations of the particle size distribution, meteorological and gas phase parameters. Events of new particle formation involving significant ultrafine particle number concentrations (> 10(4) cm(-3) in the size range 3-11 nm) were observed on 20 % of all days, pointing out that a frequent particle production from gaseous precursors can occur despite the relatively high pre-existing particle surface area in the area of investigation. The maximum in the observed particle size distributions was mostly above 3 nm, suggesting the actual particle nucleation to take place upwind of the measurement site. A particle growth analysis yielded 2.3 +/- 1.4 h as an upper limit of the time for the particles to grow from the critical cluster size till the observation of the peak in ultrafine number concentration. On 80 % of the significant events of new particle formation (though not on all), SO2 concentrations increased considerably (by an average factor of 7), most likely by entrainment from aloft. Particle surface area was, on average, higher on event days compared to non-event days, indicating only a weak competition between condensation onto the pre-existing particle surface area and the new particle formation process. The highest statistical correlation was found between the events of new particle formation and solar radiation, indicating a high degree of meteorological control.
Article
Ultrafine particulate matter (UFP; diameter <0.1 μm) concentrations are relatively high on the freeway, and time spent on freeways can contribute a significant fraction of total daily UFP exposure. We model real-time size-resolved UFP concentrations in summer time on-freeway air. Particle concentrations (32 bins, 5.5 to 600 nm) were measured on Minnesota freeways during summer 2006 and 2007 ( Johnson, J. P.; Kittelson, D. B.; Watts, W. F. Environ. Sci. Technol. 2009 , 43 , 5358 - 5364 ). Here, we develop and apply two-way stratified multilinear regressions, using an approach analogous to mobile-monitoring land-use regression but using real-time meteorological and traffic data. Our models offer the strongest predictions in the 10-100 nm size range (adj-R(2): 0.79-0.89, average adj-R(2): 0.85) and acceptable but weaker predictions in the 130-200 nm range (adj-R(2): 0.41-0.62, average adj-R(2): 0.52). The aggregate model for total particle counts performs well (adj-R(2) = 0.77). Bootstrap resampling (n = 1000) indicates that the proposed models are robust to minor perturbations in input data. The proposed models are based on readily available real-time information (traffic and meteorological parameters) and can thus be exploited to offer spatiotemporally resolved prediction of UFPs on freeways within similar geographic and meteorological environments. The approach developed here provides an important step toward modeling population exposure to UFP.
Article
There are currently no epidemiological studies on health effects of long-term exposure to ultrafine particles (UFP), largely because data on spatial exposure contrasts for UFP is lacking. The objective of this study was to develop a land use regression (LUR) model for UFP in the city of Amsterdam. Total particle number concentrations (PNC), PM10, PM2.5, and its soot content were measured directly outside 50 homes spread over the city of Amsterdam. Each home was measured during one week. Continuous measurements at a central urban background site were used to adjust the average concentration for temporal variation. Predictor variables (traffic, address density, land use) were obtained using geographic information systems. A model including the product of traffic intensity and the inverse distance to the nearest road squared, address density, and location near the port explained 67% of the variability in measured PNC. LUR models for PM2.5, soot, and coarse particles (PM10, PM2.5) explained 57%, 76%, and 37% of the variability in measured concentrations. Predictions from the PNC model correlated highly with predictions from LUR models for PM2.5, soot, and coarse particles. A LUR model for PNC has been developed, with similar validity as previous models for more commonly measured pollutants.
Article
In clinical measurement comparison of a new measurement technique with an established one is often needed to see whether they agree sufficiently for the new to replace the old. Such investigations are often analysed inappropriately, notably by using correlation coefficients. The use of correlation is misleading. An alternative approach, based on graphical techniques and simple calculations, is described, together with the relation between this analysis and the assessment of repeatability.