University of Southern California Dissertations and Theses

Physico-chemical properties and source apportionment of size-fractionated airborne particulate matter in urban areas with implications for public health

(USC Thesis Other)

Physico-chemical properties and source apportionment of size-fractionated airborne particulate matter in urban areas with implications for public health

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i PHYSICO-CHEMICAL PROPERTIES AND SOURCE APPORTIONMENT OF SIZE-FRACTIONATED AIRBORNE PARTICULATE MATTER IN URBAN AREAS WITH IMPLICATIONS FOR PUBLIC HEALTH By Sina Hasheminassab A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (ENVIRONMENTAL ENGINEERING) December 2016 ii Dedication To my beloved family for their perpetual trust, unwavering support, and unconditional love iii Acknowledgements First and foremost, I would like to express my deepest gratitude to my thesis advisor, Professor Constantinos Sioutas for his mentorship, full support, and generous guidance throughout the course of my doctoral research. His exceptional scientific knowledge, vision, and ideas have been consistent motivations for me to engage in research in aerosol science. I would also like to thank my defense committee members, Professor Rob McConnell and Professor George Ban-Weiss for their constructive feedback on my research. I am also very thankful to Drs. Schauer, Shafer, Delfino, and Ostro for their unwavering support in our collaborative studies. I owe much of my success to my former and current colleagues at the USC Aerosol Laboratory: Dr. Payam Pakbin, Dr. Kalam Cheung, Dr. Winnie Kam, Dr. Nancy Daher, Dr. Dongbin Wang, Arian Saffari, Farimah Shirmohammadi, and Mohammad Sowlat. It has been always a great pleasure to work with such a group of prominent researchers. Moreover, the successful completion of this dissertation would not have been possible without the financial support from the following funding agencies: South Coast Air Quality Management District (SCAQMD), California Air Resources Board (CARB), and California Environmental Protection Agency Office of Environmental Health Hazard Assessment (OEHHA). Last but not least, I would like to sincerely thank my family, relatives, and friends who endlessly inspired and supported me throughout this journey. Sina Hasheminassab Los Angeles, July 2016 iv Table of Contents Dedication ....................................................................................................................................... ii Acknowledgements ........................................................................................................................ iii List of Figures ............................................................................................................................... vii List of Tables ................................................................................................................................. xi Abstract ........................................................................................................................................ xiii CHAPTER 1 Introduction ........................................................................................................... 1 1.1. Background .................................................................................................................... 2 1.1.1. Air Pollution and Particulate Matter ......................................................................... 2 1.1.2. Characteristics of Particulate Matter ......................................................................... 2 1.1.3. Health Effects of Exposure to Particulate Matter ..................................................... 3 1.2. Source Apportionment ..................................................................................................... 4 1.2.1. Receptor Modeling.................................................................................................... 5 1.3. Rationale of the Proposed Research ................................................................................. 7 1.4. Objectives ........................................................................................................................ 9 1.5. Dissertation overview ..................................................................................................... 10 CHAPTER 2: Source Apportionment and Organic Compound Characterization of Ambient Ultrafine Particulate Matter in the Los Angeles Basin ........................................... 13 2.1. Introduction .................................................................................................................. 14 2.2. Methodology .................................................................................................................. 15 2.2.1. Sampling Sites Description ..................................................................................... 15 2.2.2. Sampling Method and Schedule ............................................................................. 16 2.2.3. Meteorology ............................................................................................................ 16 2.2.4. Chemical Analysis .................................................................................................. 17 2.2.5. Source Apportionment ............................................................................................ 18 2.3. Results and Discussion ................................................................................................... 19 2.3.1. Carbonaceous Species ............................................................................................. 19 2.3.2. Seasonal and Spatial Variability in Organic Compounds ....................................... 20 2.3.3. Source Contributions .............................................................................................. 28 v 2.4. Summary and Conclusions ............................................................................................. 33 2.5. Acknowledgments .......................................................................................................... 34 CHAPTER 3: Chemical characterization and source apportionment of indoor and outdoor fine particulate matter (PM 2.5) in retirement communities of the Los Angeles Basin ....... 35 3.1. Introduction ................................................................................................................... 36 3.2. Methodology ................................................................................................................. 37 3.2.1. Sampling sites and periods ........................................................................................ 37 3.2.2. Instrumentation and chemical analysis ...................................................................... 38 3.2.3. Source apportionment ................................................................................................ 39 3.2.4. Sensitivity of MM-CMB results to the vehicular emissions source profiles ............. 42 3.2.5. Meteorology and air exchange rates .......................................................................... 44 3.2.6. Data Analysis ............................................................................................................. 45 3.3. Results and discussions ................................................................................................. 46 3.3.1. Carbonaceous species ................................................................................................ 46 3.3.2. Metals and trace elements .......................................................................................... 49 3.3.3. Organic compounds ................................................................................................... 51 3.3.4. Source apportionment of PM2.5 ................................................................................. 57 3.4. Conclusions ................................................................................................................... 62 3.5. Acknowledgements ....................................................................................................... 63 CHAPTER 4: Long-term source apportionment of ambient fine particulate matter (PM2.5) in the Los Angeles Basin: A focus on emissions reduction from vehicular sources .............. 64 4.1. Introduction ................................................................................................................... 65 4.2. Methodology ................................................................................................................. 67 4.2.1. Sampling sites ............................................................................................................ 67 4.2.2. Sampling schedule and chemical analysis ................................................................. 68 4.2.3. Source apportionment ................................................................................................ 68 4.2.4. Statistical analysis...................................................................................................... 80 4.3. Results and discussion ................................................................................................... 80 4.3.1. Source identification and apportionment ................................................................... 80 4.3.2. Comparison with previous studies ............................................................................. 89 4.3.3. Emissions reduction from vehicular sources ............................................................. 91 vi 4.4. Summary and conclusions ............................................................................................. 99 4.5. Acknowledgments ....................................................................................................... 100 CHAPTER 5: Spatial and temporal variability of sources of ambient fine particulate matter (PM2.5) in California ................................................................................................... 101 5.1. Introduction ................................................................................................................. 102 5.2. Methodology ............................................................................................................... 104 5.2.1. Sampling sites .......................................................................................................... 104 5.2.2. Sampling schedule and chemical analysis ............................................................... 106 5.2.3. Source apportionment .............................................................................................. 107 5.3. Meteorology ................................................................................................................ 111 5.4. Results and discussion ................................................................................................. 113 5.4.1. Particulate mass ....................................................................................................... 113 5.4.2. Source characterization and apportionment ............................................................ 114 5.5. Summary and conclusions ........................................................................................... 138 5.6. Acknowledgements ..................................................................................................... 138 CHAPTER 6: Summary, Conclusions, and Recommendations ........................................... 139 6.1. Summary and Conclusions .......................................................................................... 140 6.2. Limitations of the Current Investigations ................................................................... 143 6.3. Recommendations for Future Research ...................................................................... 145 Bibliography ............................................................................................................................... 149 vii List of Figures Figure 2.1. Location of the 10 sampling sites. 15 Figure 2.2 a-d. Seasonal variation in the concentrations (µg m -3 ) of elemental carbon (EC), non-biomass burning water soluble organic carbon (WSOCnb), biomass burning water soluble organic carbon (WSOCbb), and water insoluble organic carbon (WIOC) in quasi-ultrafine particles (dp<0.25 µm) by site. 21 Figure 2.3 a-d. Seasonal variation in the concentration (ng m -3 ) of select Polycyclic Aromatic Hydrocarbons (PAHs) in quasi-ultrafine particles (dp< 0.25 µm) by site. 22 Figure 2.4 a-d. Seasonal variation in the concentration (ng m -3 ) of select hopanes and steranes in quasi-ultrafine particles (dp< 0.25 µm) by site. 25 Figure 2.5 a-d. Seasonal variation of n-Alkanes (C24-C33) concentration (ng m -3 ) in quasi- ultrafine particles (dp<0.25 µm) along with their carbon preference index (CPI). 26 Figure 2.6 a-d. Seasonal variation in the concentration (ng m -3 ) of levoglucosan in quasi- ultrafine particles (dp< 0.25 µm) by site. 27 Figure 2.7 a-d. Seasonal variation in source apportionment of organic carbon (OC) in quasi-ultrafine particles (dp<0.25 µm) by site. 28 Figure 2.8 a-d. Linear correlation between the concentrations of “other OC” and non- biomass burning water soluble organic carbon (WSOCnb) in (a) spring, (b) summer, (c) fall and (d) winter. 30 Figure 2.9 a-d. Seasonal variation in source apportionment of quasi-ultrafine particles (dp<0.25 µm) by site. 32 Figure 3.1. Comparison of mobile (LDV+HDV) source contribution estimates (SCE) to PM2.5 OC using source profiles from two separate studies: mobile source profile 1 (MSP1) from Kam et al. (2012) and Liacos et al. (2012); mobile source profile 2 (MSP2) from Kuhn et al. (2005), Ntziachristos et al. (2007), and Phuleria et al. (2007). Errors in the slope and intercept are standard error. 43 Figure 3.2. Average indoor-outdoor air exchange rate ([AER] hr -1 ) at each site during the warm and cold phases. Error bars correspond to one standard deviation. 45 Figure 3.3. Average indoor (IN) and outdoor (OUT) mass concentrations (µg/m 3 ) of elemental carbon (EC), water-soluble organic carbon (WSOC), and water- insoluble organic carbon (WIOC) in the fine PM size fraction by site during the warm and cold phases. Error bars correspond to one standard deviation. 46 Figure 3.4. Box plot of weekly indoor-to-outdoor (I/O) mass ratios for groups of individual organic compounds (including PAHs, Hopanes and steranes, n-alkanes, and organic acids), metals and elements, and carbonaceous species (EC, WSOC, 48 viii WIOC) during the warm and cold phases. Each box represents the data for all 3 sites pooled together. The reference line shows the I/O mass ratio of 1. Figure 3.5 a-d. Average mass concentrations (ng/m 3 ) of total (a) PAHs, (b) hopanes and steranes, (c) n-alkanes, and (d) organic acids in indoor and outdoor environments at each site during the warm and cold phases. Error bars correspond to one standard deviation. 52 Figure 3.6 a-d. Indoor-to-outdoor (I/O) mass ratios of selected (a) PAHs, (b) hopanes and steranes, (c) n-alkanes, and (d) organic acids, averaged over all sites during the warm and cold phases. Error bars correspond to one standard deviation. The reference line shows the I/O mass ratio of 1. 54 Figure 3.7 a-d. Pearson correlation coefficients of selected (a) PAHs, (b) hopanes and steranes, (c) n-alkanes, and (d) organic acids, averaged over all sites during the warm and cold phases. Error bars correspond to one standard deviation. 55 Figure 3.8. Average carbon preference index (CPI) of n-alkanes (C19-C40) at the indoor and outdoor sampling sites during the warm and cold phases. Error bars correspond to one standard deviation. 57 Figure 3.9. Average relative contribution of different sources to fine OC mass concentration (µg/m 3 ) in indoor (IN) and outdoor (OUT) environments at each site during the warm and cold phases. 60 Figure 3.10. Average relative contribution of different sources to PM2.5 mass concentration (µg/m 3 ) in indoor (IN) and outdoor (OUT) environments at each site during the warm and cold phases. 61 Figure 4.1. Location of the sampling sites in downtown Los Angeles and Rubidoux. 67 Figure 4.2 a-d. Linear regression of OC mass concentration, obtained from TOT and TOR measurement methods, versus PM2.5 mass concentration in Los Angeles and Rubidoux. Errors represent the standard error. 70 Figure 4.3 a-b. a) Linear correlation between artifact corrected OC TOT and artifact corrected OC TOR; b) linear correlation between EC TOT and EC TOR. Data were obtained from co-located measurement of STN and IMPROVE-like samplers in Los Angeles and Rubidoux. Errors correspond to standard error. 71 Figure 4.4. Linear correlation between Ion Chromatography (IC) K + , IC Na + , and IC SO4 2- , and X-ray Fluorescence (XRF) K, XRF Na, and XRF S in Los Angeles and Rubidoux. 72 Figure 4.5. PM2.5 source profiles and explained variation of each species, obtained from PMF model using two separate input datasets (2002-2006 and 2008-2012) in Los Angeles. Error bars correspond to one standard deviation obtained from the bootstrap analysis. 76 ix Figure 4.6. PM2.5 source profiles and explained variation of each species, obtained from PMF model using two separate input datasets (2002-2006 and 2008-2012) in Rubidoux. Error bars correspond to one standard deviation obtained from the bootstrap analysis. 77 Figure 4.7 a-b. Linear correlations between the measured and estimated PM 2.5 in a) Los Angeles and b) Rubidoux. Errors represent the standard error. 81 Figure 4.8. PM2.5 source profiles and explained variation (EV) of each species in Los Angeles. Error bars correspond to one standard deviation obtained from bootstrap analysis. 82 Figure 4.9. PM2.5 source profiles and explained variation (EV) of each species in Rubidoux. Error bars correspond to one standard deviation obtained from bootstrap analysis. 83 Figure 4.10 a-b. Annual average gravimetric mass concentration and estimated source contributions (µg/m 3 ) to ambient PM2.5 by year, in a) Los Angeles and b) Rubidoux. Error bars correspond to one standard error. 84 Figure 4.11 a-b. Seasonal average source contributions (µg/m 3 ) to PM2.5 in a) Los Angeles and b) Rubidoux. Error bars correspond to one standard error. 85 Figure 4.12 a-b. Average source contributions (µg/m 3 ) to PM2.5 during weekdays and weekends in a) Los Angeles and b) Rubidoux. Error bars correspond to one standard error. 86 Figure 4.13 a-d. Box plot of daily-resolved source contributions from vehicular sources, segregated by year, in a) Los Angeles and b) Rubidoux. 92 Figure 4.14 a-b. Box plot of daily-resolved source contributions from vehicular sources, segregated by year cluster, in a) Los Angeles and b) Rubidoux. 92 Figure 4.15 a-d. Box plots of daily-averaged nitrogen oxides (NOx) (ppm), segregated by year and year cluster, in Los Angeles and Rubidoux. 94 Figure 4.16 a-b. Box plots of daily-averaged traffic flow (vehicles/day), segregated by year, in a) Los Angeles and b) Rubidoux. 96 Figure 4.17 a-b. Box plots of daily-averaged traffic flow (vehicles/day), segregated by year cluster, in a) Los Angeles and b) Rubidoux. 96 Figure 4.18 a-f. Box plots of daily-averaged temperature (T) and relative humidity, as well as annual total precipitation, segregated by year, in Los Angeles and Rubidoux. 98 Figure 5.1. Location of the sampling sites. 104 Figure 5.2. Annual average concentration (µg/m 3 ) of uncorrected organic carbon (OC) from 2002 to 2007 in Los Angeles. Error bars correspond to one standard error. 109 x Figure 5.3. Scatter plot of OC mass concentration, obtained from Thermal Optical Transmittance (TOT) NIOSH 5040 method, versus PM2.5 mass concentration in Los Angeles, segregated by year. 109 Figure 5.4 a-h. PM2.5 source profiles and explained variation (EV) of each species in a) El Cajon, b) Rubidoux, c) Los Angeles, d) Simi Valley, e) Bakersfield, f) Fresno, g) San Jose, and f) Sacramento. Error bars correspond to one standard deviation obtained from bootstrap analysis. 122 Figure 5.5 a-d. Seasonal variation in the percent contribution of identified sources to ambient PM2.5, by site. 125 Figure 5.6. Seasonal average source contribution (µg/m 3 ) of vehicular emissions to ambient PM2.5, by site. Error bars correspond to one standard error. 127 Figure 5.7. Seasonal average source contribution (µg/m 3 ) of secondary ammonium nitrate to ambient PM2.5, by site. Error bars correspond to one standard error. 129 Figure 5.8. Seasonal average source contribution (µg/m 3 ) of secondary ammonium sulfate to ambient PM2.5, by site. Error bars correspond to one standard error. 130 Figure 5.9. Seasonal average source contribution (µg/m 3 ) of biomass burning to ambient PM2.5, by site. Error bars correspond to one standard error. 131 Figure 5.10. Seasonal average source contribution (µg/m 3 ) of soil to ambient PM2.5, by site. Error bars correspond to one standard error. 132 Figure 5.11. Seasonal average source contribution (µg/m 3 ) of fresh sea salt to ambient PM2.5, by site. Error bars correspond to one standard error. 134 Figure 5.12. Seasonal average source contribution (µg/m 3 ) of aged sea salt to ambient PM2.5, by site. Error bars correspond to one standard error. 134 Figure 5.13. Seasonal average source contribution (µg/m 3 ) of industrial emissions to ambient PM2.5, by site. Error bars correspond to one standard error. 137 Figure 5.14. Seasonal average contribution (µg/m 3 ) of chlorine sources to ambient PM2.5, by site. Error bars correspond to one standard error. 137 xi List of Tables Table 2.1. Select meteorological parameters at sites clusters during spring (March-May), summer (June-August), fall (September-November) and winter (December- February). 17 Table 2.2 a-b. Principal component loadings (VARIMAX normalized) of select organic species, EC, vanadium, and nickel in quasi-UFP (d p <0.25 µm) in (a) all sites and seasons pooled together and (b) all sites and seasons except spring- and summer- time data for HMS and FRE. 23 Table 3.1. Contribution of mobile sources to PM2.5 in the co-linear cases, assuming that the OC apportioned to mobile sources is emitted from 1) only LDV, 2) only HDV, 3) 25% LDV, 75% HDV, 4) 75% LDV, 25% HDV. The units are in µg/m 3 . 41 Table 3.2. Select meteorological parameters at each site during the warm and cold phases. 44 Table 3.3. Pearson correlation coefficients (R) and indoor-to-outdoor (I/O) mass ratios of elemental carbon (EC), water-soluble organic carbon (WSOC), water-insoluble organic carbon (WIOC), and selected metals and trace elements, averaged over all sites during the warm and cold phases. Errors correspond to one standard deviation. 50 Table 4.1. The U.S. EPA diesel truck emission standards for PM and NOx (g/kW.hr) along with compliance date mandated by CARB and Ports of Los Angeles and Long Beach. 66 Table 4.2. Summary statistics and mass concentrations of PM2.5 and its chemical constituents in Los Angeles and Rubidoux. Units for PM2.5, EC, OC, SO4 2- ,NH4 + , and NO3 - are in µg/m 3 , and for other species are in ng/m 3 . 73 Table 4.3 a-b. Average (± standard error) source contributions (µg/m 3 ) between 2002 and 2006, obtained from the results of the PMF model, using two separate input datasets (2002-2013 and 2002-2006) in a) Los Angeles and b) Rubidoux. 79 Table 4.4 a-b. Average (± standard error) source contributions (µg/m 3 ) between 2008 and 2012, obtained from the results of the PMF model, using two separate input datasets (2002-2013 and 2008-2012) in a) Los Angeles and b) Rubidoux. 79 Table 4.5 a-b. Average source contributions (± standard error) to PM2.5 in a) Los Angeles and b) Rubidoux, obtained from different studies. Units are in µg/m 3 . 90 Table 5.1. Estimated OC artifact values (µg/m 3 ) from Thermal Optical Transmittance (TOT) NIOSH 5040 carbon method, at each sampling site. Errors correspond to one standard error. 108 xii Table 5.2. Estimated OC artifacts (±standard errors) and the p values corresponding to the two-tailed t tests between OC artifact values in each two consecutive years in Los Angeles. Errors correspond to one standard error. 108 Table 5.3. Select meteorological parameters at each sampling site during spring, summer, fall, and winter. Seasonal averages were calculated over 6 years (from 2002 to 2007). 112 Table 5.4. Seasonal average mass concentration (± standard error) (µg/m 3 ) of ambient PM2.5 at the 8 sampling sites in the period between 2002 and 2007. 114 Table 5.5. Summary of the marker species for identified PM2.5 sources, resolved by the PMF model. 123 Table 5.6. Summary statistics of the linear regressions between daily-resolved measured ambient PM2.5 and estimated PM2.5 mass concentrations obtained from the PMF model. Errors correspond to one standard error. 123 xiii Abstract The association between adverse health effects and exposure to airborne particulate matter (PM) has been the subject of numerous epidemiological and toxicological researches. Most of these studies have used total particle mass concentration as a metric to assess the health effects of exposure to particles; however, PM is a complex mixture of several classes of chemical species, which may emanate from various sources. Thus, depending of the size of particles and their source of emission, the relative toxicity of emissions may vary significantly. Consequently, treating ambient PM in regulatory contexts solely based on total mass concentrations, regardless of chemical composition or source of emission, may not provide the information necessary to make effective decisions regarding the protection of human health; rather, a more efficient air pollution management plan requires the identification of PM sources that are detrimental to human health and to prioritize the reduction of the more toxic sources. The techniques used to identify and quantify the sources of airborne PM are called source apportionment (SA). SA is an important task in air pollution management, control, and policy options, and involves the reconstruction of PM emissions from different sources at a designated location to determine their impact. Receptor modeling is a widespread SA technique that relies heavily on statistical analysis of the chemical composition of PM at a receptor location, which makes chemical characterization of ambient PM and emission sources an important component of SA studies. In this dissertation, a series of studies were carried out in different locations (indoor and outdoor) within the state of California to characterize the chemical composition of airborne PM in different size fractions. In each study, a receptor modeling technique was used to identify and quantify the sources that contributed to total PM mass concentration at the receptor location; the spatial and temporal variations of the chemical components and the identified sources were then xiv investigated. In addition, the SA results were used in subsequent health studies to evaluate the potential associations between source-specific ambient PM and a variety of health outcomes. Furthermore, the efficacy of major emission reduction policies on ambient air quality in the Los Angeles Basin was evaluated through investigation of the annual trends of the identified sources over the past decade in this region. Finally, the research carried out in this dissertation was used to investigate the degree to which the emissions from the identified sources penetrate into indoor environments, where people spend around 85–90% of their time. The findings of this research will help federal and state regulatory agencies understand the linkage among sources, composition, and the associated health effects of size-fractionated ambient PM. They will also help to determine if there is a scientific foundation for controlling ambient PM from only a subset of PM sources with the goal of maximum benefit to human health. 1 CHAPTER 1 Introduction 2 1.1. Background 1.1.1. Air Pollution and Particulate Matter Nowadays, air pollution is a common concern in the major and growing metropolitan areas around the world. Air pollution is defined as the contamination of indoor or outdoor environments by any type of chemical, physical, and biological agents that can be harmful to the health or comfort of humans or cause damage to other living organisms (Nowak et al., 2006). The substances that cause air pollution are called air pollutants. Under the Clean Air Act, The United States Environmental Protection Agency (U.S. EPA) identified six common air pollutants of concern, named “criteria pollutants” including particulate matter (PM), ground-level ozone, carbon monoxide, sulfur oxides, nitrogen oxides, and lead. To protect human health, the U.S. EPA set the National Ambient Air Quality Standards (NAAQS) for the aforementioned criteria pollutants. Particulate matter is a general term used for a complex mixture of solid and/or liquid materials suspended in a gas medium. Particles have different sizes, shapes, and chemical characteristics and are a complex mixture of several subclasses of chemical species, including carbonaceous species, inorganic ions, metals and trace elements as well as organic compounds, some of which could be hazardous to human health. PM can either be directly emitted into the air (defined as primary PM) or formed through chemical reactions of gaseous precursors in the atmosphere (defined as secondary PM). Nonetheless, when particles are emitted, they undergo several physical and chemical processes, altering their size and chemical composition. Primary PM and the gaseous precursors have both anthropogenic and natural (biogenic) sources. Volcano, wind-blown dust, and wildfires are typical natural sources of ambient PM. Anthropogenic sources of ambient PM can be either stationary, such as coal-fired power plant, or mobile, such as vehicular sources. In addition to affecting human health, PM in the atmosphere is responsible for a variety of environmental effects including regional visibility, global climate change, soiling and damage to materials (Mathai, 1990; Seinfeld and Pandis, 2006). 1.1.2. Characteristics of Particulate Matter The size of particles, which is usually identified by the aerodynamic diameter, is the most important characteristic of PM, governing their behavior and health outcomes. According to definition, aerodynamic diameter is the diameter of a sphere with unit density that has the same terminal velocity in the air as the particle of interest (Hinds, 2012). Particles which are smaller 3 than 10 µm are usually the point of interest in the environmental health studies, mainly due to the fact that these particles are the ones that generally pass through the throat and nose and enter the lungs, while particles larger than 10 µm have very short life time in the atmosphere and near 100% deposition efficiency in human nose (Hinds, 2012). Based on their aerodynamic diameters, particles are usually classified in three separate fractions: coarse (particles smaller than 10 µm and larger than 2.5 µm, PM10-2.5), accumulation (or Aitken) (0.1-2.5 µm), and ultrafine (0.01-0.1 µm) modes. This classification is because particles show distinct behaviors and they have different emissions sources, formation mechanism, chemical composition, and residence time in the atmosphere within each size fraction. Ambient particles depending on their sizes emanate from different sources or have different formation mechanisms. Ultrafine particles are mostly emitted from fossil fuel combustion, particularly vehicular sources (Morawska et al., 2008; Sowlat et al., 2016; Vu et al., 2015; Westerdahl et al., 2005), or can be formed secondarily through photochemical reactions of gaseous precursors in the atmosphere (Kulmala et al., 2004; Sioutas et al., 2005). Likewise, particles in the accumulation mode are mostly emitted from combustion processes. These particles can be also formed by photochemical reactions, condensation of gaseous precursors, or coagulation of smaller particles in the atmosphere (Zhang et al., 2015). Lastly, coarse particles are generally mechanically generated by crushing or gridding operations (Pakbin et al., 2010). Diffusion and gravitation are respectively the main removal mechanisms of ultrafine and coarse particles, while accumulation mode particles are relatively too large to be removed by diffusion and too small to be settled down by gravity; therefore, these particles have the greatest residence time in the atmosphere and can travel larger distance (Hinds, 2012). 1.1.3. Health Effects of Exposure to Particulate Matter The most important motivation for research on particulate matter is the adverse health outcomes which are caused by exposure to PM (Kim et al., 2015). More than 3 million premature deaths are occurring annually in the world due to exposure to ambient PM (Lim et al., 2012). Numerous epidemiological and toxicological studies have found strong associations between exposure to PM and a verity of acute and chronic health outcome, including but not limited to premature deaths (Ostro et al., 2015), respiratory and cardiovascular diseases (Analitis et al., 2006; Donaldson et al., 2001; Gray et al., 2015; Hoek et al., 2013), and neurodegenerative disorders 4 (Peters et al., 2006). Most of these studies have used particle mass concentration as a metric to assess the health effects of exposure to particles; however, studies have attempted to link adverse health effects with other particle characteristics such as particle size, surface area, apparent density, number concentration, and chemical composition (Leonidas Ntziachristos et al., 2007; Rückerl et al., 2016; Saldiva et al., 2002). Since the physical and chemical properties of particles are interlinked, it is most likely that a combination of these properties may determine the overall particle toxicity. Although the exact mechanisms by which PM impacts human health are not fully resolved yet, a considerable body of evidence suggests that PM intake leads to cellular oxidative stress through the formation of reactive oxygen species (ROS) at the surface of target cells and change their redox status (Ayres et al., 2008; Kelly and Fussell, 2012; Møller et al., 2010; Nel, 2005). As noted above, particle size is one of the most important parameters mediating the health effects of PM. The NAAQS is currently limited to ambient PM10 and PM2.5 mass concentrations. However, several studies have suggested that exposure to smaller particles (i.e., ultrafine particles) is more detrimental to human health (Cho et al., 2005; Delfino et al., 2005; Leonidas Ntziachristos et al., 2007). Compared to coarse and fine particles, UFP have relatively low mass concentration, but they account for the majority of ambient PM number concentration (Cho et al., 2005). Given their increased number concentrations, therefore a large surface area, UFPs can carry a considerable amount of toxic pollutants per unit mass (Sioutas et al., 2005). In addition, due to their smaller diameters, ultrafine particles have higher pulmonary deposition efficiently. Thus exposure to UFPs may have substantial health effects; greater than or independent of those induced by larger particles. 1.2. Source Apportionment In the atmospheric sciences and air pollution area, source apportionment models aim to re- construct the impact of PM emissions from different sources at a designated location (Hopke, 2016). Source apportionment methods are generally classified in two groups: a) Source modeling: this method, which is also called dispersion modeling, utilizes the pollutant emission rate and meteorological information in a mathematical model, which describes the atmosphere, dispersion, and physical/ chemical processes within the plume to predict the pollutant’s concentration at a point of space and time (Holmes and 5 Morawska, 2006). The major drawback of these techniques is that they require detailed emission inventories that are not always available, and they are also limited by the accuracy of emission inventories, particularly when emissions from natural sources are important. CALPUFF, AERMOD, and CMAQ are examples of commonly-used dispersion models. b) Receptor modeling: receptor-oriented methods use measured ambient concentrations of pollutants at a receptor location to identify the presence of sources and to quantify their contributions in the studies area (Henry et al., 1984). Unlike source modeling, the major limitation of receptor modeling is its inability to predict the ambient concentrations and/or source contributions forward in time. The most commonly-used receptor models, which have been approved by the U.S. EPA are: Chemical Mass Balance (CMB) model, Positive Matrix Factorization (PMF), and UNMIX. The receptor modeling is more acceptable for regulatory purposes, mainly due to the direct measurement of pollutants, analyzing them in the laboratory, and quantifying (statistically) the contribution of various sources to the pollution at specific locations. On the downside, the number of measurements is few, due to the costs involved in sampling and analysis, which can be compensated by source modeling. In this dissertation different receptor modeling techniques are applied on PM chemical data sets, collected in distinctly different locations of California (indoors and outdoors), to identify and quantify major sources that contribute to the mass concentrations of airborne PM. A more detail discussion on receptor modeling will be provided in the following section. 1.2.1. Receptor Modeling The measured chemical composition of PM at a receptor site can be used to estimate the relative contribution of major sources to the measured total PM mass (Schauer et al., 1996). The fundamental principal behind this methodology, which is referred to receptor modeling, is the mass conservation of chemical species during their transport from source to receptor. Therefore, a chemical mass balance equation can be written for each chemical constituent 𝑖 at the receptor site 𝑘 , emitted from 𝑝 independent sources: 𝐶 𝑖𝑘 = ∑ 𝑎 𝑖𝑗 𝑆 𝑗 𝑘 𝑓 𝑖𝑗𝑘 𝑝 𝑗 =1 (1) 6 where 𝑐 𝑖𝑘 is the mass concentration of chemical species 𝑖 at receptor site 𝑘 originated from source 𝑗 , 𝑆 𝑗𝑘 is the mass contribution from source 𝑗 to the total mass sampled in receptor site 𝑘 , 𝑎 𝑖𝑗 is the relative concentration of chemical species 𝑖 emitted from source 𝑗 , and 𝑓 𝑖𝑘 is the coefficient of fractionation which account for the modification of 𝑎 𝑖𝑗 during the transport from source 𝑗 to receptor site 𝑘 . There are several approaches to solve equation (1) depending on what information is available. If prior information about sources and their compositions are available (i.e, 𝑎 𝑖𝑗 and 𝑝 ), then the only unknown is 𝑆 𝑗𝑘 . In this case, equation (1) is typically solved with an effective variance weighted least-squares solution, which is referred to as a chemical mass balance model (CMB) (Watson et al., 1984). On the other hand, in the absence of information about the nature of sources and their compositions, equation (1) is typically solved for two unknowns (𝑎 𝑖𝑗 and 𝑆 𝑗𝑘 ) by factor analysis models. Positive Matrix Factorization (PMF) (Paatero, 1997) and UNMIX (Henry and Kim, 1990) are the most commonly-used models to solve the factor analysis problems. PMF uses a weighted least-square approach and imposes non-negativity constraints for fitting the factor analysis model (Paatero, 1997; Paatero and Tapper, 1994). UNMIX, however, solves the chemical mass balance equation by using a principle component analysis (PCA) approach to reduce the number of dimensions in the space to the number of factors that produce the data, followed by an unique “edge detection” technique to identify “edges” defined by the data points in the space of reduced dimension (Henry, 2003). As noted above, CMB model requires source profiles, which are the mass fractions (i.e., composition) of species in source emissions, as well the ambient concentrations of species as input data. In order for CMB to provide quantitative uncertainties on source contribution estimates, both ambient concentration and source profile data sets must accompanied by their uncertainties. In addition, the species and particle size fraction measured in source profiles should match those in ambient samples to be apportioned. Since CMB assumes that all observed mass is emitted from the added sources in the model, identification of new or unknown sources is not possible. One of the other drawbacks of CMB is that it does not typically apportion secondary sources, unless it is combined with profile aging models (Watson et al., 2008). Another weakness of the CMB modeling is that chemically-similar sources may lead to collinearity in the results without more specific chemical markers (Lee et al., 2008). Nonetheless, the best advantage of CMB modeling is 7 that it can be applied on a single-sample measurement. PMF, on the other hand, requires a very large ambient data set, usually more than 100 samples in order to be able to resolve the latent factor profiles. Similarly to CMB, the observed concentrations should be allocated by appropriate uncertainties, as the application of the PMF model depends mainly on the estimated uncertainties. Unlike CMB, PMF model enables the user to identify new or unknown sources in the study area, as it does not require information on the source profiles as input data in the model. Therefore, PMF is an appropriate source apportionment approach for areas where detailed and/or accurate emission inventories are not available. 1.3. Rationale of the Proposed Research As discussed above, numerous epidemiological and toxicological studies have found a persistent association between exposure to particulate matter and several severe health outcomes (Bell et al., 2008; Dominici F et al., 2006; Peng RD et al., 2008). The majority of these studies used the PM mass concentration as the particulate pollution index. Therefore, out of all properties of particulate matter (e.g., mass, number, surface area, chemical composition), only the mass concentration of PM has so far received the utmost attention, is currently regulated by the NAAQS, and is used as a metric to assess the ambient air quality. However, it is well known that PM is a complex mixture of particles and chemical species, which may be compositionally diverse, depending on the location, particle size, emission sources, season, and the atmospheric condition. For example, particles emitted from vehicular sources are usually enriched with organic and elemental carbon and mostly apportioned in the smaller size fractions (Kuhn et al., 2005; L. Ntziachristos et al., 2007b; Phuleria et al., 2007), while soil and/or road dust are usually enriched in mineral and crustal materials and partitioned in the coarse size fraction (Pakbin et al., 2011). In addition to primary sources, secondary compounds comprise a considerable fraction of ambient PM, again depending on the location and season. Considering the significant variety in the composition of ambient PM, it is very likely that some particular components/sources of PM may be more toxic and detrimental to human health. Therefore, equally treating all components/sources of PM and the use of overall mass concentration as the only parameter in the regulatory purposes may lead to ineffective control strategies to protect human health (Thurston et al., 2005). For example, the current ambient PM standards is clearly less protective to human health in some areas of the country where the contribution of toxic sources and/or compounds are higher than other areas, while they might all attain the ambient PM mass concentration standards. As a result, if the 8 PM toxicity and the associated health outcomes could be determined for each source type, more efficient and cost-effective regulations and standards could be adopted, particularly on the sources which are more toxic to human health. Even under current regulations, nonattainment areas (i.e., areas of the country where air pollution levels persistently exceed the NAAQS levels) can focus on the most toxic components/sources of the ambient PM —if they can be identified— to meet the federal and state health-based standards, and to comply with Clean Air Act requirements. Recently, there has been a growing interest in the use of source apportionment data in epidemiological health studies with the goal of identifying the sources that are more harmful to human health relative to others (Bell et al., 2014, 2010; Dadvand et al., 2014; Gass et al., 2015; Laden et al., 2000; Mar et al., 2010; Ostro et al., 2011; Pun et al., 2014; Sarnat et al., 2008; Siponen et al., 2015). These studies have provided very valuable evidence that exposure to PM from certain sources is linked to morbidity and mortality. For example, in Barcelona, Ostro et al. (2011) found that exposure to traffic emission, sulfate, and construction dust is statistically significantly associated with all-cause and cardiovascular mortality. In Boston, Laden et al. (2000) found that exposure to PM emitted from coal combustion is associated with acute risk of mortality, while in Phoenix, Mar et al. (2000) found that mortality is associated with exposure to regional sulfate and vehicular emissions, to a lesser extent. In a more recent study in Kotka, Finland, Siponen et al. (2015) found PM2.5 emitted from biomass combustion and traffic sources, are promoters of systemic inflammation, a risk factor for cardiovascular diseases. As stated above, there are different approaches for the source apportionment of ambient PM. In 2003, the U.S. EPA sponsored a set of studies to evaluate the variability in the source apportionment results from various approaches, using the same ambient concentration data set, and then to assess the influence of this variability on the association between the estimated source contributions and daily mortality (Thurston et al., 2005). All source apportionment approaches generally identified the same set of source types, with similar composition, and the variation in their results led to only some 15% uncertainty in the mortality regression. The findings from these studies demonstrated that source apportionment results can be reliably used in epidemiological studies to assess the impact of various sources of PM on human health. 9 1.4. Objectives Findings from epidemiological and toxicological studies implicate the absolute need for identification and quantification of major sources of ambient PM. In addition, the knowledge regarding the relative toxicity of the constituents and sources of ambient particles has been identified as a crucial research gap by the National Research Council. Better understanding of the characteristics of PM sources and their contributions, along with their associations with various health effects, allows us to develop more effective control strategies to reduce the population exposure to harmful sources of airborne PM. In particular, the apportionment of sources of ambient PM in non-attainment areas, where the standard thresholds are exceeded (e.g., South Coast air basin), is of utmost importance. The main objectives of this research dissertation are therefore to:  Identify major primary and secondary sources of size-fractionated airborne PM (in indoor and outdoor environments) in different locations of the state of California, with distinct sources of PM, geography, and meteorology.  Quantify the contributions from each source to PM mass concentration, using different receptor modeling, depending on the nature of the study and the available information on ambient concentrations and composition of the sources.  Investigate the spatial and temporal trends in the estimated contributions from identified sources.  Evaluate the relationship between indoor and outdoor components of PM and investigate the degree to which indoor and outdoor sources contribute to indoor levels of PM.  Investigate the efficacy of major regulations on emissions sources (particularity on vehicular sources) on ambient air quality in the South Coast Air basin, by evaluating the contributions from the source of interest (particularly vehicular emissions) over a long period of time (e.g., over a decade)  Investigate the associations between the estimated source contributions and a variety of health outcomes, in collaboration with environmental epidemiology research groups. 