Spatial patterning in PM2.5 constituents under an inversion-focused sampling design across an urban area of complex terrain

Impact of PM2.5 filter extraction solvent on oxidative potential and chemical analysis

Fine particulate matter (PM2.5) is hypothesized to induce oxidative stress, and has been linked to acute and chronic adverse health effects. To better understand the risks and underlying mechanisms following exposure, PM2.5 is collected onto filters but prior to toxicological analysis, particles must be removed from filters. There is no standard method for filter extraction, which creates the possibility that the methods of extraction selected can alter the chemical composition and ultimately the biological implications. In this study, comparisons were made between extraction solvents (methanol (MeOH), dichloromethane (DCM), 0.9% saline, and Milli-Q water) and the results of oxidative potential and elemental concentration analysis of PM2.5 collected across sites in Arkansas, USA. Significant differences were observed between solvents, with DCM having significantly different results compared to all other extraction solvents (p ≤ 0.001). Significant correlations between element, black carbon, and PM2.5 concentrations and oxidative potential were observed. The observed correlations were extraction solvent dependent. For example, in saline extracted samples, oxidative potential had significant negative correlations with: Ba, Cd, Ce, Co, Ga, Mn and significant positive correlations with: Cr, Ni, Th, U. While in MeOH extracted samples, significant positive correlations were only between oxidative potential and Ga, U and significant negative correlations with V. This indicates that PM2.5 samples extracted with different solvents will yield different conclusions about the causal components. This study highlights the importance of filter extraction methods in interpretation of oxidative potential results and comparisons between studies.Implications: While there is no standard method for PM2.5 filter extraction, variation of extraction methods impact analytical results. This project identifies that extraction method variation, particularly extraction solvent selection, leads to discrepancies in chemical and toxicological analysis for PM2.5 collected on the same filter. This work highlights the need for methods standardization to support accurate comparisons between PM2.5 research studies, thus providing better understanding of PM2.5 across the globe.

Chemical characteristics and oxidative potential of indoor and outdoor PM2.5 in densely populated urban slums

A significant proportion of population in metropolitan cities in India live in slums which are highly dense and crowded informal housing settlements with poor environmental conditions including high exposure to air pollution. Recent studies report that toxicity is induced by oxidative processes, mediated by the water-soluble PM chemical components leading to reactive oxygen species production thereby causing inflammatory disorders. Hence, for the first time, this study assessed the chemical characteristics and oxidative potential (OP) of indoor and outdoor PM_2.5 in two slums in Mumbai, India. Daily gravimetric PM_2.5 was measured in ∼40 homes each in a low- and a high-traffic slum and analysed for 18 water-soluble elements and organic carbon (WSOC). Subsequently, OP was assessed through the Dithiothreitol (DTT) assay. Average WSOC was similar in indoor and outdoor environments while the water-soluble concentrations of total elements ranged 4.5-6.5 μg/m³ indoors and 6.4-19.2 μg/m³ outdoors, with S, Ca, K, Na and Zn being the most abundant elements. Spatial distributions of indoor concentrations were influenced by outdoor sources such as local traffic emissions for Cd, Fe, Al and Zn. The influence of outdoor-origin particles was enhanced in homes reporting high air exchange rates. OP was higher outdoors than indoors in both low-traffic slum (0.04-0.51 nmol min^-1m^-3 outdoors and 0.02-0.38 nmol min^-1m^-3 indoors) and high-traffic slum (0.03-1.06 nmol min^-1m^-3 outdoors and 0.04-0.77 nmol min^-1m^-3 indoors). Outdoor and indoor OP was also more influenced by outdoor road dust showing significant correlation with tracer elements Cu and Al (r ≥ 0.45; p < 0.05). Similar to OP, the non-carcinogenic health risk associated with indoor PM_2.5 were also higher in high-traffic slum (Hazard Index, HI = 1.60) than in low-traffic slum (HI = 0.43). Overall, this study shows that the indoor PM_2.5 and its chemical constituents in Mumbai slums are primarily of outdoor origin with higher toxicity and non-carcinogenic health risk in high-traffic slums.

Daily 1 km terrain resolving maps of surface fine particulate matter for the western United States 2003-2021

We developed daily maps of surface fine particulate matter (PM2.5) for the western United States. We used geographically weighted regression fit to air quality station observations with Moderate Resolution Imaging Spectroradiometer (MODIS) aerosol optical depth (AOD) data, and meteorological data to produce daily 1-kilometer resolution PM2.5 concentration estimates from 2003-2020. To account for impacts of stagnant air and inversions, we included estimates of inversion strength based on meteorological conditions, and inversion potential based on human activities and local topography. Model accuracy based on cross-validation was R2 = 0.66. AOD data improve the model in summer and fall during periods of high wildfire activity while the stagnation terms capture the spatial and temporal dynamics of PM2.5 in mountain valleys, particularly during winter. These data can be used to explore exposure and health outcome impacts of PM2.5 across spatiotemporal domains particularly in the intermountain western United States where measurements from monitoring station data are sparse. Furthermore, these data may facilitate analyses of inversion impacts and local topography on exposure and health outcome studies.