10 These objectives were accomplished through a series of studies conducted in different locations of the state of California. These locations covered a wide range of geographical areas with distinctly different characteristics, topology, meteorology, and sources of PM. A particular focus of this dissertation was on the characterization and source apportionment of size-fractionated ambient PM in the Los Angeles Basin. Due to its unique geography and major sources of pollution that exist in this region, this part of the state suffers from severe air pollution throughout the year. Considering its large population (over 18 million), air pollution remains to be a major concern in this region. In all of the studies presented in this dissertation, time-integrated PM samples were collected on filter substrates and chemically analyzed to quantify different components of airborne PM. Their spatial and temporal variability were then investigated. Depending on the number of samples and availability of prior information on existing sources, either CMB or PMF model was used for source apportionment analysis to identify and quantify primary and secondary sources that contribute to PM mass concentration in the study areas. The findings of this research will help federal and state regulatory agencies understand the linkage among sources, composition and associated health effects of size-fractionated PM. Information on PM emissions from a variety of sources will provide a strong scientific basis to develop cost-effective abatement strategies to protect the public health from toxic sources of PM. The results will also help determine if there is a scientific foundation for controlling PM from only a subset of PM sources. 1.5. Dissertation overview This dissertation summarizes my doctoral research under the supervision of Prof. Constantinos Sioutas with an ultimate goal of characterization and apportionment of the airborne PM in urban areas of the state of California. This dissertation comprises of 6 chapters: Chapter 1 provides an overview of urban air pollution with a particular focus on ambient PM. The characteristics and associated health effects of ambient PM are discussed. In addition, some general information on the source apportionment techniques is provided. Lastly, the rational and objectives of this dissertations are discussed. 11 Chapter 2 presents the results of a comprehensive study, in which ambient quasi-ultrafine particles (q-UFPs) were collected for a year at 10 different locations around the LA Basin. Spatial and temporal variability of carbonaceous species and organics compounds of q-UPFs are discussed. In this study, source apportionment analysis was conducted using CMB with detailed organic molecular markers (i.e., MM-CMB) along with the PCA on seasonal samples at each site to investigate the spatial and temporal variability in the sources of ambient UFPs across the megacity of Los Angeles. A UFP source apportionment study of this scale, detail, and sophistication had never before been attempted in this region. Given that ambient UPFs currently are not regulated, the results of this study provided invaluable insights on their major sources and therefore can help the progressive regulations on PM emissions in the Los Angeles metropolitan area. Chapter 3 discusses the results from an extensive sampling campaign in indoor and outdoor environments of retirement communities in the LA Basin. In this study PM2.5 were collected simultaneously indoor and outdoor of the retirement communities and the chemical compounds were quantified through a series of chemical analyses. In addition to presenting the indoor and outdoor levels of different classes of chemical compounds (i.e., carbonaceous species, inorganic ions, organic compounds, melts and elements), an MM-CMB model was applied on organic tracers—similar to the study presented in Chapter 2— and the sources that contribute to indoor and outdoor PM were quantified. The main objective of this study was to assess the degree to which indoor and outdoor sources contribute to indoor levels of PM. Chapter 4 focuses on the long-term trends of different sources that contribute to ambient PM2.5 in the Los Angeles Basin. Over the past decade, major federal, state, and local regulation were implemented to cut the emissions from vehicular sources. To evaluate the effect of these regulations on ambient air quality in the LA Basin, source apportionment analyses were conducted on PM2.5 chemical speciation data, which were acquired from the Air Quality System (AQS) for two Speciation Trends Network (STN) sampling sites in the LA area (i.e., central Los Angeles and Rubidoux) over a prolonged time period, from 2002 to 2013. In this study, a PMF receptor model was used for the source apportionment analysis. The results of this study showed a progressive improvement in the air quality of the LA Basin and a significant reduction of vehicular emissions over the past decade in this region, highlighting the efficacy of stringent regulations on vehicular sources. 12 Chapter 5 follows the same approach discussed in chapter 4 to identify and quantify major sources of ambient PM2.5 in 8 major cities of the state of California (i.e., El Cajon, Riverside, Los Angeles, Simi Valley, Bakersfield, Fresno, San Jose, and Sacramento) between 2002 and 2007. In this chapter, spatial and temporal variability of the contributions of the identified sources are thoroughly discussed. The source apportionment results from this study were also used in some epidemiological studies to investigate the associations of source-specific fine particles with a variety of health outcomes (i.e., emergency room visits, mortality, birth outcomes, etc.) Chapter 6 summarizes the major findings of this dissertation and outlines the implications of these results for future research, air quality regulations, and health effects studies. 13 CHAPTER 2 Source Apportionment and Organic Compound Characterization of Ambient Ultrafine Particulate Matter in the Los Angeles Basin This chapter is based on the following publication: Hasheminassab, S., Daher, N., Schauer, J. J., & Sioutas, C. (2013). Source apportionment and organic compound characterization of ambient ultrafine particulate matter (PM) in the Los Angeles Basin. Atmospheric Environment, 79, 529-539. 14 2.1. Introduction Persistent association between exposure to ambient PM and adverse health effects has been shown by numerous epidemiological and toxicological studies (Atkinson et al., 2001; Delfino et al., 2005; Dominici et al., 2006; Pope and Dockery, 2006). These associations were initially investigated for total suspended particulates (TSP) (Ware et al., 1986), but the majority of efforts in the past few decades extensively focused on the smaller particle size spectrum such as PM10 and PM2.5 (particles with an aerodynamic diameter smaller than 10 and 2.5 µm, respectively) (Brunekreef and Forsberg, 2005; Sørensen et al., 2003). Accordingly, current US federal ambient air quality standards are limited PM10 to PM2.5. However, population exposure to ultrafine particles (UFP, tenuously defined as particles with an aerodynamic diameter smaller than 0.1-0.2 µm (Sioutas et al., 2005)) has recently received considerable attention as the findings from recent studies have demonstrated that ultrafine particles have more mass-based toxicity potency compared to the larger particles (Cho et al., 2005; Delfino et al., 2005; L. Ntziachristos et al., 2007b). Compared to coarse and fine particles, UFP have relatively low mass concentration, but they account for the majority of ambient PM number concentration (Cho et al., 2005). Given their increased number concentration, therefore a large surface area, UFP can carry a considerable amount of toxic pollutants per unit mass (Sioutas et al., 2005). UFP are mainly emitted from fossil fuel combustion, particularly vehicular emissions(Shi et al., 1999; Westerdahl et al., 2005). They can also be formed as secondary aerosols through photochemical reactions, especially during warmer seasons and at downwind/receptor locations (Satya B. Sardar et al., 2005; Verma et al., 2009). Several studies (Hu et al., 2008a; Li et al., 2003) have suggested that the UFP toxicity is mainly driven by the organic constituent of the particles such as organic carbon (OC), which comprises the largest fraction of PM 0.25 (Daher et al., 2013), and polycyclic aromatic hydrocarbons (PAHs). Sioutas et al. (2005) have demonstrated that although coarse and accumulation modes are separated by a clear cut-point of 2.5 µm, the cut-point separating UFP and accumulation modes may vary from 0.1 to 0.2 µm, depending on season and location. In this study, we investigated the characteristics of quasi-ultrafine particles (quasi-UFP), defined as particles with aerodynamic diameters below 0.25 µm (PM0.25). According to previous studies by Hu et al. (2008) and Delfino et al. (2009), quasi-ultrafine mode particles are far more toxic than larger coarse and fine particles. 15 Organic compound characterization and source apportionment of PM0.25 have been previously investigated by in specific locations of the Los Angeles Basin (LAB) (Arhami et al., 2010; Minguillón et al., 2008). However, the main goal of this study was to investigate the seasonal and spatial variation of organic compounds in this critical size range (dp<0.25µm) and more importantly to provide a comprehensive assessment of PM0.25 source contributions in distinctly different locations of the LAB, each affected by a variety of air pollution sources. Towards that goal, we conducted a year-long sampling at 10 sites located in different regions of the LAB representing urban source, near-freeway, semi-rural receptor and desert-like areas. 2.2. Methodology 2.2.1. Sampling Sites Description 10 sampling sites, across the LAB, were selected to determine sources of quasi-UFP. The sites consisted of urban source, near-freeway, semi-rural receptor and desert-like areas. Detailed description of the sampling locations was provided in another publication (Pakbin et al., 2010). In brief, sampling sites can be grouped according to their geographical locations into Long Beach (HUD), western LA (GRD, LDS), central LA (CCL, USC), eastern LA (HMS, FRE), Riverside County (VBR, GRA) and Lancaster (LAN); in order of their increasing distance from the coast. Their locations are shown in Figure 2.1. Figure 2.1. Location of the 10 sampling sites. 16 The site in Long Beach, HUD, is considered as a “source” site in the LAB because of its location. It is situated in a mixed residential and industrial area about 2 km inland of the ports of Los Angeles and Long Beach. HUD is also located immediately to the east of Terminal Island Freeway (ca. 100 m) and 1.2 km west of I-710, which have a high volume of heavy duty diesel vehicles (HDDV) in port service (Moore et al., 2009). On the other hand, the Riverside sites (VBR and GRA) represent pollutant “receptor” areas in the basin due to their location on the prevalent air trajectory crossing the LAB from coast to inland. These sites, which are in a predominantly rural area about 80 km inland from downtown LA, are thus significantly impacted by advected and photo-chemically processed PM originating from upwind areas (Pakbin et al., 2010). Further inland, LAN represents a typical desert-like site located to the north of the LAB and over 2 km away from the nearest freeway (CA-14). The remaining sites are located in central and western LA areas. CCL is close to surface streets with significant vehicular traffic, while central USC and eastern LA (HMS, FRE) sites are within 50–800 m of nearby freeways. The western sites (GRD and LDS) represent coastal areas, with LDS located immediately to the southwest of I-405 freeway. 2.2.2. Sampling Method and Schedule Twenty-four-hour time-integrated samples were concurrently collected at the 10 sampling sites on a weekly basis during weekdays from April 2008 to March 2009. Two parallel Sioutas Personal Cascade Impactor Samplers (Sioutas TM PCIS, SKC Inc., Eighty Four, PA, USA) (Misra et al., 2002), operating at 9 lpm, were deployed at each site to collect particles in these size ranges: <0.25 µm (quasi-UFP), 0.25–2.5 µm, and 2.5–10 µm. In the current study, we only investigated the quasi-UFP fraction. The chemical speciation of coarse PM was discussed in detail in other publications(Cheung et al., 2012, 2011, Pakbin et al., 2011, 2010). For the purpose of chemical speciation, one PCIS was loaded with quartz microfiber filters (Whatman International Ltd, Maidstone, England), while the other one with Teflon filters (Pall Life Sciences, Ann Arbor, MI). Quasi-UFPs were collected on 37 mm filters. 2.2.3. Meteorology Table 2.1 presents the seasonal average of select meteorological parameters at different site clusters. Seasonally, as expected at all sampling clusters highest temperature was observed in 17 summer (19.4-27.9 o C) while minimum in winter (8.7-13 o C). Wind speed was greatest at all sampling locations in spring and summer, predominantly blowing from coast to inland. On the other hand, in winter, wind speed was lower at all locations, approaching quasi-stagnant conditions at some sites, and was mainly toward north direction except at LAN. Table 2.1. Select meteorological parameters at sites clusters during spring (March-May), summer (June-August), fall (September-November) and winter (December-February). Site Clusters Season Temperature (°C) Average ± standard deviation Wind Speed (m/s) Prevailing direction Long Beach Spring 16.3±5 1.6 SW Summer 21.3±3.8 1.9 SW Fall 19.9±4.9 1.2 W Winter 13.8±5 0.7 NW Western LA Spring 15±4.4 1.7 W Summer 19.4±2.7 0.7 W Fall 19±4.2 0.2 NE Winter 13.9±4.8 0.2 NW Central and Eastern LA Spring 17.2±5.8 2.1 SW Summer 22.8±4.8 3.6 SW Fall 20.9±5.7 1.6 SW Winter 14.1±5.6 1.5 NE Riverside Spring 17.4±6.7 2.6 NW Summer 25.6±6.4 3.6 W Fall 22±7.1 1.6 NW Winter 13.8±6.1 1.3 N Lancaster Spring 15.2±6.6 4.1 W Summer 27.9±5.4 4.3 W Fall 18.8±7.3 1.5 W Winter 8.7±4.7 1.5 W 2.2.4. Chemical Analysis To conduct chemical analyses of the quasi-UFP samples, Teflon and quartz filters were sectioned into portions. Water soluble organic carbon (WSOC) was quantified from monthly composites using a Sievers 900 Total Organic Carbon Analyzer (Stone et al., 2009). Organic speciation was conducted on seasonal quartz filter composites using gas chromatography mass spectrometry (GC-6980, quadrupole MS-5973, Agilent Technologies). In this study seasons were defined as spring (March–May), summer (June–August), fall (September–November) and winter (December–February). Filter composites were spiked with isotopically-labeled standard solutions 18 prior to extraction. Samples were then extracted in a 50/50 dichloromethane and acetone solution using Soxhlets, followed by rotary evaporation and reduction in volume under high-purity nitrogen. Further details can be found in Stone et al. (2008). Additional information about the gravimetric analysis as well as quantification of EC, OC, and total metals along with their seasonal and spatial variations have been already reported by Daher et al. (2013). 2.2.5. Source Apportionment A molecular marker-based chemical mass balance (MM-CMB) model was used to calculate the contributions from source profiles and atmospheric measurements of molecular markers. The model was mathematically solved with the U.S. Environmental Protection Agency CMB (EPA-CMB8.2) software using an effective variance weighted least squares algorithm to apportion the receptor data to the source profiles (Watson et al., 1984). Molecular marker compounds that are chemically stable during transport from source to receptor and that were detected in the quasi-UFP samples were selected as fitting species (Schauer et al., 1996). These included EC, nonacosane, hentriacontane, tritriacontane, levoglucosan, 17α(H)-22,29,30- trisnorhopane, 17α(H)-21β(H)-hopane, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(e)pyrene, indeno(1,2,3-cd)pyrene, benzo(ghi)perylene, vanadium, and nickel. The model input source profiles were based on the observed primary tracers. These profiles included diesel motor vehicles (Kam et al., 2012; Liacos et al., 2012), gasoline motor vehicles (Kam et al., 2012; Liacos et al., 2012), wood smoke (Fine et al., 2004; Sheesley et al., 2007), vegetative detritus (Rogge et al., 1993a), natural gas combustion (Rogge et al., 1993b), and ship emissions (Agrawal et al., 2008; Rogge et al., 1997). Source profiles for wood smoke, vegetative detritus, natural gas combustion and ship emissions were based on PM2.5 measurements, but were assumed to be the same for PM0.25, similarly to the approach applied by Minguillón et al. (2008). Mobile source profiles were derived from on-road measurements of quasi-UFP conducted at CA-110 and I-710 freeways in Los Angeles. Kam et al. (2012) reported an HDV composition of 3.9% and 11.3% for I-110 and I-710, respectively. These profiles, thus, suitably represent vehicular emissions in the LAB. However, inclusion of both vehicular profiles led to co-linearity problems in some CMB runs. For these samples, the model was therefore re-run using only the gasoline vehicles source profile for mobile emissions. A sensitivity analysis revealed that the mobile source contribution was not significantly different when the gasoline vehicles source profile is solely included in the 19 model compared to including both of the gasoline and diesel vehicles source profiles. As a result, mobile source contribution was estimated as the contribution of gasoline vehicles for samples exhibiting co-linearity. On the other hand, mobile source contribution was determined as the sum of diesel and gasoline vehicles source contributions for runs that included both of these profiles. In anthropogenic-influenced regions, water soluble organic carbon (WSOC) is mainly emitted from biomass burning sources and/or formed through photochemical reactions (i.e., SOA) (Weber et al., 2007). Since the tracers of secondary sources were not included in the CMB model, non-biomass burning water soluble organic carbon (WSOCnb) was calculated and used as a metric to estimate the contribution of secondary sources to PM0.25 (Snyder et al., 2009). WSOCnb was then defined as the difference between total measured WSOC and water soluble portion of OC attributed to biomass burning (WSOCbb) (Snyder et al., 2009). Further details of WSOCbb calculation will be presented later. 2.3. Results and Discussion 2.3.1. Carbonaceous Species EC was a minor constituent of PM0.25 with an annual average contribution of 4.8±1.8% to total mass. All 10 sites exhibited highest EC concentration during fall, when primary emissions are predominant (Sardar et al., 2005), while lowest during the warmer spring and summer. Unlike EC, OC constituted a large fraction of total quasi-UFP mass (30.9±6.5% on an annual average basis). Contributions of EC and OC to PM0.25 during winter at HMS and FRE were comparable to data from a previous study conducted in Pasadena (Cass et al., 2000). To better investigate the contribution of primary and secondary OC, it was further separated into water insoluble organic carbon (WIOC), biomass burning water soluble organic carbon (WSOC bb), and non-biomass burning organic carbon (WSOCnb). It has been determined by Sannigrahi et al. (2006) that about 71% of OC emitted from biomass burning is water soluble. Therefore, for each site, measured concentration of levoglucosan, an indicator of wood combustion (Simoneit et al., 1999), was divided by the levoglucosan-to-OC ratio (0.135) obtained from the wood smoke source profile (Fine et al., 2004; Sheesley et al., 2007), to estimate the OC emitted from biomass burning. Then, WSOCbb was determined as 71% of calculated OC from wood smoke. WSOCnb was hereafter determined by subtraction of WSOCbb from measured WSOC. WSOCnb was used as a metric to estimate the relative contribution of SOA to total mass (Snyder et al., 2009) prior to its estimation 20 using source apportionment, which will be discussed in the following sections. Figure 2.2 a-d shows mean seasonal concentration of EC, WIOC, WSOC bb, and WSOCnb. Overall, OC was mostly water insoluble during winter (56±5%), while it was dominated by WSOC in summer (58±11%). In winter, when mixing layer height is lowest and meteorological conditions are more stable, highest concentrations of WIOC were found at HUD and urban LA sites (1.4-3.3 µg m -3 ), while lowest levels were measured at downwind receptor locations (1.1-1.3 µg m -3 ). Conversely, during the summer season, WIOC displayed the highest concentrations at the receptor and inland locations, namely GRA, LAN and VBR (1.6-1.9 µg m -3 ). Given the lower abundance of primary emissions at the inland semi-rural locations, this trend is mainly attributed to the summertime atmospheric transport of air parcel from coast to inland, when westerly/southwesterly winds were predominant. On the other hand, highest concentrations of WSOCnb were found at the receptor sites in summer (1.6-2.0 µg m -3 ), when temperatures were highest and transport of gaseous precursors and aged PM from upwind source sites was accentuated due to stronger southwesterly wind. WSOCbb showed a significant seasonal variation across the basin with the highest concentration occurring in winter (0.39±0.11 µg m -3 ) and lowest in summer (0.03±0.01 µg m -3 ). This trend demonstrates the predominance of biomass burning, mainly wood combustion, during the cold seasons. As mentioned above, a levoglucosan-to-OC ratio of 0.135 was used to calculate WSOCbb and WSOCnb. This ratio does not vary significantly among different residential wood burning source profiles (Sheesley et al., 2007). Fine et al. (2002) also reported a range of 0.113- 0.193 for PM2.5 levoglucosan-to-OC ratio from residential wood combustion at 10 different states around the US. Applying these ratios to our data indicates that the annual average percent contribution of WSOCbb and WSOCnb to total OC varies in the ranges of 3-6% and 44-47%, respectively. 2.3.2. Seasonal and Spatial Variability in Organic Compounds 2.3.2.1. Polycyclic Aromatic Hydrocarbons (PAHs) Ambient particle-phase PAHs are mainly produced from incomplete combustion of fossil fuels (Manchester-Neesvig et al., 2003). Their concentration and composition, however, may vary significantly by several factors such as atmospheric conditions, source strength, gas-particle partitioning, and deposition processes (Polidori et al., 2007) 21 Spring HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( g/m 3 ) 0 1 2 3 4 5 6 7 EC WIOC WSOCbb WSOCnb Summer HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( g/m 3 ) 0 1 2 3 4 5 6 7 Fall HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( g/m 3 ) 0 1 2 3 4 5 6 7 Winter HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( g/m 3 ) 0 1 2 3 4 5 6 7 Figure 2.2 a-d. Seasonal variation in the concentrations (µg m -3 ) of elemental carbon (EC), non-biomass burning water soluble organic carbon (WSOCnb), biomass burning water soluble organic carbon (WSOCbb), and water insoluble organic carbon (WIOC) in quasi- ultrafine particles (dp<0.25 µm) by site. Even though the sum of all measured PAHs displayed a very low contribution to total quasi-UFP mass (0.02±0.01% on a yearly average basis), their carcinogenicity as well as an array of severe effects on human health have been reported by several studies (Boström et al., 2002; Lin et al., 2008). Seasonal and spatial variations of select PAHs, including benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(e)pyrene, indeno(1,2,3-cd)pyrene, and benzo(ghi)perylene, are shown in Figure 2.3 a-d. These species were used as fitting species in the CMB model. Generally, cumulative concentration of select PAHs displayed the highest levels in cooler seasons, while lowest or below detection limit during warmer months. Elevated winter- and fall-time PAHs concentrations, particularly at source and urban sites, are most likely due to the higher level of 22 fresh emissions from primary sources (such as wood smoke and vehicular emissions) owing to stagnant atmospheric conditions which trap the pollutants in the lower altitude. Spring HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.5 1.0 1.5 2.0 2.5 Benzo(b)fluoranthene Benzo(k)fluoranthene Benzo(e)pyrene Indeno(1,2,3-cd)pyrene Benzo(ghi)perylene Summer HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.5 1.0 1.5 2.0 2.5 Fall HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.5 1.0 1.5 2.0 2.5 Winter HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.5 1.0 1.5 2.0 2.5 Figure 2.3 a-d. Seasonal variation in the concentration (ng m -3 ) of select Polycyclic Aromatic Hydrocarbons (PAHs) in quasi-ultrafine particles (dp< 0.25 µm) by site. Moreover, cold-ignition of gasoline-powered vehicles during cool seasons may lead to an increase in the level of high molecular weight PAHs such as benzo(ghi)perylene and indeno(1,2,3- cd) pyrene (Arhami et al., 2010). Other possible sources of PAHs were further investigated by application of a principal component analysis (PCA) to select organic species, EC, nickel and vanadium. Data from all sites was combined in this analysis and a Varimax rotation was employed. According to Table 2.2 a, levoglucosan, a molecular marker of biomass burning, was clustered with some PAHs (indeno(1,2,3-cd)pyrene and benzo(ghi)perylene) in one component and presented a high loading factor (0.86), suggesting the contribution of wood smoke to PAHs, particularly during winter. 23 Table 2.2 a-b. Principal component loadings (VARIMAX normalized) of select organic species, EC, vanadium, and nickel in quasi-UFP (d p <0.25 µm) in (a) all sites and seasons pooled together and (b) all sites and seasons except spring- and summer-time data for HMS and FRE. a) Rotated Component Matrix 1 2 3 4 Hopane 0.91 0.12 0.12 0.13 ABB-20R-C29-Sitostane 0.90 0.11 0.12 0.16 ABB-20S-C29-Sitostane 0.90 0.07 0.06 0.17 22S-Homohopane 0.84 0.18 0.28 0.29 22R-Homohopane 0.78 0.16 0.12 0.36 ABB-20R-C27-Cholestane 0.76 0.26 0.04 -0.07 17 α(H)-22,29,30- trisnorhopane 0.74 0.01 0.10 -0.01 Benzo(k)fluoranthene 0.18 0.95 0.07 0.11 Benzo(b)fluoranthene 0.11 0.95 0.21 0.09 Benzo(e)pyene 0.21 0.87 0.34 0.20 Levoglucosan -0.09 0.16 0.86 -0.17 Indeno(1,2,3-cd)pyrene 0.33 0.25 0.86 0.12 Benzo(ghi)perylene 0.40 0.18 0.83 0.14 Ni 0.10 0.03 0.08 0.93 V 0.08 0.14 -0.18 0.87 EC 0.34 0.20 0.16 0.67 b) Rotated Component Matrix 1 2 3 Hopane 0.90 0.11 0.17 ABB-20R-C29-Sitostane 0.89 0.11 0.18 ABB-20S-C29-Sitostane 0.87 0.05 0.22 22S-Homohopane 0.86 0.28 0.28 22R-Homohopane 0.76 0.17 0.41 17 α(H)-22,29,30- trisnorhopane 0.75 0.01 -0.02 ABB-20R-C27-Cholestane 0.74 0.14 -0.02 Benzo(b)fluoranthene 0.01 0.92 0.22 Benzo(e)pyene 0.15 0.91 0.27 Benzo(k)fluoranthene 0.00 0.86 0.31 Indeno(1,2,3-cd)pyrene 0.53 0.72 -0.10 Levoglucosan 0.13 0.69 -0.41 Benzo(ghi)perylene 0.61 0.65 -0.08 V 0.09 0.04 0.89 Ni 0.21 0.12 0.84 EC 0.46 0.25 0.60 24 In summer, low level of PAHs could be attributed to the combined effects of the higher degree of photo-degradation of these species with oxidizing gases (ozone, nitrogen oxides, hydrogen peroxide, etc.) (Arey et al., 1988; Grosjean et al., 1983), as well as enhanced partitioning to the gas phase (Saarnio et al., 2008), due to the higher ambient temperatures and increased atmospheric dilution in that time of the year. It can be also deduced from Figure 2.3 a-d that PAHs level was generally lowest at remote LAN site. On the other hand, levels were overall highest at urban or near-freeway sites. In winter, the peak was observed at CCL site, located in south central Los Angeles close to surface streets with significant vehicular emissions (Pakbin et al., 2010), followed by near-freeway HMS site. Noticeably high PAHs concentrations were also observed in spring and summer at near-freeway HMS and FRE sites. One possible reason could be the advection of pollutants from upwind source sites to some extent, when stronger southwesterly winds in spring and summer prevail. Also, removing spring- and summer-time data of HMS and FRE from the PCA analysis resulted in reduction of one of the principle components and clustering levoglucosan with all other select PAHs into one component (Table 2.2 b). This observation suggests that there were probably some other emission sources nearby HMS and FRE during spring and summer affecting the PAHs level at these sites. 2.3.2.2. Hopanes and Steranes Hopanes and steranes, organic tracers of vehicular emissions, are mainly emitted from lubricating oil of gasoline- and diesel- fuelled vehicles in the LAB, given the lack of large scale combustion of coal and fuel oil (Schauer et al., 1996). The seasonal average concentration of all measured hopanes and steranes varied from 0.56±0.41 ng m -3 in summer to 0.91±0.57 ng m -3 in winter. While laboratory studies have suggested that hopanes and steranes may oxidize in the atmosphere when photochemical activity peaks (Roy et al., 2011), a number of source apportionment studies in LA have shown that these compounds are reasonably stable during transport from source to receptor and therefore are reliable tracers of mobile source emissions in this area (Heo et al., 2013). Figure 2.4 a-d displays the seasonal and spatial variation of select hopanes and steranes. 25 Spring HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 17A(H)-22,29,30-Trisnorhopane 17A(H)-21B(H)-Hopane 22S-Homohopane 22R-Homohopane ABB-20R-C27-Cholestane ABB-20R-C29-Sitostane ABB-20S-C29-Sitostane Summer HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fall HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Winter HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Figure 2.4 a-d. Seasonal variation in the concentration (ng m -3 ) of select hopanes and steranes in quasi-ultrafine particles (dp< 0.25 µm) by site. In winter, the peak in cumulative concentration of select hopanes and steranes was observed at source HUD site (1.2 ng m -3 ). Among LA urban locations, the coastal GRD and LDS sites displayed the lowest winter-time concentration, while the highest concentrations occurred at near-freeways CCL, USC, HMS and FRE sites. Moving inland, significantly lower concentrations were observed at receptor VBR, GRA and desert-like LAN sites during winter. Similarly to EC, another major marker of vehicular emissions, high concentrations were observed in spring/summer at near-freeway HMS, FRE, and receptors GRA and LAN sites, which could be due to increased source strength or also advection from upwind source sites, to some extent. 26 2.3.2.3. n-Alkanes Spring HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 10 20 30 40 50 CPI 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Tetracosane Pentacosane Hexacosane Heptacosane Octacosane Nonacosane Triacontane Hentriacontane Dotriacontane Tritriacontane CPI(C24-C33) Summer HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 10 20 30 40 50 CPI 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fall HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 10 20 30 40 50 CPI 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Winter HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 10 20 30 40 50 CPI 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Figure 2.5 a-d. Seasonal variation of n-Alkanes (C24-C33) concentration (ng m -3 ) in quasi- ultrafine particles (dp<0.25 µm) along with their carbon preference index (CPI). Total identified n-alkanes accounted for 0.27±0.06% of PM0.25 based on an annual average over all sites. Species with carbon number between 24 to 33 were the most abundant components with contribution of about 81% to total n-alkanes. In order to distinguish the biogenic- and anthropogenic-derived n-alkanes, carbon preference index (CPI) was calculated at each site in different seasons. CPI is defined as the sum of concentration of odd-carbon alkanes divided by that of even-carbon alkanes (Simoneit, 1986). CPI around one indicates the predominance of emissions from anthropogenic sources (e.g., fossil fuel combustion, cigarette and wood smoke), whereas 27 biogenic sources exhibit a CPI greater than 2 (Daher et al., 2011). CPI values for each site in different seasons are shown in Figure 2.5 a-d. As can be seen, CPI values spanned around unity, indicating the predominance of anthropogenic sources in the LAB. Figure 2.5 a-d illustrates that at inland sampling sites, cumulative concentrations of select n-alkanes (C24-C33) were higher in summer, probably because of long-range atmospheric transport of pollutants from upwind anthropogenic-influenced locations. On the other hand, a reverse trend was observed in winter and higher concentrations occurred at urban HUD and LA sites. This increase was pronounced at the HUD site, where winter-time concentration of cumulative select n-alkanes was about two times its corresponding concentration in summer. 2.3.2.4. Levoglucosan Spring HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 20 40 60 80 100 120 Summer HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 20 40 60 80 100 120 Fall HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 20 40 60 80 100 120 Winter HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Concentration( ng/m 3 ) 0 20 40 60 80 100 120 Figure 2.6 a-d. Seasonal variation in the concentration (ng m -3 ) of levoglucosan in quasi- ultrafine particles (dp< 0.25 µm) by site. Figure 2.6 a-d shows the seasonal and spatial variation of levoglucosan, a well-established molecular marker of biomass burning (Fraser and Lakshmanan, 2000), in quasi-UFP. 28 Levoglucosan showed a distinct seasonality, with a minimum seasonal average concentration over all sites in summer (4.9±1.7 ng m -3 ) to a maximum in winter (73.2±21.3 ng m -3 ). This trend clearly reveals the higher wood burning activities, particularly for domestic heating purposes, in cooler seasons. 2.3.3. Source Contributions 2.3.3.1. Source Apportionment of PM0.25 OC Seasonal contributions of primary sources to PM0.25 OC are presented in Figure 2.7 a-d. Source contribution estimates (SCEs) below detection limit (less than two times of their standard errors) were not included in the plots as they were not statistically significant. Spring HUD GRD LDS CCL USC HMS FRE VBR GRA LAN OC concentration( g/m 3 ) 0 1 2 3 4 5 6 Mobile Sources Vegetative detritus Natural gas combustion Wood smoke Ship emissions Other OC Summer HUD GRD LDS CCL USC HMS FRE VBR GRA LAN OC concentration( g/m 3 ) 0 1 2 3 4 5 6 Fall HUD GRD LDS CCL USC HMS FRE VBR GRA LAN OC concentration( g/m 3 ) 0 1 2 3 4 5 6 Winter HUD GRD LDS CCL USC HMS FRE VBR GRA LAN OC concentration( g/m 3 ) 0 1 2 3 4 5 6 Figure 2.7 a-d. Seasonal variation in source apportionment of organic carbon (OC) in quasi- ultrafine particles (dp<0.25 µm) by site. Mobile sources were the largest contributor to PM0.25 OC with a seasonal average SCE range of 1.1-2.3 µg m -3 accounting for 45.4-65.7% of measured OC. In winter, greatest 29 contributions from mobile sources occurred at HUD and USC (3.7 and 3.5 µg m -3 , respectively). Ship emissions, on the other hand, had a very low contribution to total OC (less than 1% on an annual average over all sites). However, spatially, highest contributions were observed at near- harbor HUD site (0.03-0.06 µg m -3 ), while significantly lower or non-detected at urban LA and inland sites. Contribution from ship emissions at LA and inland sites was lowest in winter, while highest in summer. Higher atmospheric stability in winter causes the accumulation of ship emissions in the source location (HUD), while in summer prevalent westerly onshore winds facilitate the advection of pollutants from coast to the inland valleys. Wood smoke showed a strong seasonal pattern, with seasonal average contribution to OC of 0.8% in summer to 16% in winter. This trend is mainly associated with higher biomass burning and/or wood combustion during cold seasons. Vegetative detritus accounted for 1.2% of PM0.25 OC on a yearly average basis over all sites. Natural gas combustion, with an annual average contribution of 0.54% to measured OC, was the lowest contributor to OC. Un-apportioned part of OC, denoted as “other OC”, was defined as the residual difference between the measured OC and the sum of all identified source contribution estimates. “Other OC” accounts for the contributions from unidentified primary sources (e.g., soil dust and food cooking) and also secondary organic carbon. Its contribution ranged from an average of 0.9 µg m -3 in winter to 1.3 µg m -3 in summer, accounting for 23-50% of total OC. “Other OC” levels were also greatest at inland and receptor sites (VBR, GRA, LAN) in summer (1.9-2.3 µg m - 3 ), most likely due to advection of aged and photo-chemically processed particles from upwind source regions (Sardar et al., 2005). “Other OC” was compared with seasonally averaged WSOCnb as shown in Figure 2.8 a-d. A strong correlation (R 2 =0.8) was found between “other OC” and WSOCnb in summer, whereas poor correlations were obtained in other seasons, suggesting that in summer un-apportioned OC, estimated from CMB results, mainly comprises SOC (Docherty et al., 2008; Miller-Schulze et al., 2011; Snyder et al., 2009). However, the regression slope below one in summer, as well as poor correlations between “other OC” and WSOC nb in other seasons indicate that “other OC” could be also from other primary sources of OC that were not included in the MM-CMB model, such as food cooking and soil dust, which were found to be major fine PM sources by conducting a Positive Matrix Factorization (PMF) analysis on PM2.5 in central LA and riverside (Heo et al., 2013). 30 Figure 2.8 a-d. Linear correlation between the concentrations of “other OC” and non- biomass burning water soluble organic carbon (WSOCnb) in (a) spring, (b) summer, (c) fall and (d) winter. Our findings at USC were compared to a recent PM2.5 OC apportionment conducted by Heo et al. (2013) at the same sampling location. Annual average contribution of mobile sources to PM0.25 OC was more than two times higher than PM2.5 (75% and 30%, respectively), while “other OC” showed a significant contribution to PM 2.5 OC compared to PM0.25 (56% and 21%, respectively). Wood smoke contribution to PM0.25 OC was not statistically significant in all seasons except for winter, when its contribution was about 8%, while Heo et al. (2013) reported an annual average contribution of 12% to PM2.5 OC at USC. y = 0.07x + 1.09 R² = 0.03 0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5 WSOC nb (µg/m3) Other OC (µg/m 3 ) Spring y = 0.48x + 0.75 R² = 0.82 0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5 WSOC nb (µg/m3) Other OC (µg/m 3 ) Summer y = 0.00x + 1.56 R² = 0.00 0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5 WSOC nb (µg/m3) Other OC (µg/m 3 ) Fall y = 0.33x + 0.80 R² = 0.52 0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5 WSOC nb (µg/m3) Other OC (µg/m 3 ) Winter 31 2.3.3.2. Source Apportionment of PM0.25 In order to assess the source contributions to the overall quasi-ultrafine mass, OC source apportionment results from the CMB model were converted to PM mass-based by application of OC/PM ratios obtained from each source profile. In addition to the contributions from identified sources calculated by CMB, secondary ions (sulfate, nitrate, and ammonium), and other organic matter (other OM) were considered as other contributors to PM0.25 mass as displayed in Figure 2.9 a-d. “Other OM” was calculated by multiplying “other OC” by an appropriate OM/OC ratio for each site and season. Turpin and Lim (2001) recommended the use of an average organic molecular weight per carbon weight ratio of 1.6±0.2 for urban areas and 2.1±0.2 for nonurban locations. Therefore, in spring and summer we chose a relatively high OM/OC ratio (2.1) for the receptor VBR, GRA and LAN sites, because these sites are highly affected by advected, aged and more oxygenated PM from upwind urban regions. In contrast, a relatively lower ratio of 1.8 (highest possible ratio for urban areas) was chosen for HUD and all LA sites. For fall and winter seasons, due to lower contribution of SOA, lower OM/OC ratios were selected. Thus, the ratio of 1.4 which is the lowest reasonable OM/OC ratio (Turpin and Lim, 2001) was selected for source and urban LA sites, and 1.8 was chosen for receptor VBR, GRA and LAN sites. 32 Spring HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Mass concentration( g/m 3 ) 0 2 4 6 8 10 12 14 16 18 Mobile Sources Vegetative detritus Natural gas combustion Wood smoke Ship emissions Sulfate Nitrate Ammonium Other OM Undetermined mass Summer HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Mass concentration( g/m 3 ) 0 2 4 6 8 10 12 14 16 18 Fall HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Mass concentration( g/m 3 ) 0 2 4 6 8 10 12 14 16 18 Winter HUD GRD LDS CCL USC HMS FRE VBR GRA LAN Mass concentration( g/m 3 ) 0 2 4 6 8 10 12 14 16 18 Figure 2.9 a-d. Seasonal variation in source apportionment of quasi-ultrafine particles (dp<0.25 µm) by site. Overall, apportioned mass varied from an average of 79±10% in summer to 96±17% in spring. Mobile sources were the largest contributor to PM0.25 with contribution ranging from 30.7±12.0% in summer to 56.9±16.6% in winter. ”Other OM” was the next major contributor to quasi-UFP, with an annual average contribution of 20±14%. “Other OM” exhibited a strong seasonality, with highest contributions to PM0.25 during warm spring/summer seasons, while lowest during fall/winter. “Other OM” is assumed to be mainly in the form of SOA (Zheng et al., 2002) and possible contributions from uncharacterized primary sources (e.g., soil dust and food cooking). Contribution from wood smoke to PM0.25 was ranging from 0.21±0.18% in summer to 5.6±3.2% in winter. Annual average contribution of ship emissions was largest at the near-harbor HUD, accounting for 2.1±0.4% of total measured PM0.25, with levels continually decreasing from coastal areas (1.5±0.6%) to inland LAN site (0.15±0.11%). Contribution from ship emissions to 33 PM0.25 at HUD was consistent with ranges previously reported by Minguillón et al. (2008) in the same study area. Vegetative detritus and natural gas combustion contributions, collectively, accounted for 1.3±0.8% of PM0.25 on a yearly-average basis over all sites. Among secondary ions, sulfate was the most abundant specie, accounting for 3.0±1.1% of total mass in winter and 11.4 ±3.1% in summer. Increased contribution of sulfate in summer is most likely due to enhanced photochemical activities (Khoder, 2002). Ammonium displayed similar trend as sulfate, with a minimum contribution to total mass in winter (2.5±1.1%) and maximum in summer (6.9±2.2%). Nitrate, on the other hand, had a reverse trend compared to ammonium and sulfate, with a maximum contribution to PM0.25 in winter (2.3±2.1%) and minimum in summer (1.1±0.8%). This is because the dissociation constant of ammonium nitrate is directly dependent on temperature. Higher temperatures in warm spring and summer seasons enhance the dissociation of particulate ammonium nitrate (Mozurkewich, 1993; Stelson and Seinfeld, 1982). Contribution of secondary ions to PM0.25 was consistent with that presented by Cass et al. (2000) in the LAB for particles in the size range of 0.056-0.1µm. Seasonal and spatial variations of secondary ions for this study were further investigated by Daher et al. (2013). Undetermined mass, with an annual average contribution of 16.3% to PM0.25, could be related to uncertainties in the OC to OM conversion ratios and source profiles composition. Moreover, using source profiles which were originally obtained for PM2.5 could lead to further discrepancies between the apportioned and measured mass. 2.4. Summary and Conclusions To better understand the organic composition and identify the possible sources of quasi- ultrafine particles, a sampling campaign was conducted, for an entire year, at 10 different locations of the LAB, including urban source, near-freeway, semi-rural receptor and desert-like areas. In summary, among all organic compounds, n-alkanes, which were found to be predominantly from anthropogenic sources (CPI~1), were the most abundant species in PM 0.25 with cumulative levels ranging from 9.34 to 48.08 ng m -3 over all sites and seasons. Seasonal averages of total PAHs, hopane and steranes, molecular makers of vehicular emissions, were highest in winter while lowest in summer. A Molecular marker-based chemical mass balance (MM-CMB) model was performed to estimate the relative contributions from the following primary sources: mobile sources (both gasoline and diesel vehicles), wood smoke, natural gas combustion, vegetative detritus, and ship emissions. MM-CMB results revealed that mobile sources were the major contributor to quasi- 34 UFP mass in all seasons with levels ranging from 31±12% in summer to 57±17% in winter. ”Other OC”, calculated as the difference between the measured OC and the sum of all primary source contribution estimates obtained from the MM-CMB, showed a significant correlation (R 2 =0.8) with WSOCnb in summer, suggesting that “other OC” is mainly dominated by SOC, especially during warm seasons. Other contributors to “other OC” could be the primary sources that were not included in the MM-CMB model such as soil dust and food cooking. “Other OM”, calculated by multiplying “other OC” by a suitable OM/OC ratio, was the next largest contributor to PM0.25 in all seasons (13-29%). Wood smoke, vegetative detritus, ship emissions, and natural gas combustion were minor contributors to PM0.25, with annual average contributions of 1.83%, 1.1%, 0.98%, and 0.21% to PM0.25, respectively. 2.5. Acknowledgments This research was supported by the South Coast Air Quality Management District (SCAQMD) (award #11527). We would like to acknowledge the staff at the Wisconsin State Laboratory of Hygiene for their assistance with the chemical analysis. We also wish to acknowledge the support of USC Provost's Ph.D. fellowship. 35 CHAPTER 3 Chemical characterization and source apportionment of indoor and outdoor fine particulate matter (PM 2.5) in retirement communities of the Los Angeles Basin This chapter is based on the following publication: Hasheminassab, S., Daher, N., Shafer, M. M., Schauer, J. J., Delfino, R. J., & Sioutas, C. (2014). Chemical characterization and source apportionment of indoor and outdoor fine particulate matter (PM 2.5) in retirement communities of the Los Angeles Basin. Science of the Total Environment, 490, 528-537. 36 3.1. Introduction Over the past decades, numerous epidemiological studies have reported consistent associations between exposure to particulate matter (PM) and a variety of adverse acute/chronic health effects including cardiovascular diseases (Brook et al., 2010), respiratory outcomes (Eisner et al., 2010; Holguin, 2008), and increased risk of adverse birth outcomes (Nieuwenhuijsen et al., 2013). The majority of outdoor air pollution studies largely relied on ambient air monitoring data from central sites located far from human subjects. Accordingly, air quality standards have been established for ambient environments, despite the fact that a large portion of human exposure to PM occurs indoors, where people spend most of their time (Jenkins et al., 1992; Klepeis et al., 2001). Considering the larger exposure time in different indoor micro-environments, the health effects of indoor air pollution of both indoor and outdoor origin is of considerable interest. Therefore, understanding the composition, behavior and sources of indoor PM and its relation to outdoor-generated PM are essential for personal exposure assessment. In an occupied residential building, PM is emitted from several primary sources (such as cooking, sweeping, and other human activities), but could also be formed through the reactions of gas-phase precursors emitted both indoors and outdoors (i.e., secondary sources). Indoor PM concentrations are further affected by outdoor-generated PM penetrating indoors through convective flows (e.g., open doors and windows) or diffusional flows/infiltration (e.g., cracks and fissures) (Thatcher and Layton, 1995). Penetration of particles through the building cracks strongly depends on their size (Liu and Nazaroff, 2003; Rim et al., 2010) . The results of a study conducted by Long et al. (2001) in 9 non-smoking homes of Boston, showed that ultrafine particles (UFP, particles with an aerodynamic diameter smaller than 0.1 µm) and coarse particles (PM2.5-10, particles with an aerodynamic diameter between 2.5 and 10 µm) have lower penetration efficiency compared to fine particles (PM2.5, particles with an aerodynamic diameter smaller than 2.5 µm). With the presence of indoor sources, indoor PM concentrations are often higher than their corresponding outdoor levels (Weschler and Shields, 1997). Their physical and chemical composition might also be significantly different (Lunden et al., 2003; Sarnat et al., 2006). Moreover, several epidemiological studies have demonstrated that exposure to indoor PM of outdoor origin is more deleterious compared with exposure to particles emitted indoors (Ebelt et al., 2005; Koenig et al., 2005) or exposure to the overall concentration of indoor PM (Delfino et al., 2008). Therefore, it is important to distinguish indoor from outdoor sources of PM in indoor 37 environments, as this information is vital for both health risk assessment and proper regulatory guidelines for PM. Only a few previous studies have attempted to estimate the contribution of specific sources to indoor PM using source apportionment techniques, including Positive Matrix Factorization (PMF)(Hopke, 2003; Larson et al., 2004; Minguillón et al., 2012; Ogulei et al., 2006), Chemical Mass Balance (CMB) model (Arhami et al., 2010; Kopperud et al., 2004), and Principal Component Analysis (PCA) (Koistinen et al., 2004). The present study was carried out at three retirement homes in the Los Angeles Basin (LAB), as part of the Cardiovascular Health and Air Pollution Study (CHAPS), a cohort panel study investigating the health effects of micro-environmental exposure to PM on elderly retirees with a history of coronary artery disease (Delfino et al., 2010, 2009). The objectives of the work presented in this paper are to: a) investigate the indoor/outdoor relationships of PM 2.5 chemical constituents, b) identify major sources of fine PM in both indoor and outdoor environments, and c) quantify the contribution of each source to PM2.5 mass using an MM-CMB model. 3.2. Methodology 3.2.1. Sampling sites and periods PM measurements were conducted at three retirement homes in the Los Angeles Basin (LAB), all located in the San Gabriel Valley, California. Site San Gabriel 1 (G1) was approximately 50 km east of Los Angeles, 3 km away from the nearest major freeway, located in a residential area. Site San Gabriel 2 (G2) was situated 8 km east of downtown Los Angeles, about 300 m south of a major freeway. Site San Gabriel 3 (G3) was located 55 km east of downtown Los Angeles, 2.5 and 0.15 km away from 2 busy freeways and a major street, respectively. At each site, two identical sampling stations were set up, with each being located either indoors or outdoors. At G1, the indoor station was located in a recreational area of the community’s main building. The indoor sampling station at G2 was set up in the dining room of the community’s central building, while at G3 the indoor station was close to a gym and an activity room, located in the recreational area of the main community complex. All monitored homes prohibited smoking in these indoor environments. The outdoor sampling stations at all sites were set up in movable trailers, positioned about 300 m away from the indoor stations (Polidori et al., 2007). 38 Concurrent indoor and outdoor PM sampling was conducted at each site, during two separate phases between 2005 and 2006: warm phase (P1), including summer and early fall, and cold phase (P2), including late fall and winter. In each phase, 6 weeks of sampling were conducted at each location. 3.2.2. Instrumentation and chemical analysis Size-segregated PM samples were collected daily (24-hour time-integrated) from Monday to Friday, using Sioutas Personal Cascade Impactor Samplers (Sioutas PCIS, SKC Inc., Eighty Four, PA, USA), operating at 9 lpm (Misra et al., 2002). Each PCIS was loaded with Zefluor filters (3 µm pore-size, Pall Life Sciences, Ann Arbor, Michigan, USA) and particles were collected in three size ranges, namely coarse (2.5 µm <dp<10 µm), accumulation (0.25 µm < dp < 2.5 µm), and quasi-ultrafine (dp<0.25 µm). The present study focuses only on fine PM (PM 2.5), for which data from the accumulation (PM0.25-2.5) and quasi-ultrafine (PM0.25) PM modes were combined to derive PM2.5 concentrations. The PM mass concentrations were determined by pre- and post- weighting the Zefluor filters using a microbalance (Mettler Toledo Inc., Columbus, OH, USA), after equilibration under controlled laboratory conditions (temperature of 22–24 o C and relative humidity of 40–50%). A detailed description of the chemical analysis conducted on the Zefluor filters has been previously presented by Arhami et al. (2010). Briefly, filters were composited weekly (including 5 daily samples) and 92 different organic compounds were quantified by means of gas chromatography/mass spectrometry (GC/MS) (Stone et al., 2008). To measure the concentration of trace elements, sections of weekly-composited sample filters were microwave digested in an acid mixture (containing HNO3, HF and HCl) in Teflon vessels and digestates were then analyzed by high resolution magnetic sector Inductively Coupled Plasma Mass Spectrometry (SF-ICPMS, Thermo-Finnigan Element 2) (Herner et al., 2006). Water-soluble organic carbon (WSOC) was quantified using a Sievers Total Organic Carbon analyzer (General Electric Instruments; GE Analytical Instruments, Boulder, CO, USA) (Zhang et al., 2008). Two semi-continuous OC-EC analyzers (Model 3F, Sunset Laboratory Inc.) were deployed at each site, one indoors and one outdoors, to measure the hourly mass concentration of elemental carbon (EC) and organic carbon (OC). A PM2.5 cyclone was placed at the inlet of the instruments to remove particles larger than 2.5 µm, and samples were collected at a nominal flow rate of 8 lpm. Also, a parallel plate diffusion denuder was placed upstream of each OC-EC instrument to remove 39 most of the gas-phase OC, which is known to cause positive adsorption artifacts (Arhami et al., 2006). 3.2.3. Source apportionment A molecular-marker based source apportionment model (MM-CMB) was used in this study to apportion PM2.5 organic carbon (OC) (Schauer et al., 1996). The model utilizes organic molecular markers that are source-specific tracers as fitting species and it was mathematically solved with an effective variance weighted least-squares solution (Watson et al., 1984), using the U.S. Environmental Protection Agency CMB (EPA-CMB8.2) software. Six sources were considered to have the highest contributions to fine OC in the sampling areas, including light-duty and heavy-duty vehicles (LDV and HDV, respectively) (Kam et al., 2012; Liacos et al., 2012), wood smoke (biomass burning in Western US) (Fine et al., 2004; Sheesley et al., 2007), ship emissions (Agrawal et al., 2008; Rogge et al., 1997), resuspended dust (Schauer, 1998), and vegetative detritus (Rogge et al., 1993a). Vehicular emissions source profiles were based on recent on-road studies conducted at CA-110 and I-710 freeways in Los Angeles. However, inclusion of both LDV and HDV source profiles caused co-linearity in some CMB runs (29 cases). For these samples, the “estimable linear combinations of inestimable sources” was considered as the contribution from mobile sources (Watson et al., 1997), while for the rest of the samples, mobile source contributions were determined as the sum of both LDV and HDV source contributions (Lough et al., 2007). Moreover, to evaluate the sensitivity of the MM-CMB results to the selected vehicular emissions source profile, the contribution of mobile sources to fine OC was determined using a different source profile, derived from a roadway study conducted in 2005 at the CA-110 and I-710 freeways in Los Angeles (Kuhn et al., 2005; L. Ntziachristos et al., 2007b; Phuleria et al., 2007). The sensitivity analysis revealed the stability of the MM-CMB results for the estimated contributions from mobile sources (LDV+HDV) to PM2.5 OC, using the two considered source profiles, as indicated by the slope (±standard error) of 0.97 (±0.05) and R 2 of 0.87 in Figure 3.1. A detailed discussion about this analysis and its results has been provided in the next section. A set of chemical species that are source-specific tracers and chemically stable during the transport from source to receptor was selected as the fitting species in the MM-CMB model. These species included EC, 22S-homohopane, 22R-homohopane, 17α(H)-21β(H)-hopane, 17α(H)- 22,29,30-trisnorhopane, benzo(e)pyrene, benzo(b)fluoranthene, benzo(k)fluoranthene, 40 benzo(ghi)perylene, levoglucosan, indeno(1,2,3-cd)pyrene, nonacosane, hentriacontane, tritriacontane, vanadium, nickel and aluminum. The contributions from vegetative detritus were generally not statistically significantly different from zero, and were therefore removed from our calculations. Source contributions to total PM2.5 mass were evaluated by converting the MM-CMB results for fine OC to those of PM2.5 using the OC-to-PM mass ratios obtained from each source profile. For samples displaying co- linearity for mobile sources, we assumed that the OC apportioned to mobile sources is entirely emitted from LDVs. The OC/PM ratio from the LDV source profile was therefore used for these co-linear samples. This assumption clearly has some uncertainties. To evaluate the range of variation in the mass apportionment of mobile sources for the co-linear cases, we conducted a sensitivity analysis under several different scenarios. The OC apportionment results were converted to PM mass-based assuming that the OC apportioned to mobile sources is emitted from 1) only HDV, 2) 25% LDV/75% HDV, 3) 75% LDV/25% HDV. Results were then compared to our prior assumption that OC from mobile sources is only emitted from LDVs. As can be seen in Table 3.1, results are about 25, 18, and 8% higher when assuming that OC apportioned to mobile sources is from only LDVs, compared to cases 1, 2 and 3, respectively. In areas which are affected by the anthropogenic sources, ambient WSOC is mostly emitted from biomass burning sources (Docherty et al., 2008) or is formed through photochemical reactions (Weber et al., 2007). Other water-soluble organic carbon (other WSOC) is defined as the difference between measured total WSOC and WSOC from biomass burning (WSOCbb). WSOCbb was estimated as 71% of the OC apportioned to biomass burning from the CMB output (Sannigrahi et al., 2006). Other water-soluble organic matter (other WSOM) was then calculated by multiplying other WSOC by a factor of 1.8 (μgOM/μgOC) (Turpin and Lim, 2001). Other WSOM is mainly comprised of secondary organic aerosol (SOA), particularly in outdoor environments (Snyder et al., 2009), while it also includes the emissions from other primary sources in indoor environments such as organic acids from cleaning and other consumer products (Weschler, 2004). 41 Table 3.1. Contribution of mobile sources to PM2.5 in the co-linear cases, assuming that the OC apportioned to mobile sources is emitted from 1) only LDV, 2) only HDV, 3) 25% LDV, 75% HDV, 4) 75% LDV, 25% HDV. The units are in µg/m 3 . Co-linear cases (site, phase, location, week) only LDV only HDV 25% LDV, 75% HDV 75% LDV, 25% HDV G1P1INW4 10.22 8.16 8.68 9.71 G1P2INW1 6.11 4.88 5.18 5.80 G1P2INW2 10.67 8.52 9.06 10.13 G1P2INW3 6.52 5.21 5.54 6.19 G2P1INW2 4.02 3.21 3.41 3.82 G2P2INW1 14.19 11.34 12.05 13.48 G2P2INW2 7.12 5.68 6.04 6.76 G2P2INW3 7.77 6.21 6.60 7.38 G2P2INW5 11.88 9.49 10.09 11.29 G3P1INW1 6.03 4.81 5.12 5.72 G3P1INW2 9.60 7.67 8.15 9.12 G3P1INW3 7.50 5.99 6.37 7.12 G3P1INW4 5.24 4.18 4.45 4.98 G3P1INW5 6.56 5.24 5.57 6.23 G1P1OUTW1 10.95 8.74 9.29 10.40 G1P1OUTW2 13.33 10.64 11.31 12.66 G1P1OUTW5 8.85 7.07 7.52 8.41 G1P1OUTW6 8.91 7.12 7.57 8.46 G1P2OUTW2 11.78 9.41 10.00 11.19 G1P2OUTW3 10.43 8.33 8.86 9.91 G1P2OUTW4 10.81 8.63 9.18 10.27 G1P2OUTW5 16.00 12.78 13.58 15.19 G1P2OUTW6 13.00 10.38 11.04 12.34 G2P1OUTW2 6.93 5.53 5.88 6.58 G2P1OUTW3 14.42 11.52 12.24 13.70 G2P2OUTW2 11.86 9.47 10.07 11.26 G2P2OUTW3 11.57 9.24 9.83 10.99 G3P1OUTW2 10.80 8.63 9.17 10.26 G3P1OUTW4 7.09 5.66 6.02 6.73 42 Other water-insoluble organic matter (other WIOM) corresponds to water-insoluble OM that could not be apportioned to the considered primary sources. This was estimated by multiplying other water-insoluble organic carbon (other WIOC) also by a factor of 1.8 (Turpin and Lim, 2001). Other WIOC was determined as the difference between the total concentration of WIOC (OC- WSOC) and the sum of all primary source contribution estimates (excluding biomass burning), plus the concentration of WIOC from biomass burning. In central LA and Riverside, Heo et al. (2013) found that these compounds, in ambient PM2.5, mostly originate from primary biogenic sources such as food cooking or resuspended soil. Lastly, since inorganic ions were not measured from the filters, sulfate was determined from the concentration of sulfur (S), assuming that all measured water soluble S by ICPMS is in the form of sulfate (Arhami et al., 2009). In addition to sources included in the OC apportionment, other WIOM, other WSOM and sulfate concentrations were considered in PM2.5 mass apportionment. 3.2.4. Sensitivity of MM-CMB results to the vehicular emissions source profiles To evaluate the sensitivity of the MM-CMB results to the selected vehicular emissions source profile, the contribution of mobile sources was found using a different source profile, derived from a roadway study conducted in 2005 at the CA-110 and I-710 freeways in Los Angeles (Kuhn et al., 2005; L. Ntziachristos et al., 2007b; Phuleria et al., 2007). Hereafter, the mobile source profiles (MSP) obtained from Kam et al. (2012) and Liacos et al. (2012) are referred to “MSP1”, and those developed from Kuhn et al. (2005), Ntziachristos et al. (2007), and Phuleria et al. (2007) are referred to “MSP2”. LDV and HDV source profiles were derived from the latter studies and the following species were included in the CMB model as fitting species: EC, 22S- homohopane, 22R-homohopane, 17α(H)-21β(H)-hopane, 17α(H)-22,29,30-trisnorhopane, benzo(e)pyrene, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(ghi)perylene, levoglucosan, indeno(1, 2, 3-cd)pyrene, coronene, vanadium, nickel and aluminum. 43 Figure 3.1. Comparison of mobile (LDV+HDV) source contribution estimates (SCE) to PM2.5 OC using source profiles from two separate studies: mobile source profile 1 (MSP1) from Kam et al. (2012) and Liacos et al. (2012); mobile source profile 2 (MSP2) from Kuhn et al. (2005), Ntziachristos et al. (2007), and Phuleria et al. (2007). Errors in the slope and intercept are standard error. The other source profiles included in the model were similar to those used in this study with the exception of vegetative detritus, which was excluded from the model, as its molecular markers (including nonacosane, hentriacontane, and tritriacontane) were missing in MSP2. No co- linearity was observed in the MM-CMB results using MSP2, and the contribution of mobile sources was identified as the sum of contributions from LDV and HDV. Figure 3.1 presents the comparison of mobile source contribution estimates (SCE) using MSP1 and MSP2. A very good correlation between the estimated contributions from each source profile was obtained, as suggested by the slope (± standard error) of 0.97 (±0.05) and R 2 of 0.87. In addition, a t-test at 0.05 level showed that the mobile SCEs using MSP1 is not statistically different from those obtained y = 0.97 (±0.05) x + 0.23 (±0.12) R² = 0.87 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Mobile SCE using MSP2 (µg/m 3 ) Mobile SCE using MSP1 (µg/m 3 ) 44 by MSP2 (P = 0.44). These findings reveal the stability of MM-CMB results for estimated contributions from mobile sources (LDV+HDV) using different source profiles. 3.2.5. Meteorology and air exchange rates Meteorological data, including temperature, relative humidity as well as wind speed and direction, sorted by study phases, sites, and indoor/outdoor locations are listed in Table 3.2. Table 3.2. Select meteorological parameters at each site during the warm and cold phases Phase Site Temperature ( o C) Outdoor RH a (%) Outdoor wind speed (m/s) Outdoor wind direction Indoor Outdoor Warm phase San Gabriel 1 26.1 ± 1.2 25.1 ± 2.2 59.9 ± 5.7 3.5 ± 0.1 W San Gabriel 2 23.6 ± 0.9 21.5 ± 2.1 57.8 ± 14.9 2.1 ± 0.2 SW San Gabriel 3 23.3 ± 1.1 25.9 ± 2.8 57.5 ± 10.8 2.6 ± 0.1 W Cold phase San Gabriel 1 23.4 ± 1.2 15.4 ± 2.8 58.1 ± 19.1 2.1 ± 0.3 SE San Gabriel 2 23.9 ± 0.8 14.9 ± 2.1 48.6 ± 13.5 2.7 ± 0.6 S San Gabriel 3 24.7 ± 1.2 16.6 ± 3.6 54.7 ± 10.2 1.9 ± 0.1 S a Relative humidity Mean indoor temperature showed very low variability across phases, while the average outdoor temperature was about 9 o C higher during the warm phase compared to the cold phase. Wind speed was generally higher during the warm phase, with a predominant westerly/southwesterly direction, which is typical of the LAB (Hasheminassab et al., 2013). Relative humidity showed moderate variation across sites, with slightly higher values during the warm phase compared to the cold phase (58.4±1.3% and 53.8±4.8%, respectively). Polidori et al. (2007) estimated the air exchange rates ([AER] hr -1 ) in the studied homes by monitoring the decay of indoor CO during the periods affected by a dominant indoor source (such as food cooking). Figure 3.2 shows the average AERs at each site and phase of the study. The estimated AERs were generally low and relatively similar throughout the year at all sites. The low magnitude of AERs can be explained by the common use of air-conditioning along with the low number of open doors and windows in the studied retirement communities. In a study by Suh et al. (1994), conducted in 47 homes of State College, Pennsylvania, the median AERs measured in 45 non-air-conditioned homes was about six times higher compared to air-conditioned homes. Also in the Boston area, during the summer, Long et al. (2000) reported that the measured AERs in non- air-conditioned homes were about 16 to 24 times of those AERs measured at a home equipped with a central air-conditioning system. Figure 3.2. Average indoor-outdoor air exchange rate ([AER] hr -1 ) at each site during the warm and cold phases. Error bars correspond to one standard deviation. 3.2.6. Data Analysis As mentioned in preceding sections, chemical analyses were performed on weekly- composited filters, resulting in 6 sets of chemical data, and therefore estimated source contributions, at each site and phase of the study. To determine the indoor-to-outdoor (I/O) source relationships, I/O mass ratios and correlation coefficients (R) for given species were calculated for each study site and phase. The averages (±standard deviation) of these values are reported in the following sections. High overall correlation coefficient values indicate species originating from outdoors, whereas low or negative correlation coefficients and/or higher than unity I/O values indicate species produced by indoor sources (Daher et al., 2011). In addition, for a specific phase, to investigate the statistical significance of the difference between indoor and outdoor levels of a given species (or a source), paired t-tests, at a 0.05 level of significance, were performed between each pair of datasets (N=6). When deviations from normality were observed in the data points, the significance of differences between the two datasets 0 0.2 0.4 0.6 0.8 1 G1 G2 G3 G1 G2 G3 Warm Phase Cold Phase AER (hr -1 ) 46 was evaluated by conducting the non-parametric Mann-Whitney rank sum test (U test), rather than the paired t-test (Brown and Hambley, 2002). 3.3. Results and discussions 3.3.1. Carbonaceous species Average mass concentrations of carbonaceous species, including EC, WSOC, and WIOC, at all indoor and outdoor sampling sites, during both phases of the study, are presented in Figure 3.3. The variability of their weekly I/O ratios is illustrated within box plots in Figure 3.4. Their average I/O ratios as well as the I/O Pearson correlation coefficients (R) are presented in Table 3.3. Figure 3.3. Average indoor (IN) and outdoor (OUT) mass concentrations (µg/m 3 ) of elemental carbon (EC), water-soluble organic carbon (WSOC), and water-insoluble organic carbon (WIOC) in the fine PM size fraction by site during the warm and cold phases. Error bars correspond to one standard deviation. 0 2 4 6 8 10 12 14 16 G1-IN G1-OUT G2-IN G2-OUT G3-IN G3-OUT G1-IN G1-OUT G2-IN G2-OUT G3-IN G3-OUT Warm phase Cold phase Concentration (µg/m 3 ) WIOC WSOC EC 47 I/O ratios for carbonaceous species showed very low variability during both phases of the study (Figure 3.3), indicating the relatively common origins of these species across all 3 sites. EC, which mainly originates from incomplete combustion of fossil fuels and is a tracer of pollution from diesel exhaust (Schauer, 2003), constituted a small fraction of PM2.5, with an average contribution of 8.7±2.8% to total mass, over all sites and both phases. In the RIOPA (Relationship of Indoor, Outdoor, and Personal Air) study, which was conducted in 105 homes of Los Angeles between 1999 and 2001, Polidori et al. (2006) reported that EC constituted about 7% of PM2.5 mass concentration in both indoor and outdoor environments, which is generally in agreement with the ratios reported in this study. EC did not show any significant seasonality, but its indoor concentration was comparable if somewhat lower than that outdoors, with average I/O ratios of 0.82 and 0.77 during the warm and cold phases, respectively. These I/O ratios were accompanied by high R-values (about 0.83 on average during both phases), suggesting that a substantial fraction of indoor EC is attributed to that of outdoor concentrations which infiltrated indoors. Our results are very similar to those reported by Geller et al. (2002). They reported an average I/O ratio of 0.85 and R-value of 0.84 for EC in 13 residences in Coachella Valley, California, during the winter and spring. OC accounted for about 44 and 33% of PM2.5 mass indoors and outdoors, respectively, with levels ranging from 3.1 to 13.0 µg/m 3 , across all sites and both study phases. These contributions were relatively higher than those reported in the RIOPA study. Polidori et al. (2006) found that contributions of OC to PM2.5 in indoor and outdoor environments of 105 homes of Los Angeles were around 34.4 and 21.0%, respectively. It can be readily inferred from Figure 3.3 that OC was predominantly water-insoluble (83.1±7.5%). As can be seen in Table 3.3, WIOC displayed high I/O ratios with relatively low R-values (0.34-0.47) during both phases of the study, suggesting the presence of important indoor primary sources, such as food cooking, cleaning products, and organic dusts. Outdoor WSOC, as an indicator of SOA formation (Weber et al., 2007), displayed slightly higher concentrations during the warm phase (1.14±0.04 µg/m 3 ) compared to the cold phase (0.83±0.17 µg/m 3 ), mainly because of higher photochemical activities coupled with increased advection of aged particles from upwind “source” regions, during the warmer months (Sardar et al., 2005). While WSOC exhibited I/O ratios higher than 1 during both phases, its R- value was significantly higher during the warm phase, compared to the cold phase (0.7 and 0.43, respectively). These results reflect a significant impact of outdoor sources on indoor levels of 48 WSOC during the warmer months, while they support the predominance of indoor sources (both primary and secondary to a lesser extent) in indoor environments during the cold seasons. Figure 3.4. Box plot of weekly indoor-to-outdoor (I/O) mass ratios for groups of individual organic compounds (including PAHs, Hopanes and steranes, n-alkanes, and organic acids), metals and elements, and carbonaceous species (EC, WSOC, WIOC) during the warm and cold phases. Each box represents the data for all 3 sites pooled together. The reference line shows the I/O mass ratio of 1. 49 3.3.2. Metals and trace elements The variability of weekly I/O ratios for all 47 detected metals and trace elements (TEs) across all sites is shown within box plots for each phase of the study in Figure 3.4. The average I/O ratios and the correlation coefficients (R) of selected metals and TEs at each site and phase of the study are also presented in Table 3.3. These selected species were among the most abundant elements and are toxicologically important (Oeder et al., 2012; Shi et al., 2003; Valavanidis et al., 2005). It should be noted that at some sites and phases of the study, concentrations of few metals and TEs were not detectable indoors or outdoors. To avoid obtaining inestimable indoor-to- outdoor (I/O) mass ratios, the concentration of these species were assumed as half of their limit of detection (LOD). LODs for non-detectable species ranged from 3.45 × 10 −4 to 80.27 and 5.64 × 10 −4 to 11.19 ng/m 3 in the ultrafine and accumulation modes, respectively. The box plots indicate a wide range of variability in the I/O ratios of metals and TEs during both study phases (Figure 3.4), suggesting that these species are emitted from a broad range of sources. Of all the inorganic elements, S was the most abundant species at all sites and phases of the study, with levels ranging from 392 to 1930 ng/m 3 . The outdoor concentration of S was on average 2.7 times higher during the warm phase, compared to the cold phase. Higher concentration of S during the warm phase indicates that S was mostly in the form of sulfate during the warmer months (Arhami et al., 2009). Average I/O ratios were generally below 1, except for few species during the cold phase (namely, Al, S, Ca, K, and Zn), which exceeded unity. During the warm phase, indoor concentrations of S, V, Cr, and Ni showed relatively high correlations with their corresponding outdoor levels (R-values ranging from 0.72 to 0.77), indicating a significant influence of outdoor sources on the indoor levels of these species. 50 Table 3.3. Pearson correlation coefficients (R) and indoor-to-outdoor (I/O) mass ratios of elemental carbon (EC), water-soluble organic carbon (WSOC), water-insoluble organic carbon (WIOC), and selected metals and trace elements, averaged over all sites during the warm and cold phases. Errors correspond to one standard deviation. Species Warm phase Cold phase I/O ± R ± I/O ± R ± EC 0.82 0.14 0.80 0.04 0.77 0.15 0.86 0.16 WSOC 1.07 0.37 0.70 0.18 1.08 0.51 0.43 0.70 WIOC 0.89 0.20 0.47 0.59 0.84 0.23 0.34 0.55 Mg 0.58 0.25 0.52 0.49 0.81 0.34 0.34 0.56 Al 0.62 0.31 0.22 0.30 1.18 0.74 0.10 0.72 S 0.74 0.22 0.75 0.29 1.15 1.07 0.49 0.75 K 0.97 0.59 0.30 0.46 1.08 0.45 0.09 0.16 Ca 0.79 0.33 0.44 0.41 1.13 0.76 0.31 0.68 Ti 0.67 0.33 0.36 0.53 0.78 0.29 0.30 0.62 V 0.77 0.19 0.77 0.22 0.90 0.50 0.42 0.64 Cr 1.00 0.74 0.72 0.25 0.97 0.48 -0.10 0.14 Mn 0.78 0.41 0.28 0.22 0.92 0.51 0.45 0.76 Fe 0.80 0.42 0.33 0.57 0.92 0.40 0.60 0.49 Ni 0.94 0.61 0.75 0.22 0.86 0.28 0.39 0.73 Cu 0.68 0.46 -0.08 0.33 1.00 0.65 0.14 0.75 Zn 0.75 0.21 0.47 0.50 1.12 0.38 0.71 0.10 Pb 0.61 0.23 0.62 0.28 0.88 0.36 0.46 0.39 Vanadium and sulfur are mostly associated with ship emissions, refinery operations, and residual oil combustion (Arhami et al., 2009), while Ni and Cr are known as tracers of industrial emissions (Ntziachristos et al., 2007a; Singh et al., 2002). The average I/O ratio for Zn was 0.75±0.21 and 1.12±38 during the warm and cold phases, respectively. While Zn primarily originates from outdoor sources, smoking has been found to be a dominant source of Zn at indoor environments (Jones et al., 2000). Given that the studied homes were non-smoking residences, increased indoor concentration of Zn is mostly attributed to the enhanced infiltration of these particles from outdoors. Another anthropogenically-dominated metal, Cu, showed high average I/O ratios during the warm and cold phases (0.68 and 1.00, respectively), while it was weakly correlated with its corresponding outdoor concentrations (R-values ranged from -0.08 to 0.14), indicating the presence of potential indoor sources such as resuspended mineral dust (Ibanez et al., 2010) or emissions from indoor electric universal motors such as those inside vacuum cleaners, toys, hair dryers, and blenders (Szymczak et al., 2007). More detailed discussion on indoor/outdoor 51 relationship of size-fractionated metals and TEs at all sampling sites has been provided by Polidori et al. (2009). 3.3.3. Organic compounds Individual organic constituents of PM2.5 were grouped into polycyclic aromatic hydrocarbons (PAHs), hopanes and steranes, n-alkanes, and organic acids. As implemented with the inorganic elements, concentrations of non-detectable organics were assumed as half of their LODs. LODs for non-detectable species ranged from 1.46 × 10 −2 to 0.96 and 6.81 × 10 −3 to 2.92 ng/m 3 in the ultrafine and accumulation modes, respectively. Figure 3.4 shows the box plot distributions of weekly I/O mass ratios for groups of individual organic species (including 19 PAHs, 16 hopanes and steranes, 27 n-alkanes, and 41 organic acids), during the warm and cold phases. Less variability was observed for PAHs, hopanes and steranes, suggesting that these species originate from rather similar sources across all sites, most likely vehicular emissions (Lough et al., 2007). The median I/O values for PAHs ranged from 0.94 to 0.99 during the cold and warm phases, respectively. The I/O ratios showed slightly lower values for hopanes and steranes with median levels ranging from 0.82 during the warm phase to 0.89 during the cold phase. The measured I/O values for individual PAHs and hopanes are well within the ranges reported in previous studies for the fine PM size fraction. Olson et al. (2008) reported median I/O ratios ranging from 0.7 to 1 for 7 PAHs and 1 to 1.1 for 3 hopanes in Tampa, Florida. Ohura et al. (2004) also reported median I/O ratios ranging from 0.62 to 1.27 during the summer and 0.27 to 1.09 during the winter for 19 individual PAHs in two different cities of Japan. Unlike PAHs, hopanes and steranes, n-alkanes and organic acids showed a broader range of I/O ratios, accompanied by median values slightly higher than 1, indicating a larger variability in their sources of origin in the indoor and outdoor environments over different sites. 52 Figure 3.5 a-d. Average mass concentrations (ng/m 3 ) of total (a) PAHs, (b) hopanes and steranes, (c) n-alkanes, and (d) organic acids in indoor and outdoor environments at each site during the warm and cold phases. Error bars correspond to one standard deviation. 53 Figure 3.5 a-d shows the average concentration of organic compounds in both indoor and outdoor environments, at each site and phase of the study. Moreover, the average I/O mass ratios and I/O Pearson correlation coefficients (R) were determined for selected organic species in both phases across all sites, as shown respectively in Figures 3.6 a-d and 3.7 a-d. PAHs are typically produced from incomplete combustion of fossil fuels and/or other organic matter, such as cooking at indoor environments (Abdullahi et al., 2013; Manchester- Neesvig et al., 2003). Average mass concentrations of total PAHs were overall higher during the cold phase (Figure 3.5 a), mainly due to higher atmospheric stability and lower degree of dispersion and mixing during the colder seasons. On the other hand, enhanced photo-degradation of PAHs (Miet et al., 2009) along with increased partitioning to the gas phase at higher temperatures (Mader and Pankow, 2002), can result in lower outdoor concentrations of these species during the warm phase. The highest average outdoor concentration of total PAHs in both phases was observed at G2, which was the closest sampling location to a major freeway among all sites. The average indoor concentrations of total PAHs were not statistically significantly different than the outdoor levels at all sites and both phases of the study (p values ranged from 0.06 to 0.95), suggesting likely contribution of outdoor sources to indoor particle levels. This was further corroborated by the I/O mass ratios (Figure 3.6 a) and correlation coefficients for individual PAHs (Figure 3.7 a). PAH components displayed average I/O ratios close to unity, with generally high and positive correlation coefficients (median R-value across components is 0.61 and 0.77 during the warm and cold phase, respectively), indicating a strong impact from outdoor sources (most notably diesel and gasoline exhaust) on indoor levels of PAHs. Potential indoor sources of PAHs include smoking, gas cooking and heating appliances (Liu et al., 2001; Ohura et al., 2002; Orasche et al., 2012). However, since the retirement communities in this study were non-smoking residences, the contribution of PAHs from tobacco smoke is unlikely (Arhami et al., 2010). 54 Figure 3.6 a-d. Indoor-to-outdoor (I/O) mass ratios of selected (a) PAHs, (b) hopanes and steranes, (c) n-alkanes, and (d) organic acids, averaged over all sites during the warm and cold phases. Error bars correspond to one standard deviation. The reference line shows the I/O mass ratio of 1. 55 Figure 3.7 a-d. Pearson correlation coefficients of selected (a) PAHs, (b) hopanes and steranes, (c) n-alkanes, and (d) organic acids, averaged over all sites during the warm and cold phases. Error bars correspond to one standard deviation. 56 Unlike PAHs, the summed concentration of hopanes and steranes exhibited less seasonality at all sites (Figure 3.5 b). This is likely due to the lower volatility and reactivity of these compounds compared to PAHs (Ruehl et al., 2011). Moreover, given that hopanes and steranes primarily originate from lubricating oil of gasoline- and diesel-powered vehicles, their emission rates are relatively insensitive to driving conditions (Lough et al., 2007). As can be inferred from Figure 3.6 b, the average I/O ratios of individual hopanes were slightly lower than those for steranes, yet both were close to unity (ranging from a minimum of 0.6 to a maximum of 1.1, for all components across both phases). Seasonally, the I/O ratios were relatively higher during the cold phase accompanied by larger R-values compared to the warm phase (median R-values for all components during the warm and cold phases are 0.27 and 0.64, respectively). These results indicate the strong influence of outdoor sources on indoor levels of hopanes and steranes, particularly during the cold seasons. The average cumulative concentration of measured n-alkanes was higher indoors than outdoors, with much higher levels at G3 during the cold phase. While the average I/O ratios for individual n-alkanes (C24-C40) during the warm phase spanned around unity (min=0.9, max= 1.7, median= 1.1), these values significantly increased (up to 8) during the cold phase. Furthermore, I/O correlation coefficients showed overall greater and/or positive values during the warm phase compared to the cold phase. These results, altogether, are indicative of a considerable influence of indoor sources (e.g., cooking, household products, dust, and candle burning (Fine et al., 1999; Kleeman et al., 2008; Schauer et al., 1999)) on the indoor levels of these species, particularly during the cold phase. The carbon preference index (CPI) of n-alkanes (C19-C40) was calculated for all sites and phases (Figure 3.8) to further investigate the origins of these species. CPI is defined as the ratio of summed odd-carbon number n-alkanes to even-carbon number n-alkanes (Simoneit, 1986). CPI around 1 indicates the predominance of emissions from anthropogenic sources, while emissions from biogenic sources usually exhibit CPI greater than 2 (Daher et al., 2011). Both indoor and outdoor n-alkanes showed CPI values around unity (CPI ranging from 0.73 to 0.94), suggesting an overall prevalence of anthropogenic sources in both indoor and outdoor environments. The average total concentration of measured organic acids was overall substantially higher indoors than outdoors (Figure 3.5 d). At G1 and G2 during the warm phase, in particular, average 57 indoor concentrations of total organic acids were significantly higher (more than 4 times) than their corresponding outdoor levels, with p values of 0.015 and 0.006, respectively. The I/O correlation coefficients for individual organic acids were typically low (or negative) along with high standard deviation, while they displayed I/O ratios significantly higher than 1 in both phases (Figure 3.6 d). These results confirm that the indoor levels of these species are strongly affected by indoor sources, potentially including emissions from human skin (Nicolaides, 1974), wax emission from painted walls (Naik et al., 1991), and food cooking (Abdullahi et al., 2013). Figure 3.8. Average carbon preference index (CPI) of n-alkanes (C19-C40) at the indoor and outdoor sampling sites during the warm and cold phases. Error bars correspond to one standard deviation. 3.3.4. Source apportionment of PM2.5 As noted in the methodology section, OC apportionment results from the MM-CMB model were converted to PM mass-based results using OC/PM ratios obtained from each source profile. In addition to the source contributions estimated by MM-CMB, other WSOM, other WIOM, and sulfate were considered as contributors to PM2.5 mass concentration. Source apportionment results for PM2.5 OC and PM2.5 mass concentrations are respectively illustrated in Figure 3.9 and Figure 3.10. We should note that some samples were affected by positive OC adsorption artifacts during 0 0.2 0.4 0.6 0.8 1 1.2 1.4 G1 G2 G3 G1 G2 G3 Warm phase Cold phase CPI Indoor Outdoor 58 two weeks of sampling in the cold study phase at G3. The source apportionment results for these two weeks were excluded from our calculations. Overall, mobile (vehicular) sources were found to be the major contributors to fine PM, accounting for 39±21 and 46±17% of PM2.5 mass, respectively at indoor and outdoor environments, averaged across all sites and both phases. Their contributions to PM2.5 displayed an average I/O ratio of 0.74±0.34 over all sites and phases, illustrating a significant influence of outdoor mobile source emissions on indoor levels of PM2.5. This was further supported by the results of the statistical analyses, which showed that indoor and outdoor estimated contributions from mobile sources were not statistically significantly different (p values ranged from 0.13 to 0.91) at all sites and both phases of the study (with exception of G2 during the warm phase, p= 0.014). In the RIOPA study, Meng et al. (2007) applied positive matrix factorization (PMF) receptor model on paired indoor and outdoor PM2.5 species concentrations to characterize and quantify sources of indoor and outdoor PM2.5. They reported an I/O ratio of about 0.6 for vehicular emission source contributions for 105 homes in Los Angeles, which is within the range of our findings (i.e., 0.4 to 1). In another source apportionment study conducted by Minguillón et al. (2012) in 54 homes of Barcelona, the median I/O ratio for the estimated source contributions from mobile sources was about 0.75, which is quite similar to our results (i.e., 0.74±0.34). Except for G3 during the cold phase, other WIOM, which represents uncharacterized primary sources, on average accounted for 20 and 14% of PM2.5 in indoor and outdoor environments, respectively, over all sites and both study phases. For ambient PM2.5, in central LA and Riverside, Heo et al. (2013) reported that these compounds mainly originate from primary biogenic sources such as food cooking or resuspended soil, which were not considered in the CMB model. Other WIOM displayed a significant contribution to PM2.5 at G3 during the cold phase, with levels ranging from 10.0 to 16.1 µg/m 3 and 10.1 to 16.9 µg/m 3 at indoor and outdoor environments, respectively. The high contribution of other WIOM at G3 during the cold phase suggests the presence of a very unique source at this site that also impacted the levels of n-alkanes, as shown earlier. The average contribution of sulfate to PM2.5 was 18% at indoor sites and 15% at outdoor environments. Its outdoor contribution was generally greater during the warm phase (5.0±1.0 µg/m 3 or 23±6% of PM2.5) compared to the cold phase (1.7±0.1 µg/m 3 or 7.9±1.4% of PM2.5), 59 which is consistent with the increased sulfate formation as a result of enhanced photochemical activity during warmer seasons (Khoder, 2002). Averaged over all sites and both phases, source contribution estimate of resuspended dust was 2.2±1.9 µg/m 3 , corresponding to 11±8% of fine PM. Seasonally, the average contribution of resuspended dust at indoor and outdoor environments was respectively 1.8 and 3.0 times higher in the warm phase compared to the cold phase. Occupant-related activities and road dust are the major sources of resuspended dust in indoor and outdoor environments, respectively. Other WSOM showed significant seasonality in outdoor environments, with higher contribution to PM2.5 during the warm phase (1.8±0.2 µg/m 3 or 7.6±0.7% of PM2.5) compared to the cold phase (0.8±0.6 µg/m 3 3.8±2.6% of PM2.5). This trend in outdoor environments is most likely due to increased photochemical production of WSOM in the atmosphere during warmer months. The indoor contribution of other WSOM ranged from 0.3 to 2.6 µg/m 3 (or 1 to 13% of PM2.5) over all sites and phases. Indoor concentrations of other WSOM were higher than outdoors, at all sites during the cold phase and at G1 during the warm phase. Nonetheless, none of these elevations were statistically significant (p values ranged from 0.15 to 0.70). Several studies have shown that ozone-initiated reactions with emissions from consumer products (Sarwar et al., 2004), building materials (Aoki and Tanabe, 2007), and cleaning products (Singer et al., 2006), lead to SOA formation in indoor environments. Moreover, as noted earlier, indoor other WSOM is also impacted by primary sources such as organic acids from cleaning and other consumer products (Weschler, 2004). Wood smoke accounted for about 3% of PM2.5 mass, averaged over all sites and phases. Outdoor concentrations of wood smoke displayed a significant seasonality, with more than 2 times higher contribution to ambient PM2.5 during the cold phase compared with the warm period (with the exception of G2). This trend is most likely due to increased wood burning for domestic heating purposes during the cold season. Ship emissions were the most minor primary source of PM 2.5, with less than 1% contribution to total mass, averaged over all sites and both phases of the study. Lastly, un-apportioned PM mass accounted for just 7±5% of measured PM2.5, on average. In most of the cases, the un-apportioned PM mass concentrations were not statistically different from zero. This small discrepancy in mass apportionment could be in part associated with ammonium nitrate, which was not measured in this study, but could constitute an important component of PM2.5 (Hughes et al., 2002), especially in outdoor environments. Additionally, 60 uncertainties in the source profiles composition and multiplication factors used to estimate other WIOM and other WSOM could lead to this discrepancy. Figure 3.9. Average relative contribution of different sources to fine OC mass concentration (µg/m 3 ) in indoor (IN) and outdoor (OUT) environments at each site during the warm and cold phases. 0 2 4 6 8 10 12 14 G1-IN G1-OUT G2-IN G2-OUT G3-IN G3-OUT G1-IN G1-OUT G2-IN G2-OUT G3-IN G3-OUT Warm phase Cold phase OC mass concentration (µg/m 3 ) Mobile sources Wood smoke Ship Emissions Resuspended dust Other WSOC Other WIOC 61 Figure 3.10. Average relative contribution of different sources to PM2.5 mass concentration (µg/m 3 ) in indoor (IN) and outdoor (OUT) environments at each site during the warm and cold phases. 0 5 10 15 20 25 30 G1-IN G1-OUT G2-IN G2-OUT G3-IN G3-OUT G1-IN G1-OUT G2-IN G2-OUT G3-IN G3-OUT Warm phase Cold phase PM 2.5 mass concentration (µg/m 3 ) Mobile sources Wood smoke Ship Emissions Rresuspended dust Other WSOM Other WIOM Sulfate Undetermined mass 62 3.4. Conclusions To investigate the indoor/outdoor relationships of PM2.5 and its chemical constituents, as well as to identify and quantify major sources of PM2.5 in indoor and outdoor environments, a sampling campaign was conducted between 2005 and 2006 at three retirement homes in the Los Angeles Basin. Outdoor PM2.5 levels were constantly higher than those measured indoors at all sites and phases of the study. Indoor concentrations of EC were comparable to their corresponding outdoor levels (average I/O= 0.80) and were strongly correlated (average R= 0.83), indicating a considerable impact of outdoor sources on the indoor levels of EC. Indoor levels of metals and trace elements were found to be mostly affected by outdoor sources. PAHs, hopanes and steranes, exhibited low variability in their indoor-to-outdoor (I/O) mass ratios, with median values close to unity, accompanied by relatively high I/O correlations, reflecting the significant impact of outdoor sources (most notably vehicular emissions) on their indoor levels. By contrast, n-alkanes and organic acids exhibited much higher I/O ratios along with poor correlations with their corresponding outdoor levels, implying that the indoor concentrations of these species were mostly dominated by indoor sources. Source apportionment results revealed that vehicular sources were the dominant sources of PM2.5, with an average contribution of 43±19% over all sites and phases of the study. Moreover, the contribution of mobile sources to indoor PM levels was generally comparable to their corresponding outdoor estimates (I/O= 0.74±0.34, averaged over all sites and phases). Except for G3 during the cold phase, sulfate was generally the next most abundant component, across all sites and phases. Other WIOM, which accounts for uncharacterized primary biogenic sources, showed significant contributions to indoor and outdoor PM2.5 during the cold phase at G3 (71.5 and 73.4%, respectively). Resuspended dust and other WSOM respectively contributed to 11 and 7% of PM2.5, on average across all sites and phases. Our results suggest that even though the elderly subjects of this study spend most of their time in indoor micro-environments with relatively low air exchange rates, they are exposed to considerable levels of PM of both indoor and outdoor origin. Therefore, a full understanding of the adverse health effects of exposure to indoor PM requires a detailed knowledge of human activity patterns in different micro-environments and, more importantly, information on the degree to which indoor and outdoor sources contribute to indoor levels of PM. 63 3.5. Acknowledgements This research was funded in part by NIEHS-NIH (Grant no. ES-012243) and the California Air Resources Board (ARB) (contracts no. 03-329 and 09-341). The study on the accumulation mode of the samples was funded by ARB (contract no. 09-341) and is presented in the final report. We would like to acknowledge the staff at the Wisconsin State Laboratory of Hygiene for their assistance with the chemical analysis. We also wish to acknowledge the support of USC Provost's Ph.D. fellowship. 64 CHAPTER 4 Long-term source apportionment of ambient fine particulate matter (PM 2.5 ) in the Los Angeles Basin: A focus on emissions reduction from vehicular sources This chapter is based on the following publication: Hasheminassab, S., Daher, N., Ostro, B. D., & Sioutas, C. (2014). Long-term source apportionment of ambient fine particulate matter (PM 2.5) in the Los Angeles Basin: A focus on emissions reduction from vehicular sources.Environmental Pollution, 193, 54-64. 65 4.1. Introduction Firmly established associations between exposure to ambient particulate matter (PM) and adverse health outcomes (Delfino et al., 2004; Dockery et al., 1989; Laden et al., 2000; Miller et al., 2007; Ostro et al., 2011; Pope et al., 2004) urged the U.S. Environmental Protection Agency (U.S. EPA) to include PM in the National Ambient Air Quality Standards (NAAQS) as a criteria pollutant. The NAAQS are routinely revised and in the latest revision in 2012, the annual standard for primary fine PM (PM2.5, particles with an aerodynamic diameter less than 2.5 µm) was lowered from 15 to 12 µg/m 3 . Under the Clean Air Act (CAA), each state must develop a plan and strategy describing how it will attain and maintain the NAAQS. Attaining these standards requires proper control strategies, which in the first step necessitate the identification and quantification of major sources of ambient PM. To assist states in understanding the composition of fine PM, the U.S. EPA established the Speciation Trends Network (STN), which provides nationally consistent speciated PM2.5 data in selected urban areas of the country. Interagency Monitoring of Protected Visual Environments (IMPROVE) is another long-term monitoring network, providing speciated PM2.5 data mostly in rural areas. Chemical composition datasets from these monitoring networks have been widely used in several studies over the U.S., and the major sources of ambient PM2.5 have been identified and their contributions have been assessed using different source apportionment techniques (Kim and Hopke, 2008; Lee and Hopke, 2006; Rizzo and Scheff, 2007; Zhou et al., 2009). Since 2004, the South Coast Air Basin in California has been designated as a non- attainment area (Kim et al., 2010). Therefore, the state and local air quality management agencies developed implementation plans outlining how the basin should attain the standard for fine PM by reducing air pollutant emissions contributing to PM2.5. Thus far, the majority of these regulations and implementations have targeted emissions from motor vehicles, given that many studies have linked exposure to traffic-related air pollution to various adverse health effects (de et al., 2006; Delfino et al., 2010; Ranft et al., 2009). Diesel exhaust, in particular, was declared as a toxic air contaminant in 1998 by the California Air Resources Board (CARB). In a major action by the U.S. EPA, beginning with the 2007 model year (MY), all diesel trucks must meet the 2007 PM emissions standard. The 2007 nitrogen oxides (NOx) emission standard for diesel trucks, however, was phased in over three years, based on a percent-of-sales basis: 50% from 2007 to 2010 and 100% in 2010. In other words, starting with 2007 MY until 2010, 50% of total sales for each diesel 66 engine manufacturer must meet the 2007 NOx emission standard, while after 2010 MY, all trucks must comply with the standard. The 2007 PM and NOx emission standards for diesel trucks are more than 90% lower than the prior standard levels, as shown in Table 4.1. In addition to the EPA standards, CARB’s Truck and Bus regulation went into effect in January 2012, requiring all heavy diesel trucks (greater than 26,000 lb) to use diesel particulate filters (DPFs) (California Code of Regulations, 2008). Starting in January 2014, all in-use diesel trucks operating in the state of California must meet the EPA’s 2007 emission standards. In an innovative local policy, starting in October 2008, the ports of Los Angeles and Long Beach banned all heavy-duty diesel trucks with pre-1989 MY engines from entering the port terminals. Moreover, starting in January 2010, 1989- 1993 MY trucks were banned in addition to 1994-2003 MY trucks that had not been retrofitted. In the latest phase of the program, since January 2012, all trucks entering the ports must meet the EPA promulgated emissions standards for 2007 MY trucks (POBL, 2010). Table 4.1 summarizes the EPA diesel truck emission standards for PM and NOx along with the compliance date mandated by the CARB and the Ports of Los Angeles and Long Beach. Table 4.1. The U.S. EPA diesel truck emission standards for PM and NOx (g/kW.hr) along with compliance date mandated by CARB and Ports of Los Angeles and Long Beach. Diesel truck Model year (MY) PM NOx CARB Ports 1998 0.13 5.4 2004 0.13 3.2 2004 2010 2007 0.013 0.27 (50%) * 2014 2012 >2010 0.013 0.27 (100%) * 2023 no date * 50% of manufactured trucks must meet the 2007 emissions standards until 2010, while 100% must comply with the standards after 2010 MY. So far, several source apportionment studies have been conducted in the Los Angeles Basin (LAB), using different techniques, to identify and quantify major sources of ambient PM (Hasheminassab et al., 2013; Heo et al., 2013; Schauer et al., 1996). The majority of these studies, however, were limited to a single year, with only a few investigating source apportionment of PM over a longer period. In the south coast area, source attribution of ambient PM 2.5 was conducted, using STN data, over a few years, ranging from 2001 to 2004 (Kim and Hopke, 2007a) and 2003 to 2005 (Kim et al., 2010). In the present study, by comparison, the source apportionment of ambient PM2.5 is conducted over an almost 12-year period ranging from 2002 to 2013, using STN 67 data collected in downtown Los Angeles and Rubidoux. Having a comprehensive dataset for near 12 years has offered us a unique opportunity to investigate and track the changes in contributions from different sources, particularly vehicular emissions, considering that during this time period (i.e., 2002-2013) major local and federal regulations on reduction of PM from vehicular sources went into effect, as noted above. The results of this study will be also used as input parameters to examine the associations between the estimated source contributions and a variety of health outcomes, in an epidemiological investigation led by California EPA (Cal EPA). 