Exposure models for particulate matter elemental concentrations in Southern California

Due to a scarcity of routine monitoring of speciated particulate matter (PM), there has been limited capability to develop exposure models that robustly estimate component-specific concentrations. This paper presents the largest such study conducted in a single urban area. Using samples that were collected at 220 locations over two seasons, quasi-ultrafine (PM_0.2), accumulation mode fine (PM_0.2-2.5), and coarse (PM_2.5-10) particulate matter concentrations were used to develop spatiotemporal regression, machine learning models that enabled predictions of 24 elemental components in eight Southern California communities. We used supervised variable selection of over 150 variables, largely from publicly available sources, including meteorological, roadway and traffic characteristics, land use, and dispersion model estimates of traffic emissions. PM components that have high oxidative potential (and potentially large health effects) or are otherwise important markers for major PM sources were the primary focus. We present results for copper, iron, and zinc (as non-tailpipe vehicle emissions); elemental carbon (diesel emissions); vanadium (ship emissions); calcium (soil dust); and sodium (sea salt). Spatiotemporal linear regression models with 17 to 36 predictor variables including meteorology; distance to different classifications of roads; intersections and off ramps within a given buffer distance; truck and vehicle traffic volumes; and near-roadway dispersion model estimates produced superior predictions over the machine learning approaches (cross validation R-squares ranged from 0.76 to 0.92). Our models are easily interpretable and appear to have more effectively captured spatial gradients in the metallic portion of PM than other comparably large studies, particularly near roadways for the non-tailpipe emissions. Furthermore, we demonstrated the importance of including spatiotemporally resolved meteorology in our models as it helped to provide key insights into spatial patterns and allowed us to make temporal predictions.

Inter- and Intra-Individual Variability of Personal Health Risk of Combined Particle and Gaseous Pollutants across Selected Urban Microenvironments

Exposure surrogates, such as air quality measured at a fixed-site monitor (FSM) or residence, are typically used for health estimates. However, people spend various amounts of time in different microenvironments, including the home, office, outdoors and in transit, where they are exposed to different magnitudes of particle and gaseous air pollutants. Health risks caused by air pollution exposure differ among individuals due to differences in activity, microenvironmental concentration, as well as the toxicity of pollutants. We evaluated individual and combined added health risks (AR) of exposure to PM_2.5, NO_2, and O₃ for 21 participants in their daily life based on real-world personal exposure measurements. Exposure errors from using surrogates were quantified. Inter- and intra-individual variability in health risks and key contributors in variations were investigated using linear mixed-effects models and correlation analysis, respectively. Substantial errors were found between personal exposure concentrations and ambient concentrations when using air quality measurements at either FSM or the residence location. The mean exposure errors based on the measurements taken at either the FSM or residence as exposure surrogates was higher for NO₂ than PM_2.5, because of the larger spatial variability in NO₂ concentrations in urban areas. The daily time-integrated AR for the combined PM_2.5, NO₂, and O₃ (TIAR_combine) ranged by a factor of 2.5 among participants and by a factor up to 2.5 for a given person across measured days. Inter- and intra-individual variability in TIAR_combine is almost equally important. Several factors were identified to be significantly correlated with daily TIAR_combine, with the top five factors, including PM_2.5, NO₂ and O₃ concentrations at ‘home indoor’, O₃ concentrations at ‘office indoor’ and ambient PM_2.5 concentrations. The results on the contributors of variability in the daily TIAR_combine could help in targeting interventions to reduce daily health damage related to air pollutants.

Elemental Composition of PM2.5 and PM10 and Health Risks Assessment in the Industrial Districts of Chelyabinsk, South Ural Region, Russia

Air pollution impacts all populations globally, indiscriminately and has site-specific variation and characteristics. Airborne particulate matter (PM) levels were monitored in a typical industrial Russian city, Chelyabinsk in three destinations, one characterized by high traffic volumes and two by industrial zone emissions. The mass concentration and trace metal content of PM2.5 and PM10 were obtained from samples collected during four distinct seasons of 2020. The mean 24-h PM10 ranged between 6 and 64 μg/m3. 24-h PM2.5 levels were reported from 5 to 56 μg/m3. About half of the 24-h PM10 and most of the PM2.5 values in Chelyabinsk were higher than the WHO recommendations. The mean PM2.5/PM10 ratio was measured at 0.85, indicative of anthropogenic input. To evaluate the Al, Fe, As, Cd, Co, Cr, Cu, Mn, Ni, Pb, and Zn concentration in PM2.5 and PM10, inductively coupled plasma mass spectrometry (ICP-MS) was used. Fe (337-732 ng/m3) was the most abundant component in PM2.5 and PM10 samples while Zn (77-206 ng/m3), Mn (10-96 ng/m3), and Pb (11-41 ng/m3) had the highest concentrations among trace elements. Total non-carcinogenic risks for children were found higher than 1, indicating possible health hazards. This study also presents that the carcinogenic risk for As, Cr, Co, Cd, Ni, and Pb were observed higher than the acceptable limit (1 × 10-6).