4.2. Methodology 4.2.1. Sampling sites Sampling was conducted at two STN sites in the LAB. The location of the sampling sites and adjacent major freeways is presented in Figure 4.1. One of the sites is located on North Main Street in downtown Los Angeles, a typical urban area impacted mostly by primary emissions (Moore et al., 2007). The other sampling site is situated 60 km inland from downtown LA, in Rubidoux, which is typically subject to aged and photo-chemically processed particulate plumes, advected from the upwind regions of west and central Los Angeles. Rubidoux is also located downwind of Chino dairy farms and livestock, with significant emissions of ammonia, leading to high concentrations of ammonium nitrate after atmospheric reactions (Hughes et al., 1999). Figure 4.1. Location of the sampling sites in downtown Los Angeles and Rubidoux. 68 4.2.2. Sampling schedule and chemical analysis Time-integrated 24-hr PM2.5 samples were collected once every three days from 1/2/2002 to 8/5/2013 in Rubidoux. In Los Angeles, sampling was conducted between 6/1/2002 and 8/11/2013. At the latter site, samples were collected once every six days from 2002 to 2011, while every other three days after 2011. Samples were chemically analyzed based on the EPA protocol specified for STN field sampling. PM2.5 mass concentration was measured gravimetrically by pre- and post- weighing the Teflon filters. Teflon filters were also analyzed by energy-dispersive X- ray fluorescence (EDXRF) to quantity the concentration of elements. Anions (nitrate and sulfate) and cations (ammonium, sodium, and potassium) were measured by Ion Chromatography (IC). Lastly, thermal/optical analysis was conducted on quartz filters to measure the concentration of elemental carbon (EC) and organic carbon (OC) (Kim et al., 2005). Until 2007, the PM2.5 carbon data had the greatest discrepancy between the STN and IMPROVE networks, due to differences in sampling and analytical methods. The STN used a Total Optical Transmittance (TOT) NIOSH 5040 carbon method (Birch and Cary, 1996), while the IMPROVE used Total Optical Reflectance (TOR) method (Chow et al., 2001). In order to improve data comparability, the STN method was converted to IMPROVE-like carbon sampling and analysis protocol. The carbon channel of the speciation sampler was therefore replaced with the URG 3000N sampler that is similar to the IMPROVE sampler except for the addition of active flow control (Chow et al., 2010). 4.2.3. Source apportionment The measured chemical composition of PM at a receptor site can be used to estimate the relative contribution of major sources to total PM mass. The fundamental principle of this methodology, which is referred to as receptor modeling, is the mass conservation of chemical species during their transport from source to receptor (Watson et al., 2008). In this study, the EPA positive matrix factorization (PMF) model (version 3.0.2.2) was applied to identify and quantify major sources of ambient PM2.5 at each site. PMF is a factor analysis model that solves the chemical mass balance equations using a weighted least-squares algorithm and by imposing non-negativity constrains on the factors (Reff et al., 2007). 69 4.2.3.1.Data screening As noted earlier, the method of carbon measurement was changed in 2007. The OC artifact was estimated for both TOR and TOT methods, using the intercept of the regression of OC against PM2.5 mass concentration (Kim et al., 2005), as shown in Figure 4.2 a-d. OC concentrations were then corrected by subtracting the OC artifact concentrations. In addition, for 2007-2013, artifact- corrected OC TOR data were amended to match the artifact-corrected OC TOT data, using linear correlation equations obtained from co-located measurements of STN and IMPROVE-like samplers (Figure 4.3 a). Likewise, EC TOR was corrected to match EC TOT data, using linear correlation equation obtained from their co-located measurements (Figure 4.3 b). It should be noted that due to the limited number of co-located data points in Los Angeles, data from both LA and Rubidoux sampling sites were combined together in the regression analysis between co- located measurements of STN and IMPROVE-like samplers. 70 Figure 4.2 a-d. Linear regression of OC mass concentration, obtained from TOT and TOR measurement methods, versus PM2.5 mass concentration in Los Angeles and Rubidoux. Errors represent the standard error. 71 Figure 4.3 a-b. a) Linear correlation between artifact corrected OC TOT and artifact corrected OC TOR; b) linear correlation between EC TOT and EC TOR. Data were obtained from co-located measurement of STN and IMPROVE-like samplers in Los Angeles and Rubidoux. Errors correspond to standard error. To avoid double counting of species, the linear correlations in each pair of S/SO4 2- , Na/Na + , and K/K + were examined. Because they showed good correlations (R 2 >0.6, slope of about 3 for S/SO4 2- and near unity for Na/Na + , K/K + ) (Figure 4.4), and because of higher IC analytical precision (Kim and Hopke, 2008), IC SO4 2- , IC Na + , and IC K + were included in the PMF analyses. Measured concentrations below the detection limit (BDL) were replaced by half of the detection limit (DL) values, and their uncertainties were set as 5/6 of the DL values (Polissar et al., 1998). In addition, missing concentrations were replaced by the geometric mean of the species concentration, and their accompanying uncertainties were set at four times this geometric mean concentration. Table 4.2 presents a summary of statistics and mass concentrations of PM2.5 and its chemical constituents in Los Angeles and Rubidoux. y = 1.15( ±0.05) x R² = 0.84 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Artifact corrected OC TOT (µg/m 3 ) Artifact corrected OC TOR (µg/m 3 ) y = 0.92( ±0.04) x + 0.14( ±0.04) R² = 0.83 0 1 2 3 4 0 1 2 3 4 EC TOT (µg/m 3 ) EC TOR (µg/m 3 ) a) b) 72 Figure 4.4. Linear correlation between Ion Chromatography (IC) K + , IC Na + , and IC SO4 2- , and X-ray Fluorescence (XRF) K, XRF Na, and XRF S in Los Angeles and Rubidoux. 73 Table 4.2. Summary statistics and mass concentrations of PM2.5 and its chemical constituents in Los Angeles and Rubidoux. Units for PM2.5, EC, OC, SO4 2- ,NH4 + , and NO3 - are in µg/m 3 , and for other species are in ng/m 3 . a Below detection limit b Signal-to-noise ratio, calculated by EPA PMF Los Angeles Rubidoux Species Arithmetic mean Geometric mean Minimum Maximum BDL a (%) S/N b Arithmetic mean Geometric mean Minimum Maximum BDL a (%) S/N b PM2.5 17.08 14.86 2.20 68.50 0.1% 18.3 19.98 16.24 1.40 112.60 0.0% 18.5 EC 1.30 1.13 0.12 4.35 0.2% 1.4 1.19 0.97 0.12 5.15 2.8% 2.0 OC 2.16 1.40 0.12 15.29 10.3% 1.6 2.42 1.40 0.12 14.47 14.8% 2.6 SO4 2- 2.67 1.77 0.02 12.20 0.0% 7.3 2.28 1.57 0.01 9.86 0.1% 9.7 NH4 + 2.06 1.16 0.01 18.10 0.4% 8.9 2.78 1.57 0.01 21.70 0.2% 11.1 NO3 - 4.85 3.25 0.01 41.00 0.1% 7.7 7.11 4.25 0.05 53.90 0.0% 10.8 Al 39.22 23.40 3.55 364.00 40.5% 3.3 57.81 102.15 3.55 1200.00 27.7% 5.4 Br 5.02 4.11 0.55 22.40 10.5% 3.9 5.16 11.81 0.50 30.90 12.2% 4.1 Ca 68.77 54.38 3.70 413.00 0.7% 12.3 118.33 82.89 2.25 1680.00 0.4% 12.8 Cl 72.64 27.00 2.35 1200.00 22.1% 11.6 74.33 28.86 2.35 966.00 21.8% 11.7 Cr 2.73 1.75 0.75 31.40 65.9% 2.0 - - - - - - Cu 12.42 8.07 0.65 58.30 5.8% 8.8 7.40 5.29 0.60 39.90 12.3% 6.2 Fe 208.69 165.74 17.60 1060.00 0.0% 12.9 155.12 358.91 1.05 1250.00 0.1% 12.8 Pb 3.78 2.81 0.90 22.10 68.6% 1.0 4.16 2.97 0.90 53.30 67.6% 1.4 Mg 33.25 18.35 4.75 182.00 53.4% 2.4 26.52 47.10 4.75 254.00 59.2% 1.7 Mn 4.75 3.22 0.80 27.50 28.7% 4.0 3.00 6.28 0.80 28.50 42.0% 2.4 Ni 3.10 1.48 0.50 32.60 52.8% 3.0 1.86 1.22 0.50 53.20 61.8% 2.8 Si 122.29 87.08 4.75 1120.00 3.8% 10.0 170.20 123.84 4.80 2340.00 2.7% 10.3 K+ 100.46 32.72 1.00 269.00 33.1% 3.8 78.52 96.87 3.40 512.00 36.2% 5.2 NA+ 373.19 249.94 15.00 2200.00 1.5% 5.4 286.43 188.15 14.20 1880.00 4.6% 5.1 Sr 3.10 1.75 0.85 17.30 69.8% 0.9 - - - - - - Ti 9.53 6.53 1.75 57.40 34.4% 3.5 7.89 5.53 0.90 93.60 40.6% 3.1 V 4.16 2.88 0.75 30.00 58.5% 2.1 3.85 2.70 0.75 25.10 59.7% 2.0 Zn 15.73 11.42 0.85 66.30 4.0% 9.6 17.10 9.41 0.85 348.00 9.7% 11.2 74 Chemical species with more than 70% BDL values as well as samples with missing mass and/or all of the elemental concentrations were excluded from the model. Lastly, to decrease the influence of extreme and episodic events, samples collected around July 4 th and New Year eves, in which K + mass concentrations were unusually high, due to fireworks emissions, were excluded. The excluded samples accounted for less than 4% of total number of samples in each sampling site. Finally, 779 samples and 24 species (including PM2.5 mass concentration) in Los Angeles, and 1323 samples and 22 species (including PM2.5 mass concentration) in Rubidoux were included in the PMF models. 4.2.3.2.Positive Matrix Factorization (PMF) model To reduce the weight in the solution, species that have a signal-to-noise (S/N) ratio between 0.2 and 2, and those that have BDL values more than 50% were considered as weak variables (i.e., their uncertainties were increased by a factor of 3). In addition, to directly obtain the mass apportionment results without the usual multiple regression, PM2.5 mass concentrations were included in the model as a total variable (set as a weak species) (Lee et al., 2011). The model was ran in the default robust mode to decrease the influence of extreme values on the PMF solution, and the FPEAK parameter was applied to control rotational ambiguity (Paatero et al., 2002). In order to determine the optimal solutions, different numbers of sources were explored by applying a trial and error method. The PMF model provides an additional feature to add “Extra Modeling Uncertainty (0- 25%)”, which is applied to all species. This value encompasses various errors not considered in measurement or lab errors and blank values (Norris et al., 2008). By conducting several sensitivity tests with various extra modeling uncertainty values, the most stable and realistic solutions were obtained by a value of 5% extra modeling uncertainty. The final solutions were chosen based on the evaluation of the deduced source profiles and the quality of the chemical species fits by testing different numbers of factors. Moreover, uncertainties in the source profiles were estimated by a bootstrap procedure, provided by the EPA PMF software (Norris et al., 2008). The estimated uncertainty by the bootstrap analysis accounts for several sources of error, including temporal variation of PM source profiles, measurement errors, and errors in the modeling process, such as rotational ambiguity and mis-specified number of factors (Reff et al., 2007). 500 runs were considered for the bootstrap 75 analysis and a solution was considered valid when the occurrence of unmapped factors was less than 10% of the total runs. 4.2.3.3.Sensitivity of the PMF model to input dataset To evaluate the sensitivity of the PMF results to the input chemical dataset, the PMF model at each sampling site was performed once with the chemical dataset from 2002 to 2006 (pooled together) and once with the chemical dataset from 2008 to 2012 (pooled together). In order to maintain consistency in the number of years before and after the implementation of the emissions standards, 2007 and 2013 chemical datasets were intentionally excluded for this analysis. Figures 4.5 and 4.6 show the side-by-side source profiles as well as the explained variations of species in each identified factor, obtained from the PMF model performed on the two separate datasets (i.e., 2002-2006 and 2008-2012) in Los Angeles and Rubidoux, respectively. In Rubidoux, the PMF model conducted on the chemical datasets before and after 2007 resolved a similar number of sources with analogous composition (Figure 4.6). In Los Angeles, however, the PMF model was able to discern two separate source profiles for diesel and gasoline vehicles before 2007 (i.e., using 2002-2006 dataset), while this split was not achieved in the PMF model performed on the chemical dataset after 2007 (i.e., using 2008-2012 dataset) (Figure 4.5). This difference is mainly due to the changes in the composition of diesel exhaust after the implementation of emissions standards in 2007, which made the separation of gasoline and diesel emissions rather complicated and ambiguous. With the exception of mobile sources, the PMF model before and after 2007 in Los Angeles resolved source profiles with similar composition (Figure 4.5). In particular, the source profiles/composition of secondary aerosols (i.e., secondary ammonium nitrate and ammonium sulfate) at both sites remained unchanged after 2007, indicating that the changes in the traffic- related PM2.5 did not affect the composition of secondary particles. Nonetheless, as will be discussed in the following sections, the contributions of secondary aerosols to total PM2.5 mass significantly decreased over the years in the LA Basin, which could be in part attributed to the reduction of their gaseous precursors (e.g., NOx) , emitted from mobile sources. 76 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn 0 20 40 60 80 100 0.001 0.01 0.1 1 Secondary nitrate 0 20 40 60 80 100 0.001 0.01 0.1 1 Soil 0 20 40 60 80 100 0.001 0.01 0.1 1 Aged sea salt 0 20 40 60 80 100 0.001 0.01 0.1 1 Secondary sulfate 0 20 40 60 80 100 0.001 0.01 0.1 1 Biomass burning 0 20 40 60 80 100 0.001 0.01 0.1 1 Fresh sea salt 0 20 40 60 80 100 0.001 0.01 0.1 1 Gasoline vehicles 0 20 40 60 80 100 0.001 0.01 0.1 1 Diesel vehicles 0 20 40 60 80 100 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Mixed vehicular Concentration (µg/µg) Explained variation (%) Figure 4.5. PM2.5 source profiles and explained variation of each species, obtained from PMF model using two separate input datasets (2002-2006 and 2008-2012) in Los Angeles. Error bars correspond to one standard deviation obtained from the bootstrap analysis. 76 77 0 20 40 60 80 100 0.001 0.01 0.1 1 Diesel vehicles 0 20 40 60 80 100 0.001 0.01 0.1 1 Gasoline vehicles 0 20 40 60 80 100 0.001 0.01 0.1 1 Industrial 0 20 40 60 80 100 0.001 0.01 0.1 1 Aged sea salt 0 20 40 60 80 100 0.001 0.01 0.1 1 Secodnary nitrate 0 20 40 60 80 100 0.001 0.01 0.1 1 Fresh sea salt 0 20 40 60 80 100 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Secondary sulfate Concentration (µg/µg) Explained variation (%) OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn 0 20 40 60 80 100 0.001 0.01 0.1 1 Soil Figure 4.6. PM2.5 source profiles and explained variation of each species, obtained from PMF model using two separate input datasets (2002-2006 and 2008-2012) in Rubidoux. Error bars correspond to one standard deviation obtained from the bootstrap analysis. 78 Moreover, to evaluate whether the contributions from identified sources are biased when the PMF model is performed on the combined chemical dataset from 2002 to 2013, average source contributions during the periods of interest (i.e., 2002 to 2006 and 2008 to 2012) were first calculated from the output results of the PMF model conducted on the combined 2002-2013 dataset at each site. These values were then compared to the average contributions of identified sources, obtained from the output of the PMF model performed separately on the pooled chemical datasets of each time period (2002-2006 and 2008-2012). As shown in Tables 4.3 a-b and 4.4 a-b, it can be deduced that, for both sampling sites, running the PMF model with the combined chemical dataset from 2002 to 2013 results in very similar source contributions to the case when PMF was ran separately with each of the clustered dataset (i.e., 2002-2006 and 2008-2012). The difference between the source contributions averaged 12±10 and 18±14% over all sources and both periods, in Los Angeles and Rubidoux, respectively. This was also supported by Mann-Whitney rank sum tests which showed that none of these differences are statistically significant (p>0.05). In Los Angeles, in particular, even though the split between gasoline and diesel disappears when PMF is performed on the combined 2002-2013 dataset, the total contribution from vehicular emissions is still conserved during both time periods (16 and 2% difference for 2002 to 2006 and 2008 to 2012, respectively). Given that the scope of this study is to focus on the changes in total mobile source contributions, these results indicate that performing the PMF analysis on combined chemical data from 2002 to 2013 was reasonable. 79 Dataset used in PMF model 2002-2013 2002-2006 Secondary nitrate 6.03 ± 0.45 5.37 ± 0.42 Secondary sulfate 4.74 ± 0.28 4.82 ± 0.29 Biomass burning 1.03 ± 0.07 1.24 ± 0.09 Aged sea salt 2.68 ± 0.13 2.11 ± 0.11 Fresh sea salt 0.2 ± 0.03 0.34 ± 0.05 Soil 1.06 ± 0.07 1.06 ± 0.08 Vehicular emissions (gasoline + diesel) 3.62 ± 0.17 4.20 ± 0.15 Dataset used in PMF model 2002-2013 2002-2006 Secondary nitrate 12.12 ± 0.52 11.11 ± 0.51 Secondary sulfate 3.18 ± 0.12 3.37 ± 0.13 Biomass burning 0.72 ± 0.04 0.91 ± 0.05 Aged sea salt 1.43 ± 0.06 2.16 ± 0.08 Fresh sea salt 0.59 ± 0.04 1.19 ± 0.09 Soil 0.85 ± 0.05 0.88 ± 0.05 Vehicular emissions (gasoline + diesel) 4.33 ± 0.12 4.11 ± 0.12 Industrial emissions 0.20 ± 0.01 0.27 ± 0.02 Table 4.3 a-b. Average (± standard error) source contributions (µg/m 3 ) between 2002 and 2006, obtained from the results of the PMF model, using two separate input datasets (2002-2013 and 2002-2006) in a) Los Angeles and b) Rubidoux. Dataset used in PMF model 2002-2013 2008-2012 Secondary nitrate 3.57 ± 0.22 3.74 ± 0.24 Secondary sulfate 2.09 ± 0.13 1.98 ± 0.11 Biomass burning 1.12 ± 0.05 0.91 ± 0.04 Aged sea salt 2.98 ± 0.13 3.39 ± 0.14 Fresh sea salt 0.44 ± 0.04 0.28 ± 0.02 Soil 0.86 ± 0.05 0.86 ± 0.05 Vehicular emissions (gasoline + diesel) 3.00 ± 0.12 3.07 ± 0.13 Dataset used in PMF model 2002-2013 2008-2012 Secondary nitrate 6.03 ± 0.30 6.08 ± 0.29 Secondary sulfate 1.89 ± 0.07 1.12 ± 0.04 Biomass burning 0.92 ± 0.04 0.77 ± 0.03 Aged sea salt 1.67 ± 0.07 2.1 ± 0.08 Fresh sea salt 0.56 ± 0.04 0.46 ± 0.04 Soil 0.85 ± 0.03 0.9 ± 0.04 Vehicular emissions (gasoline + diesel) 3.31 ± 0.09 3.39 ± 0.09 Industrial emissions 0.08 ± 0.00 0.25 ± 0.01 a) b) a) Table 4.4 a-b. Average (± standard error) source contributions (µg/m 3 ) between 2008 and 2012, obtained from the results of the PMF model, using two separate input datasets (2002-2013 and 2008-2012) in a) Los Angeles and b) Rubidoux. b) 80 4.2.4. Statistical analysis In the following sections, for a given parameter (e.g., source contributions and/or traffic flow), to investigate the statistical significance of the difference between two groups, normality in the distribution of the datasets was first analyzed by Shapiro–Wilk W-test, using SigmaPlot (v 11.0). Given that normality was not achieved in any statistical test, the non-parametric Mann- Whitney rank sum test (U test) was applied to assess the statistical significance (at a 0.05 level) of the difference between each two datasets (Brown and Hambley, 2002). 4.3. Results and discussion 4.3.1. Source identification and apportionment Seven and nine sources, with FPEAK=0, were deduced from the PMF models in Los Angeles and Rubidoux, respectively. Source profiles along with the explained variation (EV) of each species in Los Angeles and Rubidoux are respectively shown in Figures 4.8 and 4.9. Grey bars represent the concentration of each species normalized to the mass concentration of PM 2.5 apportioned to that factor, while the black dots represent the percent of each species apportioned to that factor (Lee et al., 1999). The comparisons of measured and reconstructed PM2.5 (Figure 4.7 a-b) revealed that the PMF models were able to effectively estimate the daily measured PM2.5 in Los Angeles (slope= 0.89±0.01, R 2 = 0.86) and Rubidoux (slope= 0.92±0.01, R 2 = 0.94). The annual average gravimetric PM2.5 mass concentrations as well as contributions from the identified sources, segregated by year, are presented in Figure 4.10 a-b. Moreover, the seasonal and weekday/weekend variations of source contributions are illustrated in Figures 4.11 and 4.12, respectively. Overall, annual average PM2.5 mass concentration has decreased by approximately half from 2002 to 2013 (Figure 4.10 a-b), at both sampling sites. As will be discussed in detail below, these reductions were mainly driven by the reduction of secondary aerosols (including both secondary ammonium nitrate and ammonium sulfate) as well as the contributions from vehicular sources, over the past almost decade in the LA Basin. 81 Figure 4 a-b. Linear correlations between the measured and estimated PM2.5 in a) Los Angeles and b) Rubidoux. Errors represent the standard error. Figure 4.7 a-b. Linear correlations between the measured and estimated PM2.5 in a) Los Angeles and b) Rubidoux. Errors represent the standard error. y = 0.92(±0.01) x + 1.05(±0.15) R² = 0.94 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Estimated PM 2.5 (µg/m 3 ) Measured PM 2.5 (µg/m 3 ) y = 0.89( ±0.01) x + 1.14( ±0.24) R² = 0.86 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Estimated PM 2.5 (µg/m 3 ) Measured PM 2.5 (µg/m 3 ) a) b) 82 Figure 4.8. PM2.5 source profiles and explained variation (EV) of each species in Los Angeles. Error bars correspond to one standard deviation obtained from bootstrap analysis. 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Fresh sea salt Concentration (µg/µg) Explained Variation (%) 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Soil 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Aged sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Vehicular emissions 83 Figure 4.9. PM2.5 source profiles and explained variation (EV) of each species in Rubidoux. Error bars correspond to one standard deviation obtained from bootstrap analysis. 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Fresh sea salt Concentration (µg/µg) Explained variation (%) 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Diesel vehicles 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Industrial 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Aged sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Gasoline vehicles 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Soil 83 84 Figure 4.10 a-b. Annual average gravimetric mass concentration and estimated source contributions (µg/m 3 ) to ambient PM2.5 by year, in a) Los Angeles and b) Rubidoux. Error bars correspond to one standard error. 0 5 10 15 20 25 30 35 Mass concentration (µg/m 3 ) 0 5 10 15 20 25 30 35 Mass concentration (µg/m 3 ) Industrial Biomass burning Soil Fresh sea salt Aged sea salt Secondary sulfate Secondary nitrate Vehicular emissions Gravimetric PM2.5 mass a) b) 85 Figure 4.11 a-b. Seasonal average source contributions (µg/m 3 ) to PM2.5 in a) Los Angeles and b) Rubidoux. Error bars correspond to one standard error. 0 2 4 6 8 10 12 Mass concentration (µg/m 3 ) Spring Summer Fall Winter a) 0 2 4 6 8 10 12 Mass concentration (µg/m 3 ) b) 86 Figure 4.12 a-b. Average source contributions (µg/m 3 ) to PM2.5 during weekdays and weekends in a) Los Angeles and b) Rubidoux. Error bars correspond to one standard error. The secondary ammonium nitrate factor is characterized by high concentrations of NO3 - and NH4 + . This source was the major contributor to PM2.5 at both sites, accounting for about 27 and 44% of total PM2.5 in Los Angeles and Rubidoux, respectively. This observation is in agreement with several other studies conducted in the Los Angeles Basin (Christoforou et al., 2000; Kim et al., 2010; Kim and Hopke, 2007a). The formation of secondary ammonium nitrate depends on the emissions of its gaseous precursors (i.e., nitrogen oxides (NOx) and ammonia (NH3)), temperature, relative humidity, and the presence of hydroxyl radical (OH-) (Heo et al., 2009; Seinfeld and Pandis, 2006). Given that in the LA Basin vehicular traffic is a major source of NOx (McDonald et al., 2012), control strategies on NOx emissions from vehicular sources have the potential to reduce the concentration of secondary nitrate (Millstein et al., 2008). On an average basis over all years, secondary ammonium nitrate showed more than two times higher contribution to ambient PM2.5 in Rubidoux compared with Los Angeles. This is mainly because of the location 0 2 4 6 8 10 Mass concentration (µg/m 3 ) Weekday Weekend a) 0 2 4 6 8 10 Mass concentration (µg/m 3 ) b) 87 of Rubidoux immediately downwind of Chino dairy farms, which constitute a major source of ammonia (Hughes et al., 1999), as noted earlier. In Los Angeles, this source showed a strong seasonality, with highest concentration during fall and winter, while lowest in summer. Considering that the dissociation constant of ammonium nitrate directly depends on temperature and relative humidity, elevated concentration of secondary ammonium nitrate during the cold seasons is due to the increased partitioning of ammonium nitrate into the particle phase, favored by lower wintertime temperatures and higher relative humidity (Mozurkewich, 1993). This trend, however, was inverse in Rubidoux, with higher concentration during the summer compared to winter. This is most likely due to increased advection of ammonia from upwind Chino area, caused by stronger westerly/southwesterly wind speed during the summer in the LA Basin (Daher et al., 2013). Increased photochemical production of nitric acid, which reacts with fugitive ammonia, also leads to high concentration of ammonium nitrate in summer in this area (Hughes et al., 1999). The annual average trend for secondary ammonium nitrate shows a clear decrease in its concentration, with an approximate 60 and 70% reduction from 2002 to 2013 in Los Angeles and Rubidoux, respectively. This substantial reduction could be in part attributed to a major reduction in NOx emissions in the LA Basin (Fujita et al., 2013), following the implementation of emissions standards after 2007, which targeted NOx as well as PM emissions from diesel trucks. A detailed discussion on the yearly variation of ambient NOx concentrations is provided in the following sections. The secondary ammonium sulfate source profile is identified by high concentrations of SO4 2- and NH4 + and it accounts for, on average, 18 and 13% of PM2.5 in Los Angeles and Rubidoux, respectively. At both sites, the contribution of secondary ammonium sulfate displayed wintertime minima while summertime peaks, mainly due to increased photochemical activities during the summer. The contributions from this source also displayed major reductions over the years at both sites. Annual average source contributions of secondary ammonium sulfate showed 78 and 58 % reductions from 2002 to 2013, in Los Angeles and Rubidoux, respectively. The biomass burning source profile consisted primarily of EC, OC, and K+. On an average over all years, this source accounted for 6 and 4% of ambient fine PM in Los Angeles and Rubidoux, respectively. It can be inferred from Figure 4.12 a-b that at both sites the contribution of biomass burning was higher during the weekends compared to weekdays, suggesting that this source mostly represents residential wood burning. Figure 4.11 a-b illustrates that the contribution 88 of this source at both sites was highest during the winter. In Rubidoux, however, the summertime contribution of biomass burning was comparable to the wintertime level, suggesting a possible additional contribution from wildfires, which are typical in southern California during the hot and dry summer. Among all years, the highest contribution of biomass burning at both sites was observed in 2009. This is most likely because of the substantial contributions from the largest wildfire in the history of Los Angeles County, which occurred in late summer of 2009 (Wonaschütz et al., 2011). Soil source profile is identified on the basis of high levels of Al, Ca, Fe, Si, and Ti. This source, which respectively accounted for 6 and 5% of PM2.5 in Los Angeles and Rubidoux, did not show any appreciable seasonal variations. Fresh sea salt has high concentrations of Na + and Cl - . This source accounted for about 2 and 3% of ambient PM2.5 in Los Angeles and Rubidoux, receptively. This source showed the highest contribution during spring at both sites, consistent with increased southwesterly winds (Cheung et al., 2011). Aged sea salt is characterized by high loadings of Na + , SO4 2- , and NO3 - . Unlike fresh sea salt, this profile lacks the concentration of chlorine due to Cl - displacement by acidic gases during long range transport of sea salt aerosols (Song and Carmichael, 1999). Aged sea salt displayed a summer-high seasonal pattern at both sites. Vehicular emissions source profile in Los Angeles was identified by high loadings of EC and OC, as well as few elemental tracers such as Fe, Cu, and Zn. In Rubidoux, the PMF model was able to discern two separate source profiles for diesel and gasoline vehicles. These source profiles are characterized by high loadings of EC and OC, respectively. In Rubidoux, the EC/OC ratios in the resolved gasoline and diesel vehicles source profiles were 0.4 and 2.2, respectively. These results are in agreement with ratios reported in previous studies (Fujita et al., 1998; Heo et al., 2009; Liu et al., 2006; Watson et al., 1998). It should be noted that diesel vehicles operating at very low speed and in stop-and-go traffic usually produce similar EC/OC ratios to typical gasoline vehicles (Shah et al., 2004). This could explain the inability of PMF to unambiguously distinguish the gasoline and diesel emissions in Los Angeles, considering that downtown LA has typically high levels of traffic congestion, in which diesel vehicles cannot move at high speed. In addition, the diesel emissions source profile that was obtained in Rubidoux may represent only diesel vehicles driving in relatively constant speed in fluid traffic conditions and the diesel emissions 89 from stop-and-go traffic could be apportioned into the gasoline vehicles category at this site. To overcome this ambiguity and be able to compare the results with those obtained for Los Angeles, the contributions from diesel and gasoline vehicles were added together in Rubidoux and referred to as vehicular emissions. On an average basis over all years, vehicular emissions accounted for near 20% of ambient PM2.5 at both sites. They also displayed a strong seasonal and weekday/weekend variability with higher contributions during the winter as well as weekdays. A detailed discussion of year-to-year variations in the contributions from vehicular sources will be provided in the following sections. At Rubidoux, a source profile was deduced with high loadings of Zn, Pb, EC, and OC, which is most likely attributed to local industrial emissions in the surrounding areas. Similar source profile was also obtained in previous studies in this area (Kim et al., 2010; Kim and Hopke, 2007a). Even though this source overall contributed to less than 1% of measured PM 2.5 in Rubidoux, it showed very strong winter-high and weekday-high variations. The contribution of this source to total PM2.5 showed a clear reduction over the years, varying from 0.23 µg/m 3 in 2002 to 0.07 µg/m 3 in 2013 (i.e., more than a 70% reduction). 4.3.2. Comparison with previous studies To validate our results, the average contributions from the identified sources were calculated between 2003 and 2005, and compared to those reported in previous studies for the same sampling sites in LA and Rubidoux (Kim et al., 2010; Kim and Hopke, 2007a), as presented in Table 4.5 a-b. Kim et al. (2010) applied the EPA PMF model to STN data collected between 2003 and 2005, while Kim and Hopke (2007a) analyzed STN data, collected between 2001 and 2004, through the application of a two-way factor analysis PMF2 model (Paatero, 1997). Although both PMF2 and EPA PMF models are based on a weighted least-squares method, there are major algorithmic differences between these two models, as discussed in detail by Kim and Hopke (2007b). Similar sources were identified in all three studies, except for biomass burning which was not resolved in Los Angeles in the analysis by Kim and Hopke (2007a). Moreover, it should be noted that separate source profiles were identified for gasoline and diesel exhaust in both previous studies. However, for the purpose of comparison, the sum of these two sources was calculated and denoted as “vehicular emissions”. Overall, the estimated contributions, particularly from major sources of PM2.5, in this study were in good agreement with those reported by Kim et al. (2010). 90 For example, the average contributions of secondary nitrate, between 2003 and 2005, obtained in LA and Rubidoux in this study were about 8 and 15% higher than those reported by Kim et al. (2010), respectively. Vehicular emissions, which are the main focus of this study, showed only 15 and 10% differences compared to those obtained by Kim et al. (2010) in Los Angeles and Rubidoux, respectively. Table 4.5 a-b. Average source contributions (± standard error) to PM2.5 in a) Los Angeles and b) Rubidoux, obtained from different studies. Units are in µg/m 3 . Study Current study Kim et al. (2010) Kim and Hopke (2007a) Period 2003-2005 2003-2005 2001-2004 Software used EPA PMF EPA PMF PMF2 Secondary nitrate 6.28 ± 0.65 5.83 ± 0.75 6.39 ± 0.52 Secondary sulfate 5.15 ± 0.40 3.77 ± 0.40 4.49 ± 0.49 Vehicular emissions (Gasoline+Diesel) 3.62 ± 0.22 3.15 ± 0.19 6.63 ± 0.40 Biomass burning 1.07 ± 0.10 2.15 ± 0.22 - Soil 0.88 ± 0.07 1.97 ± 0.20 1.45 ± 0.18 Fresh sea salt 0.17 ± 0.04 0.96 ± 0.28 0.79 ± 0.15 Aged sea salt 2.68 ± 0.17 1.33 ± 0.13 0.69 ± 0.60 Study Current study Kim et al. (2010) Kim and Hopke (2007a) Period 2003-2005 2003-2005 2001-2004 Software used EPA PMF EPA PMF PMF2 Secondary nitrate 11.31 ± 0.73 9.79 ± 0.76 14.06 ± 0.67 Secondary sulfate 3.27 ± 0.17 3.65 ± 0.20 3.68 ± 0.17 Vehicular emissions (Gasoline+Diesel) 4.56 ± 0.21 4.15 ± 0.17 5.79 ± 0.18 Biomass burning 0.69 ± 0.05 1.30 ± 0.10 0.79 ± 0.06 Soil 0.82 ± 0.08 1.88 ± 0.16 1.86 ± 0.10 Fresh sea salt 0.50 ± 0.06 1.21 ± 0.15 1.71 ± 0.16 Aged sea salt 1.32 ± 0.08 1.80 ± 0.10 1.78 ± 0.11 Industrial 0.21 ± 0.02 0.37 ± 0.04 0.77 ± 0.06 a) b) 91 Higher discrepancies were observed when comparing our results to those obtained by Kim and Hopke (2007a). Excluding aged sea salt in Los Angeles, which showed the largest difference (more than 250%), the percent differences across sources ranged from 2 to 78 and 11 to 73%, respectively in Los Angeles and Rubidoux. These differences may be because: a) the averaging time periods are not exactly the same; b) the PMF algorithms that were used are not identical. Nonetheless, the overall agreement between our findings and those reported by Kim et al. (2010), during the same period of time and via similar PMF software, corroborates our results. 4.3.3. Emissions reduction from vehicular sources The daily-resolved source contributions of vehicular emissions, segregated by year, are presented in box plots for each site in Figure 4.13 a-b. In addition, the trends in the number of vehicles as well as the concentration of NOx, an important tracer of vehicular emissions particularly in the LAB (McDonald et al., 2012), were evaluated over the years from 2002 to 2013. Daily-averaged concentrations of NOx were acquired from the online database of the CARB, at the same sampling sites (Figure 4.15 a-d). Daily-averaged number of vehicles were obtained from the freeway performance measurement system (PeMS), operated by the California Department of Transportation (CalTrans). As shown in Figure 4.1, the sampling site in downtown Los Angeles is surrounded by three major freeways, namely I-110, I-5, and US 101. Therefore, the nearest vehicle detection stations (VDS) to the sampling site on each freeway (only one direction) were selected and the sum of daily-averaged number of vehicles from all three VDSs was calculated, as presented in Figure 4.16 a. In Rubidoux, the traffic data were obtained from the nearest VDS to the sampling site, located on the eastbound of State Route (SR) 60 (Figure 4.16 b). 92 Figure 4.13 a-d. Box plot of daily-resolved source contributions from vehicular sources, segregated by year, in a) Los Angeles and b) Rubidoux. Dotted lines represent annual arithmetic Figure 4.14 a-b. Box plot of daily-resolved source contributions from vehicular sources, segregated by year cluster, in a) Los Angeles and b) Rubidoux. Dotted lines represent arithmetic means. The black dots correspond to the 5 th and 95 th percentiles. Values above each box represent the median value ± 95% confidence interval (in µg/m 3 ) of each dataset. *p= 0.001, **p<0.001, statistical significant difference compared with (2002-2006) dataset, by Mann-Whitney rank sum test a) b) a) b) 93 As can be seen in Figure 4.13 a-b, during 2002 to 2004, the median and annual average values of the daily-resolved vehicular emissions source contributions were relatively constant and there was not any statistically significant difference between each two consecutive years, at both sites. During these years, the median values ranged from 2.70 to 2.9 µg/m 3 and 3.2 to 3.3 µg/m 3 in LA and Rubidoux, respectively. In Rubidoux, the highest annual average and median value of the daily-resolved source contributions from vehicular emissions were observed in 2005 (5.3 and 3.9 µg/m 3 , respectively). Thereafter, the median values trended downward and reached a minimum of 1.9 µg/m 3 in 2013 (a near 52% reduction compared with its corresponding value in 2005). In LA, however, the contributions from vehicular sources peaked in 2007 (annual average and median values were 4.2 and 3.6 µg/m 3 , respectively). Following 2007, the median and annual average values of the daily-resolved vehicular emissions source contributions displayed a continual decline until 2013, with an almost 70% reduction in the median value in 2013 compared to 2007. It should be noted that the increase in the median values from 2011 to 2012, in Los Angeles (Figure 4.13 a), was not statistically significant (p= 0.175). Reductions in PM2.5 emissions from vehicular sources is also supported by the steady decrease in median values of daily-averaged NOx concentrations from 2006 to 2013 in Los Angeles (~47% reduction) and Rubidoux (~34% reduction), as illustrated in Figures 4.15 a-b. It is noteworthy that in spite of reductions in PM2.5 emissions from vehicular sources as well as ambient NOx concentrations, the flow of vehicles overall increased after 2005 at both sampling sites (Figure 4.16 a-b). These observations strongly suggest that the reductions in the emissions from vehicular sources are mainly attributed to the implementation of state and local regulations on vehicular emissions. 94 Figure 4.15 a-d. Box plots of daily-averaged nitrogen oxides (NOx) (ppm), segregated by year and year cluster, in Los Angeles and Rubidoux. Dotted lines represent the annual arithmetic means. The black dots correspond to the 5 th and 95 th percentiles. Values above each box represent the median value ± 95% confidence interval (in ppm) of each dataset. *p<0.001, statistical significant difference compared with (2002-2006) dataset, by Mann-Whitney rank sum test a) b) c) d) Los Angeles Rubidoux 95 As pointed out earlier, starting 2007 MY, all manufactured diesel trucks have to comply with the U.S. EPA 2007 emission standards, which target PM and NOx emissions. In addition, following 2007, several major regulations by the CARB and ports of LA and Long Beach went into effect to reduce the PM emissions from diesel trucks, as discussed earlier. Considering 2007 as the transition year, to further evaluate the effect of the implementation of these regulations on PM emissions from vehicular sources, which in our case is the sum of contributions from both gasoline and diesel vehicles, source contribution estimates of vehicular emissions before and after 2007 were compared at both sampling sites. Therefore, for each site, daily-resolved source contributions of vehicular emissions from 2002 to 2006 were pooled together and then compared to the combination of 2008 to 2012 datasets (Figure 4.15 a-b). Given that the 2013 dataset does not cover a whole year and also to maintain consistency in the number of years investigated prior to and after the implementation of the standards starting in 2007, the results for 2013 were not included in this analysis. As shown in Figure 4.15 a-b, compared to 2002-2006 period, the median values of vehicular source contributions in 2008-2012 period dropped by 24 and 21%, in LA and Rubidoux, respectively. U-tests, between 2002-2006 (pooled together) and 2008-2012 (pooled together) datasets showed that the reductions in PM2.5 emissions from vehicular sources between these clusters of years were highly statistically significant in both LA (p= 0.001) and Rubidoux (p <0.001). Moreover, despite these reductions, the daily-averaged flow of vehicles in downtown LA displayed a small but statistically significant increase (p<0.001, ~5% increase in the median values) from 2002-2006 (pooled together) to 2008-2012 (pooled together), as shown in Figure 4.17 a, which further substantiates that these reductions in PM emissions from vehicular sources are mainly due to the implementation of the aforementioned standards after 2007. 96 Figure 4.16 a-b. Box plots of daily-averaged traffic flow (vehicles/day), segregated by year, in a) Los Angeles and b) Rubidoux. Dotted lines represent the annual arithmetic means. The black dots correspond to the 5 th and 95 th percentiles. Traffic data retrieved from the nearest vehicle detection stations (VDS) on I-110 (southbound), I-5 (northbound), US-101 (northbound), and SR-60 (eastbound). Figure 4.17 a-b. Box plots of daily-averaged traffic flow (vehicles/day), segregated by year cluster, in a) Los Angeles and b) Rubidoux. Dotted lines represent the annual arithmetic means. The black dots correspond to the 5 th and 95 th percentiles. Traffic data retrieved from the nearest vehicle detection stations (VDS) on I-110 (southbound), I-5 (northbound), US- 101 (northbound), and SR-60 (eastbound). Values above each box represent the median value ± 95% confidence interval of each dataset. *p<0.001, statistical significant difference compared with (2002-2006) dataset, by Mann-Whitney rank sum test a) b) a) b) 97 It is noteworthy that even though the economic crisis, which began in late 2007, resulted in the reduction of number of trucks serving the ports of LA and Long Beach (Lee et al., 2012), the total number of vehicles in downtown LA showed an overall increase after 2007 (Figure 4.17 a), suggesting the low impact of the economic recession on the number of vehicles driving in the LA area. The effectiveness of PM regulations is also noted at Rubidoux where a reduction in PM2.5 emissions from vehicular sources is observed in spite of the comparable vehicular flow in both periods (Figure 4.17 b). As noted earlier, the U.S. EPA 2007 emission standards target the NOx emissions from diesel trucks as well. A similar comparison between 2002-2006 and 2008-2012 datasets for daily-averaged ambient NOx concentrations (Figure 4.15 c,d) shows that the median values statistically significantly decreased (p<0.001) by more than 30% after 2007, at both sites, which further supports the effectiveness of the 2007 emission standards. Meteorological conditions were also taken into account and the yearly variations of important parameters, including temperature, relative humidity, and precipitation were evaluated, as shown in Figure 4.18 a-f. Overall, annual average temperatures and relative humidity were consistent from 2002 to 2013 over the basin (T= 18.1±0.7 o C, RH= 59.9±3.7%, averaged over both sites). On the other hand, relatively high year-to-year fluctuations were observed for the annual total precipitation with minima in 2007 and maxima in 2005. Nonetheless, the influence of changes in the meteorological conditions on the annually averaged levels and trends of vehicular emissions source contributions is expected to be negligible. 98 b) a) c) d) Los Angeles Rubidoux e) f) Figure 4.18 a-f. Box plots of daily-averaged temperature (T) and relative humidity, as well as annual total precipitation, segregated by year, in Los Angeles and Rubidoux. Dotted lines represent the annual arithmetic means. The black dots correspond to the 5 th and 95 th percentile. 98 99 These findings, altogether, indicate that although the number of vehicles (both LDV and HDV) increased or was relatively constant over the years after 2007, their emissions contributions to ambient PM2.5 generally declined, implying that this reduction is due to the implementation of stringent PM standards on vehicular emissions, particularly diesel trucks. As discussed earlier and shown in Table 4.1, starting January 2014, all in-use diesel trucks in the state of California must comply with the 2007 emission standards. As a result, more substantial reductions in PM and NOx emissions are expected beyond 2014. More recent studies in the LA Basin have also shown that the emissions reduction from vehicular sources, particularly diesel trucks, has been progressive after 2009, as a result of the gradual turnover of polluting trucks to cleaner ones, as well as the implementation of more stringent regulations after 2009. In a mobile measurement platform study, conducted between 2009 and 2011 on truck-dominated freeways in southern California, Kozawa et al. (2014) found NOx and black carbon (BC) emissions were reduced by about 40 and 70%, respectively. Between 2008 and 2012, Bishop et al. (2013) observed over 50% reduction in NOx emissions and infrared opacity (a measure of particulate matter), for trucks serving the Port of Los Angeles. Similarly, at the port of Oakland, another major shipping port in California, Kuwayama et al. (2013) reported a near 75% reduction in fine primary PM emissions (EC+OC) from heavy-duty diesel trucks after the implementation of the Comprehensive Truck Management Plan in 2010. 4.4. Summary and conclusions A long-term source apportionment study was conducted in the Los Angeles Basin. PM2.5 chemical composition data were obtained from the Speciation Trends Network (STN), collected in central Los Angeles and Rubidoux between 2002 and 2013. Positive Matrix Factorization (PMF) receptor model was utilized to identify and quantify major sources of ambient PM2.5 at both sampling sites. Vehicular emissions, which were found to be the second major contributor to ambient PM2.5, following secondary particles, displayed significant reductions after 2007, at both sampling sites, mainly because of the implementation of major federal, state and local regulations during this period of time. The PMF results revealed that the contributions from vehicular sources (including both diesel and gasoline emissions) peaked in 2007 and 2005, respectively in Los Angeles and Rubidoux, while trended downward afterward until 2013. In 2013, the median values of the daily-resolved source contributions from vehicular emissions in Los Angeles and Rubidoux 100 were respectively 70 and 52% lower than their corresponding values in 2007 and 2005. To evaluate the effect of the implementation of the U.S. EPA 2007 emission standards as well as several major regulations mandated by the CARB and ports of LA and Long Beach after 2007, the daily-resolved vehicular emissions source contributions from 2002 to 2006 were pooled together and then compared to the combination of 2008 to 2012 datasets. Statistically significant differences, using U-tests, were observed between the median values of these two datasets in both Los Angeles (24% reduction, p= 0.001) and Rubidoux (21% reduction, p<0.001). On the other hand, an overall increase (about 5%) was observed in the median values of the number of vehicles in downtown Los Angeles after 2007, while the number of vehicles was comparable before and after 2007, at Rubidoux. Overall, our findings underscore effectiveness of strict regulations in reducing PM emissions from vehicular sources in the LA Basin over the past almost decade. In addition, the results presented in this paper provide a baseline for projected additional reductions in coming years. 4.5. Acknowledgments The work presented in this paper was funded by the California Environmental Protection Agency (Cal EPA), Office of Environmental Health Hazard Assessment (OEHHA), under award number 12-E0021. 101 CHAPTER 5 Spatial and temporal variability of sources of ambient fine particulate matter (PM 2.5 ) in California This chapter is based on the following publication: Hasheminassab, S., Daher, N., Saffari, A., Wang, D., Ostro, B. D., & Sioutas, C. (2014). Spatial and temporal variability of sources of ambient fine particulate matter (PM2.5) in California. Atmospheric Chemistry and Physics, 14(22), 12085-12097. 102 5.1. Introduction Exposure to ambient airborne particulate matter (PM) is one of the leading causes of morbidity and mortality, contributing to more than 3 million premature deaths in the world annually, based on a recent global burden of disease study (Lim et al., 2012). PM inhalation has been linked to a wide range of adverse health effects, such as respiratory inflammation (Araujo et al., 2008), cardiovascular diseases (Delfino et al., 2005; Ostro et al., 2014), and most recently neurodegenerative and neurodevelopmental disorders (Davis et al., 2013a, 2013b). During the past few decades, California has been constantly suffering from high concentrations of ambient PM, among the highest levels recorded within the United States, with estimated rates of PM-related morbidity and mortality exceeding any other state in the country (Fann et al., 2012). Ambient PM in California originates from a large number of diverse sources (Hu et al., 2014) and is a complex mixture of different chemical components, the composition of which may change drastically with PM size (Hu et al., 2008b), location, and season (Cheung et al., 2011; Daher et al., 2013). Current PM regulations in California target PM10 and PM2.5 (particles with aerodynamic diameter less than 10 and 2.5 µm, respectively) mass concentrations, with PM2.5 being of major concern due to the higher rate of PM2.5-related morbidity and mortality in the state compared to PM10 (Ostro et al., 2006; Woodruff et al., 2006). These regulations only target PM mass concentration, regardless of their sources of emission and/or toxico-chemical characteristics. There is, however, strong evidence that the level of toxicity and health-related characteristics of PM are significantly affected by their chemical composition and therefore by their emission sources (Rohr and Wyzga, 2012; Saffari et al., 2013; Stanek et al., 2011; Zhang et al., 2008). Recently, there has been growing interest in using source apportionment data in epidemiological health studies (Laden et al., 2000; Mar et al., 2000; Ostro et al., 2011; Özkaynak and Thurston, 1987; Sarnat et al., 2008). These studies have provided significant evidence that exposure to PM from certain sources is linked to mortality. In a recent study in Barcelona, Ostro et al. (2011) found that exposure to several sources, including traffic emissions, sulfate from ship emissions and long- range transport, as well as construction dust, is statistically significantly associated with all-cause and cardiovascular mortality. Nonetheless, to draw firm conclusions and develop more effective control strategies to reduce population exposure to harmful sources of airborne PM, further epidemiological studies that use source apportionment data are warranted. 103 To date, several source apportionment studies have been conducted in California, using source-oriented (DeNero, 2012; Hu et al., 2014; Kleeman and Cass, 2001; Zhang et al., 2014) and receptor models (Ham and Kleeman, 2011; Hasheminassab et al., 2013; Hwang and Hopke, 2006; Kim et al., 2010; Kim and Hopke, 2007b; Schauer and Cass, 2000). Source-oriented models focus on the transport, dilution, and transformation of pollutants from the source of emission to the receptor site; thereby providing an overall estimation regarding the spatial distribution of source contributions. Receptor models, on the other hand, focus on the behavior of ambient environments at the point of impact (Hopke, 2003). Even though these studies have provided important insights on the characteristics of sources of ambient PM as well as their relative contributions, they have been mostly conducted in a limited number of sampling locations and/or within a relatively short period of time. As a result, spatial and temporal variability of the identified sources have not been extensively examined. For instance, Kim et al. (2010) analyzed the PM2.5 speciation data collected between 2003 and 2005 at two sampling sites in southern California (i.e., Los Angeles (LA) and Rubidoux) to identify and quantify major PM2.5 sources, by application of a PMF model. Using similar source apportionment approach, Hwang and Hopke (2006) evaluated the sources of ambient PM2.5 at two sampling sites in San Jose, using the STN data collected between 2000 and 2005. In a more comprehensive study, Chen et al. (2007) applied several receptor models to the chemically speciated PM2.5 measurements collected for one year (between 2000 and 2001) at 23 sites, all located in California’s San Joaquin Valley (SJV), to estimate PM2.5 source contributions. In this study, positive matrix factorization (PMF), one of the most widely-used receptor- oriented source apportionment techniques (Paatero and Tapper, 1994), was employed in order to provide a detailed and long-term (from 2002 to 2007) quantification of the contributions of different emission sources to ambient PM2.5 mass concentration in California, at 8 distinct locations spanning southern, central, and northern regions of the state. The association between PM-related mortality and PM2.5 mass concentration as well as individual PM2.5 chemical components has been investigated in previous epidemiological studies in California (Ostro et al., 2007, 2006). The results of this study will be used as an input for future epidemiological studies conducted by California Environmental Protection Agency (Cal EPA), in order to further expand the current epidemiological knowledge, by establishing the relationship between PM-related adverse health effects and specific source contributions. These findings will be crucial in establishing targeted and cost-effective regulations on PM2.5 emissions in the state of California. 104 5.2. Methodology 5.2.1. Sampling sites Sampling was conducted at eight Speciation Trends Network (STN) sampling sites, established by the United States Environmental Protection Agency (U.S. EPA), located in distinctly different cities all over California, including El Cajon, Rubidoux, Los Angeles, Simi Valley, Bakersfield, Fresno, San Jose and Sacramento. The studied sampling sites comprise a mixture of urban and semi-rural communities, with El Cajon and Rubidoux being located in semi- rural areas, while the rest of sampling sites being situated in densely developed urban regions of the state. Figure 5.1 shows the location of all sampling sites. Figure 5.1. Location of the sampling sites. 105 The Sacramento sampling site is located next to a park in a residential area with commercial establishments and high-density residential homes in the surrounding neighborhood. It is also about 3 km southeast of a major freeway (I-80). The sampling site in San Jose is located 46 km east of the Pacific Ocean and 14 km southeast of the San Francisco Bay. It is also surrounded by primary commercial facilities (Hwang and Hopke, 2006). Cities of Fresno and Bakersfield are located in California’s heavily SJV (Zhao et al., 2011). These two cities are relatively far from the Pacific Ocean and are mostly impacted by secondary aerosols formed by emissions from upwind areas (Ying and Kleeman, 2006a). Moreover, this part of the state usually suffers from severe particulate pollution, especially during the colder seasons (Kleeman et al., 2009). The northern parts of the SJV are dominated by agricultural activities, while the southern regions are mostly impacted by oil production (Held et al., 2004). The sampling site in Bakersfield is located about 6.5 km southwest of downtown, in a residential neighborhood and 2 km away from the nearest freeway (State Route (SR) 99). The sampling site in Fresno is about 5.5 km northeast of the downtown commercial district (Watson et al., 2000), next to a four-lane artery with moderate traffic level. Simi Valley is located 50 km northwest of downtown LA, in Ventura county, and the sampling site in this city is situated 500 m south of SR 118 (Kim and Hopke, 2007a). Two sampling locations in the South Coast Air Basin were considered in this study; Los Angeles and Rubidoux. The sampling site in downtown LA is surrounded by three major freeways (i.e., I-110, I-5, and US-101) and is 30 km away from the ports of LA and Long Beach, both of which are the busiest ports in the U.S. (Minguillón et al., 2008). This sampling site is therefore heavily impacted by primary emissions. Rubidoux is situated 60 km inland from downtown LA and is typically subject to aged and photo-chemically processed particulate plumes advected from upwind regions (Sardar et al., 2005). Previous studies have reported high concentration of ammonium nitrate in this region, which is mostly formed by the atmospheric reaction of nitric acid with ammonia from Chino dairy farms and livestock in upwind regions (Hughes et al., 1999). Lastly, the El Cajon sampling site is located in an inland valley, downwind of a heavily populated coastal zone, in San Diego County. This site is also impacted by emissions from I-8 freeway, situated 500 m to its north. 106 5.2.2. Sampling schedule and chemical analysis Time-integrated 24 h PM2.5 samples were collected between 2002 and 2007 at all sampling sites, except for LA and Rubidoux, at which the STN data collected from 2002 to 2013 was used as the input file when running the PMF model (Hasheminassab et al., 2014). In the present study, in order to compare the results with those obtained for the rest of sampling sites, we calculated the average source contributions between 2002 and 2007 from the output of the same PMF runs which were originally conducted using the 2002-2013 chemical dataset. By performing a sensitivity analysis, Hasheminassab et al. (2014) showed that the results of the PMF model performed on the entire chemical dataset (i.e., 2002-2013) is comparable to the output of the PMF model conducted separately on 2002-2006 and 2008-2012 datasets, in terms of the sources identified (similar number of sources with almost identical compositions) and the absolute source contributions (less than 18% difference in average source contributions among all sources). The outcome of the sensitivity analysis thus indicated that the daily-resolved source contributions between 2002 and 2007 are not significantly biased when the chemical data between 2008 and 2013 are also included into the PMF input file. During the studied period (i.e., 2002 to 2007), PM2.5 samples were collected every third day in Sacramento, San Jose, Fresno, Bakersfield, Rubidoux, and El Cajon sites, while every sixth day in Simi Valley and Los Angeles sites. Filter weighing and chemical analyses were performed according to the U.S. EPA Quality Assurance Project Plan (QAPP) (EPA-454/R-01-001) adopted for the STN field sampling. According to the QAPP, filters were tested, equilibrated, and weighted in the U.S. EPA contract laboratories, and then they were shipped to the field. After sampling, filters bearing PM2.5 deposits were promptly shipped back to the laboratories for weight determination and other chemical analyses. PM2.5 mass concentration was determined gravimetrically by pre- and post-weighing the Teflon filters. Concentration of elements on Teflon filter samples was quantified by energy- dispersive X-ray fluorescence (ED-XRF) (RTI, 2009). Major ions, including nitrate, sulfate, ammonium, sodium, and potassium, were measured by Ion Chromatography (IC) (RTI, 2009b, 2009c). Elemental carbon (EC) and organic carbon (OC) were quantified from quartz filters, using Thermal Optical Transmittance (TOT) NIOSH 5040 carbon method (Birch and Cary, 1996). 107 5.2.3. Source apportionment In this study, the EPA PMF receptor model (version 3.0.2.2) was performed at each sampling site separately to identify the major sources of ambient PM2.5 and quantify their relative contributions to total PM2.5 mass. PMF is a factor analysis model that solves the chemical mass balance equations using a weighted least-squares algorithm and by imposing non-negativity constrains on the factors (Reff et al., 2007). 5.2.3.1.Data screening The first step of data screening was correcting the OC data to account for sampling artifacts, caused by adsorption and/or desorption of organic vapors on quartz filters (Chow et al., 2010). For each sampling site, the OC artifact was estimated using the intercept of the linear regression of OC against PM2.5 mass concentration (Kim et al., 2005). OC concentrations were then corrected by subtracting the OC artifact concentrations. The estimated OC artifact values (± standard errors) at each site are presented in Table 5.1. The estimated OC artifacts may have some time variability. In our study, however, the variations of OC artifact among different years were negligible and for this reason for each sampling site we used an average OC artifact value for the entire study period (the approach similar to most of the past long-term PMF investigations). To further clarify this point, we picked the sampling site with the maximum estimated OC artifact (i.e., Los Angeles) and evaluated the year-to-year variability of the artifact values. At this sampling site, the annual average concentration of uncorrected OC (i.e., comprising the artifact) decreased from near 7µg/m 3 in 2002 to about 5µg/m 3 in 2007 (Figure 5.2). Figure 5.3 shows the scatter plot of OC TOT versus measured PM2.5 mass concentration, in which the data points corresponding to each year are specified with a distinct marker/color. As described in the paper, OC artifact is estimated using the intercept of the linear regression of OC against PM2.5 mass concentration, following the method of Kim et al. (2005). Table 5.2 presents the estimated OC artifacts (±standard errors), segregated by year, in Los Angeles. As can be seen, the OC artifact values do not show a significant year-to- year variability. This can be inferred from the levels of significance corresponding to the two- tailed t tests performed on the OC artifact values between each two consecutive years, as shown in Table 5.2 (p values ranging from 0.23 to 0.69), indicating that there is no statistically significant difference between OC artifacts among different years. Furthermore, the maximum difference between the individual estimated OC artifact and the average value calculated over all 6 years (i.e., ~3.7 µg/m 3 ) is less than 17% of the average value. Considering these observations, for each 108 sampling site we estimated one OC artifact, using the interpret of OC TOT vs PM2.5, over the entire study period (i.e., 2002-2007), similarly to the approach taken in many other studies in the literature, using the PM2.5 chemical speciation data over a long period of time (Hwang and Hopke, 2006; Kim and Hopke, 2008). Table 5.1. Estimated OC artifact values (µg/m 3 ) from Thermal Optical Transmittance (TOT) NIOSH 5040 carbon method, at each sampling site. Errors correspond to one standard error. Monitoring site OC artifact value (µg/m 3 ) El Cajon 1.84 ± 0.23 Rubidoux 3.33 ± 0.13 Los Angeles 3.61 ± 0.22 Simi Valley 2.30 ± 0.15 Bakersfield 3.24 ± 0.14 Fresno 2.02 ± 0.15 San Jose 1.09 ± 0.12 Sacramento 0.83 ± 0.13 Table 5.2. Estimated OC artifacts (±standard errors) and the p values corresponding to the two-tailed t tests between OC artifact values in each two consecutive years in Los Angeles. Errors correspond to one standard error. Year Estimated OC artifact (µg/m 3 ) p value 2002 4.37 ± 1.08 0.51 2003 3.69 ± 0.48 0.57 2004 4.08 ± 0.47 0.23 2005 3.32 ± 0.41 0.53 2006 3.70 ± 0.45 0.69 2007 3.32 ± 1.04 109 Figure 5.2. Annual average concentration (µg/m 3 ) of uncorrected organic carbon (OC) from 2002 to 2007 in Los Angeles. Error bars correspond to one standard error. Figure 5.3. Scatter plot of OC mass concentration, obtained from Thermal Optical Transmittance (TOT) NIOSH 5040 method, versus PM2.5 mass concentration in Los Angeles, segregated by year. 0 2 4 6 8 10 12 14 16 0 20 40 60 80 OC TOT (µg/m 3 ) PM 2.5 (µg/m 3 ) 2002 2003 2004 2005 2006 2007 4 4.5 5 5.5 6 6.5 7 7.5 2002 2003 2004 2005 2006 2007 OC TOT (µg/m 3 ) year 110 To avoid double-counting of species, the linear correlations in each pair of S/SO4 2- , Na/Na + , and K/K + were examined. Depending on the goodness of fit and the percent number of samples below detection limit (BDL) (threshold of 70%), either IC SO4 2- , Na + , K + or ED-XRF S, Na, K data were included in the PMF analyses. Measured BDL concentrations were replaced by half of the detection limit (DL) values, and their uncertainties were set as 5/6 of the DL values (Polissar et al., 1998). Missing values were replaced by the geometric mean of the existing concentrations, and their accompanying uncertainties were set as four times this geometric mean concentration. Species with more than 70% BDL values as well as samples with missing mass and/or all of the elemental concentrations were excluded from the model. Lastly, occasional samples with unusually high concentrations of a few chemical species, such as those collected around July 4th and/or New Year eves with extremely high concentrations of K and/or K + were discarded. 5.2.3.2.PMF model The uncertainties used in the PMF model were the estimated uncertainties reported in the Air Quality System (AQS) for the PM2.5 chemical speciation network. The uncertainties reported by STN include both the analytical uncertainties and uncertainties associated with the field sampling component (Flanagan et al., 2006). The uncertainties of elements, measured by the ED- XRF method, go through a comprehensive calculation procedure that harmonizes the uncertainties between different instruments and accounts for filter matrix effect, in addition to the field sampling and handling uncertainty (Gutknecht et al., 2010). For the other species, uncertainty is estimated as the analytical uncertainty of the instrument, augmented by 5% of the calculated concentration, assuming that this 5% is representing the total “field” variability (Flanagan et al., 2006). Species with a signal-to-noise (S/N) ratio between 0.2-2, as well as those that have BDL values more than 50% of total samples were considered as weak variables and their uncertainties were increased by a factor of 3. In order to directly apportion the total PM mass, PM2.5 mass concentrations were included in the data matrix as a “total variable” in the PMF model (Lee et al., 2011). To ensure that the inclusion of total PM mass concentration does not affect the resulting PMF solution, their uncertainties were increased by a factor of 3, similarly to a weak variable (Reff et al., 2007). The model was performed in the default robust mode to diminish the influence of extreme values on the PMF solution, and the FPEAK parameter was applied to control rotational 111 ambiguity (Paatero et al., 2002). Furthermore, a value of 5% extra modeling uncertainty was applied. Uncertainties in the source profiles were estimated by a bootstrap procedure (Norris et al., 2008). 500 runs were considered for the bootstrap analysis in this study, and a solution was considered valid when the occurrence of unmapped factors was less than 10% of the total runs. The final solutions were chosen based on the evaluation of the deduced source profiles and the quality of the chemical species fits by testing different numbers of factors. 5.3. Meteorology Select meteorological parameters data, including temperature, relative humidity (RH), precipitation, as well as vector-average wind speed and direction were acquired from the online database of the California Air Resources Board (CARB). Table 5.3 presents the seasonal averages of these parameters at all studied sampling sites. In this study, seasons were defined as spring (March–May), summer (June–August), fall (September–November) and winter (December–February), and seasonal/annual averages of all parameters reported in the following sections and shown in the figures and tables were calculated over all 6 years (i.e., 2002 to 2007). In addition, the standard errors accompanying the seasonal averages were calculated based on all daily-resolved source contributions that fall within a given season. Lastly, in all of the figures and tables presented in this study, sampling sites were ranked according to their latitude, from south to north (i.e., from El Cajon to Sacramento). 112 Table 5.3. Select meteorological parameters at each sampling site during spring, summer, fall, and winter. Seasonal averages were calculated over 6 years (from 2002 to 2007). Sampling Site Season Temperature ( o C) Relative Humidity (%) Precipitation (cm) Wind (vector-average) Average ± Stdev Average ± Stdev Average of yearly total ± Stdev Speed(mph) (% calm) Prevailing direction El Cajon Spring 15.8 ± 2.8 73.3 ± 13.5 N/A a 2.1 (33.9) SW Summer 21.6 ± 2.7 76.8 ± 9.8 N/A 2.4 (32.4) SW Fall 18.1 ± 3.7 71.0 ± 16.1 N/A 1.4 (49.0) SW Winter 12.1 ± 2.2 69.6 ± 17.3 N/A 0.7 (57.8) SW Rubidoux Spring 17.1 ± 3.7 67.9 ± 17.5 5.0 ± 3.9 2.5 (20.0) W Summer 24.7 ± 3.3 65.4 ± 12.4 0.1 ± 0.2 3.4 (16.1) W Fall 19.7 ± 4.7 59.9 ± 21.0 2.4 ± 2.4 1.5 (24.1) NW Winter 13.4 ± 3.1 57.5 ± 24.0 11.1 ± 10.8 1.3 (23.1) N Los Angeles Spring 17.0 ± 3.1 59.2 ± 17.0 7.5 ± 6.6 2.4 (2.8) SW Summer 22.3 ± 2.4 53.3 ± 25.0 0.0 ± 0.0 4.0 (2.1) SW Fall 19.2 ± 3.7 50.8 ± 26.6 5.6 ± 3.3 0.6 (3.3) W Winter 14.1 ± 2.7 54.1 ± 19.6 21.7 ± 17.1 1.2 (2.9) NE Simi Valley Spring 14.8 ± 3.7 60.4 ± 18.6 5.5 ± 6.9 1.9 (18.1) W Summer 20.5 ± 3.1 58.5 ± 14.9 0.0 ± 0.0 3.0 (17.6) W Fall 18.0 ± 4.4 53.0 ± 22.6 2.3 ± 2.2 0.8 (25.3) NW Winter 13.3 ± 3.8 50.3 ± 24.7 14.1 ± 13.0 1.7 (19.3) NE Bakersfield Spring 18.2 ± 4.8 53.9 ± 13.4 3.8 ± 3.4 2.3 (1.8) NW Summer 28.3 ± 3.2 39.5 ± 7.3 0.0 ± 0.0 2.8 (1.4) NW Fall 18.7 ± 5.9 58.2 ± 15.3 2.0 ± 1.5 1.1 (3.6) NW Winter 10.2 ± 3.1 73.4 ± 13.0 7.1 ± 3.6 0.4 (3.3) NW Fresno Spring 17.3 ± 4.8 58.0 ± 13.0 9.1 ± 7.0 3.0 (1.4) NW Summer 27.1 ± 3.3 42.1 ± 7.0 0.3 ± 0.5 4.4 (0.9) NW Fall 18.3 ± 5.7 59.6 ± 14.8 2.7 ± 3.1 1.8 (3.6) NW Winter 9.7 ± 2.9 77.9 ± 10.1 10.7 ± 3.8 0.2 (2.9) E San Jose Spring 14.4 ± 2.9 66.3 ± 9.1 8.3 ± 5.3 N/A N/A Summer 19.5 ± 2.5 65.7 ± 7.7 0.1 ± 0.2 N/A N/A Fall 16.1 ± 3.5 63.9 ± 12.9 3.7 ± 1.7 N/A N/A Winter 10.5 ± 2.4 75.2 ± 10.2 16.3 ± 3.9 N/A N/A Sacramento Spring 15.4 ± 4 67.7 ± 12.7 11.7 ± 7.4 1.9 (9.2) S Summer 23.4 ± 3.3 55.6 ± 10.2 0.5 ± 0.8 3.5 (5.3) S Fall 16.3 ± 4.9 67.6 ± 15.7 5.9 ± 4.4 0.9 (22.7) SW Winter 9.0 ± 2.7 78.6 ± 22.7 23.4 ± 7.1 0.6 (18.7) S a N/A, data was not available. 113 Most intense seasonality in temperature and RH was observed at the inland areas of the SJV, in Fresno and Bakersfield. These two sites experience the hottest and driest summertime weather across the state (temperature over 25 o C and RH below 40%), while during winter, the mean temperature in these cities is within the lowest levels among all sites (below 10 o C) and the RH reaches about 75%, comparable to levels in other sites in the northern region of the state (i.e., San Jose and Sacramento). Unlike northern areas, RH exhibited more moderate seasonality in southern California, displaying minima in fall/winter (50-71%) and maxima in spring/summer (59- 77%). At all sampling locations, the average of yearly total precipitation was negligible in summer, but greatest in winter. During the studied period, Sacramento showed the highest total precipitation in winter, followed by LA, San Jose, and Simi Valley (23.4±7.1, 21.7±17.1, 16.3±3.9, and 14.1±13.0 cm, respectively). Additionally, wind speeds were generally much stronger in summer compared with fall/winter. During spring and summer, wind blows mostly from coast to inland in the southern part of the state (i.e., El Cajon, Rubidoux, LA, and Simi Valley), with a predominant westerly/southwesterly direction, while it shifts in winter and has a predominantly northerly origin at all sites, with the exception of El Cajon. In Bakersfield and Fresno, wind constantly blows from northwest throughout the year, except for Fresno in winter, when wind has an easterly direction. Lastly, in Sacramento, the prevailing wind direction is southerly/southwesterly throughout the year. 5.4. Results and discussion 5.4.1. Particulate mass Seasonal average mass concentration of ambient PM2.5 at each sampling site is presented in Table 5.4. Overall, mass concentrations spanned a broad range of 8.2 to 36.6 µg/m 3 across the studied sites and all seasons. PM2.5 mass concentration showed a very strong seasonality in central and northern parts of the state (i.e., Bakersfield, Fresno, San Jose, and Sacramento), with 2 to 4 times higher concentrations in winter compared with summer. This trend is typical of the California’s Central Valley, which usually experiences the most severe particulate pollution during winter in the U.S. (Ying and Kleeman, 2009). In winter, ambient PM2.5 mass concentrations peaked at Bakersfield and Fresno (32.0±1.8 and 36.6±1.5 µg/m 3 , respectively). Severe stagnation periods and decreased mixing height are mostly responsible for elevated particulate pollution during winter in this part of the state. As it will be discussed in the following section, secondary ammonium 114 nitrate and emissions from biomass burning were mainly responsible for elevated PM 2.5 mass concentrations in these two cities during winter. In summer, on the other hand, highest mass concentrations were observed in sampling sites located in the Los Angeles Basin (i.e., LA and Rubidoux). Rubidoux displayed highest mass concentration in fall, followed by summer and spring. In addition to local sources, this region of the state is typically subject to transported plumes from upwind regions in west and central LA (Daher et al., 2013; Sardar et al., 2005), particularly during the warm seasons when the westerly wind prevails (Table 5.3). Table 5.4. Seasonal average mass concentration (± standard error) (µg/m 3 ) of ambient PM2.5 at the 8 sampling sites in the period between 2002 and 2007. El Cajon Rubidoux Los Angeles Simi Valley Bakersfield Fresno San Jose Sacramento Spring 12.0 ± 0.5 23.6 ± 1.3 18.1 ± 1.5 12.8 ± 0.8 11.8 ± 0.5 16.4 ± 1.1 9.7 ± 0.4 8.2 ± 0.3 Summer 13.1 ± 0.4 25.6 ± 0.9 20.2 ± 0.7 15.9 ± 0.5 13.5 ± 0.4 9.7 ± 0.3 9.6 ± 0.4 9.2 ± 0.4 Fall 14.5 ± 0.5 27.4 ± 1.5 20.8 ± 1.2 14.4 ± 0.9 24.6 ± 1.7 13.7 ± 0.6 14.8 ± 0.8 15.1 ± 0.9 Winter 17.1 ± 0.7 20.0 ± 1.1 20.4 ± 1.6 9.8 ± 0.8 32.0 ± 1.8 36.6 ± 1.5 18.6 ± 1.2 23.5 ± 1.2 5.4.2. Source characterization and apportionment 5.4.2.1.Overview Between five to nine particle sources were identified at each sampling site. Resolved source profiles along with the explained variation (EV) of each species are shown in Figure 5.4 a-h, for all studied sampling sites. Gray bars represent the normalized concentration of each species to the mass concentration of PM2.5 apportioned to that factor, while the black dots represent the percent of each species apportioned to that factor (Lee et al., 1999). Table 5.5 summarizes the marker species which were used to identify each source profile. Several sources, including secondary ammonium nitrate, secondary ammonium sulfate, vehicular emissions, biomass burning, soil, fresh and aged sea salt were commonly identified at multiple sites. Few minor sources were exclusively identified at some of the sites, depending on the site location and nearby emission sources. These sources, however, accounted for a small fraction of the total mass (1 to 15% across the state, on an annual average basis). 115 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Soil 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Fresh sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Aged sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Vehicular emissions 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Copper smelters a) El Cajon 116 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Fresh sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Diesel vehicles 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Aged sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Mixed Industrial 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Gasoline vehicles 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Pb Mg Mn Ni Si K+ NA+ Ti V Zn Explained variation (%) Concentration (µg/µg) Soil b) Rubidoux 117 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Fresh sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Vehicular emissions 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Soil 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cr Cu Fe Pb Mg Mn Ni Si K+ NA+ Sr Ti V Zn Explained Variation (%) Concentration (µg/µg) Aged sea salt c) Los Angeles 118 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ NA+ Ti Zn Explained variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ NA+ Ti Zn Explained variation (%) Concentration (µg/µg) Vehicular emissions 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ NA+ Ti Zn Explained variation (%) Concentration (µg/µg) Aged sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ NA+ Ti Zn Explained variation (%) Concentration (µg/µg) Fresh sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ NA+ Ti Zn Explained variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ NA+ Ti Zn Explained variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Br Ca Cl Cu Fe Mg Mn Ni Si K+ NA+ Ti Zn Explained variation (%) Concentration (µg/µg) Soil d) Simi Valley 119 120 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn Si K+ Ti Zn Explained Variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn Si K+ Ti Zn Explained Variation (%) Concentration (µg/µg) Chlorine sources 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn Si K+ Ti Zn Explained Variation (%) Concentration (µg/µg) Sulfate-bearing road dust 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn Si K+ Ti Zn Explained Variation (%) Concentration (µg/µg) Vehicular emissions 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn Si K+ Ti Zn Explained Variation (%) Concentration (µg/µg) Secondary nitrate e) Fresno 121 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC S O 4… NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Fresh sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Vehicular emissions 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Aged sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Ni-related industrial sources 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC SO42- NO3- NH4+ Al Ba Br Ca Cl Cr Cu Fe Pb Mg Mn Ni K Si Na+ Ti V Zn Explained Variation (%) Concentration (µg/µg) Soil f) San Jose 122 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn K+ Na+ Si Ti Zn Explained variation (%) Concentration (µg/µg) Vehicular emissions 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn K+ Na+ Si Ti Zn Explained variation (%) Concentration (µg/µg) Secondary sulfate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn K+ Na+ Si Ti Zn Explained variation (%) Concentration (µg/µg) Soil 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn K+ Na+ Si Ti Zn Explained variation (%) Concentration (µg/µg) Biomass burning 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn K+ Na+ Si Ti Zn Explained variation (%) Concentration (µg/µg) Secondary nitrate 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn K+ Na+ Si Ti Zn Explained variation (%) Concentration (µg/µg) Aged sea salt 0 20 40 60 80 100 0.0001 0.001 0.01 0.1 1 OC EC NH4 NO3 SO4 Al Br Ca Cl Cu Fe Mn K+ Na+ Si Ti Zn Explained variation (%) Concentration (µg/µg) Chlorine sources g) Sacramento Figure 5.4 a-h. PM2.5 source profiles and explained variation (EV) of each species in a) El Cajon, b) Rubidoux, c) Los Angeles, d) Simi Valley, e) Bakersfield, f) Fresno, g) San Jose, and f) Sacramento. Error bars correspond to one standard deviation obtained from bootstrap analysis. 122 123 Table 5.5. Summary of the marker species for identified PM2.5 sources, resolved by the PMF model. Source Marker species Vehicular emissions EC, OC, Fe, Cu, Zn, Pb, Mn Secondary ammonium nitrate NO 3 + , NH 4 + Secondary ammonium sulfate SO 4 2- , NH 4 + Soil Al, Si, Ca, Fe, Ti Fresh sea salt Na + , Cl - Aged sea salt Na + , NO 3 + , SO 4 2- Biomass burning EC, OC, K/K + Copper smelters Cu, EC Mixed industrial EC, OC, Zn, Pb Chlorine sources Cl - Sulfate-bearing road dust EC, OC, SO 4 2- ,Fe, Ca, Mn, Si, Ti Ni-related industrial sources Ni, Mn, Mg Table 5.6. Summary statistics of the linear regressions between daily-resolved measured ambient PM2.5 and estimated PM2.5 mass concentrations obtained from the PMF model. Errors correspond to one standard error. R 2 Slope Intercept (µg/m 3 ) El Cajon 0.85 0.91 ± 0.02 0.89 ± 0.26 Rubidoux 0.96 0.91 ± 0.01 1.30 ± 1.22 Los Angeles 0.86 0.88 ± 0.02 1.58 ± 0.47 Simi Valley 0.91 0.91 ± 0.02 0.84 ± 0.23 Bakersfield 0.95 0.91 ± 0.01 0.95 ± 0.24 Fresno 0.94 0.91 ± 0.01 1.01 ± 0.23 San Jose 0.88 0.85 ± 0.01 1.35 ± 0.23 Sacramento 0.91 0.83 ± 0.01 1.47 ± 0.18 124 Table 5.6 presents the slope, intercept, and R 2 of the linear regressions between daily- resolved measured ambient PM2.5 and estimated PM2.5 mass concentrations, calculated by the sum of PM mass apportioned to each identified factor. It can be inferred that the PMF model was able to effectively estimate the measured PM2.5 mass concentrations at all sites (slope varying from 0.83 to 0.91 and R 2 ranging from 0.85 to 0.96). Year-to-year variability in the source contributions was overall quite small for almost all identified sources. This can be deduced from the relatively small standard errors in the 6-year seasonal average source contributions (median relative standard error of 8%, across all sites, seasons, and sources). Identified sources, on the other hand, displayed distinct seasonal and spatial variability. The percent contributions from these sources to PM2.5 mass are presented in Figure 5.5. Overall, secondary aerosols (including secondary ammonium nitrate and ammonium sulfate) collectively comprised the largest fraction of ambient PM2.5 at all sampling sites (except for San Jose), accounting for 26 to 63% of total mass across all sites, on an annual average basis. Vehicular emissions were the second major contributor to PM2.5 at all sites (11 to 25% annual average contribution, across the state), except for San Jose and Fresno, at which biomass burning was the dominant primary source of PM2.5 (35 and 27% annual average contribution, respectively). “Other sources” in Figure 5.5 are associated with those sources which were exclusively identified at some specific locations. These contributed to < 15% of the mass, on an annual average basis. The unapportioned mass, which is the difference between the seasonal average PM2.5 mass and the sum of the seasonal average source contributions from each factor, accounted for 3 to 6% of total mass across the state, on an annual average basis. The unapportioned mass represents the fraction that could not be resolved by the model. 125 Figure 5.5 a-d. Seasonal variation in the percent contribution of identified sources to ambient PM2.5, by site. 126 5.4.2.2.Vehicular emissions Vehicular emissions source profiles were identified by high concentrations of carbonaceous species (i.e., EC and OC). Elevated loadings of several non-exhaust PM tracers (e.g., Fe, Cu, Zn, Pb, Mn) indicate that these sources are affected by particles emitted from brake and tire wear, road surface abrasion, and resuspension of road surface dust (Dall’Osto et al., 2014; Pant and Harrison, 2013). Only at Rubidoux, the PMF model was able to determine two separate source profiles for diesel and gasoline vehicles (Figure 5.4 b). These source profiles are characterized by high loadings of EC and OC, respectively, with EC/OC ratios being 0.4 in gasoline source profile, while 2.2 in diesel vehicles source profile. These ratios are within the ranges reported in previous studies (Fujita et al., 1998; Heo et al., 2009; Liu et al., 2006; Watson et al., 1998). Diesel vehicles operating at very low speed and in stop-and-go traffic usually produce similar EC/OC ratios to typical gasoline vehicles(Shah et al., 2004). As a result, the diesel emissions source profile that was obtained in Rubidoux may represent only diesel vehicles driving in relatively constant speed in fluid traffic conditions and the diesel emissions from stop-and-go traffic could be apportioned into the gasoline vehicles category. To overcome this uncertainty and also be able to compare the results with those obtained at other sampling sites, the contributions from diesel and gasoline vehicles were combined together at Rubidoux and referred to as vehicular emissions throughout the discussion. As can be seen in Figure 5.6, across the state, estimated PM2.5 mass attributed to vehicular sources (including diesel and gasoline vehicles) displayed highest levels at Rubidoux, LA, and Sacramento, with annual average (±standard error) contributions of 4.3±0.1, 3.6±0.1, and 3.5±0.1 µg/m 3 , respectively. Spatial pattern of PM2.5 emissions from mobile sources across the state is in a good agreement with the findings of a recent study by Hu et al. (2014), in which they applied a source-oriented air quality model to predict primary PM2.5 source contributions across the state of California between 2000 and 2006. Vehicular emissions displayed similar seasonal patterns at all sampling sites, with higher contributions in fall and winter compared to spring and summer. In spring, summer, and fall, highest vehicular emissions source contributions were observed at Rubidoux. In contrast, during winter, when particulate pollution is confined within the emission area due to higher atmospheric stability and lower mixing height, vehicular source contribution exhibited the highest value in downtown LA. 127 Figure 5.6. Seasonal average source contribution (µg/m 3 ) of vehicular emissions to ambient PM2.5, by site. Error bars correspond to one standard error. This trend is typical of the LA Basin, in which downwind “receptor” areas are generally impacted by emissions from upwind “source” regions, when westerly/south-westerly onshore winds prevail (Table 5.3) (Daher et al., 2013). Several previous studies have reported similar trends in the LA Basin (Hasheminassab et al., 2013; Heo et al., 2013). It should be noted that after 2007 until 2012, contributions of vehicular emissions to ambient PM2.5 in the LA Basin statistically significantly decreased by 20 to 25%, following the implementation of major federal, state, and local regulations on vehicular emissions, particularly on diesel trucks (Hasheminassab et al., 2014). Among the studied locations in the California’s Central Valley, vehicular emissions displayed the highest levels in Sacramento, while lowest in San Jose, accounting for nearly 30 and 10% of total mass, respectively, on an average over 6 years. Vehicular emissions were comparable at Bakersfield and Fresno during spring and summer, whereas levels were slightly higher at 0 1 2 3 4 5 6 7 Mass concentration (µg/m 3 ) Vehicular emissions Spring Summer Fall winter 128 Bakersfield in fall and winter. Schauer and Cass (2000) conducted a 4-day sampling in Bakersfield during the winter of 1995 to quantify the sources of ambient PM2.5, using chemical mass balance receptor model. Average wintertime level of vehicular emissions in our study at Bakersfield (3.0±0.2 µg/m 3 ) was about half of that reported by Schauer and Cass (2000) (6.3±0.4 µg/m 3 ), whereas the percent contributions of this source to total mass were comparable in both studies (10 and 12%, respectively). This finding suggests that vehicular emissions have decreased by almost half after almost a decade in Bakersfield. 5.4.2.3.Secondary aerosols Secondary ammonium nitrate source profile was identified by high concentrations of NO3 - and NH4 + (Figure 5.4 a-h). Its contribution ranged from 0.2 to 16.8 µg/m 3 , accounting for 3 to 55% of ambient PM2.5 mass, among all sites and seasons, as displayed in Figure 5.7. Seasonally, the contribution of secondary ammonium nitrate was largest in winter while lowest during summer, with statewide average contribution of 8.4 and 3.2 µg/m 3 , respectively. Elevated concentration of secondary ammonium nitrate during the cold seasons is mainly due to the increased partitioning of ammonium nitrate into the particle phase, favored by lower wintertime temperatures and higher RH (Ying, 2011). This source displayed considerably higher contribution at Fresno and Bakersfield in winter (16.8±1.3 and 15.8±1.0 µg/m 3 , respectively). Ying and Kleeman (2006) stated that diesel engines and catalyst equipped gasoline vehicles are important local sources that contribute to secondary nitrate in the SJV. Unlike all other sites, the seasonal trend of secondary ammonium nitrate was reverse at Rubidoux, with higher concentration in summer compared to winter (12.5±0.8 and 8.9±0.8 µg/m 3 , respectively). This is probably due to increased advection of ammonia from the upwind Chino area, caused by stronger westerly/southwesterly wind speed during summer in the LA Basin (Hasheminassab et al., 2013) combined with the increased photochemical production of nitric acid in summer, which reacts with fugitive ammonia to produce high concentrations of ammonium nitrate in summer in this area (Hughes et al., 2002). 129 Figure 5.7. Seasonal average source contribution (µg/m 3 ) of secondary ammonium nitrate to ambient PM2.5, by site. Error bars correspond to one standard error. The characterized secondary ammonium sulfate source profiles have high loadings of SO4 2- and NH4 + (Figure 5.4 a-h). This source was identified at all sites, except at Fresno, where sulfate largely partitioned in a source named “sulfate-bearing road dust” along with a few other components, which will be discussed in further detail below. Annual average contributions of this source ranged from 1.3 to 4.6 µg/m 3 (or 10 to 24% of total mass) among all sites, indicating that this source is a smaller contributor to total mass compared with secondary ammonium nitrate. Secondary ammonium sulfate exhibited a similar seasonal trend at all monitoring sites, displaying wintertime minima while summertime peaks due to increased photochemical activity that forms this species. Levels were also overall higher in the southern part of the state, compared to the upper regions (Figure 5.8). As argued by Ying and Kleeman (2006), the majority of secondary aerosols 0 2 4 6 8 10 12 14 16 18 20 Mass concentration (µg/m 3 ) Secondary nitrate Spring Summer Fall winter 130 formed in southern California are formed from locally emitted precursors, whereas in the SJV secondary PM is mostly impacted by emissions from upwind areas (i.e., regional sources). Figure 5.8. Seasonal average source contribution (µg/m 3 ) of secondary ammonium sulfate to ambient PM2.5, by site. Error bars correspond to one standard error. 5.4.2.4.Biomass burning Identified biomass burning source profiles consisted primarily of EC, OC, and either K or K + (Figure 5.4 a-h). Biomass burning includes emissions from wildfires and residential wood combustion. This source showed distinct seasonal and spatial variability, with highest levels observed during winter and also in upper parts of the state. Higher concentrations associated with biomass burning in winter are mainly due to the higher residential wood burning during the colder seasons. Central and northern parts of the state usually experience colder winters compared to southern regions (Table 5.3), therefore higher biomass burning is expected in these geographical locations, as shown in many previous studies (Chen et al., 2007; Hu et al., 2014). Biomass burning 0 1 2 3 4 5 6 7 8 9 Mass concentration (µg/m 3 ) Secondary sulfate Spring Summer Fall winter 131 was the major primary source of ambient PM2.5 at Fresno and San Jose during all seasons, with levels ranging from 2.4 to 10.4 µg/m 3 (or 22 to 30% of PM2.5) at Fresno and from 2.2 to 8.0 µg/m 3 (or 22 to 43% of PM2.5) in San Jose (Figure 5.9). This source was also the dominant primary contributor to ambient PM2.5 in Bakersfield and Sacramento during winter (12 and 31% of PM2.5, respectively), consistent with the findings of many previous studies in this area (Chow et al., 2007; Gorin et al., 2006; Schauer and Cass, 2000). Figure 5.9. Seasonal average source contribution (µg/m 3 ) of biomass burning to ambient PM2.5, by site. Error bars correspond to one standard error. 0 2 4 6 8 10 12 Mass concentration (µg/m 3 ) Biomass burning Spring Summer Fall winter 132 5.4.2.5.Soil Resolved soil source profiles were dominated by crustal elements such as Al, Ca, Fe, Si, and Ti (Figure 5.4 a-h). These profiles generally lacked the contributions from EC and OC, indicating that they are not majorly impacted by emissions of road dust. As stated above, road dust was partially apportioned in the resolved vehicular emissions source profiles. A distinct source profile attributable to soil was not identified at Fresno. Instead, crustal elements partitioned in a separate source profile, along with high loadings of sulfate, EC, and OC, which was characterized as “sulfate-bearing road dust”. Figure 5.10. Seasonal average source contribution (µg/m 3 ) of soil to ambient PM2.5, by site. Error bars correspond to one standard error. 0 0.5 1 1.5 2 2.5 3 Mass concentration (µg/m 3 ) Soil Spring Summer Fall winter 133 Across the state, soil exhibited lower concentrations in northern regions, namely at San Jose and Sacramento (Figure 5.10). This is likely attributed to increased precipitation and higher RH in this part of the state (Table 5.3), which limit the wind-induced resuspension of soil (Harrison et al., 2001). Soil, in contrast, accounted for a large fraction of PM2.5 at Bakersfield, in concert with the findings of Chen et al. (2007) . During summer, in particular, contribution of soil to total mass was near 20% at Bakersfield, which could be mainly due to the lack of precipitation and low RH in this area (Table 5.3). As discussed by Chen et al. (2007), farm lands, pasture lands, and unpaved roads are major sources of soil and windblown dust in the SJV. 5.4.2.6.Fresh and aged sea salt Sources with high concentrations of Na + and Cl - were characterized as fresh sea salt (Figure 5.4 a-h). Aged sea salt source profiles, on the other hand, were dominated by loadings of Na + , SO4 2- , and NO3 - . Unlike fresh sea salt, chlorine has a negligible or near-zero contribution to aged sea salt source profile. Chlorine is typically depleted due to reactions of sea salt with acidic gases during the long range transport of sea salt aerosols from the point of emission (Song and Carmichael, 1999). Aged sea salt overall accounted for a lager fraction (2 to 27%) of ambient PM2.5 compared to fresh sea salt (1 to 13%), in all sites and seasons (Figures 5.5, 5.11, and 5.12). Aged sea salt showed a clear seasonal pattern at all sites, with higher concentrations in summer, consistent with increasing onshore winds (Table 5.3), while lowest during winter. It is also noteworthy that the PMF model did not apportion a separate factor for ship emissions or a source related to ocean goods transport. However, high loadings of Ni and V (tracers of ship emissions (Arhami et al., 2009)) in secondary ammonium sulfate and aged sea salt source profiles for the sampling sites in the LA Basin, suggest that these sources are affected in part by emissions from ships serving the ports of LA and Long Beach (Hwang and Hopke, 2007). 