Characterization of Ambient Particulate Matters in an Industry-Intensive Area in Central Taiwan

Atmospheric particulate matters (PMs) were measured in an industry-intensive region in central Taiwan in order to investigate the characteristics and possible sources of PMs. The samplings were simultaneously conducted using a 10- and 3-stage Micro Orifice Uniform Deposit Impactor (MOUDI) from 2017 to 2018. In this study, the characteristics of PMs in this region were evaluated by measuring the mass concentration of PMs and analyzing water-soluble ions and metallic elements, as well as dioxins. Additionally, principal component analysis (PCA) was used to identify the potential sources of PMs. The results showed that the mean concentration of coarse (>1.8 μm), fine (0.1–1.8 μm), and ultrafine (<0.1 μm) particles were 13.60, 14.38, and 3.44 μg/m3, respectively. In the industry-intensive region, the size distribution of ambient particles showed a bi-modal distribution with a high concentration of coarse particles in the spring and summer, while fine particles were dominant in the autumn and winter. The most abundant water-soluble ions of PMs were NO3−, Cl−, and SO42−, while the majority of metallic elements were Na, Fe, Ca, Al, and Mg in different particle sizes. The results of Pearson’s correlation analysis for metals indicated that the particles in the collected air samples were related to the iron and steelmaking industries, coal burning, vehicle exhausts, and high-tech industries. The dioxin concentration ranged from 0.0006 to 0.0017 pg I-TEQ/Nm3. Principal component analysis (PCA) revealed that the contribution to PMs was associated with sea salt, secondary pollutants, and industrial process.

Predicting intraurban PM2.5 concentrations using enhanced machine learning approaches and incorporating human activity patterns

Urban areas contribute substantially to human exposure to ambient air pollution. Numerous statistical prediction models have been used to estimate ambient concentrations of fine particulate matter (PM_2.5) and other pollutants in urban environments, with some incorporating machine learning (ML) algorithms to improve predictive power. However, many ML approaches for predicting ambient pollutant concentrations to date have used principal component analysis (PCA) with traditional regression algorithms to explore linear correlations between variables and to reduce the dimensionality of the data. Moreover, while most urban air quality prediction models have traditionally incorporated explanatory variables such as meteorological, land use, transportation/mobility, and/or co-pollutant factors, recent research has shown that local emissions from building infrastructure may also be useful factors to consider in estimating urban pollutant concentrations. Here we propose an enhanced ML approach for predicting urban ambient PM_2.5 concentrations that hybridizes cascade and PCA methods to reduce the dimensionality of the data-space and explore nonlinear effects between variables. We test the approach using different durations of time series air quality datasets of hourly PM_2.5 concentrations from three air quality monitoring sites in different urban neighborhoods in Chicago, IL to explore the influence of dynamic human-related factors, including mobility (i.e., traffic) and building occupancy patterns, on model performance. We test 9 state-of-the-art ML algorithms to find the most effective algorithm for modeling intraurban PM_2.5 variations and we explore the relative importance of all sets of factors on intraurban air quality model performance. Results demonstrate that Gaussian-kernel support vector regression (SVR) was the most effective ML algorithm tested, improving accuracy by 118% compared to a traditional multiple linear regression (MLR) approach. Incorporating the enhanced approach with SVR algorithm increased model performance up to 18.4% for yearlong and 98.7% for month-long hourly datasets, respectively. Incorporating assumptions for human occupancy patterns in dominant building typologies resulted in improvements in model performance by between 4% and 37%. Combined, these innovations can be used to improve the performance and accuracy of urban air quality prediction models compared to conventional approaches.

Long-Term Ambient Air Pollution Exposures and Circulating and Stimulated Inflammatory Mediators in a Cohort of Midlife Adults

BACKGROUND

Chronic exposure to air pollution may prime the immune system to be reactive, increasing inflammatory responses to immune stimulation and providing a pathway to increased risk for inflammatory diseases, including asthma and cardiovascular disease. Although long-term exposure to ambient air pollution has been associated with increased circulating markers of inflammation, it is unknown whether it also relates to the magnitude of inflammatory response.

OBJECTIVES

The aim of this study was to examine associations between chronic ambient pollution exposures and circulating and stimulated levels of inflammatory mediators in a cohort of healthy adults.

METHODS

Circulating interleukin (IL)-6, C-reactive protein (CRP) (n=392), and lipopolysaccharide stimulated production of IL-1β, IL-6, and tumor necrosis factor (TNF)-α (n=379) were measured in the Adult Health and Behavior II cohort. Fine particulate matter [particulate matter with aerodynamic diameter less than or equal to 2.5 μm (PM2.5)] and constituents [black carbon (BC), and lead (Pb), manganese (Mn), zinc (Zn), and iron (Fe)] were estimated for each residential address using hybrid dispersion land use regression models. Associations between pollutant exposures and inflammatory measures were examined using linear regression; models were adjusted for age, sex, race, education, smoking, body mass index, and month of blood draw.

RESULTS

There were no significant correlations between circulating and stimulated measures of inflammation. Significant positive associations were found between exposure to PM2.5 and BC with stimulated production of IL-6, IL-1β, and TNF-α. Pb, Mn, Fe, and Zn exposures were positively associated with stimulated production of IL-1β and TNF-α. No pollutants were associated with circulating IL-6 or CRP levels.

DISCUSSION

Exposure to PM2.5, BC, Pb, Mn, Fe, and Zn was associated with increased production of inflammatory mediators by stimulated immune cells. In contrast, pollutant exposure was not related to circulating markers of inflammation. These results suggest that chronic exposure to some pollutants may prime immune cells to mount larger inflammatory responses, possibly contributing to increased risk for inflammatory disease. https://doi.org/10.1289/EHP7089.

Spatiotemporal dynamics and impacts of socioeconomic and natural conditions on PM2.5 in the Yangtze River Economic Belt

The determination of the spatiotemporal patterns and driving factors of PM_2.5 is of great interest to the atmospheric and climate science community, who aim to understand and better control the atmospheric linkage indicators. However, most previous studies have been conducted on pollution-sensitive cities, and there is a lack of large-scale and long-term systematic analyses. In this study, we investigated the spatiotemporal evolution of PM_2.5 and its influencing factors by using an exploratory spatiotemporal data analysis (ESTDA) technique and spatial econometric model based on remote sensing imagery inversion data of the Yangtze River Economic Belt (YREB), China, between 2000 and 2016. The results showed that 1) the annual value of PM_2.5 was in the range of 23.49-37.67 μg/m³ with an inverted U-shaped change trend, and the PM_2.5 distribution presented distinct spatial heterogeneity; 2) there was a strong local spatial dependence and dynamic PM_2.5 growth process, and the spatial agglomeration of PM_2.5 exhibited higher path-dependence and spatial locking characteristics; and 3) the endogenous interaction effect of PM_2.5 was significant, where each 1% increase in the neighbouring PM_2.5 levels caused the local PM_2.5 to increase by at least 0.4%. Natural and anthropogenic factors directly and indirectly influenced the PM_2.5 levels. Our results provide spatial decision references for coordinated trans-regional air pollution governance as well as support for further studies which can inform sustainable development strategies in the YREB.

The delayed effect of wildfire season particulate matter on subsequent influenza season in a mountain west region of the USA

Particularly in rural settings, there has been little research regarding the health impacts of fine particulate matter (PM2.5) during the wildfire season smoke exposure period on respiratory diseases, such as influenza, and their associated outbreaks months later. We examined the delayed effects of PM2.5 concentrations for the short-lag (1-4 weeks prior) and the long-lag (during the prior wildfire season months) on the following winter influenza season in Montana, a mountainous state in the western United States. We created gridded maps of surface PM2.5 for the state of Montana from 2009 to 2018 using spatial regression models fit with station observations and Moderate Resolution Imaging Spectroradiometer (MODIS) aerosol optical thickness data. We used a seasonal quasi-Poisson model with generalized estimating equations to estimate weekly, county-specific, influenza counts for Montana, associated with delayed PM2.5 concentration periods (short-lag and long-lag effects), adjusted for temperature and seasonal trend. We did not detect an acute, short-lag PM2.5 effect nor short-lag temperature effect on influenza in Montana. Higher daily average PM2.5 concentrations during the wildfire season was positively associated with increased influenza in the following winter influenza season (expected 16% or 22% increase in influenza rate per 1 μg/m3 increase in average daily summer PM2.5 based on two analyses, p = 0.04 or 0.008). This is one of the first observations of a relationship between PM2.5 during wildfire season and influenza months later.

Evaluation of metal aerosols in four communities adjacent to metal recyclers in Houston, Texas, USA

The metal recycling industry provides jobs, generates revenue in local communities and conserves energy and resources. Nonetheless, possible negative impacts of metal recyclers (MRs) include the potential for emissions of metal aerosols and other dusts, noise, traffic and fire during operations. In Houston, Texas, there were more than 180 resident complaints about air quality related to MRs from 2006 to 2011 that were reported to the city's 311 call system. As a part of a community-based participatory research study, Metal Air Pollution Partnership Solutions (MAPPS), we evaluated the impact of metal emissions from MRs on air quality over two years in four environmental justice communities. We simultaneously collected samples of inhalable particles (aerodynamic particle size less than 10 µm, PM10) using a sampling strategy to capture emissions from the MRs while they were in operation at four locations within each community: an upwind location, the fence line of MR and two downwind locations and analyzed the samples for 10 metals. The highest values of iron (Fe), manganese (Mn), nickel (Ni), lead (Pb), arsenic (As) and chromium (Cr) were detected at the fence lines of MRs. The normalized ratios of these metals at near and far neighborhood locations were 0.01 to 0.64 and 0.01 to 0.34, respectively, as compared with the metals at the fence line. The concentrations of metals rapidly decreased by 57-70% within 100 meters and reached similar levels at upwind (background) locations at approximately 600 meters. After adjusting the measured data for wind direction, rain and operating hours, we calculated non-carcinogenic hazard index values and carcinogenic risks for adult residents from breathing metals emitted from the facilities. Estimated inhalation cancer risks ranged from 0.12 case to 24 cases in 1 million people and the hazard index values ranged from 0.04 to 11.Implications: In Houston, Texas, residents complained about air quality related to metal recyclers from 2006 to 2011. Using a community-based participatory research method, metal emissions were characterized at four environmental justice communities. The results indicate that metal concentrations were the highest at the fence line and decreased by 57-70% within 100 meters and reached similar levels of background at 600 meters. After adjusting the measured data for meteorological parameters and operating hours, estimated inhalation cancer risks ranged from 0.12 cases to 24 cases in 1 million people and hazard index values ranged from 0.04 to 11.

Evaluation of the Efficiency of Arundo donax L. Leaves as Biomonitors for Atmospheric Element Concentrations in an Urban and Industrial Area of Central Italy

Washed and unwashed Arundo donax L. (A. donax) leaves were analyzed for elements, and results were compared with element concentrations detected in river water and particulate matter (PM) Samples were collected along a river in an urban and industrial hot spot of Central Italy, where element concentrations show relevant spatial gradients both in air and river water. The aim of this study is to identify the role of the two environmental matrices on leaves composition. Element concentrations of washed and unwashed leaves were compared to differentiate between the superficial deposition and the uptake into leaf tissues of elements. Water-soluble and -insoluble element concentrations were measured in PM10 samples collected on membrane filters by using innovative high spatial resolution samplers. The comparison among leaf and atmospheric concentrations of PM10 elements showed a similar trend for Ni, Mo, Cr, Ti, and Fe, which are reliable tracers of the PM10 contribution by steel plant and vehicular traffic. Soluble species appeared to be mainly bounded into leaf tissues, while insoluble species were deposited on their surface. On the other hand, element concentrations detected in washed A. donax leaves were poorly correlated with those measured in river water samples. The obtained results proved that A. donax leaves can be used as reliable biomonitors for the evaluation of the atmospheric concentrations of some PM10 elemental components.

Urban Oxidation Flow Reactor Measurements Reveal Significant Secondary Organic Aerosol Contributions from Volatile Emissions of Emerging Importance

Mobile sampling studies have revealed enhanced levels of secondary organic aerosol (SOA) in source-rich urban environments. While these enhancements can be from rapidly reacting vehicular emissions, it was recently hypothesized that nontraditional emissions (volatile chemical products and upstream emissions) are emerging as important sources of urban SOA. We tested this hypothesis by using gas and aerosol mass spectrometry coupled with an oxidation flow reactor (OFR) to characterize pollution levels and SOA potentials in environments influenced by traditional emissions (vehicular, biogenic), and nontraditional emissions (e.g., paint fumes). We used two SOA models to assess contributions of vehicular and biogenic emissions to our observed SOA. The largest gap between observed and modeled SOA potential occurs in the morning-time urban street canyon environment, for which our model can only explain half of our observation. Contributions from VCP emissions (e.g., personal care products) are highest in this environment, suggesting that VCPs are an important missing source of precursors that would close the gap between modeled and observed SOA potential. Targeted OFR oxidation of nontraditional emissions shows that these emissions have SOA potentials that are similar, if not larger, compared to vehicular emissions. Laboratory experiments reveal large differences in SOA potentials of VCPs, implying the need for further characterization of these nontraditional emissions.

Evaluating deciduous tree leaves as biomonitors for ambient particulate matter pollution in Pittsburgh, PA, USA

Fine particulate matter (PM_2.5) air pollution varies spatially and temporally in concentration and composition and has been shown to cause or exacerbate adverse effects on human and ecological health. Biomonitoring using airborne tree leaf deposition as a proxy for particulate matter (PM) pollution has been explored using a variety of study designs, tree species, sampling strategies, and analytical methods. In the USA, relatively few have applied these methods using co-located fine particulate measurements for comparison and relying on one tree species with extensive spatial coverage, to capture spatial variation in ambient air pollution across an urban area. Here, we evaluate the utility of this approach, using a spatial saturation design and pairing tree leaf samples with filter-based PM_2.5 across Pittsburgh, Pennsylvania, with the goal of distinguishing mobile and stationary sources using PM_2.5 composition. Co-located filter and leaf-based measurements revealed some significant associations with traffic and roadway proximity indicators. We compared filter and leaf samples with differing protection from the elements (e.g., meteorology) and PM collection time, which may account for some variance in PM source and/or particle size capture between samples. To our knowledge, this study is among the first to use deciduous tree leaves from a single tree species as biomonitors for urban PM_2.5 pollution in the northeastern USA.

The influence of spatial patterning on modeling PM2.5 constituents in Eastern Massachusetts

Geostatistical exposure methods for air pollution have inherent uncertainties, resulting in varying levels of exposure misclassification. In this study, we propose that areas representing clusters of PM_2.5 elements are potential predictor variables to be included in spatial models for particle composition. The inclusion of these clusters may minimize the exposure misclassification. We evaluated the influence of spatial patterning on modeling of 10 components of ambient PM_2.5, which included Al, Cu, Fe, K, Ni, Pb, S, Ti, V, and Zn. This study was performed in three stages. First, we applied a hybrid approach (combination of Empirical Bayesian Kriging and land use regression) to estimate spatial variability for each one of the 10 components of ambient PM_2.5. In this stage, we accounted for numerous predictors representing land use, transportation, demographic, and geographical characteristics. In the second stage, we applied the same hybrid approach adding clusters of each PM_2.5 component to the set of predictor variables. The clusters here were estimated by a multivariate clustering approach based on k means. Finally, in the last stage, we compared the estimates obtained from the model without clusters (first stage) and the model with clusters (second stage). Overall, our findings suggest significant influence of spatial clusters on modeling some PM_2.5 components. We observed that the clusters may affect the error of the prediction values and especially the proportion of explained variance for most of the PM_2.5 constituents evaluated in this study. The model with cluster presented a better performance for all PM_2.5 components, except for Pb, which the R² value decreased 8.51% when we included the clusters in the analysis; and for V, which the R² value did not change with the clusters. Models for Cu and Fe explained the highest concentration variance. The R² value for the model without cluster was 0.55 for both pollutants. When we accounted for clusters, R² value increased 13 and 7% for Cu (R² = 0.62) and Fe (R² = 0.59), respectively. The models for K and S presented the lowest performance for both models with and without cluster (although the model with cluster improved substantially the R² values). Better knowledge of the influence of spatial patterns on air pollution modeling should be of interest to policy makers to devise future strategies to improve human exposure assessment to air particulates while controlling for spatial patterns of ambient PM_2.5 elemental concentration.

Multivariate spatial patterns of ambient PM2.5 elemental concentrations in Eastern Massachusetts

Understanding the factors that affect spatial differences in PM_2.5 composition is crucial for implementing emissions control and health policies. Although previous studies have explored modeling of spatial patterns as a tool to improve human exposure assessment, little work has employed a multivariate clustering approach to identify spatial patterns in particle composition. In this study, we used this approach to assess the spatial patterns of ambient PM_2.5 elemental concentrations in Eastern Massachusetts in the United States. To distinguish one cluster of sites from another, we considered air pollution sources and geodemographic variables. We evaluated spatial patterns for 11 elemental components of ambient PM_2.5, which included S, K, Ca, Fe, Zn, Cu, Ti, Al, Pb, V, and Ni. The analyses for S, Ca, Cu, Ti, Al, and Pb resulted in: 2 clusters for Fe, Zn, V, and Ni; 3 clusters; and for 12 clusters for K. Overall, our findings suggest substantial variation of clusters among PM_2.5 components. In addition, land use, population density, and daily traffic were used as variables to more effectively characterize clusters of sites. We used R² values to estimate the effectiveness of each variable in characterizing clusters. Larger R² values indicate better the discrimination among the sites. For example, population density had the highest R² value when the analysis was performed for S, Ca, Zn, Ti, Al, Pb, and V; land use presented the highest R² value for Cu, V, and Ni; and, traffic showed the highest R² value for PM_2.5 mass concentration. This study improves the ability to model both the between- and within-area variability of source emissions and pollution regime, using concentrations of PM_2.5 components.

Evaluation of predictive capabilities of ordinary geostatistical interpolation, hybrid interpolation, and machine learning methods for estimating PM2.5 constituents over space

Numerous modeling approaches to estimate concentrations of PM_2.5 components have been developed to derive better exposures for health studies, including geostatistical interpolation approaches, land use regression models and, models based on remote sensing technology. Recently, there have been some efforts to develop models based on machine learning algorithms. Each one of these exposure assessment methods has inherent uncertainties resulting in varying levels of exposure misclassification. To date, only a few studies have attempted to systematically compare exposure estimates from different PM_2.5 constituent models. Our research addresses this gap, by comparing the predictive capabilities of ordinary geostatistical interpolation (Ordinary Kriging - OK), hybrid interpolation (combination of Empirical Bayesian Kriging and land use regression), and machine learning techniques (forest-based regression) for estimating PM_2.5 constituents in Eastern Massachusetts in the United States. We compared the estimates of 10 ambient PM_2.5 components, which included Al, Cu, Fe, K, Ni, Pb, S, Ti, V, and Zn. The OK model performed poorest for all PM_2.5 components, with an R² under 0.30. The hybrid model presented a slight improvement, especially for Cu and Fe, for which the R² value increased to 0.62 and 0.59, respectively. These elements presented the highest R² value from the hybrid model. The forest model presented the best performance, with R² values higher than 0.7 for most of the particle components, including Cu, Fe, Ni, Pb, Ti, and V. Same as observed with the hybrid model, the forest model for Cu and Fe explained the highest concentration variance, with a R² value equal to 0.88 and 0.92, respectively. The forest model for K, S, and Zn performed poorest with an R² value of 0.54, 0.37, and 0.44, respectively. The results presented here can be useful for the environmental health community to more accurately estimate PM_2.5 constituents over space.

Hybrid land use regression modeling for estimating spatio-temporal exposures to PM2.5, BC, and metal components across a metropolitan area of complex terrain and industrial sources

Land use regression (LUR) modeling has become a common method for predicting pollutant concentrations and assigning exposure estimates in epidemiological studies. However, few LUR models have been developed for metal constituents of fine particulate matter (PM_2.5) or have incorporated source-specific dispersion covariates in locations with major point sources. We developed hybrid AERMOD LUR models for PM_2.5, black carbon (BC), and steel-related PM_2.5 constituents lead, manganese, iron, and zinc, using fine-scale air pollution data from 37 sites across the Pittsburgh area. These models were designed with the aim of developing exposure estimates for time periods of interest in epidemiology studies. We found that the hybrid LUR models explained greater variability in PM_2.5 (R² = 0.79) compared to BC (R² = 0.59) and metal constituents (R² = 0.34-0.55). Approximately 70% of variation in PM_2.5 was attributable to temporal variance, compared to 36% for BC, and 17-26% for metals. An AERMOD dispersion covariate developed using PM_2.5 industrial emissions data for 207 sources was significant in PM_2.5 and BC models; all metals models contained a steel mill-specific PM_2.5 emissions AERMOD term. Other significant covariates included industrial land use, commercial and industrial land use, percent impervious surface, and summed railroad length.

Assessment of Spatial Variability across Multiple Pollutants in Auckland, New Zealand

Spatial saturation studies using source-specific chemical tracers are commonly used to examine intra-urban variation in exposures and source impacts, for epidemiology and policy purposes. Most such studies, however, has been performed in North America and Europe, with substantial regional combustion-source contributions. In contrast, Auckland, New Zealand, a large western city, is relatively isolated in the south Pacific, with minimal impact from long-range combustion sources. However, fluctuating wind patterns, complex terrain, and an adjacent major port complicate pollution patterns within the central business district (CBD). We monitored multiple pollutants (fine particulate matter (PM_2.5), black carbon (BC), elemental composition, organic diesel tracers (polycyclic aromatic hydrocarbons (PAHs), hopanes, steranes), and nitrogen dioxide (NO₂)) at 12 sites across the ~5 km² CBD during autumn 2014, to capture spatial variation in traffic, diesel, and proximity to the port. PM_2.5 concentrations varied 2.5-fold and NO₂ concentrations 2.9-fold across the CBD, though constituents varied more dramatically. The highest-concentration constituent was sodium (Na), a distinct non-combustion-related tracer for sea salt (µ = 197.8 ng/m³ (SD = 163.1 ng/m³)). BC, often used as a diesel-emissions tracer, varied more than five-fold across sites. Vanadium (V), higher near the ports, varied more than 40-fold across sites. Concentrations of most combustion-related constituents were higher near heavy traffic, truck, or bus activity, and near the port. Wind speed modified absolute concentrations, and wind direction modified spatial patterns in concentrations (i.e., ports impacts were more notable with winds from the northeast).

Separating spatial patterns in pollution attributable to woodsmoke and other sources, during daytime and nighttime hours, in Christchurch, New Zealand

During winter nights, woodsmoke may be a predominant source of air pollution, even in cities with many sources. Since two major earthquakes resulted in major structural damage in 2010 and 2011, reliance on woodburning for home heating has increased substantially in Christchurch, New Zealand (NZ), along with intensive construction/demolition activities. Further, because NZ is a relatively isolated western country, it offers the unique opportunity to disentangle multiple source impacts in the absence of long-range transport pollution. Finally, although many spatial saturation studies have been published, and levoglucosan is an established tracer for woodburning emissions, few studies have monitored multiple sites simultaneously for this or other organic constituents, with the ability to distinguish spatial patterns for daytime vs. nighttime hours, in complex urban settings. We captured seven-day integrated samples of PM_2.5, and elemental and organic tracers of woodsmoke and diesel emissions, during "daytime" (7 a.m. - 5:30 p.m.) and "nighttime" (7 p.m. - 5:30 a.m.) hours, at nine sites across commercial and residential areas, over three weeks in early winter (May 2014). At a subset of six sites, we also sampled during hypothesized "peak" woodburning hours (7 p.m. - 12 a.m.), to differentiate emissions during "active" residential woodburning, vs. overnight smouldering. Concentrations of PM_2.5 were, on average, were twice as high during nighttime than daytime [µ = 18.4 (SD = 6.13) vs. 9.21 (SD = 6.13) µg/m³], with much greater differences in woodsmoke tracers (i.e., levoglucosan [µ = 1.83 (SD = 0.82) vs. 0.34 (SD = 0.17) µg/m³], potassium) and indicators of treated- or painted-wood burning (e.g., arsenic, lead). Only nitrogen dioxide, calcium, iron, and manganese (tracers of vehicular emissions) were higher during daytime. Levoglucosan and most PAHs were higher during "active" woodburning, vs. overnight smouldering. Our time-stratified spatial saturation detected strong spatial variability throughout the study area, which distinctly differed during daytime vs. night time hours, and quantified the substantial contribution of woodsmoke to overnight spatial variation in PM_2.5 across Christchurch. Daytime vs. nighttime differences were greater than those observed across sites. Traffic, especially diesel, contributed substantially to daytime NO₂ and spatial gradients in non-woodsmoke constituents.

Fine-Scale Source Apportionment Including Diesel-Related Elemental and Organic Constituents of PM2.5 across Downtown Pittsburgh

Health effects of fine particulate matter (PM_2.5) may vary by composition, and the characterization of constituents may help to identify key PM_2.5 sources, such as diesel, distributed across an urban area. The composition of diesel particulate matter (DPM) is complicated, and elemental and organic carbon are often used as surrogates. Examining multiple elemental and organic constituents across urban sites, however, may better capture variation in diesel-related impacts, and help to more clearly separate diesel from other sources. We designed a “super-saturation” monitoring campaign of 36 sites to capture spatial variance in PM_2.5 and elemental and organic constituents across the downtown Pittsburgh core (~2.8 km²). Elemental composition was assessed via inductively-coupled plasma mass spectrometry (ICP-MS), organic and elemental carbon via thermal-optical reflectance, and organic compounds via thermal desorption gas-chromatography mass-spectrometry (TD-GCMS). Factor analysis was performed including all constituents—both stratified by, and merged across, seasons. Spatial patterning in the resultant factors was examined using land use regression (LUR) modelling to corroborate factor interpretations. We identified diesel-related factors in both seasons; for winter, we identified a five-factor solution, describing a bus and truck-related factor [black carbon (BC), fluoranthene, nitrogen dioxide (NO₂), pyrene, total carbon] and a fuel oil combustion factor (nickel, vanadium). For summer, we identified a nine-factor solution, which included a bus-related factor (benzo[ghi]fluoranthene, chromium, chrysene, fluoranthene, manganese, pyrene, total carbon, total elemental carbon, zinc) and a truck-related factor (benz[a]anthracene, BC, hopanes, NO₂, total PAHs, total steranes). Geographic information system (GIS)-based emissions source covariates identified via LUR modelling roughly corroborated factor interpretations.

Spatial Patterns in Rush-Hour vs. Work-Week Diesel-Related Pollution across a Downtown Core

Despite advances in monitoring and modelling of intra-urban variation in multiple pollutants, few studies have attempted to separate spatial patterns by time of day, or incorporated organic tracers into spatial monitoring studies. Due to varying emissions sources from diesel and gasoline vehicular traffic, as well as within-day temporal variation in source mix and intensity (e.g., rush-hours vs. full-day measures), accurately assessing diesel-related air pollution within an urban core can be challenging. We allocated 24 sampling sites across downtown Pittsburgh, Pennsylvania (2.8 km²) to capture fine-scale variation in diesel-related pollutants, and to compare these patterns by sampling interval (i.e., “rush-hours” vs. “work-week” concentrations), and by season. Using geographic information system (GIS)-based methods, we allocated sampling sites to capture spatial variation in key traffic-related pollution sources (i.e., truck, bus, overall traffic densities). Programmable monitors were used to collect integrated work-week and rush-hour samples of fine particulate matter (PM_2.5), black carbon (BC), trace elements, and diesel-related organics (polycyclic aromatic hydrocarbons (PAHs), hopanes, steranes), in summer and winter 2014. Land use regression (LUR) models were created for PM_2.5, BC, total elemental carbon (EC), total organic carbon (OC), elemental (Al, Ca, Fe), and organic constituents (total PAHs, total hopanes), and compared by sampling interval and season. We hypothesized higher pollution concentrations and greater spatial contrast in rush-hour, compared to full work-week samples, with variation by season and pollutant. Rush-hour sampling produced slightly higher total PM_2.5 and BC concentrations in both seasons, compared to work-week sampling, but no evident difference in spatial patterns. We also found substantial spatial variability in most trace elements and organic compounds, with comparable spatial patterns using both sampling paradigms. Overall, we found higher concentrations of traffic-related trace elements and organic compounds in rush-hour samples, and higher concentrations of coal-related elements (e.g., As, Se) in work-week samples. Mean bus density was the strongest LUR predictor in most models, in both seasons, under each sampling paradigm. Within each season and constituent, the bus-related terms explained similar proportions of variance in the rush-hour and work-week samples. Rush-hour and work-week LUR models explained similar proportions of spatial variation in pollutants, suggesting that the majority of emissions may be produced during rush-hour traffic across downtown. Results suggest that rush-hour emissions may predominantly shape overall spatial variance in diesel-related pollutants.

Determinants of personal exposure to fine particulate matter (PM2.5) in adult subjects in Hong Kong

Personal monitoring for fine particulate matter (PM_2.5) was conducted for adults (48 subjects, 18-63years of age) in Hong Kong during the summer and winter of 2014-2015. All filters were analyzed for PM_2.5 mass and constituents (including carbonaceous aerosols, water-soluble ions, and elements). We found that season (p=0.02) and occupation (p<0.001) were significant factors affecting the strength of the personal-ambient PM_2.5 associations. We applied mixed-effects models to investigate the determinants of personal exposure to PM_2.5 mass and constituents, along with within- and between-individual variance components. Ambient PM_2.5 was the dominant predictor of (R²=0.12-0.59, p<0.01) and the largest contributor (>37.3%) to personal exposures for PM_2.5 mass and most components. For all subjects, a one-unit (2.72μg/m³) increase in ambient PM_2.5 was associated with a 0.75μg/m³ (95% CI: 0.59-0.94μg/m³) increase in personal PM_2.5 exposure. The adjusted mixed-effects models included information extracted from individual's activity diaries as covariates. The results showed that season, occupation, time indoors at home, in transit, and cleaning were significant determinants for PM_2.5 components in personal exposure (R²_β=0.06-0.63, p<0.05), contributing to 3.0-70.4% of the variability. For one-hour extra time spent at home, in transit, and cleaning an average increase of 1.7-3.6% (ammonium, sulfate, nitrate, sulfur), 2.7-12.3% (elemental carbon, ammonium, titanium, iron), and 8.7-19.4% (ammonium, magnesium ions, vanadium) in components of personal PM_2.5 were observed, respectively. In this research, the within-individual variance component dominated the total variability for all investigated exposure data except PM_2.5 and EC. Results from this study indicate that performing long-term personal monitoring is needed for examining the associations of mass and constituents of personal PM_2.5 with health outcomes in epidemiological studies by describing the impacts of individual-specific data on personal exposures.

Intraurban Variation of Fine Particle Elemental Concentrations in New York City

Few past studies have collected and analyzed within-city variation of fine particulate matter (PM2.5) elements. We developed land-use regression (LUR) models to characterize spatial variation of 15 PM2.5 elements collected at 150 street-level locations in New York City during December 2008-November 2009: aluminum, bromine, calcium, copper, iron, potassium, manganese, sodium, nickel, lead, sulfur, silicon, titanium, vanadium, and zinc. Summer- and winter-only data available at 99 locations in the subsequent 3 years, up to November 2012, were analyzed to examine variation of LUR results across years. Spatial variation of each element was modeled in LUR including six major emission indicators: boilers burning residual oil; traffic density; industrial structures; construction/demolition (these four indicators in buffers of 50 to 1000 m), commercial cooking based on a dispersion model; and ship traffic based on inverse distance to navigation path weighted by associated port berth volume. All the elements except sodium were associated with at least one source, with R(2) ranging from 0.2 to 0.8. Strong source-element associations, persistent across years, were found for residual oil burning (nickel, zinc), near-road traffic (copper, iron, and titanium), and ship traffic (vanadium). These emission source indicators were also significant and consistent predictors of PM2.5 concentrations across years.