134 Figure 5.11. Seasonal average source contribution (µg/m 3 ) of fresh sea salt to ambient PM2.5, by site. Error bars correspond to one standard error. Figure 5.12. Seasonal average source contribution (µg/m 3 ) of aged sea salt to ambient PM2.5, by site. Error bars correspond to one standard error. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Mass concentration (µg/m 3 ) Fresh sea salt Spring Summer Fall winter 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Mass concentration (µg/m 3 ) Aged sea salt Spring Summer Fall winter 135 5.4.2.7.Other sources As noted above, few sources were exclusively identified at some sites, with relatively low annual contributions to total mass (1 to 15%, across the sites). At Rubidoux, a source profile was deduced with high loadings of Zn, Pb, EC, and OC (Figure 5.4 b), which is most likely attributed to local “mixed industrial” emissions in the surrounding areas. A similar source profile was also obtained in previous studies in this area (Kim et al., 2010; Kim and Hopke, 2007a). At San Jose, a source profile dominated by Ni was identified, which likely indicates the contribution from nearby Ni-related industrial sources. Hwang and Hopke (2006) reported similar findings at the same sampling location, by application of the PMF model on STN data, collected between 2002 and 2005. This source, nonetheless, accounted for less than 2% of the total mass, on an annual average basis. Copper smelters source profile, with a very high loading of Cu (>80%) and a slight contribution of EC, was identified in El Cajon and Bakersfield sampling sites (Figure 5.4 a,e). This source accounted for about 1 and 4% of total mass, over all years, in Bakersfield and El Cajon, respectively. Figure 5.13 shows the seasonal trends of industrial emissions in locations where these sources were identified. In El Cajon and Rubidoux, contributions of the identified industrial sources peaked in winter, while in Bakersfield and San Jose, maximum emissions from copper smelters and Ni-related sources were observed in summer. It is important to note that although the contributions from the identified industrial sources to total PM mass were overall trivial (<4%), these sources and the related elements may be important contributors to the overall particle toxicity (Dall’Osto et al., 2008; Saffari et al., 2013; Toledo et al., 2007; von Schneidemesser et al., 2010). At Fresno, a source profile with a high loading of sulfate along with road dust tracers, such as OC, EC, Fe, Ca, Mn, Si and Ti, was resolved (Figure 5.4 f). These road dust tracers most likely originate from the re-suspension of deposited soil and road dust enriched with vehicular emissions and lubricating oils (Dall’Osto et al., 2014; Pant and Harrison, 2013). This source was therefore named “sulfate-bearing road dust” (Katrinak et al., 1995). As mentioned above, separate source profiles for secondary ammonium sulfate and soil were not identified at Fresno. Nonetheless, the relatively high loadings of sulfate and a few crustal elements (e.g., Al, Ca, Fe, Si), along with the modest contribution of ammonium, suggest that these two sources are partially apportioned into this source profile. On an average basis over all 6 years, “sulfate-bearing road dust” accounted for 136 about 15% of total mass at Fresno and its contribution was highest in summer among all seasons (2.7±0.1 µg/m 3 ). Relatively similar source profiles, with high loadings of chlorine, were obtained at Fresno, Bakersfield, and Sacramento, with annual average contributions of about 5, 2, and 1% to total mass, respectively (Figure 5.4 e, f, and h). This source, which was denoted as “chlorine sources”, was mostly detected during fall and winter at Fresno and Bakersfield, in the SJV, while it displayed the maximum seasonal average value in summer at Sacramento (Figure 5.14). 137 Figure 5.13. Seasonal average source contribution (µg/m 3 ) of industrial emissions to ambient PM2.5, by site. Error bars correspond to one standard error. Figure 5.14. Seasonal average contribution (µg/m 3 ) of chlorine sources to ambient PM2.5, by site. Error bars correspond to one standard error. 0 0.2 0.4 0.6 0.8 1 Mass concentration (µg/m 3 ) Industrial emissions Spring Summer Fall winter * ** * *** 0 0.5 1 1.5 2 2.5 3 3.5 Mass concentration (µg/m 3 ) Chlorine sources Spring Summer Fall Winter 138 5.5. Summary and conclusions Source apportionment analyses were conducted using PMF receptor model applied on chemical speciation datasets, obtained from 8 different STN sampling sites throughout the state of California, between 2002 and 2007. Five-to-nine major sources contributing to ambient PM2.5 were identified at each site, with several of which being common in multiple locations. Overall, secondary aerosols (including secondary ammonium nitrate and ammonium sulfate) were collectively the main contributor to PM2.5 mass at all sampling sites. Annual average source contribution of secondary ammonium nitrate and ammonium sulfate ranged from 3.1 to 12 µg/m 3 (or 16 to 50% of total mass) and 1.3 to 4.6 µg/m 3 (or 10 to 23% of total mass) across the state, respectively. On an annual average basis, vehicular emissions (including both diesel and gasoline vehicles) were the largest primary sources of PM2.5 at all sampling sites in the southern part of the state (i.e., El Cajon, Rubidoux, LA, and Simi Valley), with 17-18% contribution total PM mass. In Fresno and San Jose, on the other hand, biomass burning was the dominant primary source of ambient PM2.5, contributing to 27 and 35% of total mass, on average over all years. In Bakersfield and Sacramento, biomass burning and vehicular emissions equally contributed to PM2.5 mass with near 12 and 25% annual contributions, respectively. Other sources commonly identified at all sites were minor contributors to PM2.5, including aged and fresh sea salt as well as soil, which contributed to 0.5-13%, 2-27%, and 1-19% of total mass, respectively, across all sites and seasons. Furthermore, a few sources (including chlorine sources, sulfate-bearing road dust, and different types of industrial emissions), which overall accounted for a small fraction of total mass (1 to 15%, on an annual average basis), were solely identified at some of the sites. 5.6. Acknowledgements This study was supported by the California Environmental Protection Agency (Cal EPA), Office of Environmental Health Hazard Assessment (OEHHA) (award number 12-E0021). 139 CHAPTER 6 Summary, Conclusions, and Recommendations 140 6.1. Summary and Conclusions In chapter 2, the results of a study were presented, in which quasi-UFP (PM0.25, dp < 0.25 μm) samples were collected for 24 h once per week from April 2008 to March 2009 at 10 different locations in the Los Angeles Basin. Samples were chemically analyzed and organic constituents of PM0.25 were grouped into polycyclic aromatic hydrocarbon (PAHs), hopanes and steranes, n-alkanes, and levoglucosan. A molecular marker-based chemical mass balance (MM- CMB) model was applied to estimate the relative contributions from the following primary sources: mobile sources (combined gasoline and diesel vehicles), wood smoke, natural gas combustion, vegetative detritus, and ship emissions. Secondary organic aerosol (SOA) tracers were not included in the model; however their contributions were estimated from non-biomass burning water soluble organic carbon (WSOC nb) and un-apportioned OC from MM-CMB model (“other OC”). High correlation (R 2 = 0.8) between “other OC” and WSOCnb in summer suggested that “other OC” is highly impacted by SOA, however un-apportioned primary sources may contribute to “other OC” as well. Mobile sources were expectedly the major primary contributor to PM0.25, with seasonal average contributions of 31 ± 12% in summer and 57 ± 17% in winter. “Other organic matter” was the second largest contributor to PM0.25 in all seasons, across the basin, with substantially higher contribution during warmer spring and summer seasons (27%), while lowest during cold seasons (13%). Wood smoke was the third major contributor to PM 0.25 in winter, whereas its contribution was lowest in summer. As expected, ship emissions displayed the highest contribution at the near-harbor HUD site, and their levels continually decreased as a function of distance from coast. Two other primary sources, vegetative detritus and natural gas combustion, collectively contributed to 1.3 ± 0.8% of PM0.25 on an annual average basis over all sites. In chapter 3, the focus was switched to a larger size fraction (PM2.5) with the main objective of investigating the degree to which indoor and outdoor sources contribute to indoor levels of PM. Therefore, in the study presented in Chapter 3 concurrent indoor and outdoor measurements of fine particulate matter (PM2.5) were conducted at three retirement homes in the Los Angeles Basin during two separate phases (cold and warm) between 2005 and 2006. Indoor-to-outdoor relationships of PM2.5 chemical constituents were determined and sources of indoor and outdoor PM2.5were evaluated using a molecular marker-based chemical mass balance (MM-CMB) model. 141 Indoor levels of elemental carbon (EC) along with metals and trace elements were found to be significantly affected by outdoor sources. EC, in particular, displayed very high indoor-to-outdoor (I/O) mass ratios accompanied by strong I/O correlations, illustrating the significant impact of outdoor sources on indoor levels of EC. Similarly, indoor levels of polycyclic aromatic hydrocarbons (PAHs), hopanes, and steranes were strongly correlated with their outdoor components and displayed I/O ratios close to unity. On the other hand, concentrations of n-alkanes and organic acids inside the retirement communities were dominated by indoor sources (e.g., food cooking and consumer products), as indicated by their I/O ratios, which exceeded unity. Source apportionment results revealed that vehicular emissions were the major contributor to both indoor and outdoor PM2.5, accounting for 39 and 46% of total mass, respectively. Moreover, the contribution of vehicular sources to indoor levels was generally comparable to its corresponding outdoor estimate. Other water-insoluble organic matter (other WIOM), which accounts for emissions from uncharacterized primary biogenic sources, displayed a wider range of contributions, varying from 2 to 73% of PM2.5, across all sites and phases of the study. Lastly, higher indoor than outdoor contribution of other water-soluble organic matter (other WSOM) was evident at some of the sites, suggesting the production of secondary aerosols as well as direct emissions from primary sources (including cleaning or other consumer products) at the indoor environments. To determine the effect of major federal, state, and local regulations on vehicular emissions, in Chapter 4 long-term changes in the concentration levels and source contributions of ambient fine particulate matter (PM2.5) were investigated in the Los Angeles Basin (LAB) between 2002 and 2013. Positive Matrix Factorization (PMF) receptor model was utilized to identify and quantify sources of ambient PM2.5 in central Los Angeles (LA) and Rubidoux, a receptor site in Riverside County, using 24-hr PM2.5 chemical composition data acquired from the U.S. Environmental Protection Agency (U.S. EPA) Speciation Trends Network (STN) and collected between 2002 and 2013. Over an almost 12-year average, vehicular emissions (including both gasoline and diesel vehicles) were found to be the second major contributor to PM 2.5, following secondary aerosols, with a near 20% contribution to overall mass in both LA and Rubidoux. During 2002 to 2004, the median and annual average values of daily-resolved source contributions of vehicular emissions were relatively constant, and did not show any statistically significant difference in each two consecutive years, at both sites. Vehicular emissions source 142 contributions peaked in 2007 and 2005, respectively, in LA and Rubidoux, while trended downward afterward until 2013. In 2013, the median values of daily-resolved vehicular emissions source contributions in LA and Rubidoux were respectively 70 and 52% lower than their corresponding values in 2007 and 2005. Starting in 2007, all manufactured diesel trucks must meet the U.S. EPA 2007 emission standard, requiring a 90% reduction in emissions of PM. In addition, after 2007, several major steps were taken by the California Air Resources Board as well as the ports of Los Angeles and Long Beach to reduce emissions from vehicular sources, particularly from diesel trucks. In order to assess the effect of these regulations, daily-resolved vehicular emissions source contributions from 2002 to 2006 were pooled together and compared to the combination of 2008 to 2012 datasets. Compared to the 2002-2006 dataset, the median values of daily-resolved source contributions from vehicular emissions in 2008-2012 were 24 and 21% lower in LA and Rubidoux, respectively. Mann-Whitney rank sum tests showed that these reductions at both sampling sites were statistically significant with p values ≤0.001. These reductions were noted despite an overall increase (of about 5%) in the median value of the daily flow of vehicles in downtown LA after 2007, while the traffic counts were comparable before and after 2007 in Rubidoux. Overall, these findings demonstrate the effectiveness of stringent regulations in reducing PM emissions from vehicular sources in the LA basin over the past decade. The study discussed in Chapter 4 was expanded to 6 other cities in the state of California and as discussed in Chapter 5 major sources of ambient PM2.5 were identified and quantified using PMF receptor modeling on STN data, collected between 2002 and 2007. The sampling sites included El Cajon, Rubidoux, Los Angeles, Simi Valley, Bakersfield, Fresno, San Jose, and Sacramento. Between five to nine sources of fine PM were identified at each sampling site, several of which were common among multiple locations. Secondary aerosols, including secondary ammonium nitrate and ammonium sulfate, were the most abundant contributor to ambient PM2.5 mass at all sampling sites, except for San Jose, with an annual average cumulative contribution of 26 to 63%, across the state. On an annual average basis, vehicular emissions (including both diesel and gasoline vehicles) were the largest primary source of fine PM at all sampling sites in southern California (17–18% of total mass), whereas in Fresno and San Jose, biomass burning was the most dominant primary contributor to ambient PM2.5 (27 and 35% of total mass, respectively), in general agreement with the results of previous source apportionment studies in California. In Bakersfield and Sacramento, vehicular emissions and biomass burning 143 displayed relatively equal annual contributions to ambient PM2.5 mass (12 and 25%, respectively). Other commonly identified sources at all sites included aged and fresh sea salt and soil, which contributed to 0.5–13%, 2–27%, and 1–19% of the total mass, respectively, across all sites and seasons. In addition, a few minor sources were identified exclusively at some of the sites (e.g., chlorine sources, sulfate-bearing road dust, and different types of industrial emissions). These sources overall accounted for a small fraction of the total PM mass across the sampling locations (1 to 15%, on an annual average basis). 6.2. Limitations of the Current Investigations The findings from the case studies presented in this dissertation provided invaluable insight on the physicochemical properties and source apportionment of size-fractionated airborne PM in major urban areas in the state of California. Although these studies were unique in their own ways in terms of the sample size, number of measurement locations, and sampling durations, a careful consideration of limitations in the examined studies would enable us to enhance our understanding of the characteristics of airborne PM, and the issues discussed in this section can be the subject of further research in the future. Therefore, in the next section, several research ideas are proposed to address the issues discussed in this section. In all the studies presented in this dissertation, the sampling methodology was based on time-integrated filter-based measurements. Given the nature of this methodology, sampling needs to be prolonged for a considerable amount of time in order to collect enough particle mass for the desired chemical analyses. In all the documented studies in this dissertation, the time resolution of the samples was 24 hours, with sampling frequencies varying between 3 and 6 days. Although the adopted time-integrated approach is suitable for routine air quality monitoring and long-term investigation of PM characteristics, more highly time-resolved measurements will enhance our understanding of the diurnal variation of PM along with its constituents and provide valuable insights on possible PM formation mechanisms, atmospheric aging, and their source contributions. This is, of course, a more important issue for PM sources/constituents that show very strong diurnal variability. For instance, a recent source apportionment study on number size distribution of ambient PM in central Los Angeles showed that the contribution of vehicular emissions to total PM number concentration can vary by a factor of two to three during a day (Sowlat et al., 2016). Such strong diurnal variability, which is vital for an acute exposure assessment, is unlikely to be 144 captured by time-integrated sampling methodologies. As will be thoroughly discussed in the next section, this issue can be addressed by the use of more advanced sampling monitors that provide detailed information on the chemical and microphysical properties of airborne PM with high temporal resolution. Although there is a rich history of the development of apparatuses for continuous measurement of physical properties of PM (e.g., size distribution, mass concentration, number concentration, light absorption, etc.), over the past decade there has been a growing interest in the aerosol science community in developing new techniques to achieve near-continuous measurement of chemical components of airborne PM. Examples of such instruments are aerosol mass spectrometers for measurements of non-refractory materials and some new online measurement techniques for the measurement of refractory species (Canagaratna et al., 2007; Ng et al., 2011; Park et al., 2014; Wang et al., 2016, 2015). In addition to temporal resolution, another point of consideration is the spatial resolution of the samples in the studies presented in this dissertation. The studies documented in Chapters 2– 5 were unique in terms of the number of sampling sites considered in each study. Generally, due to the time-, resource-, and labor-intensive nature of such field campaigns, inclusion of too many sampling locations is not practically feasible. Although such networks of fixed sampling stations provide valuable “big picture” insights on the characteristics of ambient PM, often at background locations, they are not capable of capturing detailed spatial variability within a city, which is essential for accurate exposure assessment, especially in major urban areas. The study presented in Chapter 2, which consisted of a network of 10 sampling locations in the LA Basin, was one of the most comprehensive studies of this type conducted in LA, investigating the spatial and temporal variability of ambient UFP. Nonetheless, given the vast surface area and the population of the LA Basin, 10 sampling locations may not fully represent the spatial variability of ambient PM at a fine scale. As the results from this study attest, chemical composition and source contributions of ambient UFPs show robust spatial variability within the 10 study sampling locations. A common historical approach to address this issue in the aerosol scientific community has been ambient PM modeling. These approaches, however, have considerable inherent uncertainties, mainly due to crude approximations used in input conditions (Vardoulakis et al., 2002). An alternative and indeed emerging approach is the use of a network consisting of a significant number of low-cost air quality sensors for a better understating of finer-grained spatial 145 variability of ambient PM in metropolitan areas (Kumar et al., 2015). More detailed discussion on this matter will be provided in the next section. Lastly, the results presented in Chapters 2, 4, and 5 were mainly focused on outdoor PM, whereas a detailed exposure assessment study requires measurements at a more personal scale. The study discussed in Chapter 3 addressed this issue to some extent, but, as noted above, the filter-based measurements precluded the high time resolution of the measured parameters inside the retirement communities where the instantaneous activities of the residents directly impact their exposure levels. As mentioned before, the advent of new-generation low-cost, easy-to-use, portable air pollution monitors (sensors) that provide high time-resolution data in near real-time has been changing the paradigm of air pollution monitoring and motivating finer-grained and more personalized air monitoring data collection. As will be discussed in the next section, an integrated network of miniaturized air quality sensors would provide a platform for more precise personal exposure assessments. 6.3. Recommendations for Future Research The studies carried out for this doctoral dissertation demonstrate a blueprint for further research in the future. The documented studies shed light on the spatiotemporal variability of physical and chemical properties of ambient PM and also demonstrated the degree to which these particles can penetrate into indoor environments. The findings verified that the chemical composition of ambient PM and its sources show strong spatial and temporal variability. This suggests that future studies of this kind need to shift their focus toward measurements with higher spatial and temporal resolutions. As discussed above, the filter-based measurement of airborne PM precludes investigation of variations in the physicochemical properties and sources of PM at a fine time scale. To overcome this issue, in recent years, development of instruments that provide near-continuous information about the physical and chemical properties of aerosols has received considerable attention. Aerosol mass spectrometry, in particular, has advanced increasingly in recent years for highly time-resolved chemical characterization of aerosols (Canagaratna et al., 2007). Aerodyne Research aerosol mass spectrometers (e.g., AMS and ACSM) are the most widely used thermal desorption-based mass spectrometers in aerosol research (Jayne et al., 2000; Ng et al., 2011). The AMS can quantify mass concentrations of non-refractory materials including sulfate, nitrate, 146 ammonium, chloride, and total organic matter—via thermal vaporization and ionization—with a time resolution down to a few minutes (Canagaratna et al., 2007). Other techniques have also been developed to continuously measure the mass concentrations of refractory species such as EC (e.g., measured by EC-OC Sunset Lab) or BC (e.g., measured by Aethalometer), and metals (e.g., measured by techniques developed by Wang et al. (2016)) in ambient PM. A combination of the aforementioned instruments along with aerosol sizing apparatuses (e.g., SMPS, OPS, APS, etc.) can provide detailed information about the microphysical and chemical properties of aerosols at very high temporal resolution. Subsequently, factor analysis of these time and compositionally resolved data enables the extraction of source profiles that contribute to mass and number concentrations of ambient PM. The diurnal variability in the physical/chemical properties and the resolved factor profiles of ambient PM can then enhance our fundamental knowledge regarding aerosol chemistry, mechanisms of particle formation, and atmospheric aging. Furthermore, given the large amount of time spent by people indoors and the considerable number of sources that exist exclusively in such environments (e.g., emissions from materials and human skin, surface chemistry, etc.) the newly developed instruments should be deployed in indoor environments to measure the physicochemical properties and indoor chemistry of aerosols at high time resolution. Considering the major differences in building designs, materials, ventilation, and indoor sources in different parts of the world, research on indoor chemistry of aerosols needs to be pursued on a global scale (Jimenez, 2016). In addition to high time resolution of ambient PM sampling, another point of consideration in future studies is increasing the spatial resolution of PM measurements. Obviously, achieving this goal by setting up numerous fixed sampling stations equipped with rather expensive and sophisticated air quality monitors is cost- and labor- prohibitive and not practically feasible. A more effective approach is the deployment of a large number of low-cost portable air quality monitors that provide high time-resolution data in near real time. Integrated sensor networks have the potential to provide information on fine-scale spatial and temporal variability of air pollutants and deliver improved estimates of the emission profiles near and around specific sources (Jovašević-Stojanović et al., 2015). In addition, these networks can lead to creating emission inventories of air pollutants and detecting pollution hotspots, as well as allowing real-time personal exposure assessment for designing health effects studies. A series of such portable sensors, coupled with advanced computing and communication systems, can be used to form interactive, 147 participatory sensor networks that could facilitate citizen and professional users to jointly collect, analyze, and share a wide range of real-time data (Piedrahita et al., 2014). These networks assist health researchers in advancing scientific understanding of the symptoms and exacerbations of air pollution health effects. This also leads to timely interventions and treatments and the ability to provide active feedback to patients and caregivers. Another application of these portable sensors which are capable of providing real-time data is in traffic management systems where they can be used to track the impacts of regulations such as low-emissions zones. Nonetheless, it is noteworthy that the new generation of air quality measurement using miniaturized sensors is still in its earliest stages and some major challenges in their production and large-scale deployment in urban environments need to be addressed. This new paradigm of air pollution monitoring has great potential to complement and enhance air quality assessment and forecasting. Therefore, future source apportionment studies need to shift their focus toward using the considerably larger amount of data derived from these networks to investigate the contributions of various sources to ambient PM at higher spatial and temporal resolutions. In addition to spatial and temporal resolution discussed above, some other areas of research that require further investigation are summarized below:  The existing profiles in the literature for a majority of sources are outdated and pertain to a certain geographical area which may not be applicable in source apportionment studies conducted in other parts of the world. Therefore, up-to-date and locally relevant source profiles, required for CMB modeling, are not widely available and should be further investigated in future studies.  Despite major advances in source apportionment studies, there is still a large uncertainty associated with the apportionment of gasoline and diesel emissions separately. This warrants more detailed investigation, particularly by using separated molecular markers groups. The split between diesel and gasoline emissions is vital in evaluating the impact of emission reduction policies, which are mostly implemented on diesel trucks.  Lastly, a particular attention needs to be given to the “non-traditional” PM sources which are not fully characterized yet. As the emissions from major traditional sources (e.g. vehicular and ship emissions) have declined over the past 2 decades, 148 the emissions of other uncontrolled sources are becoming increasingly important as a fraction of the total PM mass. Food cooking, in particular, is an important source that contributes to both indoor and outdoor PM. Some recent investigations have even identified cooking as rivaling emissions from traffic sources (Pandis et al., 2016). Non-tailpipe traffic emissions (e.g. road dust, vehicular abrasion, etc.) are other poorly characterized sources whose relative importance is increasing, despite rapid reduction of tailpipe emissions (Shirmohammadi et al., 2016). The precise estimates of contributions from these “non-traditional” sources will help guide the development of more targeted and cost-effective strategies that maximize improvements in public health. 149 Bibliography Abdullahi, K.L., Delgado-Saborit, J.M., Harrison, R.M., 2013. Emissions and indoor concentrations of particulate matter and its specific chemical components from cooking: A review. Atmos. Environ. 71, 260–294. doi:10.1016/j.atmosenv.2013.01.061 Agrawal, H., Malloy, Q.G.J., Welch, W.A., Wayne, M., Cocker, I., 2008. In-use gaseous and particulate matter emissions from a modern ocean going container vessel. Atmos. Environ. 42, 5504–5510. doi:10.1016/j.atmosenv.2008.02.053 Analitis, A., Katsouyanni, K., Dimakopoulou, K., Samoli, E., Nikoloulopoulos, A.K., Petasakis, Y., Touloumi, G., Schwartz, J., Anderson, H.R., Cambra, K., Forastiere, F., Zmirou, D., Vonk, J.M., Clancy, L., Kriz, B., Bobvos, J., Pekkanen, J., 2006. Short-Term Effects of Ambient Particles on Cardiovascular and Respiratory Mortality: Epidemiology 17, 230– 233. doi:10.1097/01.ede.0000199439.57655.6b Aoki, T., Tanabe, S. -i., 2007. Generation of sub-micron particles and secondary pollutants from building materials by ozone reaction. Atmos. Environ. 41, 3139–3150. doi:10.1016/j.atmosenv.2006.07.053 Araujo, J.A., Barajas, B., Kleinman, M., Wang, X., Bennett, B.J., Gong, K.W., Navab, M., Harkema, J., Sioutas, C., Lusis, A.J., Nel, A.E., 2008. Ambient Particulate Pollutants in the Ultrafine Range Promote Early Atherosclerosis and Systemic Oxidative Stress. Circ. Res. 102, 589–596. doi:10.1161/CIRCRESAHA.107.164970 Arey, J., Zielinska, B., Atkinson, R., Winer, A.M., 1988. Formation of nitroarenes during ambient high-volume sampling. Environ. Sci. Technol. 22, 457–462. Arhami, M., Kuhn, T., Fine, P.M., Delfino, R.J., Sioutas, C., 2006. Effects of sampling artifacts and operating parameters on the performance of a semicontinuous particulate elemental carbon/organic carbon monitor. Environ. Sci. Technol. 40, 945–954. doi:10.1021/es0510313 Arhami, M., Minguillón, M.C., Polidori, A., Schauer, J.J., Delfino, R.J., Sioutas, C., 2010. Organic compound characterization and source apportionment of indoor and outdoor quasi- ultrafine particulate matter in retirement homes of the Los Angeles Basin. Indoor Air 20, 17–30. doi:10.1111/j.1600-0668.2009.00620.x Arhami, M., Sillanpää, M., Hu, S., Olson, M.R., Schauer, J.J., Sioutas, C., 2009. Size-segregated inorganic and organic components of PM in the communities of the Los Angeles Harbor. Aerosol Sci. Technol. 43, 145–160. doi:10.1080/02786820802534757 Atkinson, R.W., Anderson, H.R., Sunyer, J., Ayres, J., Baccini, M., Vonk, J.M., Boumghar, A., Forastiere, F., Forsberg, B., Touloumi, G., Schwartz, J., Katsouyanni, K., 2001. Acute effects of particulate air pollution on respiratory admissions: Results from APHEA 2 project. Am. J. Respir. Crit. Care Med. 164, 1860–1866. Ayres, J.G., Borm, P., Cassee, F.R., Castranova, V., Donaldson, K., Ghio, A., Harrison, R.M., Hider, R., Kelly, F., Kooter, I.M., Marano, F., Maynard, R.L., Mudway, I., Nel, A., Sioutas, C., Smith, S., Baeza-Squiban, A., Cho, A., Duggan, S., Froines, J., 2008. Evaluating the Toxicity of Airborne Particulate Matter and Nanoparticles by Measuring Oxidative Stress Potential—A Workshop Report and Consensus Statement. Inhal. Toxicol. 20, 75–99. doi:10.1080/08958370701665517 Bell, M.L., Belanger, K., Ebisu, K., Gent, J.F., Lee, H.J., Koutrakis, P., Leaderer, B.P., 2010. Prenatal Exposure to Fine Particulate Matter and Birth Weight. Epidemiol. Camb. Mass 21, 884–891. doi:10.1097/EDE.0b013e3181f2f405 150 Bell, M.L., Ebisu, K., Leaderer, B.P., Gent, J.F., Lee, H.J., Koutrakis, P., Wang, Y., Dominici, F., Peng, R.D., 2014. Associations of PM2.5 Constituents and Sources with Hospital Admissions: Analysis of Four Counties in Connecticut and Massachusetts (USA) for Persons > or = 65 Years of Age. Environ. Health Perspect. Online 122, 138. doi:http://dx.doi.org/10.1289/ehp.1306656 Bell, M.L., Ebisu, K., Peng, R.D., Walker, J., Samet, J.M., Zeger, S.L., Dominici, F., 2008. Seasonal and Regional Short-term Effects of Fine Particles on Hospital Admissions in 202 US Counties, 1999–2005. Am. J. Epidemiol. 168, 1301–1310. doi:10.1093/aje/kwn252 Birch, M.E., Cary, R.A., 1996. Elemental carbon-based method for monitoring occupational exposures to particulate diesel exhaust. Aerosol Sci. Technol. 25, 221–241. Bishop, G.A., Schuchmann, B.G., Stedman, D.H., 2013. Heavy-duty truck emissions in the south coast air basin of California. Environ. Sci. Technol. 47, 9523–9529. doi:10.1021/es401487b Boström, C.-E., Gerde, P., Hanberg, A., Jernström, B., Johansson, C., Kyrklund, T., Rannug, A., Törnqvist, M., Victorin, K., Westerholm, R., 2002. Cancer risk assessment, indicators, and guidelines for polycyclic aromatic hydrocarbons in the ambient air. Environ. Health Perspect. 110, 451–488. Brook, R.D., Rajagopalan, S., Pope, C.A., Brook, J.R., Bhatnagar, A., Diez-Roux, A.V., Holguin, F., Hong, Y., Luepker, R.V., Mittleman, M.A., Peters, A., Siscovick, D., Smith, S.C., Whitsel, L., Kaufman, J.D., 2010. Particulate matter air pollution and cardiovascular disease: An update to the scientific statement from the american heart association. Circulation 121, 2331–2378. doi:10.1161/CIR.0b013e3181dbece1 Brown, P.M.B.L.C., Hambley, D.F., 2002. Statistics for environmental engineers. Env. Eng Geosci 8, 244–245. Brunekreef, B., Forsberg, B., 2005. Epidemiological evidence of effects of coarse airborne particles on health. Eur. Respir. J. 26, 309–318. doi:10.1183/09031936.05.00001805 California Code of Regulations, 2008. Regulation to Reduce Emissions of Diesel Particulate Matter, Oxides of Nitrogen and Other Criteria Pollutants, from In-use Heavy-duty Diesel- fueled Vehicles, Title 13, Art. 4.5, Chap. 1, x 2025. Calif. Environ. Prot. Agency Air Resour. Board, Sacramento. Canagaratna, M. r., Jayne, J. t., Jimenez, J. l., Allan, J. d., Alfarra, M. r., Zhang, Q., Onasch, T. b., Drewnick, F., Coe, H., Middlebrook, A., Delia, A., Williams, L. r., Trimborn, A. m., Northway, M. j., DeCarlo, P. f., Kolb, C. e., Davidovits, P., Worsnop, D. r., 2007. Chemical and microphysical characterization of ambient aerosols with the aerodyne aerosol mass spectrometer. Mass Spectrom. Rev. 26, 185–222. doi:10.1002/mas.20115 Cass, G.R., Hughes, L.A., Bhave, P., Kleeman, M.J., Allen, J.O., Salmon, L.G., 2000. The chemical composition of atmospheric ultrafine particles. Philos. Trans. R. Soc. Math. Phys. Eng. Sci. 358, 2581–2592. Chen, L.-W.A., Watson, J.G., Chow, J.C., Magliano, K.L., 2007. Quantifying PM2.5 Source Contributions for the San Joaquin Valley with Multivariate Receptor Models. Environ. Sci. Technol. 41, 2818–2826. doi:10.1021/es0525105 Cheung, K., Daher, N., Kam, W., Shafer, M.M., Ning, Z., Schauer, J.J., Sioutas, C., 2011. Spatial and temporal variation of chemical composition and mass closure of ambient coarse particulate matter (PM10-2.5) in the Los Angeles area. Atmos. Environ. 45, 2651–2662. doi:10.1016/j.atmosenv.2011.02.066 151 Cheung, K., Olson, M.R., Shelton, B., Schauer, J.J., Sioutas, C., 2012. Seasonal and spatial variations of individual organic compounds of coarse particulate matter in the Los Angeles Basin. Atmos. Environ. 59, 1–10. doi:10.1016/j.atmosenv.2012.05.037 Cho, A.K., Sioutas, C., Miguel, A.H., Kumagai, Y., Schmitz, D.A., Singh, M., Eiguren-Fernandez, A., Froines, J.R., 2005. Redox activity of airborne particulate matter at different sites in the Los Angeles Basin. Environ. Res. 99, 40–47. doi:10.1016/j.envres.2005.01.003 Chow, J.C., Watson, J.G., Chen, L.-W.A., Rice, J., Frank, N.H., 2010. Quantification of PM2.5 organic carbon sampling artifacts in US networks. Atmospheric Chem. Phys. 10, 5223– 5239. doi:10.5194/acp-10-5223-2010 Chow, J.C., Watson, J.G., Crow, D., Lowenthal, D.H., Merrifield, T., 2001. Comparison of IMPROVE and NIOSH carbon measurements. Aerosol Sci. Technol. 34, 23–34. Chow, J.C., Watson, J.G., Lowenthal, D.H., Chen, L.W.A., Zielinska, B., Mazzoleni, L.R., Magliano, K.L., 2007. Evaluation of organic markers for chemical mass balance source apportionment at the Fresno Supersite. Atmos Chem Phys 7, 1741–1754. doi:10.5194/acp- 7-1741-2007 Christoforou, C.S., Salmon, L.G., Hannigan, M.P., Solomon, P.A., Cass, G.R., 2000. Trends in fine particle concentration and chemical composition in Southern California. J. Air Waste Manag. Assoc. 50, 43–53. Dadvand, P., Ostro, B., Amato, F., Figueras, F., Minguillón, M.-C., Martinez, D., Basagaña, X., Querol, X., Nieuwenhuijsen, M., 2014. Particulate air pollution and preeclampsia: a source-based analysis. Occup. Environ. Med. 71, 570–577. doi:10.1136/oemed-2013- 101693 Daher, N., Hasheminassab, S., Shafer, M.M., Schauer, J.J., Sioutas, C., 2013. Seasonal and spatial variability in chemical composition and mass closure of ambient ultrafine particles in the megacity of Los Angeles. Environ. Sci. Process. Impacts 15, 283–295. Daher, N., Ruprecht, A., Invernizzi, G., De, M., Miller-Schulze, J., Heo, J.B., Shafer, M.M., Schauer, J.J., Sioutas, C., 2011. Chemical characterization and source apportionment of fine and coarse particulate matter inside the refectory of santa maria delle grazie church, home of Leonardo da Vincis “last supper.” Environ. Sci. Technol. 45, 10344–10353. doi:10.1021/es202736a Dall’Osto, M., Beddows, D.C.S., Gietl, J.K., Olatunbosun, O.A., Yang, X., Harrison, R.M., 2014. Characteristics of tyre dust in polluted air: Studies by single particle mass spectrometry (ATOFMS). Atmos. Environ. 94, 224–230. doi:10.1016/j.atmosenv.2014.05.026 Dall’Osto, M., Booth, M.J., Smith, W., Fisher, R., Harrison, R.M., 2008. A Study of the Size Distributions and the Chemical Characterization of Airborne Particles in the Vicinity of a Large Integrated Steelworks. Aerosol Sci. Technol. 42, 981–991. doi:10.1080/02786820802339587 Davis, D.A., Akopian, G., Walsh, J.P., Sioutas, C., Morgan, T.E., Finch, C.E., 2013a. Urban air pollutants reduce synaptic function of CA1 neurons via an NMDA/NȮ pathway in vitro. J. Neurochem. 127, 509–519. doi:10.1111/jnc.12395 Davis, D.A., Bortolato, M., Godar, S.C., Sander, T.K., Iwata, N., Pakbin, P., Shih, J.C., Berhane, K., McConnell, R., Sioutas, C., Finch, C.E., Morgan, T.E., 2013b. Prenatal Exposure to Urban Air Nanoparticles in Mice Causes Altered Neuronal Differentiation and Depression- Like Responses. PLOS ONE 8, e64128. doi:10.1371/journal.pone.0064128 152 de, K., Driece, H.A.L., Hogervorst, J.G.F., Briedé, J.J., 2006. Toxicological assessment of ambient and traffic-related particulate matter: A review of recent studies. Mutat. Res. - Rev. Mutat. Res. 613, 103–122. doi:10.1016/j.mrrev.2006.07.001 Delfino, R.J., Quintana, P.J.E., Floro, J., Gastañaga, V.M., Samimi, B.S., Kleinman, M.T., Liu, L.-J.S., Bufalino, C., Wu, C.-F., McLaren, C.E., 2004. Association of FEV1 in asthmatic children with personal and microenvironmental exposure to airborne particulate matter. Environ. Health Perspect. 112, 932–941. Delfino, R.J., Sioutas, C., Malik, S., 2005. Potential role of ultrafine particles in associations between airborne particle mass and cardiovascular health. Environ. Health Perspect. 113, 934–946. doi:10.1289/ehp.7938 Delfino, R.J., Staimer, N., Tjoa, T., Gillen, D.L., Polidori, A., Arhami, M., Kleinman, M.T., Vaziri, N.D., Longhurst, J., Sioutas, C., 2009. Air pollution exposures and circulating biomarkers of effect in a susceptible population: Clues to potential causal component mixtures and mechanisms. Environ. Health Perspect. 117, 1232–1238. doi:10.1289/journla.ehp.0800194 Delfino, R.J., Staimer, N., Tjoa, T., Polidori, A., Arhami, M., Gillen, D.L., Kleinman, M.T., Vaziri, N.D., Longhurst, J., Zaldivar, F., Sioutas, C., 2008. Circulating biomarkers of inflammation, antioxidant activity, and platelet activation are associated with primary combustion aerosols in subjects with coronary artery disease. Environ. Health Perspect. 116, 898–906. doi:10.1289/ehp.11189 Delfino, R.J., Tjoa, T., Gillen, D.L., Staimer, N., Polidori, A., Arhami, M., Jamner, L., Sioutas, C., Longhurst, J., 2010. Traffic-related Air Pollution and Blood Pressure in Elderly Subjects With Coronary Artery Disease. Epidemiol. Camb. Mass 21. doi:10.1097/EDE.0b013e3181d5e19b DeNero, S.P., 2012. Development of a source oriented version of the WRF-Chem model and its application to the California Regional PM10/PM2.5 air quality study. UNIVERSITY OF CALIFORNIA, DAVIS. Docherty, K.S., Stone, E.A., Ulbrich, I.M., DeCarlo, P.F., Snyder, D.C., Schauer, J.J., Peltier, R.E., Weber, R.J., Murphy, S.M., Seinfeld, J.H., Grover, B.D., Eatough, D.J., Jimenez, J.L., 2008. Apportionment of primary and secondary organic aerosols in Southern California during the 2005 Study of Organic Aerosols in Riverside (SOAR-1). Environ. Sci. Technol. 42, 7655–7662. doi:10.1021/es8008166 Dockery, D.W., Speizer, F.E., Stram, D.O., Ware, J.H., Spengler, J.D., Ferris, J., 1989. Effects of inhalable particles on respiratory health of children. Am. Rev. Respir. Dis. 139, 587–594. Dominici, F., Peng, R.D., Bell, M.L., Pham, L., McDermott, A., Zeger, S.L., Samet, J.M., 2006. Fine particulate air pollution and hospital admission for cardiovascular and respiratory diseases. J. Am. Med. Assoc. 295, 1127–1134. doi:10.1001/jama.295.10.1127 Dominici F, Peng RD, Bell ML, et al, 2006. FIne particulate air pollution and hospital admission for cardiovascular and respiratory diseases. JAMA 295, 1127–1134. doi:10.1001/jama.295.10.1127 Donaldson, K., Stone, V., Seaton, A., MacNee, W., 2001. Ambient particle inhalation and the cardiovascular system: potential mechanisms. Environ. Health Perspect. 109, 523–527. Ebelt, S.T., Wilson, W.E., Brauer, M., 2005. Exposure to ambient and nonambient components of particulate matter: A comparison of health effects. Epidemiology 16, 396–405. doi:10.1097/01.ede.0000158918.57071.3e Eisner, M.D., Anthonisen, N., Coultas, D., Kuenzli, N., Perez-Padilla, R., Postma, D., Romieu, I., Silverman, E.K., Balmes, J.R., 2010. An official American Thoracic Society public policy 153 statement: Novel risk factors and the global burden of chronic obstructive pulmonary disease. Am. J. Respir. Crit. Care Med. 182, 693–718. doi:10.1164/rccm.200811-1757ST Fann, N., Lamson, A.D., Anenberg, S.C., Wesson, K., Risley, D., Hubbell, B.J., 2012. Estimating the National Public Health Burden Associated with Exposure to Ambient PM2.5 and Ozone. Risk Anal. 32, 81–95. doi:10.1111/j.1539-6924.2011.01630.x Fine, P.M., Cass, G.R., Simoneit, B.R.T., 2004. Chemical Characterization of Fine Particle Emissions from the Fireplace Combustion of Wood Types Grown in the Midwestern and Western United States. Environ. Eng. Sci. 21, 387–409. doi:10.1089/109287504323067021 Fine, P.M., Cass, G.R., Simoneit, B.R.T., 2002. Organic compounds in biomass smoke from residential wood combustion: Emissions characterization at a continental scale. J. Geophys. Res. Atmospheres 107, 8349. doi:10.1029/2001JD000661 Fine, P.M., Cass, G.R., Simoneit, B.R.T., 1999. Characterization of fine particle emissions from burning church candles. Environ. Sci. Technol. 33, 2352–2362. doi:10.1021/es981039v Flanagan, J.B., Jayanty, R.K.M., Edward E. Rickman, J., Peterson, M.R., 2006. PM2.5 Speciation Trends Network: Evaluation of Whole-System Uncertainties Using Data from Sites with Collocated Samplers. J. Air Waste Manag. Assoc. 56, 492–499. doi:10.1080/10473289.2006.10464516 Fraser, M.P., Lakshmanan, K., 2000. Using levoglucosan as a molecular marker for the long-range transport of biomass combustion aerosols. Environ. Sci. Technol. 34, 4560–4564. doi:10.1021/es991229l Fujita, E.M., Campbell, D.E., Stockwell, W.R., Lawson, D.R., 2013. Past and future ozone trends in California’s South Coast Air Basin: Reconciliation of ambient measurements with past and projected emission inventories. J. Air Waste Manag. Assoc. 63, 54–69. doi:10.1080/10962247.2012.735211 Fujita, E.M., Watson, J.G., Chow, J.C., Robinson, N.F., Richards, L.W., Kumar, N., 1998. Northern Front Range Air Quality Study. Volume C: Source Apportionment and Simulation Methods and Evaluation. Prepared for Colorado State University. Coop. Inst. Res. Atmosphere Ft Collins CO Desert Res. Inst. Reno NV. Gass, K., Balachandran, S., Chang, H.H., Russell, A.G., Strickland, M.J., 2015. Ensemble-Based Source Apportionment of Fine Particulate Matter and Emergency Department Visits for Pediatric Asthma. Am. J. Epidemiol. 181, 504–512. doi:10.1093/aje/kwu305 Geller, M.D., Chang, M., Sioutas, C., Ostro, B.D., Lipsett, M.J., 2002. Indoor/outdoor relationship and chemical composition of fine and coarse particles in the southern California deserts. Atmos. Environ. 36, 1099–1110. doi:10.1016/S1352-2310(01)00340-5 Gorin, C.A., Jr, J.L.C., Herckes, P., 2006. Wood Smoke Contribution to Winter Aerosol in Fresno, CA. J. Air Waste Manag. Assoc. 56, 1584–1590. doi:10.1080/10473289.2006.10464558 Gray, D.L., Wallace, L.A., Brinkman, M.C., Buehler, S.S., La Londe, C., 2015. Respiratory and Cardiovascular Effects of Metals in Ambient Particulate Matter: A Critical Review, in: Whitacre, D.M. (Ed.), Reviews of Environmental Contamination and Toxicology. Springer International Publishing, Cham, pp. 135–203. Grosjean, D., Fung, K., Harrison, J., 1983. Interactions of polycyclic aromatic hydrocarbons with atmospheric pollutants. Environ. Sci. Technol. 17, 673–679. Gutknecht, W., Flanagan, J., McWilliams, A., Jayanty, R.K.M., Kellogg, R., Rice, J., Duda, P., Sarver, R.H., 2010. Harmonization of Uncertainties of X-Ray Fluorescence Data for PM2.5 154 Air Filter Analysis. J. Air Waste Manag. Assoc. 60, 184–194. doi:10.3155/1047- 3289.60.2.184 Ham, W.A., Kleeman, M.J., 2011. Size-resolved source apportionment of carbonaceous particulate matter in urban and rural sites in central California. Atmos. Environ. 45, 3988– 3995. doi:10.1016/j.atmosenv.2011.04.063 Harrison, R.M., Yin, J., Mark, D., Stedman, J., Appleby, R.S., Booker, J., Moorcroft, S., 2001. Studies of the coarse particle (2.5–10 μm) component in UK urban atmospheres. Atmos. Environ. 35, 3667–3679. doi:10.1016/S1352-2310(00)00526-4 Hasheminassab, S., Daher, N., Ostro, B.D., Sioutas, C., 2014. Long-term source apportionment of ambient fine particulate matter (PM2.5) in the Los Angeles Basin: A focus on emissions reduction from vehicular sources. Environ. Pollut. 193, 54–64. doi:10.1016/j.envpol.2014.06.012 Hasheminassab, S., Daher, N., Schauer, J.J., Sioutas, C., 2013. Source apportionment and organic compound characterization of ambient ultrafine particulate matter (PM) in the Los Angeles Basin. Atmos. Environ. 79, 529–539. doi:10.1016/j.atmosenv.2013.07.040 Held, T., Ying, Q., Kaduwela, A., Kleeman, M., 2004. Modeling particulate matter in the San Joaquin Valley with a source-oriented externally mixed three-dimensional photochemical grid model. Atmos. Environ. 38, 3689–3711. doi:10.1016/j.atmosenv.2004.02.053 Henry, R.C., 2003. Multivariate receptor modeling by N-dimensional edge detection. Chemom. Intell. Lab. Syst. 65, 179–189. doi:10.1016/S0169-7439(02)00108-9 Henry, R.C., Kim, B.M., 1990. Extension of self-modeling curve resolution to mixtures of more than three components. Chemom. Intell. Lab. Syst. 8, 205–216. doi:10.1016/0169- 7439(90)80136-T Henry, R.C., Lewis, C.W., Hopke, P.K., Williamson, H.J., 1984. Review of receptor model fundamentals. Atmospheric Environ. 1967 18, 1507–1515. doi:10.1016/0004- 6981(84)90375-5 Heo, J., Dulger, M., Olson, M.R., McGinnis, J.E., Shelton, B.R., Matsunaga, A., Sioutas, C., Schauer, J.J., 2013. Source apportionments of PM2.5 organic carbon using molecular marker Positive Matrix Factorization and comparison of results from different receptor models. Atmos. Environ. 73, 51–61. doi:10.1016/j.atmosenv.2013.03.004 Heo, J.-B., Hopke, P.K., Yi, S.-M., 2009. Source apportionment of PM2.5 in seoul, korea. Atmospheric Chem. Phys. 9, 4957–4971. Herner, J.D., Green, P.G., Kleeman, M.J., 2006. Measuring the trace elemental composition of size-resolved airborne particles. Environ. Sci. Technol. 40, 1925–1933. doi:10.1021/es052315q Hinds, W.C., 2012. Aerosol Technology: Properties, Behavior, and Measurement of Airborne Particles. John Wiley & Sons. Hoek, G., Krishnan, R.M., Beelen, R., Peters, A., Ostro, B., Brunekreef, B., Kaufman, J.D., 2013. Long-term air pollution exposure and cardio- respiratory mortality: a review. Environ. Health 12, 43. doi:10.1186/1476-069X-12-43 Holguin, F., 2008. Traffic, Outdoor Air Pollution, and Asthma. Immunol. Allergy Clin. North Am. 28, 577–588. doi:10.1016/j.iac.2008.03.008 Holmes, N.S., Morawska, L., 2006. A review of dispersion modelling and its application to the dispersion of particles: An overview of different dispersion models available. Atmos. Environ. 40, 5902–5928. doi:10.1016/j.atmosenv.2006.06.003 155 Hopke, P.K., 2016. Review of receptor modeling methods for source apportionment. J. Air Waste Manag. Assoc. 66, 237–259. doi:10.1080/10962247.2016.1140693 Hopke, P.K., 2003. Recent developments in receptor modeling. J. Chemom. 17, 255–265. doi:10.1002/cem.796 Hu, J., Zhang, H., Chen, S., Ying, Q., Wiedinmyer, C., Vandenberghe, F., Kleeman, M.J., 2014. Identifying PM2.5 and PM0.1 Sources for Epidemiological Studies in California. Environ. Sci. Technol. 48, 4980–4990. doi:10.1021/es404810z Hu, S., Polidori, A., Arhami, M., Shafer, M.M., Schauer, J.J., Cho, A., Sioutas, C., 2008a. Redox activity and chemical speciation of size fractioned PM in the communities of the Los Angeles-Long Beach harbor. Atmos Chem Phys 8, 6439–6451. doi:10.5194/acp-8-6439- 2008 Hu, S., Polidori, A., Arhami, M., Shafer, M.M., Schauer, J.J., Cho, A., Sioutas, C., 2008b. Redox activity and chemical speciation of size fractioned PM in the communities of the Los Angeles-Long Beach harbor. Atmos Chem Phys 8, 6439–6451. doi:10.5194/acp-8-6439- 2008 Hughes, L.S., Allen, J.O., Kleeman, M.J., Johnson, R.J., Cass, G.R., Gross, D.S., Gard, E.E., Gälli, M.E., Morrical, B.D., Fergenson, D.P., Dienes, T., Noble, C.A., Liu, D.-Y., Silva, P.J., Prather, K.A., 1999. Size and composition distribution of atmospheric particles in Southern California. Environ. Sci. Technol. 33, 3506–3515. doi:10.1021/es980884a Hughes, L.S., Allen, J.O., Salmon, L.G., Mayo, P.R., Johnson, R.J., Cass, G.R., 2002. Evolution of nitrogen species air pollutants along trajectories crossing the Los Angeles area. Environ. Sci. Technol. 36, 3928–3935. doi:10.1021/es0110630 Hwang, I., Hopke, P.K., 2007. Estimation of source apportionment and potential source locations of PM2.5 at a west coastal IMPROVE site. Atmos. Environ. 41, 506–518. doi:10.1016/j.atmosenv.2006.08.043 Hwang, I., Hopke, P.K., 2006. Comparison of Source Apportionments of Fine Particulate Matter at Two San Jose Speciation Trends Network Sites. J. Air Waste Manag. Assoc. 56, 1287– 300. Ibanez, Y., Bot, B.L., Glorennec, P., 2010. House-dust metal content and bioaccessibility: a review. Eur. J. Mineral. 22, 629–637. doi:10.1127/0935-1221/2010/0022-2010 Jayne, J.T., Leard, D.C., Zhang, X., Davidovits, P., Smith, K.A., Kolb, C.E., Worsnop, D.R., 2000. Development of an Aerosol Mass Spectrometer for Size and Composition Analysis of Submicron Particles. Aerosol Sci. Technol. 33, 49–70. doi:10.1080/027868200410840 Jenkins, P.L., Phillips, T.J., Mulberg, E.J., Hui, S.P., 1992. Activity patterns of Californians: Use of and proximity to indoor pollutant sources. Atmospheric Environ. Part Gen. Top. 26, 2141–2148. doi:10.1016/0960-1686(92)90402-7 Jimenez, J.L., 2016. Concluding remarks: Faraday Discussion on chemistry in the urban atmosphere. Faraday Discuss. doi:10.1039/C6FD90019D Jones, N.C., Thornton, C.A., Mark, D., Harrison, R.M., 2000. Indoor/outdoor relationships of particulate matter in domestic homes with roadside, urban and rural locations. Atmos. Environ. 34, 2603–2612. doi:10.1016/S1352-2310(99)00489-6 Jovašević-Stojanović, M., Bartonova, A., Topalović, D., Lazović, I., Pokrić, B., Ristovski, Z., 2015. On the use of small and cheaper sensors and devices for indicative citizen-based monitoring of respirable particulate matter. Environ. Pollut. 206, 696–704. doi:10.1016/j.envpol.2015.08.035 156 Kam, W., Liacos, J.W., Schauer, J.J., Delfino, R.J., Sioutas, C., 2012. Size-segregated composition of particulate matter (PM) in major roadways and surface streets. Atmos. Environ. 55, 90– 97. doi:10.1016/j.atmosenv.2012.03.028 Katrinak, K.A., Anderson, J.R., Buseck, P.R., 1995. Individual Particle Types in the Aerosol of Phoenix, Arizona. Environ. Sci. Technol. 29, 321–329. doi:10.1021/es00002a007 Kelly, F.J., Fussell, J.C., 2012. Size, source and chemical composition as determinants of toxicity attributable to ambient particulate matter. Atmos. Environ. 60, 504–526. doi:10.1016/j.atmosenv.2012.06.039 Khoder, M.., 2002. Atmospheric conversion of sulfur dioxide to particulate sulfate and nitrogen dioxide to particulate nitrate and gaseous nitric acid in an urban area. Chemosphere 49, 675–684. doi:10.1016/S0045-6535(02)00391-0 Kim, E., Hopke, P.K., 2008. Source characterization of ambient fine particles at multiple sites in the Seattle area. Atmos. Environ. 42, 6047–6056. doi:10.1016/j.atmosenv.2008.03.032 Kim, E., Hopke, P.K., 2007a. Source characterization of ambient fine particles in the Los Angeles basin. J. Environ. Eng. Sci. 6, 343–353. doi:10.1139/S06-054 Kim, E., Hopke, P.K., 2007b. Source identifications of airborne fine particles using positive matrix factorization and U.S. Environmental Protection Agency positive matrix factorization. J. Air Waste Manag. Assoc. 57, 811–819. doi:10.3155/1047-3289.57.7.811 Kim, E., Hopke, P.K., Qin, Y., 2005. Estimation of organic carbon blank values and error structures of the speciation trends network data for source apportionment. J. Air Waste Manag. Assoc. 55, 1190–1199. Kim, E., Turkiewicz, K., Zulawnick, S.A., Magliano, K.L., 2010. Sources of fine particles in the South Coast area, California. Atmos. Environ. 44, 3095–3100. doi:10.1016/j.atmosenv.2010.05.037 Kim, K.-H., Kabir, E., Kabir, S., 2015. A review on the human health impact of airborne particulate matter. Environ. Int. 74, 136–143. doi:10.1016/j.envint.2014.10.005 Kleeman, M.J., Cass, G.R., 2001. A 3D Eulerian Source-Oriented Model for an Externally Mixed Aerosol. Environ. Sci. Technol. 35, 4834–4848. doi:10.1021/es010886m Kleeman, M.J., Riddle, S.G., Robert, M.A., Jakober, C.A., Fine, P.M., Hays, M.D., Schauer, J.J., Hannigan, M.P., 2009. Source Apportionment of Fine (PM1.8) and Ultrafine (PM0.1) Airborne Particulate Matter during a Severe Winter Pollution Episode. Environ. Sci. Technol. 43, 272–279. doi:10.1021/es800400m Kleeman, M.J., Robert, M.A., Riddle, S.G., Fine, P.M., Hays, M.D., Schauer, J.J., Hannigan, M.P., 2008. Erratum to “Size distribution of trace organic species emitted from biomass combustion and meat charbroiling” [Atoms. Environ. 42 (2008) 3059-3075] (DOI:10.1016/j.atmosenv.2007.12.044). Atmos. Environ. 42, 6152–6154. doi:10.1016/j.atmosenv.2007.12.075 Klepeis, N.E., Nelson, W.C., Ott, W.R., Robinson, J.P., Tsang, A.M., Switzer, P., Behar, J.V., Hern, S.C., Engelmann, W.H., 2001. The National Human Activity Pattern Survey (NHAPS): A resource for assessing exposure to environmental pollutants. J. Expo. Anal. Environ. Epidemiol. 11, 231–252. doi:10.1038/sj.jea.7500165 Koenig, J.Q., Mar, T.F., Allen, R.W., Jansen, K., Lumley, T., Sullivan, J.H., Trenga, C.A., Larson, T.V., Liu, L.-J.S., 2005. Pulmonary effects of indoor- and outdoor-generated particles in children with asthma. Environ. Health Perspect. 113, 499–503. doi:10.1289/ehp.7511 Koistinen, K.J., Edwards, R.D., Mathys, P., Ruuskanen, J., Künzli, N., Jantunen, M.J., 2004. Sources of fine particulate matter in personal exposures and residential indoor, residential 157 outdoor and workplace microenvironments in the Helsinki phase of the EXPOLIS study. Scand. J. Work. Environ. Health 30, 36–46. Kopperud, R.J., Ferro, A.R., Hildemann, L.M., 2004. Outdoor versus indoor contributions to indoor particulate matter (PM) determined by mass balance methods. J. Air Waste Manag. Assoc. 54, 1188–1196. Kozawa, K.H., Park, S.S., Mara, S.L., Herner, J.D., 2014. Verifying emission reductions from heavy-duty diesel trucks operating on southern California freeways. Environ. Sci. Technol. 48, 1475–1483. doi:10.1021/es4044177 Kuhn, T., Biswas, S., Sioutas, C., 2005. Diurnal and seasonal characteristics of particle volatility and chemical composition in the vicinity of a light-duty vehicle freeway. Atmos. Environ. 39, 7154–7166. doi:10.1016/j.atmosenv.2005.08.025 Kulmala, M., Vehkamäki, H., Petäjä, T., Dal Maso, M., Lauri, A., Kerminen, V.-M., Birmili, W., McMurry, P.H., 2004. Formation and growth rates of ultrafine atmospheric particles: a review of observations. J. Aerosol Sci. 35, 143–176. doi:10.1016/j.jaerosci.2003.10.003 Kumar, P., Morawska, L., Martani, C., Biskos, G., Neophytou, M., Di Sabatino, S., Bell, M., Norford, L., Britter, R., 2015. The rise of low-cost sensing for managing air pollution in cities. Environ. Int. 75, 199–205. doi:10.1016/j.envint.2014.11.019 Kuwayama, T., Schwartz, J.R., Harley, R.A., Kleeman, M.J., 2013. Particulate matter emissions reductions due to adoption of clean diesel technology at a major shipping port. Aerosol Sci. Technol. 47, 29–36. doi:10.1080/02786826.2012.720049 Laden, F., Neas, L.M., Dockery, D.W., Schwartz, J., 2000. Association of fine particulate matter from different sources with daily mortality in six U.S. cities. Environ. Health Perspect. 108, 941–947. Larson, T., Gould, T., Simpson, C., Liu, L.-J.S., Claiborn, C., Lewtas, J., 2004. Source apportionment of indoor, outdoor, and personal PM2.5 in Seattle, Washington, using positive matrix factorization. J. Air Waste Manag. Assoc. 54, 1175–1187. Lee, E., Chan, C.K., Paatero, P., 1999. Application of positive matrix factorization in source apportionment of particulate pollutants in Hong Kong. Atmos. Environ. 33, 3201–3212. doi:10.1016/S1352-2310(99)00113-2 Lee, G., You, S.I., Ritchie, S.G., Saphores, J.-D., Jayakrishnan, R., Ogunseitan, O., 2012. Assessing air quality and health benefits of the Clean Truck Program in the Alameda corridor, CA. Transp. Res. Part Policy Pract. 46, 1177–1193. doi:10.1016/j.tra.2012.05.005 Lee, H.J., Gent, J.F., Leaderer, B.P., Koutrakis, P., 2011. Spatial and temporal variability of fine particle composition and source types in five cities of Connecticut and Massachusetts. Sci. Total Environ. 409, 2133–2142. doi:10.1016/j.scitotenv.2011.02.025 Lee, J.H., Hopke, P.K., 2006. Apportioning sources of PM2.5 in St. Louis, MO using speciation trends network data. Atmos. Environ. 40, 360–377. doi:10.1016/j.atmosenv.2005.11.074 Lee, S., Liu, W., Wang, Y., Russell, A.G., Edgerton, E.S., 2008. Source apportionment of PM2.5: Comparing PMF and CMB results for four ambient monitoring sites in the southeastern United States. Atmos. Environ. 42, 4126–4137. doi:10.1016/j.atmosenv.2008.01.025 Li, N., Sioutas, C., Cho, A., Schmitz, D., Misra, C., Sempf, J., Wang, M., Oberley, T., Froines, J., Nel, A., 2003. Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage. Environ. Health Perspect. 111, 455–460. Liacos, J.W., Kam, W., Delfino, R.J., Schauer, J.J., Sioutas, C., 2012. Characterization of organic, metal and trace element PM2.5 species and derivation of freeway-based emission rates in 158 Los Angeles, CA. Sci. Total Environ. 435–436, 159–166. doi:10.1016/j.scitotenv.2012.06.106 Lim, S.S., Vos, T., Flaxman, A.D., Danaei, G., Shibuya, K., Adair-Rohani, H., AlMazroa, M.A., Amann, M., Anderson, H.R., Andrews, K.G., Aryee, M., Atkinson, C., Bacchus, L.J., Bahalim, A.N., Balakrishnan, K., Balmes, J., Barker-Collo, S., Baxter, A., Bell, M.L., Blore, J.D., Blyth, F., Bonner, C., Borges, G., Bourne, R., Boussinesq, M., Brauer, M., Brooks, P., Bruce, N.G., Brunekreef, B., Bryan-Hancock, C., Bucello, C., Buchbinder, R., Bull, F., Burnett, R.T., Byers, T.E., Calabria, B., Carapetis, J., Carnahan, E., Chafe, Z., Charlson, F., Chen, H., Chen, J.S., Cheng, A.T.-A., Child, J.C., Cohen, A., Colson, K.E., Cowie, B.C., Darby, S., Darling, S., Davis, A., Degenhardt, L., Dentener, F., Des Jarlais, D.C., Devries, K., Dherani, M., Ding, E.L., Dorsey, E.R., Driscoll, T., Edmond, K., Ali, S.E., Engell, R.E., Erwin, P.J., Fahimi, S., Falder, G., Farzadfar, F., Ferrari, A., Finucane, M.M., Flaxman, S., Fowkes, F.G.R., Freedman, G., Freeman, M.K., Gakidou, E., Ghosh, S., Giovannucci, E., Gmel, G., Graham, K., Grainger, R., Grant, B., Gunnell, D., Gutierrez, H.R., Hall, W., Hoek, H.W., Hogan, A., Hosgood III, H.D., Hoy, D., Hu, H., Hubbell, B.J., Hutchings, S.J., Ibeanusi, S.E., Jacklyn, G.L., Jasrasaria, R., Jonas, J.B., Kan, H., Kanis, J.A., Kassebaum, N., Kawakami, N., Khang, Y.-H., Khatibzadeh, S., Khoo, J.-P., Kok, C., Laden, F., Lalloo, R., Lan, Q., Lathlean, T., Leasher, J.L., Leigh, J., Li, Y., Lin, J.K., Lipshultz, S.E., London, S., Lozano, R., Lu, Y., Mak, J., Malekzadeh, R., Mallinger, L., Marcenes, W., March, L., Marks, R., Martin, R., McGale, P., McGrath, J., Mehta, S., Memish, Z.A., Mensah, G.A., Merriman, T.R., Micha, R., Michaud, C., Mishra, V., Hanafiah, K.M., Mokdad, A.A., Morawska, L., Mozaffarian, D., Murphy, T., Naghavi, M., Neal, B., Nelson, P.K., Nolla, J.M., Norman, R., Olives, C., Omer, S.B., Orchard, J., Osborne, R., Ostro, B., Page, A., Pandey, K.D., Parry, C.D., Passmore, E., Patra, J., Pearce, N., Pelizzari, P.M., Petzold, M., Phillips, M.R., Pope, D., Pope III, C.A., Powles, J., Rao, M., Razavi, H., Rehfuess, E.A., Rehm, J.T., Ritz, B., Rivara, F.P., Roberts, T., Robinson, C., Rodriguez-Portales, J.A., Romieu, I., Room, R., Rosenfeld, L.C., Roy, A., Rushton, L., Salomon, J.A., Sampson, U., Sanchez-Riera, L., Sanman, E., Sapkota, A., Seedat, S., Shi, P., Shield, K., Shivakoti, R., Singh, G.M., Sleet, D.A., Smith, E., Smith, K.R., Stapelberg, N.J., Steenland, K., Stöckl, H., Stovner, L.J., Straif, K., Straney, L., Thurston, G.D., Tran, J.H., Van Dingenen, R., van Donkelaar, A., Veerman, J.L., Vijayakumar, L., Weintraub, R., Weissman, M.M., White, R.A., Whiteford, H., Wiersma, S.T., Wilkinson, J.D., Williams, H.C., Williams, W., Wilson, N., Woolf, A.D., Yip, P., Zielinski, J.M., Lopez, A.D., Murray, C.J., Ezzati, M., 2012. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet 380, 2224– 2260. doi:10.1016/S0140-6736(12)61766-8 Lin, C.-C., Chen, S.-J., Huang, K.-L., Lee, W.-J., Lin, W.-Y., Tsai, J.-H., Chaung, H.-C., 2008. PAHs, PAH-induced carcinogenic potency, and particle-extract-induced cytotoxicity of traffic-related nano/ultrafine particles. Environ. Sci. Technol. 42, 4229–4235. doi:10.1021/es703107w Liu, D.-L., Nazaroff, W.W., 2003. Particle penetration through building cracks. Aerosol Sci. Technol. 37, 565–573. doi:10.1080/02786820300927 Liu, W., Wang, Y., Russell, A., Edgerton, E.S., 2006. Enhanced source identification of southeast aerosols using temperature-resolved carbon fractions and gas phase components. Atmos. Environ. 40, 445–466. doi:10.1016/j.atmosenv.2005.11.079 159 Liu, Y., Zhu, L., Shen, X., 2001. Polycyclic aromatic hydrocarbons (PAHs) in indoor and outdoor air of Hangzhou, China. Environ. Sci. Technol. 35, 840–844. doi:10.1021/es001354t Long, C.M., Suh, H.H., Catalano, P.J., Koutrakis, P., 2001. Using time- and size-resolved particulate data to ouantify indoor penetration and deposition behavior. Environ. Sci. Technol. 35, 2089–2099. doi:10.1021/es001477d Long, C.M., Suh, H.H., Koutrakis, P., 2000. Characterization of indoor particle sources using continuous mass and size monitors. J. Air Waste Manag. Assoc. 50, 1236–1250. Lough, G.C., Christensen, C.G., Schauer, J.J., Tortorelli, J., Mani, E., Lawson, D.R., Clark, N.N., Gabele, P.A., 2007. Development of molecular marker source profiles for emissions from on-road gasoline and diesel vehicle fleets. J. Air Waste Manag. Assoc. 57, 1190–1199. doi:10.3155/1047-3289.57.10.1190 Lunden, M.M., Thatcher, T.L., Hering, S.V., Brown, N.J., 2003. Use of time- and chemically resolved particulate data to characterize the infiltration of outdoor PM2.5 into a residence in the San Joaquin Valley. Environ. Sci. Technol. 37, 4724–4732. doi:10.1021/es026387i Mader, B.T., Pankow, J.F., 2002. Study of the effects of particle-phase carbon on the gas/particle partitioning of semivolatile organic compounds in the atmosphere using controlled field experiments. Environ. Sci. Technol. 36, 5218–5228. doi:10.1021/es011048v Manchester-Neesvig, J.B., Schauer, J.J., Cass, G.R., 2003. The distribution of particle-phase organic compounds in the atmosphere and their use for source apportionment during the Southern California children’s health study. J. Air Waste Manag. Assoc. 53, 1065–1079. Mar, T.F., Koenig, J.Q., Primomo, J., 2010. Associations between asthma emergency visits and particulate matter sources, including diesel emissions from stationary generators in Tacoma, Washington. Inhal. Toxicol. 22, 445–448. doi:10.3109/08958370903575774 Mar, T.F., Norris, G.A., Koenig, J.Q., Larson, T.V., 2000. Associations between air pollution and mortality in Phoenix, 1995-1997. Environ. Health Perspect. 108, 347–353. Mathai, C.V., 1990. Visibility and fine particles : a summary of the A&WMA/EPA international speciality conference. J. Air Waste Manag. Assoc. 40, 1486–1494. McDonald, B.C., Dallmann, T.R., Martin, E.W., Harley, R.A., 2012. Long-term trends in nitrogen oxide emissions from motor vehicles at national, state, and air basin scales. J. Geophys. Res. Atmospheres 117, D00V18. doi:10.1029/2012JD018304 Meng, Q.Y., Turpin, B.J., Lee, J.H., Polidori, A., Weisel, C.P., Morandi, M., Colome, S., Zhang, J., Stock, T., Winer, A., 2007. How Does Infiltration Behavior Modify the Composition of Ambient PM2.5 in Indoor Spaces? An Analysis of RIOPA Data. Environ. Sci. Technol. 41, 7315–7321. doi:10.1021/es070037k Miet, K., Le, M., Flaud, P.-M., Budzinski, H., Villenave, E., 2009. Heterogeneous reactivity of pyrene and 1-nitropyrene with NO2: Kinetics, product yields and mechanism. Atmos. Environ. 43, 837–843. doi:10.1016/j.atmosenv.2008.10.041 Miller, K.A., Siscovick, D.S., Sheppard, L., Shepherd, K., Sullivan, J.H., Anderson, G.L., Kaufman, J.D., 2007. Long-term exposure to air pollution and incidence of cardiovascular events in women. N. Engl. J. Med. 356, 447–458. doi:10.1056/NEJMoa054409 Miller-Schulze, J.P., Shafer, M.M., Schauer, J.J., Solomon, P.A., Lantz, J., Artamonova, M., Chen, B., Imashev, S., Sverdlik, L., Carmichael, G.R., Deminter, J.T., 2011. Characteristics of fine particle carbonaceous aerosol at two remote sites in Central Asia. Atmos. Environ. 45, 6955–6964. doi:10.1016/j.atmosenv.2011.09.026 Millstein, D.E., Harley, R.A., Hering, S.V., 2008. Weekly cycles in fine particulate nitrate. Atmos. Environ. 42, 632–641. doi:10.1016/j.atmosenv.2007.10.010 160 Minguillón, M.C., Arhami, M., Schauer, J.J., Sioutas, C., 2008. Seasonal and spatial variations of sources of fine and quasi-ultrafine particulate matter in neighborhoods near the Los Angeles-Long Beach harbor. Atmos. Environ. 42, 7317–7328. doi:10.1016/j.atmosenv.2008.07.036 Minguillón, M.C., Schembari, A., Triguero-Mas, M., de, N., Dadvand, P., Figueras, F., Salvado, J.A., Grimalt, J.O., Nieuwenhuijsen, M., Querol, X., 2012. Source apportionment of indoor, outdoor and personal PM2.5 exposure of pregnant women in Barcelona, Spain. Atmos. Environ. 59, 426–436. doi:10.1016/j.atmosenv.2012.04.052 Misra, C., Singh, M., Shen, S., Sioutas, C., Hall, P.M., 2002. Development and evaluation of a personal cascade impactor sampler (PCIS). J. Aerosol Sci. 33, 1027–1047. doi:10.1016/S0021-8502(02)00055-1 Møller, P., Jacobsen, N.R., Folkmann, J.K., Danielsen, P.H., Mikkelsen, L., Hemmingsen, J.G., Vesterdal, L.K., Forchhammer, L., Wallin, H., Loft, S., 2010. Role of oxidative damage in toxicity of particulates. Free Radic. Res. 44, 1–46. doi:10.3109/10715760903300691 Moore, K., Krudysz, M., Pakbin, P., Hudda, N., Sioutas, C., 2009. Intra-community variability in total particle number concentrations in the san pedro harbor area (Los Angeles, California). Aerosol Sci. Technol. 43, 587–603. doi:10.1080/02786820902800900 Moore, K.F., Ning, Z., Ntziachristos, L., Schauer, J.J., Sioutas, C., 2007. Daily variation in the properties of urban ultrafine aerosol-Part I: Physical characterization and volatility. Atmos. Environ. 41, 8633–8646. doi:10.1016/j.atmosenv.2007.07.030 Morawska, L., Ristovski, Z., Jayaratne, E.R., Keogh, D.U., Ling, X., 2008. Ambient nano and ultrafine particles from motor vehicle emissions: Characteristics, ambient processing and implications on human exposure. Atmos. Environ. 42, 8113–8138. doi:10.1016/j.atmosenv.2008.07.050 Mozurkewich, M., 1993. The dissociation constant of ammonium nitrate and its dependence on temperature, relative humidity and particle size. Atmospheric Environ. Part Gen. Top. 27, 261–270. doi:10.1016/0960-1686(93)90356-4 Naik, D.V., Weschler, C.J., Shields, H.C., 1991. Indoor and outdoor concentrations of organic compounds associated with airborne particles: Results using a novel solvent system. Indoor Air Pollut. Radon Bioaerosols VOCs 67. Nel, A., 2005. Air Pollution-Related Illness: Effects of Particles. Science 308, 804–806. doi:10.1126/science.1108752 Ng, N.L., Herndon, S.C., Trimborn, A., Canagaratna, M.R., Croteau, P.L., Onasch, T.B., Sueper, D., Worsnop, D.R., Zhang, Q., Sun, Y.L., Jayne, J.T., 2011. An Aerosol Chemical Speciation Monitor (ACSM) for Routine Monitoring of the Composition and Mass Concentrations of Ambient Aerosol. Aerosol Sci. Technol. 45, 780–794. doi:10.1080/02786826.2011.560211 Nicolaides, N., 1974. Skin lipids: their biochemical uniqueness. Science 186, 19–26. Nieuwenhuijsen, M.J., Dadvand, P., Grellier, J., Martinez, D., Vrijheid, M., 2013. Environmental risk factors of pregnancy outcomes: a summary of recent meta-analyses of epidemiological studies. Environ. Health 12, 6. doi:10.1186/1476-069X-12-6 Norris, G.A., Vedantham, R., Wade, K., Brown, S., Prouty, J., Foley, C., 2008. EPA Positive Matrix Factorization (PMF) 3.0 Fundamentals & User Guide. Prepared for the US Environmental Protection Agency, Washington, DC, by the National Exposure Research Laboratory, Research Triangle Park. 161 Nowak, D.J., Crane, D.E., Stevens, J.C., 2006. Air pollution removal by urban trees and shrubs in the United States. Urban For. Urban Green. 4, 115–123. doi:10.1016/j.ufug.2006.01.007 Ntziachristos, L., Froines, J.R., Cho, A.K., Sioutas, C., 2007. Relationship between redox activity and chemical speciation of size-fractionated particulate matter. Part. Fibre Toxicol. 4, 5. doi:10.1186/1743-8977-4-5 Ntziachristos, L., Ning, Z., Geller, M.D., Sheesley, R.J., Schauer, J.J., Sioutas, C., 2007a. Fine, ultrafine and nanoparticle trace element compositions near a major freeway with a high heavy-duty diesel fraction. Atmos. Environ. 41, 5684–5696. doi:10.1016/j.atmosenv.2007.02.043 Ntziachristos, L., Ning, Z., Geller, M.D., Sioutas, C., 2007b. Particle concentration and characteristics near a major freeway with heavy-duty diesel traffic. Environ. Sci. Technol. 41, 2223–2230. doi:10.1021/es062590s Oeder, S., Dietrich, S., Weichenmeier, I., Schober, W., Pusch, G., Jörres, R.A., Schierl, R., Nowak, D., Fromme, H., Behrendt, H., Buters, J.T.M., 2012. Toxicity and elemental composition of particulate matter from outdoor and indoor air of elementary schools in Munich, Germany. Indoor Air 22, 148–158. doi:10.1111/j.1600-0668.2011.00743.x Ogulei, D., Hopke, P.K., Wallace, L.A., 2006. Analysis of indoor particle size distributions in an occupied townhouse using positive matrix factorization. Indoor Air 16, 204–215. doi:10.1111/j.1600-0668.2006.00418.x Ohura, T., Amagai, T., Sugiyama, T., Fusaya, M., Matsushita, H., 2004. Characteristics of particle matter and associated polycyclic aromatic hydrocarbons in indoor and outdoor air in two cities in Shizuoka, Japan. Atmos. Environ. 38, 2045–2054. doi:10.1016/j.atmosenv.2004.01.038 Ohura, T., Sugiyama, T., Amagai, T., Fusaya, M., Matsushita, H., 2002. Simultaneous liquid chromatographic determination of 39 polycyclic aromatic hydrocarbons in indoor and outdoor air and application to a survey on indoor air pollution in Fuji, Japan. J. AOAC Int. 85, 188–202. Olson, D.A., Turlington, J., Duvall, R.M., McDow, S.R., Stevens, C.D., Williams, R., 2008. Indoor and outdoor concentrations of organic and inorganic molecular markers: Source apportionment of PM2.5 using low-volume samples. Atmos. Environ. 42, 1742–1751. doi:10.1016/j.atmosenv.2007.11.035 Orasche, J., Seidel, T., Hartmann, H., Schnelle-Kreis, J., Chow, J.C., Ruppert, H., Zimmermann, R., 2012. Comparison of emissions from wood combustion. part 1: Emission factors and characteristics from different small-scale residential heating appliances considering particulate matter and polycyclic aromatic hydrocarbon (PAH)-related toxicological potential of particle-bound organic species. Energy Fuels 26, 6695–6704. doi:10.1021/ef301295k Ostro, B., Broadwin, R., Green, S., Feng, W.-Y., Lipsett, M., 2006. Fine Particulate Air Pollution and Mortality in Nine California Counties: Results from CALFINE. Environ. Health Perspect. 114, 29–33. doi:10.1289/ehp.8335 Ostro, B., Feng, W.-Y., Broadwin, R., Green, S., Lipsett, M., 2007. The effects of components of fine particulate air pollution on mortality in California : Results from CALFINE. Environ. Health Perspect. 115, 13–19. Ostro, B., Hu, J., Goldberg, D., Reynolds, P., Hertz, A., Bernstein, L., Kleeman, M.J., 2015. Associations of Mortality with Long-Term Exposures to Fine and Ultrafine Particles, 162 Species and Sources: Results from the California Teachers Study Cohort. Environ. Health Perspect. Online 123, 549. doi:http://dx.doi.org/10.1289/ehp.1408565 Ostro, B., Malig, B., Broadwin, R., Basu, R., Gold, E.B., Bromberger, J.T., Derby, C., Feinstein, S., Greendale, G.A., Jackson, E.A., Kravitz, H.M., Matthews, K.A., Sternfeld, B., Tomey, K., Green, R.R., Green, R., 2014. Chronic PM2.5 exposure and inflammation: Determining sensitive subgroups in mid-life women. Environ. Res. 132, 168–175. doi:10.1016/j.envres.2014.03.042 Ostro, B., Tobias, A., Querol, X., Alastuey, A., Amato, F., Pey, J., Pérez, N., Sunyer, J., 2011. The Effects of Particulate Matter Sources on Daily Mortality: A Case-Crossover Study of Barcelona, Spain. Environ. Health Perspect. 119, 1781–7. Özkaynak, H., Thurston, G.D., 1987. Associations Between 1980 U.S. Mortality Rates and Alternative Measures of Airborne Particle Concentration. Risk Anal. 7, 449–461. doi:10.1111/j.1539-6924.1987.tb00482.x Paatero, P., 1997. Least squares formulation of robust non-negative factor analysis. Chemom. Intell. Lab. Syst. 37, 23–35. doi:10.1016/S0169-7439(96)00044-5 Paatero, P., Hopke, P.K., Song, X.-H., Ramadan, Z., 2002. Understanding and controlling rotations in factor analytic models. Chemom. Intell. Lab. Syst. 60, 253–264. doi:10.1016/S0169- 7439(01)00200-3 Paatero, P., Tapper, U., 1994. Positive matrix factorization: A non-negative factor model with optimal utilization of error estimates of data values. Environmetrics 5, 111–126. doi:10.1002/env.3170050203 Pakbin, P., Hudda, N., Cheung, K.L., Moore, K.F., Sioutas, C., 2010. Spatial and temporal variability of coarse (PM10-2.5) particulate matter concentrations in the Los Angeles Area. Aerosol Sci. Technol. 44, 514–525. doi:10.1080/02786821003749509 Pakbin, P., Ning, Z., Shafer, M.M., Schauer, J.J., Sioutas, C., 2011. Seasonal and spatial coarse particle elemental concentrations in the Los Angeles area. Aerosol Sci Technol 45, 949– 963. Pandis, S.N., Skyllakou, K., Florou, K., Kostenidou, E., Kaltsonoudis, C., Hasa, E., Presto, A.A., 2016. Urban particulate matter pollution: a tale of five cities. Faraday Discuss. doi:10.1039/C5FD00212E Pant, P., Harrison, R.M., 2013. Estimation of the contribution of road traffic emissions to particulate matter concentrations from field measurements: A review. Atmos. Environ. 77, 78–97. doi:10.1016/j.atmosenv.2013.04.028 Park, S., Cho, S., Jo, M., Gong, B., Park, J., Lee, S., 2014. Field evaluation of a near–real time elemental monitor and identification of element sources observed at an air monitoring supersite in Korea. Atmospheric Pollut. Res. 5, 119–128. doi:10.5094/APR.2014.015 Peng RD, Chang HH, Bell ML, et al, 2008. COarse particulate matter air pollution and hospital admissions for cardiovascular and respiratory diseases among medicare patients. JAMA 299, 2172–2179. doi:10.1001/jama.299.18.2172 Peters, A., Veronesi, B., Calderón-Garcidueñas, L., Gehr, P., Chen, L.C., Geiser, M., Reed, W., Rothen-Rutishauser, B., Schürch, S., Schulz, H., 2006. Translocation and potential neurological effects of fine and ultrafine particles a critical update. Part. Fibre Toxicol. 3, 13. doi:10.1186/1743-8977-3-13 Phuleria, H.C., Sheesley, R.J., Schauer, J.J., Fine, P.M., Sioutas, C., 2007. Roadside measurements of size-segregated particulate organic compounds near gasoline and diesel-dominated 163 freeways in Los Angeles, CA. Atmos. Environ. 41, 4653–4671. doi:10.1016/j.atmosenv.2007.03.031 Piedrahita, R., Xiang, Y., Masson, N., Ortega, J., Collier, A., Jiang, Y., Li, K., Dick, R.P., Lv, Q., Hannigan, M., Shang, L., 2014. The next generation of low-cost personal air quality sensors for quantitative exposure monitoring. Atmospheric Meas. Tech. 7, 3325. doi:http://dx.doi.org/10.5194/amt-7-3325-2014 POBL, 2010. San Pedro Bay Ports Clean Air Action Plan 2010 Update. Polidori, A., Cheung, K.L., Arhami, M., Delfino, R.J., Schauer, J.J., Sioutas, C., 2009. Relationships between size-fractionated indoor and outdoor trace elements at four retirement communities in southern California. Atmos Chem Phys 9, 4521–4536. doi:10.5194/acp-9-4521-2009 Polidori, A., Hu, S., Biswas, S., Delfino, R.J., Sioutas, C., 2007. Real-time characterization of particle-bound polycyclic aromatic hydrocarbons in ambient aerosols and from motor- vehicle exhaust. Atmospheric Chem. Phys. 7, 17475–17510. Polidori, A., Turpin, B., Meng, Q.Y., Lee, J.H., Weisel, C., Morandi, M., Colome, S., Stock, T., Winer, A., Zhang, J., Kwon, J., Alimokhtari, S., Shendell, D., Jones, J., Farrar, C., Maberti, S., 2006. Fine organic particulate matter dominates indoor-generated PM2.5 in RIOPA homes. J. Expo. Sci. Environ. Epidemiol. 16, 321–331. doi:10.1038/sj.jes.7500476 Polissar, A.V., Hopke, P.K., Paatero, P., Malm, W.C., Sisler, J.F., 1998. Atmospheric aerosol over Alaska 2. Elemental composition and sources. J. Geophys. Res. Atmospheres 103, 19045– 19057. Pope, I., Burnett, R.T., Thurston, G.D., Thun, M.J., Calle, E.E., Krewski, D., Godleski, J.J., 2004. Cardiovascular Mortality and Long-Term Exposure to Particulate Air Pollution: Epidemiological Evidence of General Pathophysiological Pathways of Disease. Circulation 109, 71–77. doi:10.1161/01.CIR.0000108927.80044.7F Pope, I., Dockery, D.W., 2006. Health effects of fine particulate air pollution: Lines that connect. J. Air Waste Manag. Assoc. 56, 709–742. Pun, V.C., Yu, I.T., Ho, K., Qiu, H., Sun, Z., Tian, L., 2014. Differential Effects of Source-Specific Particulate Matter on Emergency Hospitalizations for Ischemic Heart Disease in Hong Kong. Environ. Health Perspect. 122, 391–396. doi:10.1289/ehp.1307213 Ranft, U., Schikowski, T., Sugiri, D., Krutmann, J., Krämer, U., 2009. Long-term exposure to traffic-related particulate matter impairs cognitive function in the elderly. Environ. Res. 109, 1004–1011. doi:10.1016/j.envres.2009.08.003 Reff, A., Eberly, S.I., Bhave, P.V., 2007. Receptor modeling of ambient particulate matter data using positive matrix factorization: Review of existing methods. J. Air Waste Manag. Assoc. 57, 146–154. Rim, D., Wallace, L., Persily, A., 2010. Infiltration of Outdoor Ultrafine Particles into a Test House. Environ. Sci. Technol. 44, 5908–5913. doi:10.1021/es101202a Rizzo, M.J., Scheff, P.A., 2007. Fine particulate source apportionment using data from the USEPA speciation trends network in Chicago, Illinois: Comparison of two source apportionment models. Atmos. Environ. 41, 6276–6288. doi:10.1016/j.atmosenv.2007.03.055 Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1997. Sources of fine organic aerosol. 8. Boilers burning no. 2 Distillate fuel oil. Environ. Sci. Technol. 31, 2731–2737. doi:10.1021/es9609563 164 Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993a. Sources of fine organic aerosol. 4. Particulate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol. 27, 2700–2711. Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993b. Sources of fine organic aerosol. 5. Natural gas home appliances. Environ. Sci. Technol. 27, 2736– 2744. Rohr, A.C., Wyzga, R.E., 2012. Attributing health effects to individual particulate matter constituents. Atmos. Environ. 62, 130–152. doi:10.1016/j.atmosenv.2012.07.036 Roy, A.A., Wagstrom, K.M., Adams, P.J., Pandis, S.N., Robinson, A.L., 2011. Quantification of the effects of molecular marker oxidation on source apportionment estimates for motor vehicles. Atmos. Environ. 45, 3132–3140. doi:10.1016/j.atmosenv.2011.03.020 RTI, 2009. Standard Operating Procedure for the X-ray Fluorescence Analysis of Particulate Matter Deposits on Teflon Filters, Research Triangle Park, NC. Rückerl, R., Schneider, A., Hampel, R., Breitner, S., Cyrys, J., Kraus, U., Gu, J., Soentgen, J., Koenig, W., Peters, A., 2016. Association of novel metrics of particulate matter with vascular markers of inflammation and coagulation in susceptible populations –results from a panel study. Environ. Res. 150, 337–347. doi:10.1016/j.envres.2016.05.037 Ruehl, C.R., Ham, W.A., Kleeman, M.J., 2011. Temperature-induced volatility of molecular markers in ambient airborne particulate matter. Atmos Chem Phys 11, 67–76. doi:10.5194/acp-11-67-2011 Saarnio, K., Sillanpää, M., Hillamo, R., Sandell, E., Pennanen, A.S., Salonen, R.O., 2008. Polycyclic aromatic hydrocarbons in size-segregated particulate matter from six urban sites in Europe. Atmos. Environ. 42, 9087–9097. doi:10.1016/j.atmosenv.2008.09.022 Saffari, A., Daher, N., Shafer, M.M., Schauer, J.J., Sioutas, C., 2013. Seasonal and spatial variation in reactive oxygen species activity of quasi-ultrafine particles (PM0.25) in the Los Angeles metropolitan area and its association with chemical composition. Atmos. Environ. 79, 566– 575. doi:10.1016/j.atmosenv.2013.07.058 Saldiva, P.H.N., Clarke, R.W., Coull, B.A., Stearns, R.C., Lawrence, J., Murthy, G.G.K., Diaz, E., Koutrakis, P., Suh, H., Tsuda, A., Godleski, J.J., 2002. Lung Inflammation Induced by Concentrated Ambient Air Particles Is Related to Particle Composition. Am. J. Respir. Crit. Care Med. 165, 1610–1617. doi:10.1164/rccm.2106102 Sannigrahi, P., Sullivan, A.P., Weber, R.J., Ingall, E.D., 2006. Characterization of water-soluble organic carbon in urban atmospheric aerosols using solid-state13C NMR spectroscopy. Environ. Sci. Technol. 40, 666–672. doi:10.1021/es051150i Sardar, S.B., Fine, P.M., Mayo, P.R., Sioutas, C., 2005. Size-fractionated measurements of ambient ultrafine particle chemical composition in Los Angeles using the NanoMOUDI. Environ. Sci. Technol. 39, 932–944. doi:10.1021/es049478j Sardar, S.B., Fine, P.M., Sioutas, C., 2005. Seasonal and spatial variability of the size-resolved chemical composition of particulate matter (PM10) in the Los Angeles Basin. J. Geophys. Res. Atmospheres 110, D07S08. doi:10.1029/2004JD004627 Sarnat, J.A., Marmur, A., Klein, M., Kim, E., Russell, A.G., Sarnat, S.E., Mulholland, J.A., Hopke, P.K., Tolbert, P.E., 2008. Fine Particle Sources and Cardiorespiratory Morbidity: An Application of Chemical Mass Balance and Factor Analytical Source-Apportionment Methods. Environ. Health Perspect. 116, 459–66. 165 Sarnat, S.E., Coull, B.A., Ruiz, P.A., Koutrakis, P., Suh, H.H., 2006. The Influences of Ambient Particle Composition and Size on Particle Infiltration in Los Angeles, CA, Residences. J. Air Waste Manag. Assoc. 56, 186–196. doi:10.1080/10473289.2006.10464449 Sarwar, G., Olson, D.A., Corsi, R.L., Weschler, C.J., 2004. Indoor Fine Particles: The Role of Terpene Emissions from Consumer Products. J. Air Waste Manag. Assoc. 54, 367–377. doi:10.1080/10473289.2004.10470910 Schauer, J.J., 2003. Evaluation of elemental carbon as a marker for diesel particulate matter. J. Expo. Sci. Environ. Epidemiol. 13, 443–453. doi:10.1038/sj.jea.7500298 Schauer, J.J., 1998. Source contributions to atmospheric organic compound concentrations : emissions measurements and model predictions (phd). California Institute of Technology. Schauer, J.J., Cass, G.R., 2000. Source Apportionment of Wintertime Gas-Phase and Particle- Phase Air Pollutants Using Organic Compounds as Tracers. Environ. Sci. Technol. 34, 1821–1832. doi:10.1021/es981312t Schauer, J.J., Kleeman, M.J., Cass, G.R., Simoneit, B.R.T., 1999. Measurement of Emissions from Air Pollution Sources. 1. C1 through C29 Organic Compounds from Meat Charbroiling. Environ. Sci. Technol. 33, 1566–1577. doi:10.1021/es980076j Schauer, J.J., Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1996. Source apportionment of airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30, 3837–3855. doi:10.1016/1352-2310(96)00085-4 Seinfeld, J.H., Pandis, S.N., 2006. Atmospheric Chemistry and Physics: from Air Pollution to Climate Change, Second. ed. John Wiley and Sons. Shah, S.D., Cocker, I., Miller, J.W., Norbeck, J.M., 2004. Emission Rates of Particulate Matter and Elemental and Organic Carbon from In-Use Diesel Engines. Environ. Sci. Technol. 38, 2544–2550. doi:10.1021/es0350583 Sheesley, R.J., Schauer, J.J., Zheng, M., Wang, B., 2007. Sensitivity of molecular marker-based CMB models to biomass burning source profiles. Atmos. Environ. 41, 9050–9063. doi:10.1016/j.atmosenv.2007.08.011 Shi, J.P., Khan, A.A., Harrison, R.M., 1999. Measurements of ultrafine particle concentration and size distribution in the urban atmosphere. Sci. Total Environ. 235, 51–64. doi:10.1016/S0048-9697(99)00189-8 Shi, T., Schins, R.P.F., Knaapen, A.M., Kuhlbusch, T., Pitz, M., Heinrich, J., Borm, P.J.A., 2003. Hydroxyl radical generation by electron paramagnetic resonance as a new method to monitor ambient particulate matter composition. J. Environ. Monit. 5, 550–556. doi:10.1039/B303928P Shirmohammadi, F., Hasheminassab, S., Wang, D., Schauer, J.J., Shafer, M.M., Delfino, R.J., Sioutas, C., 2016. The relative importance of tailpipe and non-tailpipe emissions on the oxidative potential of ambient particles in Los Angeles, CA. Faraday Discuss. doi:10.1039/C5FD00166H Simoneit, B.R.T., 1986. Characterization of Organic Constituents in Aerosols in Relation to Their rigin and Transport: A Review. Int. J. Environ. Anal. Chem. 23, 207–237. doi:10.1080/03067318608076446 Simoneit, B.R.T., Schauer, J.J., Nolte, C.G., Oros, D.R., Elias, V.O., Fraser, M.P., Rogge, W.F., Cass, G.R., 1999. Levoglucosan, a tracer for cellulose in biomass burning and atmospheric particles. Atmos. Environ. 33, 173–182. Singer, B.C., Coleman, B.K., Destaillats, H., Hodgson, A.T., Lunden, M.M., Weschler, C.J., Nazaroff, W.W., 2006. Indoor secondary pollutants from cleaning product and air 166 freshener use in the presence of ozone. Atmos. Environ. 40, 6696–6710. doi:10.1016/j.atmosenv.2006.06.005 Singh, M., Jaques, P.A., Sioutas, C., 2002. Size distribution and diurnal characteristics of particle- bound metals in source and receptor sites of the Los Angeles Basin. Atmos. Environ., NADP 2000 - Ten Years After the Clean Air Act Amendments 36, 1675–1689. doi:10.1016/S1352-2310(02)00166-8 Sioutas, C., Delfino, R.J., Singh, M., 2005. Exposure assessment for atmospheric Ultrafine Particles (UFPs) and implications in epidemiologic research. Environ. Health Perspect. 113, 947–955. doi:10.1289/ehp.7939 Siponen, T., Yli-Tuomi, T., Aurela, M., Dufva, H., Hillamo, R., Hirvonen, M.-R., Huttunen, K., Pekkanen, J., Pennanen, A., Salonen, I., Tiittanen, P., Salonen, R.O., Lanki, T., 2015. Source-specific fine particulate air pollution and systemic inflammation in ischaemic heart disease patients. Occup. Environ. Med. 72, 277–283. doi:10.1136/oemed-2014-102240 Snyder, D.C., Rutter, A.P., Collins, R., Worley, C., Schauer, J.J., 2009. Insights into the origin of water soluble organic carbon in atmospheric fine particulate matter. Aerosol Sci. Technol. 43, 1099–1107. doi:10.1080/02786820903188701 Song, C.H., Carmichael, G.R., 1999. The aging process of naturally emitted aerosol (sea-salt and mineral aerosol) during long range transport. Atmos. Environ. 33, 2203–2218. doi:10.1016/S1352-2310(98)00301-X Sørensen, M., Daneshvar, B., Hansen, M., Dragsted, L.O., Hertel, O., Knudsen, L., Loft, S., 2003. Personal PM2.5 exposure and markers of oxidative stress in blood. Environ. Health Perspect. 111, 161–166. Sowlat, M.H., Hasheminassab, S., Sioutas, C., 2016. Source apportionment of ambient particle number concentrations in central Los Angeles using positive matrix factorization (PMF). Atmos Chem Phys 16, 4849–4866. doi:10.5194/acp-16-4849-2016 Stanek, L.W., Sacks, J.D., Dutton, S.J., Dubois, J.-J.B., 2011. Attributing health effects to apportioned components and sources of particulate matter: An evaluation of collective results. Atmos. Environ. 45, 5655–5663. doi:10.1016/j.atmosenv.2011.07.023 Stelson, A.W., Seinfeld, J.H., 1982. Relative humidity and temperature dependence of the ammonium nitrate dissociation constant. Atmospheric Environ. 1967 16, 983–992. doi:10.1016/0004-6981(82)90184-6 Stone, E.A., Hedman, C.J., Sheesley, R.J., Shafer, M.M., Schauer, J.J., 2009. Investigating the chemical nature of humic-like substances (HULIS) in North American atmospheric aerosols by liquid chromatography tandem mass spectrometry. Atmos. Environ. 43, 4205– 4213. doi:10.1016/j.atmosenv.2009.05.030 Stone, E.A., Snyder, D.C., Sheesley, R.J., Sullivan, A.P., Weber, R.J., Schauer, J.J., 2008. Source apportionment of fine organic aerosol in Mexico City during the MILAGRO experiment 2006. Atmospheric Chem. Phys. 8, 1249–1259. Suh, H.H., Koutrakis, P., Spengler, J.D., 1994. The relationship between airborne acidity and ammonia in indoor environments. J. Expo. Anal. Environ. Epidemiol. 4, 1–22. Szymczak, W., Menzel, N., Keck, L., 2007. Emission of ultrafine copper particles by universal motors controlled by phase angle modulation. J. Aerosol Sci. 38, 520–531. doi:10.1016/j.jaerosci.2007.03.002 Thatcher, T.L., Layton, D.W., 1995. Deposition, resuspension, and penetration of particles within a residence. Atmos. Environ. 29, 1487–1497. doi:10.1016/1352-2310(95)00016-R 167 Thurston, G.D., Ito, K., Mar, T., Christensen, W.F., Eatough, D.J., Henry, R.C., Kim, E., Laden, F., Lall, R., Larson, T.V., Liu, H., Neas, L., Pinto, J., Stölzel, M., Suh, H., Hopke, P.K., 2005. Workgroup Report: Workshop on Source Apportionment of Particulate Matter Health Effects-Intercomparison of Results and Implications. Environ. Health Perspect. 113, 1768–1774. Toledo, V.E., Júnior, P.B. de A., Quiterio, S.L., Arbilla, G., Moreira, A., Escaleira, V., Moreira, J.C., 2007. Evaluation of levels, sources and distribution of toxic elements in PM10 in a suburban industrial region, Rio de Janeiro, Brazil. Environ. Monit. Assess. 139, 49–59. doi:10.1007/s10661-007-9815-y Turpin, B.J., Lim, H.-J., 2001. Species contributions to PM2.5 mass concentrations: Revisiting common assumptions for estimating organic mass. Aerosol Sci. Technol. 35, 602–610. Valavanidis, A., Fiotakis, K., Bakeas, E., Vlahogianni, T., 2005. Electron paramagnetic resonance study of the generation of reactive oxygen species catalysed by transition metals and quinoid redox cycling by inhalable ambient particulate matter. Redox Rep. 10, 37–51. doi:10.1179/135100005X21606 Vardoulakis, S., Fisher, B.E.A., Gonzalez-Flesca, N., Pericleous, K., 2002. Model sensitivity and uncertainty analysis using roadside air quality measurements. Atmos. Environ. 36, 2121– 2134. doi:10.1016/S1352-2310(02)00201-7 Verma, V., Ning, Z., Cho, A.K., Schauer, J.J., Shafer, M.M., Sioutas, C., 2009. Redox activity of urban quasi-ultrafine particles from primary and secondary sources. Atmos. Environ. 43, 6360–6368. doi:10.1016/j.atmosenv.2009.09.019 von Schneidemesser, E., Stone, E.A., Quraishi, T.A., Shafer, M.M., Schauer, J.J., 2010. Toxic metals in the atmosphere in Lahore, Pakistan. Sci. Total Environ. 408, 1640–1648. doi:10.1016/j.scitotenv.2009.12.022 Vu, T.V., Delgado-Saborit, J.M., Harrison, R.M., 2015. Review: Particle number size distributions from seven major sources and implications for source apportionment studies. Atmos. Environ. 122, 114–132. doi:10.1016/j.atmosenv.2015.09.027 Wang, D., Shafer, M.M., Schauer, J.J., Sioutas, C., 2015. A new technique for online measurement of total and water-soluble copper (Cu) in coarse particulate matter (PM). Environ. Pollut. 199, 227–234. doi:10.1016/j.envpol.2015.02.006 Wang, D., Sowlat, M.H., Shafer, M.M., Schauer, J.J., Sioutas, C., 2016. Development and evaluation of a novel monitor for online measurement of iron, manganese, and chromium in ambient particulate matter (PM). Sci. Total Environ. 565, 123–131. doi:10.1016/j.scitotenv.2016.04.164 Ware, J.H., Ferris, J., Dockery, D.W., Spengler, J.D., Stram, D.O., Speizer, F.E., 1986. Effects of ambient sulfur oxides and suspended particles on respiratory health of preadolescent children. Am. Rev. Respir. Dis. 133, 834–842. Watson, J.G., Chen, L.-W.A., Chow, J.C., Doraiswamy, P., Lowenthal, D.H., 2008. Source Apportionment: Findings from the U.S. Supersites Program. J. Air Waste Manag. Assoc. 58, 265–288. doi:10.3155/1047-3289.58.2.265 Watson, J.G., Chow, J.C., Bowen, J.L., Lowenthal, D.H., Hering, S., Ouchida, P., Oslund, W., 2000. Air Quality Measurements from the Fresno Supersite. J. Air Waste Manag. Assoc. 50, 1321–1334. doi:10.1080/10473289.2000.10464184 Watson, J.G., Cooper, J.A., Huntzicker, J.J., 1984. The effective variance weighting for least squares calculations applied to the mass balance receptor model. Atmospheric Environ. 1967 18, 1347–1355. doi:10.1016/0004-6981(84)90043-X 168 Watson, J.G., Fujita, E.M., Chow, J.C., Zielinska, B., Richards, L.W., Neff, W., Dietrich, D., 1998. Northern Front Range Air Quality Study. Final Rep. Prep. Colo. State Univ. Coop. Inst. Res. Atmosphere Fort Collins CO Desert Res. Inst. Reno NV. Watson, J.G., Robinson, N.F., Lewis, C.W., Coulter, T., 1997. Chemical mass balance receptor model version 8 (CMB8) user’s manual. Document no. 1808.1D1. Reno, NV: Desert Research Institute. Weber, R.J., Sullivan, A.P., Peltier, R.E., Russell, A., Yan, B., Zheng, M., de Gouw, J., Warneke, C., Brock, C., Holloway, J.S., Atlas, E.L., Edgerton, E., 2007. A study of secondary organic aerosol formation in the anthropogenic-influenced southeastern United States. J. Geophys. Res. Atmospheres 112, D13302. doi:10.1029/2007JD008408 Weschler, C.J., 2004. Chemical reactions among indoor pollutants: what we’ve learned in the new millennium. Indoor Air 14, 184–194. doi:10.1111/j.1600-0668.2004.00287.x Weschler, C.J., Shields, H.C., 1997. Potential reactions among indoor pollutants. Atmos. Environ. 31, 3487–3495. doi:10.1016/S1352-2310(97)00219-7 Westerdahl, D., Fruin, S., Sax, T., Fine, P.M., Sioutas, C., 2005. Mobile platform measurements of ultrafine particles and associated pollutant concentrations on freeways and residential streets in Los Angeles. Atmos. Environ. 39, 3597–3610. doi:10.1016/j.atmosenv.2005.02.034 Wonaschütz, A., Hersey, S.P., Sorooshian, A., Craven, J.S., Metcalf, A.R., Flagan, R.C., Seinfeld, J.H., 2011. Impact of a large wildfire on water-soluble organic aerosol in a major urban area: The 2009 station fire in Los Angeles County. Atmospheric Chem. Phys. 11, 8257– 8270. doi:10.5194/acp-11-8257-2011 Woodruff, T.J., Parker, J.D., Schoendorf, K.C., 2006. Fine particulate matter (PM2.5) air pollution and selected causes of postneonatal infant mortality in California. Environ. Health Perspect. 114, 786–90. Ying, Q., 2011. Physical and chemical processes of wintertime secondary nitrate aerosol formation. Front. Environ. Sci. Eng. China 5, 348–361. doi:10.1007/s11783-011-0343-1 Ying, Q., Kleeman, M., 2009. Regional contributions to airborne particulate matter in central California during a severe pollution episode. Atmos. Environ. 43, 1218–1228. doi:10.1016/j.atmosenv.2008.11.019 Ying, Q., Kleeman, M.J., 2006a. Source contributions to the regional distribution of secondary particulate matter in California. Atmos. Environ. 40, 736–752. doi:10.1016/j.atmosenv.2005.10.007 Ying, Q., Kleeman, M.J., 2006b. Source contributions to the regional distribution of secondary particulate matter in California. Atmos. Environ. 40, 736–752. doi:10.1016/j.atmosenv.2005.10.007 Zhang, H., DeNero, S.P., Joe, D.K., Lee, H.-H., Chen, S.-H., Michalakes, J., Kleeman, M.J., 2014. Development of a source oriented version of the WRF/Chem model and its application to the California regional PM10 / PM2.5 air quality study. Atmospheric Chem. Phys. 14, 485. doi:http://dx.doi.org/10.5194/acp-14-485-2014 Zhang, R., Wang, G., Guo, S., Zamora, M.L., Ying, Q., Lin, Y., Wang, W., Hu, M., Wang, Y., 2015. Formation of Urban Fine Particulate Matter. Chem. Rev. 115, 3803–3855. doi:10.1021/acs.chemrev.5b00067 Zhang, Y., Schauer, J.J., Shafer, M.M., Hannigan, M.P., Dutton, S.J., 2008. Source Apportionment of in Vitro Reactive Oxygen Species Bioassay Activity from Atmospheric Particulate Matter. Environ. Sci. Technol. 42, 7502–7509. doi:10.1021/es800126y 169 Zhao, Z., Chen, S.-H., Kleeman, M.J., Tyree, M., Cayan, D., 2011. The Impact of Climate Change on Air Quality–Related Meteorological Conditions in California. Part I: Present Time Simulation Analysis. J. Clim. 24, 3344–3361. doi:10.1175/2011JCLI3849.1 Zheng, M., Cass, G.R., Schauer, J.J., Edgerton, E.S., 2002. Source apportionment of PM2.5 in the southeastern United States using solvent-extractable organic compounds as tracers. Environ. Sci. Technol. 36, 2361–2371. doi:10.1021/es011275x Zhou, L., Hopke, P.K., Zhao, W., 2009. Source apportionment of airborne particulate matter for the speciation trends network site in Cleveland, OH. J. Air Waste Manag. Assoc. 59, 321– 331. doi:10.3155/1047-3289.59.3.321

Abstract (if available)

Abstract The association between adverse health effects and exposure to airborne particulate matter (PM) has been the subject of numerous epidemiological and toxicological researches. Most of these studies have used total particle mass concentration as a metric to assess the health effects of exposure to particles

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Size-resolved particulate matter (PM) in urban areas: toxico-chemical characteristics, sources, trends and health implications

PDF

Impact of urban source emissions on ambient particulate matter (PM) composition and associated toxicity in local and regional scales using source appointment models

PDF

Investigating the temporal trends, sources, and toxicity of ambient particulate matter (PM) in different metropolitan environments, and development of a novel aerosol generation setup for inhalat...

$Exposure assessment and source apportionment of size fractions of airborne particulate matter$

PDF

Exposure assessment and source apportionment of size fractions of airborne particulate matter

PDF

Investigation of physico-chemical characteristics of particulate matter from vehicular sources

PDF

Physico-chemical characteristics and sources of ambient PM mass and number concentrations and their associated toxicity, and development of novel techniques for high time-resolution measurement o...

$Chemical and toxicological characteristics and historical trends of size-fractioned particulate matter from traffic-related emissions in Los Angeles$

PDF

Chemical and toxicological characteristics and historical trends of size-fractioned particulate matter from traffic-related emissions in Los Angeles

PDF

Chemical and toxicological characteristics of particulate matter in urban environments with a focus on its sources, associated health impacts and mitigation policies

PDF

Oxidative potential of urban atmospheric particles: spatiotemporal trends and associations with source-specific chemical components

PDF

Temporal, spatial and toxicological characteristics of coarse particulate matter in an urban area and relation to sources and regulations

PDF

Assessing indoor and outdoor air quality by characterizing the physicochemical and toxicological properties of particulate matter using…

PDF

Development of novel techniques for evaluating physical, chemical and toxicological properties of particulate matter in ambient air

PDF

Characterization of black carbon: from source to evolution of physical and optical properties in the atmosphere

PDF

Toxicity of urban particulate matter: long-term health risks, influences of surrounding geography, and diurnal variation in chemical composition and the cellular oxidative stress response

PDF

Toxicological characteristics of particulate matter in an urban environment and their linkage to the source-specific constituents

PDF

Analysis of sources and profiles of organic carbon in ambient particulate matter across fine and coarse sizes and introducing an optical technique for real-time urban dust measurement

PDF

Particulate matter (PM) exposure for commuters in Los Angeles: chemical characterization and implications to public health

PDF

Measurement and methods of assessing the impact of prevalent particulate matter sources on air quality in southern California

PDF

Investigating the role of urban emission sources on redox-active PM compounds and the chemical analysis of the standardized diesel exhaust particles

PDF

Mechanics of abrasion emissions of particulate matter and microplastics

Asset Metadata

Creator Hasheminassab, Sina (author)

Core Title Physico-chemical properties and source apportionment of size-fractionated airborne particulate matter in urban areas with implications for public health

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Environmental Engineering

Publication Date 09/28/2016

Defense Date 07/27/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag health effects,Los Angeles Basin,molecular marker-based chemical mass balance model,OAI-PMH Harvest,particulate matter,positive matrix factorization,source apportionment,vehicular emissions

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sioutas, Constantinos (committee chair), Ban-Weiss, George (committee member), McConnell, Rob (committee member)

Creator Email hashemin@usc.edu,sina.hasheminassab@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-308764

Unique identifier UC11280621

Identifier etd-Hasheminas-4828.pdf (filename),usctheses-c40-308764 (legacy record id)

Legacy Identifier etd-Hasheminas-4828.pdf

Dmrecord 308764

Document Type Dissertation

Format application/pdf (imt)

Rights Hasheminassab, Sina

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA