University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1
IMPACT OF URBAN SOURCE EMISSIONS ON AMBIENT
PARTICULATE MATTER (PM) COMPOSITION AND
ASSOCIATED TOXCITIY IN LOCAL AND REGIONAL SCALES
USING SOURCE APPORTIONMENT MODELS
by
Amirhosein Mousavi Nasabi Shams
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
ENGINEERING
(ENVIRONMENTAL ENGINEERING)
May 2020
Copyright 2020 Amirhosein Mousavi Nasabi Shams
ii

Dedication

To my beloved parents and sister, things that never changes in my journey, the
ones I can always count on for unconditional love and support.

&

To all my mentors for inspiring me to pursue my goals with hard work and
dedication.

iii

Acknowledgments

First and foremost, I would like to extend my sincere thanks to my PhD advisor, Prof. Constantinos
Sioutas, for giving me an opportunity to be part of his group, and for his invaluable guidance and
unwavering support at every stage of my research.
Thanks to Prof. Jamie Schauer and Dr. Martin Shafer at the University of Wisconsin-Madison, for
providing me with their insightful advice over the years.
Thanks to the members of my candidacy and defense committees at USC, Prof. George Ban-Weiss,
Prof. Rob McConnell, Prof. Kelly Sanders, Prof. Amy Childress and Prof. Frank Glilland, for
taking the time to review my thesis proposal and dissertation and providing me with their
constructive feedback.
Thanks to our research collaborators at Fondazione IRCCS Milan, Dr. Ario Ruprecht group and
University of Bern Switzerland Prof. Sönke Szidat group.
Thanks to the South Coast Air Quality Management district (SCAQMD) employees, Dr. Andrea
Polidori, Dr. Sina Hasheminassab and Dr. Olga Pekinova for their cooperation and help in
providing essential data.
Thanks to my past and present group mates at the USC aerosols laboratory, Dr. Arian Saffari, Dr.
Farimah Shirmohammadi, Dr, Mohammad H. Sowlat, Dr. Chris Lovett, Sina Taghvaee, Milad
Pirhadi, Ehsan Soleimanian, Maryam Hakimzadeh and Abdulmalik Altuwayjiri, for being such
wonderful colleagues and friends.
Finally, thanks to the large group of collaborators at USC and elsewhere who I had the privilege
of working with and who contributed to this dissertation.
iv

Table of Contents

Dedication ................................................................................................................ ii
Acknowledgements................................................................................................. iii
List of Tables ........................................................................................................... ix
List of Figures .......................................................................................................... xi
Abstract .................................................................................................................. xii
Chapter 1. Research background and objectives ................................................. 1
1.1. Airborne Particulate Matter (PM) .................................................................. 1
1.2. Research Motivation ......................................................................................... 3
1.3. Research Chapters ............................................................................................. 3
Chapter 2. ships, locomotives, and freeways emissions across Los Angeles....... 7
2.1. Introduction ..................................................................................................... 7
2.2. Methodology .................................................................................................. 10
2.2.1. Study area………………. ...................................................................................................10
2.2.2. Instrumentation for on-road measurements .........................................................................15
2.2.3. Instrumentation for stationary/mobile measurements at the POLA and POLB ..................17
2.2.4. Instrumentation for Locomotive emissions .........................................................................21
2.2.5. Calculation of emission factors and rates ............................................................................22
2.2.5.1. Emission factors………. ................................................................................................................................. 22
2.2.5.2. Emission rates ................................................................................................................................................. 23
2.3. Results and Discussion .................................................................................. 28
v

2.3.1. PM2.5, PN, and BC emission factors ....................................................................................28
2.3.2. PN, PM2.5, and BC concentrations on freeways and at the San Pedro Bay ports ................31
2.3.3. Impact of POLB and POLA emissions at the local and regional scales…………………..35
2.4. Summary and Conclusions ............................................................................42
2.5. Acknowledgements ………… ……… … ………… ……… … ………… ….. ...43

Chapter 3. Impact of urban activity source emissions on the ambient PM0.25
oxidative potential across the Los Angeles County …… …… ………… …….. ...44
3.1. Introduction …… … ………… ……… … ………… ……… … ………… ……44
3.2. Methodology ...................................................................................................46
3.2.1. Sampling sites and meteorology…………………………………………………………. 46
3.2.2. Collection schedule and method…………………………………………………………. 51
3.2.3. Gravimetric and chemical analysis……………………………………………………… 51
3.2.4. Assessment of PM0.25 oxidative potential by means of the alveolar macrophage (AM)
assay……………………………………………………………………………………… 52
3.2.5. Source apportionment of PM0.25 related oxidative potential……………………………... 53
3.3. Results and Discussion. .................................................................................54
3.3.1. Mass concentration and chemical composition of PM0.25……………………………………………… 54
3.3.1.1. Concentrations of PM 0.25 mass and carbonaceous compounds……………………………………………... 54
3.3.1.2. Elemental content of PM 0.25…………………………………………………………………………………………………………………………….. 55
3.3.1.3. Organic species……………………………………………………………………………………………... 57
3.3.2. PM0.25 Oxidative Potential………………………………………………………………. 59
3.3.3. PM0.25 source apportionment and associated oxidative potential of dominant sources…...60
3.3.3.1. Correlation between individual species and oxidative potential……………………………………………. 60
3.3.3.2. MLR analysis and oxidative potential source apportionment………………………………………………. 66
3.4. Summary and Conclusions ……… …… ………… ……… … ………… …… .71
vi

3.5. Acknowledgements ………… ……… … ………… ……… … ………… ….. . 72
Chapter 4. Source contributions and temporal trends of Redox-Active Metals
in central Los Angeles …… ……… …… ………… ……… … ………… ……….. .73
4.1. Introduction …… … ………… ……… … ………… ……… … ………… ….. . 73
4.2. Methods …… ……… ………… ……… … ………… ……… … ………… …. 75
4.2.1. Sampling site and study period…………………………………………………………... 75
4.2.2. Instrumentation…………………………………………………………………………... 76
4.2.3. Auxiliary variables………………………………………………………………………. 77
4.2.4. PMF run…………………………………………………………………………………. 79
4.3. Results and Discussion …… ……… …… ………… ……… … ………… …. 79
4.3.1. Overview of the data……………………………………………………………………...79
4.3.2. Number of factors……………………………………………………………………….. 82
4.3.3. Identification of factors………………………………………………………………….. 85
4.4. Summary and Conclusions ……… …… ………… ……… … ………… ….. 90
4.5. A ckn o w l ed g em en ts ………… ……… … ………… ……… … ………… … … 9 1

Chapter 5. Spatio-temporal trends and source apportionment of black carbon
(BC) in the Los Angeles Basin: fossil fuel and biomass burning origin
particles ……… …… ………… ……… … ………… ……… … ………… ……….92
5.1. Introduction …… … ………… ……… … ………… ……… … ………… … ..92
5.2. Methodology ............................................................................................... ..95
5.2.1. Sampling sites and period……………………………………………………………......96
5.2.2. Instrumentation... ………………………………………….………………………….....98
5.2.3. BC source apportionment principles ...............................................................................100
vii

5.2.4. Tracer analysis………………………………………………………………………….102
5.3. Results and Discussion …… ……… …… ………… ……… … ………….. .102
5.3.1.Data overview and average BC concentrations………………………………………... 102
5.3.2.MAC values and seasonal trends of sources…………………………………………… 105
5.3.3.Diurnal variation of BCff and BCbb………………………………………………………………………………. 111
5.3.4.BC vs fossil fuel/biomass burning tracers……………………………………………… 116
5.3.5.Sensitivity analysis of the Aethalometer model for biomass burning fraction of BC…..127
5.4. Summary and Conclusions ……… …… ………… ……… … ………… …. 129
5.5. Acknowledgements ..................................................................................... .130
Chapter 6. Source apportionment of black carbon (BC) from fossil fuel and
biomass burning in metropolitan Milan, I ta l y ….. .. .. .. .. .. .. .. . .. .. . .. .. .. .. .. .. .. .. .. .. .. ..131
6.1. Introduction ................................................................................................... 131
6.2. Methodology .................................................................................................. 136
6.2.1. Sampling sites and collection periods ...............................................................................136
6.2.2. Instrumentation………………………………………………………………………….. 138
6.2.3. Source apportionment of BC .............................................................................................139
6.2.4.
14
C analysis………………………………………………………………………………142

6.3. Results and Discussion .................................................................................. 144
6.3.1. Data overview………………………………………………………………………….... 144
6.3.2. MAC values and the temporal trends in the source contributions ....................................146
6.3.2.1. Overview of the Aethalometer model result and MAC values ..................................................................... 146
6.3.2.2. Spatial and temporal variation of BC ff and BC bb ........................................................................................................................................ 149
6.3.3. Impact of local events on the BCff and BCbb concentrations .............................................152
viii

6.3.4. Diurnal variations of BCff and BCbb........................................................................................................................ 152
6.4. Summary and Conclusions......................................................................... 153
6.5. Acknowledgements ..................................................................................... 154
Chapter 7. Concluding remarks ......................................................................... 155
7.1. Major findings ............................................................................................. 155
7.2. Recommendations and future research ideas .......................................... 156

Bibliography … …… ………… ……… … ………… ……… … ………… ……….159

ix

List of Tables

Table 2.1 Meteorological conditions during the sampling campaigns of
the study………………………………………………………………..20
Table 2.2 Vehicle kilometers traveled per hour (VKT) and traffic flow
data during the on-road sampling campaign for I-110 and I -710 freeways.
Errors represent one standard deviation………………………………..20
Table 2.3 Instruments used at each site and sampling campaign to
measure PN, PM, BC and CO2 concentrations………………………..23
Table 2.4 Median PM2.5, particle number (PN), and black carbon (BC)
emission factors (EF) for different pollution sources near the Port of Los
Angeles (POLA) and Long Beach (POLB). Errors represent one standard
deviation (SD)………………………………………………………….31
Table 2.5 Daily emission rates of PM2.5, particle number (PN), and black
carbon (BC) for different sources at the local and regional spatial scales.
Errors represent the uncertainty associated with each value…………..38
Table 3.1 Summary of the meteorological data (average ± standard
deviation) at the three sampling locations for: (a) current study; (b) Hu et
al. (2008). Data were obtained from the website of the California Air
Resources Board (CARB) for the sites that were closest to our sampling
locations………………………………………………………………..51
Table 3.2 Spearman rank correlation coefficients (R) between the
concentrations of the chemical species (ng.m
-3
) and the oxidative potential
(µg of Zymosan.m
-3
) of PM0.25 in the study area: (a) elements and metals
(N= 15); and (b) WSOC and organic compounds (N = 6). R values above
0.5 are bolded………………………………………………………….64
Table 3.4 Results of the multiple linear regression (MLR) analysis
performed between: (a) PM0.25 associated oxidative potential as
dependent variables, and PM0.25 organic chemical species as
independent variables; (b) total PAH concentrations as dependent variable
and individual PAHs concentrations as independent variables; using the
combined dataset (N=6) including WSOC and organic
species……………………………………………………………….....69
x

Table 4.1 Inputs of the PMF model along with their time resolution and
measurement method/collection source………………………………..80
Table 4.2 Summary statistics for the parameters inserted into the model
by season (N=181)……………………………………………………..84
Table 4.3 Correlation matrix between the factor contributions from the
two separate runs for the chemical species and size distribution…... …89
Table 5.1 Geographical coordinates and characteristics of the four
sampling sites along with the parameters measured/collected in each
site…………………………………………………………………….103
Table 6.1 Average meteorological data during the sampling campaigns in
Milan and Bareggio…………………………………….......................147
xi

List of Figures

Figure 2.1 Map of the spatial scales at which emissions from Ports of Los
Angeles (POLA) and Long Beach (POLB) and those from freeways were
compared: a) Local (i.e., impact zone); and b) regional (i.e., the whole Los
Angeles County)………………………………………………………..12
Figure 2.2 Wind rose of the Port of Los Angeles (POLA) and Long Beach
(POLB) during the: a) on-road sampling campaign; b) stationary/mobile
measurements campaign; and c) locomotive measurements campaign.
Wind data were obtained from the Air Quality Management District
(AQMD) weather station at the Port of Long Beach……………………14
Figure 2.3 Particle emission factor as a function of SO2 emission factor
for ship emissions from 2000 to 2015. Error bars represent one standard
deviation (SD) for EFSO2 and EFPN…………………………………..32
Figure 2.4 Box-and-whisker plots of variations in the concentrations of
the target pollutants on-road of freeways and at the Port of Long Beach
(POLB) for: a) PM2.5; b) particle number (PN); c) black carbon (BC);
and d) CO2. The (×) sign represents the arithmetic mean. Lines inside the
box represent the median values, and the lower and higher lines of the box
show the 25th and 75th percentiles, respectively. Whiskers represent the
5th and 95th percentiles……………………………………………….. 36
Figure 2.5 (a-d). Relative contributions of major sources at the Ports of
Los Angeles (POLA) and Long Beach (POLB) to PM2.5, particle number
(PN), and black carbon (BC) emissions: a) at the local scale; and b) at the
regional scale. 39
Figure 3.1 (a) Map of the study area with respect to the location of the
three sampling sites. Wind roses during the sampling period at: (b) PRT;
(c) NLB; and (d) USC……………………………………………….....49
Figure 3.2 Average PM0.25 concentrations measured in the current study
compared to the study conducted in 2007 at the same sampling locations.
Error bars represent one standard deviation (SD)………………………57
xii

Figure 3.3 Average concentrations of water-soluble organic carbon
(WSOC) and individual organic SOA tracers at the three sampling
locations. Error bars represent one standard deviation (SD)……………57
Figure 3.4 Average concentrations of metals and trace elements at each
sampling site during the study period. Error bars represent one standard
deviation (SD)……………………………………………………….....59
Figure 3.5 Average concentrations of organic species: (a) polycyclic
aromatic hydrocarbons (PAHs); (b) hopanes and steranes; and (c) n-
alkanes. Error bars represent total standard deviation (SD) of
compound………………………………………………………............61
Figure 3.6 Oxidative potential of PM0.25 at the three locations
normalized per m3 of air volume. Error bars represent one standard
deviation (SD)………………………………………………………….63
Figure 3.7 Average concentrations of individual PAH compounds used
as marker species for: a) vehicular emissions; b) emissions from port-
related activities; at the three sampling sites. Error bars represent one
standard deviation (SD)………………………………………………...67
Figure 3.8 Contributions of different sources to the oxidative potential of
ambient PM0.25 at the three sampling sites……………………………70
Figure 4.1 Map of the study area………………………………………77
Figure 4.2 Diurnal variations of important meteorological parameters in
the cold and warm phases. Error bars correspond to one standard error.81
Figure 4.3 Diurnal variations of: (a) PM2.5 mass, (b) Fe, (c) Cr, (d) Cu,
and (e) Mn concentrations during the entire campaign. Error bars
correspond to one standard error
(SE)…………………………………....................................................86
Figure 4.4 Diurnal variations of mass fractions of: (a) Fe, (b) Cr, (c) Cu,
and (d) Mn during the entire campaign. Error bars correspond to one
standard
error(SE)…………………………………………………………….....87
Figure 4.5 Factor profiles resolved by the PMF model for redox-active
metals and auxiliary variables: (a) Fresh traffic; (b) Urban background
aerosol; (c) Secondary aerosol; and (d) soil/road dust…………………90
ix

Figure 4.6 Factor profiles resolved by the model for the PM10 volume
size distribution data: (a) Nucleation; (b) Fresh traffic; (c) Urban
background aerosol; (d) Secondary aerosol; and (e) Soil/Road dust.
Volume/mass concentrations in each size channel are represented by solid
lines (primary Y axis), while the explained variation is represented by
triangles (secondary Y axis)……………………………………............91
Figure 4.7 Relative contributions of each factor to: (a) Fe, (b), Cr, (c) Cu,
(d), Mn, and (e) PM2.5 mass concentrations…………………………...92
Figure 4.8 Absolute contributions (by cold or warm phases) of each factor
to the total PM2.5 mass concentrations………………………………...93
Figure 4.9 Diurnal variations for the contribution of each of the factors to
the total PM2.5 concentrations in the warm and cold seasons: (a) Fresh
traffic; (b) Urban background aerosol; (c) Secondary aerosol; and (d)
soil/road dust……………………………………………………….......94
Figure 5.1 Map of the Los Angeles Basin along with the location of
sampling sites………………………………………………………....105
Figure 5.2 BC vs. EC linear regression line using data from all of the sites
in 2012-2013 sampling campaign…………………………………….108
Figure 5.3 Seasonal variations of total BC concentrations at the 4
sampling sites in 2012-2013 and 2016-2017. Error bars represent one
standard deviation (SD)………………………………………….........113
Figure 5.4 Seasonal variations of Mass Absorption Cross-section (MAC)
in the 2012-2013 and 2016-2017 campaigns at different sites: a) MAC
370 nm and b) MAC 880 nm. Error bars represent one standard deviation
(SD)……………………………………………………………...........116
Figure 5.5 Seasonal variation of BCff and BCbb concentrations at the 4
sampling sites in the 2012-2013 and 2016-2017 campaigns: a) BCff, and
b) BCbb. Error bars represent one standard deviation (SD)…………..118
Figure 5.6 Seasonal variation in the BC fossil fuel (BCff) and BC biomass
burning (BCbb) contributions at the 4 sampling sites in the 2012-2013 and
2016-2017 campaigns: a) BCff % b) BCbb %. Error bars represent one
standard deviation (SD)……………………………………………….119
x

Figure 5.7 Diurnal variations of BCff% and BCbb% at all sites during the
entire study period in: a) warm period, b) cold period. Error bars represent
one standard deviation (SD)…………………………………………..121
Figure 5.8 Diurnal variations of BCff (ng.m-3) at all sites during the
entire study period in: a) warm period, b) cold period. Error bars represent
one standard deviation (SD).deviation of 7 bi-weekly composited
samples………………………………………………………………..123
Figure 5.9 Diurnal variations of BCbb (ng.m-3) at all sites during the
entire study period in: a) warm period, b) cold period. Error bars represent
one standard deviation (SD)…………………………………..............124
Figure 5.10 Monthly variations of BCff, NOx, and CO concentrations in:
a) CELA, and b) Riverside at 2012-2013. Error bars represent one
standard deviation (SD)…………………………………………….....126
Figure 5.11 Monthly variations of BCff, NOx, and CO concentrations
during the sampling at 2016-2017 campaign in: a) CELA, and b)
Riverside. Error bars represent one standard deviation (SD)…………127
Figure 5.12 Diurnal variations of a) BCff, b) NOx, and c) CO in the cold
and warm phases in CELA and Riverside sites in the 2012-2103
campaign. Error bars represent one standard deviation (SD)…………128
Figure 5.13 Diurnal variations of a) BCff, b) NOx, and c) CO in the cold
and warm phases in CELA and Riverside sites in the 2016-2107
campaign. Error bars represent one standard deviation (SD)………….129
Figure 5.14 Linear regression between BCff and fossil fuel tracers (NOx
and CO) in the cold and warm periods in: a) CELA and b) Riverside at
2012- 2013 sampling campaign. The regression lines for NOx are plotted
based on seasonally-averaged data points for diurnal variations……...130
Figure 5.15 Linear regression between BCff and fossil fuel tracers (NOx
and CO) in the cold and warm periods of the 2016-2017 campaign in: a)
CELA, and b) Riverside. The regression lines for NOx are plotted based
on seasonally-averaged data points for diurnal variations……………132
Figure 5.16 Regression lines of BCbb and K+/K (weekly-averaged data)
in the cold and warm phases for: a) CELA 2012-2013, b) Riverside at
2012- 2013………………………………………………………........135
xi

Figure 5.17 Regression lines of BCbb and K+/K (weekly-averaged data)
in the cold and warm periods of the 2016-2017 sampling campaign in: a)
CELA, and b) Riverside………………………………………………136
Figure 5.18 Regression lines of BCbb and levoglucosan (weekly
samples) sites in the cold periods of the 2012-2013 campaign: a) CELA
and b) Anaheim……………………………………………………….137
Figure 5.19 (a) Babs370 and (b) Levoglucosan reduced major axis
(RMA) intercept as a function of αff when αbb varies between 1.8-2.2.139
Figure 6.1 a) Map of the metropolitan areas within the Po valley (star sign
indicates Milan); and b) locations of the two sampling sites within the
metropolitan Milan area………………………………………………146
Figure 6.2 Seasonal variations of BC concentrations in Milan and
Bareggio. Error bars represent standard error (SE). SENATO and
PASCAL stations are the closest ARPA stations near the Milan and
Bareggio sites, respectively…………………………………………...154
Figure 6.3 Fig. 3. Residuals of BCff/BC compared to ECff/EC
(ΔBCff=BC) as a function of measured ECff/EC for the 4 analyzed
samples using different combinations of αbb when αff=0.9………….157
Figure 6.4 Seasonal variations of mass absorption cross-section (MAC)
in Milan and Bareggio: a) MAC 370 nm and b) MAC 880 nm. Error bars
represent one standard error (SE)…………………………………......157
Figure 6.5 Seasonal variations of: a) BCff; and b) BCbb concentrations
in Milan and Bareggio. Error bars represent one standard error (SE)….160
Figure 6.6 Diurnal variations of BCff and BCbb during the whole
campaign in: a) Milan; and b) Bareggio. Error bars represent one standard
error (SE). …………………………………………………………….162

xii

Abstract

There is compelling evidence indicating strong associations between airborne particulate matter
(PM) and increased risk of a wide range of adverse health outcomes in humans mostly attributable
to PM oxidative potential (i.e., capacity of PM species to induce cellular oxidative stress in
biological systems). Despite the comprehensive documentation of PM-related health effects, the
state of knowledge regarding the exact causative components and the extent to which each PM
source is contributing to its severity is uncertain. Accordingly, current PM regulations mainly
target PM mass concentration, which may not be a good representative for the PM-induced health
effects. The research gaps in understanding the PM source-specific adverse health effects as well
as the insufficient target PM criteria pollutant urge the study of the physico-chemical and
toxicological characteristics of ambient PM sources. This proposed research would provide
significant insight into the implementation of more efficient and targeted source-specific
mitigation policies to control adverse health impacts on the exposed population.
The main objective of the presented dissertation is to quantify the effect of different urban activity
source emissions on physical, chemical and toxicological properties of ambient particulate matter
(PM) in local and regional scales. To this end, a series of comprehensive on-road and off-road air
sampling campaigns were carried on in Los Angeles and Milan metropolitan areas, as an example
of a complex urban environment impacted by a variety of PM sources. Mobile source emissions
within the area (major freeways, ports, railways and airports) were measured and calculated in
local and regional scales. Accordingly, physicochemical properties of collected ambient PM
samples demonstrated the temporal and spatial variation of important unregulated PM components
such as particle number (PN), redox-active metals and black carbon (BC). Subsequently, statistical
xiii

principal component analysis (PCA), regression modeling and distinct particle optical properties
were deployed to quantify the contribution of sources to important PM compounds (i.e., metal
particles and BC) as well as measured associated oxidative potential (i.e., capacity of PM species
to oxidize target molecules. Findings of this work advance our knowledge of complex source
emission impacts on the PM toxicity and physicochemical composition in different
microenvironments and provide valuable insights for more targeted and cost-effective air pollution
control schemes in polluted areas around the globe.
1

Chapter 1:

Research background and objectives

1.1. Airborne Particulate Matter (PM)

The rapid pace of urbanization and industrialization has created a wide range of environmental
challenges in metropolitan areas around the globe, and air pollution is among the most difficult
ones to overcome. Ambient particulate matter (PM) is of particular importance, due to its distinct
characteristics and diverse health impacts. Ambient PM can come from a variety of sources,
including, but not limited to, vehicles (exhaust and non-exhaust), industries, biomass burning,
wildfires, and resuspension of road dust. It can also be formed secondarily due to
chemical/photochemical reactions in the atmosphere. Finally, ambient PM can possess a wide
range of chemical components, including carbonaceous materials, elements and metals, water-
soluble components, and organic species. All of these characteristics have made ambient PM a
distinct air pollutant, which has become the target of thousands of scientific research studies
around the globe.
There is a rapidly growing body of toxicological as well as epidemiological evidence identifying
major health impacts associated with population exposure to particulate matter (PM), including
respiratory and cardiovascular diseases, hospitalization, premature death, and neurodegenerative
effects (Brunekreef and Forsberg, 2005; Dockery and Stone, 2007; Gauderman et al., 2015; Miller
et al., 2007; Pope et al., 2004b; Pope Iii et al., 2002). According to an earlier global burden of
disease (GBD) study, over 3 million premature deaths annually occur around the globe due to
exposure to ambient PM (Lim et al., 2013). However, according to the most recent GBD results,
2

exposure to air pollution now ranks above AIDS, malaria, and malnutrition as a cause of disease
burden, and coming in a close second to cigarette smoking, with an estimated 6.5 million premature
deaths in 2015 around the world (Wang et al., 2016c). 4.5 million of these premature deaths were
attributable to ambient PM. Studies have indicated that ultrafine particles (UFP, i.e. particles with
an aerodynamic diameter of <100 nm) have higher toxicity per unit mass (Donaldson et al., 1998;
Li et al., 2003; Nel et al., 2006; Oberdörster et al., 2002), have higher deposition efficiencies in
the lung (Venkataraman, 1999), and penetrate deeper into the alveolar regions of lungs (Sioutas et
al., 2005b) than larger particles. Several studies have also found that PM number concentrations
(mostly UFPs) can be associated with adverse effects on human health, particularly for
cardiovascular diseases (Delfino et al., 2005; Peters et al., 1997; Wichmann et al., 2000). In
addition, rather than the total mass of ambient PM, specific chemical components of PM, including
water-soluble organic carbon, black carbon (BC), and redox-active metals have been found to
cause adverse human health effects as well (Delfino et al., 2005; Donaldson et al., 2003; Li et al.,
2009; Peters et al., 2006; Tao et al., 2003). Other chemical components, such as secondary
ammonium nitrate and sulfate, are considered relatively harmless PM species, although they
contribute to a major fraction of ambient PM, especially in the PM 2.5 size fraction (due to their
secondary nature).
This makes the study of the physical, chemical and toxicological characteristics of ambient PM
quite critical, as it can provide significant insight into the contributing sources of ambient PM most
responsible for its toxicity and the resulting adverse health impacts. This information is crucial for
designing air pollution control schemes to effectively mitigate air pollution problems in an area.
The research presented herein provides valuable information on the spatio-temporal trends and
sources of ambient PM components and associated toxicity in different atmospheric and urban
3

settings.
1.2. Research Motivation

Despite commendable progress in PM-related toxicological research during the past decade,
appropriate extrapolation of the clinical data to the actual real-world PM exposure conditions has, so
far, been hindered, largely due to the yet incomplete knowledge regarding the specific chemical
components and associated PM emission/formation sources that dominantly drive the PM-induced
oxidative potential in the atmosphere. The regulatory measures in urban areas have, therefore, been
focused on the PM mass as the sole surrogate of the PM-related health effects, with minimal
consideration of the chemical and toxicological properties associated with the specific emission
sources. In the current regulatory framework, same mass emissions from all different sources are
treated as similar, while the relative toxicological potency of various sources can be significantly
different. The main purpose of this dissertation is to fill some of the knowledge gaps between the
current clinical knowledge regarding the PM-induced toxicity and the actual sources and chemical
components that drive the PM toxicity in the urban atmosphere in local and regional scales, with the
aim of promoting current regulations toward a more targeted and effective approach.

1.3. Research Chapters

Presented dissertation is consisted of five research chapters as described below:

Chapter 2: This chapter provides detailed assessment on the importance of urban activity source
emissions in local and regional scales within the Los Angeles Air Basin. Ambient fine particulate
matter (PM2.5) mass, PM2.5 particle number (PN), and black carbon (BC) from of-road and on-road
mobile sources at the Ports of Los Angeles (POLA) and Long Beach (POLB), including ships,
4

cargo-handling equipment (CHE), locomotives, and heavy-duty vehicles (HDVs) operating at the
port terminals, as well as from vehicles on the major freeways. Findings of this chapter revealed
the dominant emission sources of various PM pollutants (i.e., PM2.5, PN and BC) in the adjacent
local communities (i.e., impact zone) shedding more light upon selecting effective criteria
pollutants in different urban metropolitan areas.

Chapter 3: In this chapter chemical composition and oxidative potential of ambient PM0.25 particles
within identified impact zone in chapter 1 were evaluated. To this end, weekly filter samples were
collected in three contrasting locations, including central Los Angeles, North Long Beach, and the
Port of Long Beach. The oxidative potential of the collected samples was quantified by means of an
in vitro cell-based alveolar macrophage (AM) assay. Subsequently, Multiple linear regression
(MLR) analysis was used to link individual measured toxic PM species concentrations, used as
source markers, to the oxidative potential of the ambient PM0.25 across the monitoring sites within
the impact zone. Results from the MLR analysis compared the impact of regional urban PM
sources such as vehicular emissions and secondary organic aerosols (SOA) to that more localized
emissions from port area on PM0.25 oxidative potential.

Chapter 4: This chapter includes comprehensive identification and quantification of source
contributions of four redox-active trace metals (i.e., Iron (Fe), Chromium (Cr), Copper (Cu), and
Manganese (Mn) in central Los Angeles, California, by employing time-resolved measurements
(i.e., a time resolution of 2 hrs) with a recently developed online metals monitor and Positive
Matrix Factorization (PMF). The size distribution of ambient PN (14 nm to 10 µm) as well as
auxiliary variables were collected, including elemental (EC) and organic carbon (OC), gaseous
5

pollutants (NO2 and O3), meteorological parameters (including relative humidity (RH) and
temperature), and traffic data (for heavy- (HDVs) and light-duty vehicles (LDVs)). Novel
measurement techniques used to measure time-resolved PM-bound metals concentration, might
enhance our understanding of the toxic metal emission sources in urban environments by providing
us with measurements with finer temporal resolution that otherwise would not have been possible
using traditional filter-based measurement techniques.

Chapter 5: Spatial and temporal trends of BC, a well-known PM toxic species in the Los Angeles
Basin were evaluated in this chapter. Equivalent black carbon concentrations (eBC) were measured
at seven wavelengths using Aethalometers (AE33) at four sites, including central Los Angeles
(CELA), Anaheim, Fontana, and Riverside, during a warm and cold period. Sources of BC (i.e.,
Fossil fuel and biomass burning) estimated based on the spectral dependence of light absorption
of different sources. Results of the study were validated using regression analysis between the
source specific PM species concentrations and estimated eBC from biomass burning and fossil
fuel sources.

Chapter 6: Final research chapter of this dissertation provides temporal variations of sources of
BC concentrations in the metropolitan area of Milan, Italy, during three distinct seasons to compare
the result with those of chapter 5 in contrasting areas in terms of dominant seasonal sources.
Similar to the chapter 5, eBC concentrations were measured via Aethalometers at two sampling
sites, one in the city center of Milan, and one in the less densely populated suburb of Bareggio,
approximately 14 km to the west of Milan’s urban center. In addition to the use of similar source
contribution estimate model to chapter 5, radiocarbon
14
C analysis was performed on the PM
6

samples, allowing us to refine our estimates of the contributions of both fossil fuel and biomass
burning sources to total BC concentrations and assess the dependence of the mass absorption cross
section (MAC) of BC particles on its source.

Chapter 7: This chapter summarizes the major findings of this dissertation and provides directions
for future research on this topic, as well as recommendations for implementing more cost-effective
and targeted criteria pollutant regulations in different urban settings
7

Chapter 2:

Ships, locomotives, and freeways emissions across Los Angeles

2.1. Introduction

Emissions from combustion sources related to goods movement, including ships, diesel trucks,
and locomotives, contribute significantly to air pollution all around the globe. Although shipping
is considered as one of the most energy-efficient modes of goods transportation (Alfoldy et al.,
2013), according to recent studies, it has been estimated that ocean-going vessels (OGVs) release
0.9 Million metric tons of particulate matter (PM) into the atmosphere, and contribute 20-28% to
the total gaseous pollutant (e.g. CO2, SO2, and NOx) emissions from the transportation sector
(Corbett et al., 2007a;Lack et al., 2009). The high rate of PM emissions from ships and vessels is
a direct result of the high sulfur content of the fuel used by the ships’ engines, which is called the
“heavy fuel oil” (HFO) (Agrawal et al., 2008). Similarly, to ships and vessels, railway locomotives
are also considered as a relatively clean mode of goods transportation. However, their increasingly
widespread use worldwide has raised concerns on their overall contribution to air quality
deterioration, particularly in areas adjacent to railways (Johnson et al., 2013). For instance, studies
performed on locomotive emissions have indicated significantly elevated levels of important air
pollutants, including PM, black carbon (BC), CO2, SOx, NOx, and volatile organic compounds
(VOCs), in areas near or downwind of railroads (Tang et al., 2015;Krasowsky et al., 2015;Johnson
et al., 2013).
In addition to the climate change impacts of ship and locomotive emissions due to their release of
CO2 into the atmosphere, emissions from these vehicles contain many important air pollutants,
8

including PM, BC, NOx, and SOx, for which significant adverse human health impacts have been
reported in the literature. PM exposure has been linked to a wide range of adverse health effects,
including hospitalization and premature death due to cardiovascular and respiratory diseases
(Delfino et al., 2005; Neupane et al., 2010; Sun et al., 2010;Ito et al., 2011;Anderson et al.,
2012;Shah et al., 2013). Most of the particles emitted from these two combustion sources are
ultrafine particles (UFP) with aerodynamic diameters smaller than approximately 200 nm
(Kleeman et al., 2007;Johnson et al., 2013;Petzold et al., 2003;Petzold et al., 2008). UFPs,
compared to larger sizes of PM, are believed to be more harmful to human health, because they
have a larger surface area and can penetrate deeper into the lungs and may directly enter the
bloodstream and cells (Beck-Speier et al., 2005;Brown et al., 2001; Li et al., 2003). Black carbon
(BC) is another important particulate pollutant released by ships, HDVs and locomotives, which
is known to pose cancer risks via tumor-promoting mechanisms and associated with several
cardiovascular and respiratory diseases (Sauvain et al., 2003;Suglia et al., 2007;McCracken et al.,
2010;Janssen et al., 2011;WHO, 2012). In addition, since the majority of ship and locomotive
emissions occur within or near residential areas of highly dense population, they could potentially
have a significant impact on the communities adjacent to them. Consequently, numerous studies
have attempted to characterize the composition of these emissions and evaluate their impact on
areas downwind of the sources (Moldanová et al., 2009;Kasper et al., 2007;Alfoldy et al.,
2013;Johnson et al., 2014).
The Ports of Los Angeles (POLA) and Long Beach (POLB), together known as the San Pedro Bay
Ports, represent the biggest seaway cargo-movement/handling facility in the U.S. in terms of both
cargo value and container traffic (Knatz, 2006). POLA and POLB are located adjacent to the Long
Beach community, which has a population density of nearly 9191 persons/square mile based on
9

the US Census report. A total population of approximately 500,000 people live within the “impact
zone” of POLA and POLB, an area confined to the south by the ports, to the east and west by I-
710 and I-110 freeways, and to the north by 105 freeway (Figure 1a). This area is impacted by a
mixture of major freeway emissions (e.g., I-110, I-710, I-405, and I-105), as well as emissions
from the ports that relate to the transport of goods. The latter includes emissions from ships, cargo
handling equipment (CHE), heavy duty vehicles (HDVs), and locomotives operating at the ports
terminals and nearby communities (i.e., the Alameda corridor) (POLB, 2015). These emissions
contribute significantly to the mass concentration of ambient PM in the communities adjacent to
the POLA and POLB, as indicated by the results of previous source apportionment studies in the
area (Hu et al., 2008;Minguillón et al., 2008;Daher et al., 2013). However, most of these studies
have evaluated the contributions of sources to PM mass using ambient PM source apportionment
techniques, such as the chemical mass balance (CMB) model. Since there are no profiles available
for some of the sources at the POLA and POLB, such as locomotive and CHE emissions, results
from these studies may not fully reflect the exact break-down of sources and their contributions to
ambient PM in the area. In addition, most of these studies have focused on source apportionment
of PM mass concentrations, but less attention has been paid to specific components of PM, such
as BC, or PM characteristics other than mass, such as particle numbers.
The main goal of this study was to evaluate the emission rates of PM2.5 mass, particle number (PN)
(as representative of ultrafine PM), and BC from different mobile sources at or near the POLA and
POLB, including ships, locomotives, CHE, and HDVs at the ports, and vehicles on freeways, and
to assess the impact of these emissions on air quality in the “impact zone” of POLA and POLB,
and on the air quality in the Los Angeles County. In order to calculate the emission rates for the
vehicles on the adjacent freeways, on-road measurements were performed from May to June of
10

2017 inside the I-710 and I-110 freeways, which are the two major ground transportation routes
north of the POLA and POLB. In addition, emission factors and rates for ships and vessels at the
POLA and POLB were calculated based on the results of a campaign carried out by the Chalmers
University of Technology and FluxSense Inc. in October and November 2015 to capture
ships/vessels plumes (Mellqvist et al., 2017; Mellqvist et al., 2016). Emission factors and rates for
all other source categories were either obtained or calculated using pertinent values reported in the
literature (CARB, 2008;POLB, 2015; Krasowsky et al., 2015;Bergin et al., 2012;POLB, 2006,
2007a;POLA, 2015).

2.2. Methodology

2.2.1. Study area

The POLA and POLB, located along the southern border of Los Angles air basin, are the most
crowded cargo movement facilities in the U.S and one of the busiest seaports in the world with
more than 10000 vessel arrival/departures in 2017. POLA and POLB in total consist of 7000 and
12000 acres of land and water, respectively, which include 50 terminals and 150 berths. For the
purposes of this study, emissions from the POLA and POLB are categorized into four separate
categories: a) emissions from ships and vessels during arrival, departure, or at berth; b) emissions
from locomotives operating at-or within the zone of impact of POLA and POLB for goods
movement; c) emissions from HDVs operating at the POLA and POLB for goods movement (this
is a separate category from HDVs on the adjacent freeways); and d) emissions from the cargo-
handling equipment at the POLA and POLB.
Figure 4.1(a-b) illustrates the two spatial scales at which ships, locomotives, CHE, HDVs, and
freeway emissions are evaluated and compared. At the local scale, emissions from POLA and
11

POLB and adjacent freeways were compared in an area limited to the impact zone of the ports,
which is defined as the area surrounded by I-105, I-110, I-710, and the ports (Figure 4.1a), with a
major southerly wind direction (Figure 4.2). At the regional scale, emissions from ships,
locomotives, HDVs, CHE, and the sum of all freeways in the Los Angeles County were evaluated
and compared, as illustrated in Figure 1(b). More detailed information on each of the sampling
campaigns can be found in the subsequent sections.
Figure 2.1 Map of the spatial scales at which emissions from Ports of Los Angeles (POLA) and
Long Beach (POLB) and those from freeways were compared: a) Local (i.e., impact zone); and
b) regional (i.e., the whole Los Angeles County).
(a)
12

(b)

13

Figure 2.2 Wind rose of the Port of Los Angeles (POLA) and Long Beach (POLB) during the: a)
on-road sampling campaign; b) stationary/mobile measurements campaign; and c) locomotive
measurements campaign. Wind data were obtained from the Air Quality Management District
(AQMD) weather station at the Port of Long Beach.
a)

14

b)

c)

15

2.2.2. Instrumentation for on-road measurements

On-road measurements were performed inside two major ground transportation routes north of the
POLA and POLB, i.e. the I-710 and I-110 freeways, from May to June of 2017. In order to assess
the validity of the results and their agreement with values reported in the literature, the 2017
campaign results were compared to those from an earlier study performed in 2016
(Shirmohammadi et al., 2017) at the same freeways and during a similar season. Freeway
measurements covered 40 km of I-110 and I-710, between their intersections with the I-105
freeway and the terminal island exit for both freeways. On-road freeway emission measurement
were carried out using a mobile platform with portable, battery-powered instruments. Sampling
was performed on 9 randomly chosen weekdays over 4 weeks, with relatively low variations in
the meteorological conditions. The mean (±SD) values for temperature, wind speed, and relative
humidity (RH) during the sampling days were 21.5 ± 3.2°C, 2.87 ± 1.1 m. s
-1
, and 57% ± 6%,
respectively (Table 4.1). The prevailing wind during the on-road sampling was southerly (Figure
4.2). The meteorological data were obtained from the California Air Resources Board’s (CARB)
website (https://www.arb.ca.gov), for three meteorological stations spread across the impact zone
of POLA and POLB. Freeway sampling took place between 10 AM to 2 PM to avoid the impact
of rush-hour traffic emissions on our measurements, and to be consistent with the previous on-
road measurement campaign performed by Shirmohammadi et al. (2017). This was also done to
reduce the effect of emissions from the vehicles immediately in front of the sampling vehicle on
our measurements, a situation that might occur frequently during congested traffic condition on
the freeway. This ensures that we measured the ambient concentrations of the target pollutants
inside the freeways rather than capture the plumes coming from the tailpipe of the car immediately
in front of the sampling vehicle. Several studies in the literature have used the same approach to
16

perform measurements inside freeways with the ultimate goal of calculating emission factors and
emission rates (Park et al., 2011;Frey et al., 2003;Vogt et al., 2003;Shirmohammadi et al.,
2017;Liacos et al., 2012). Traffic flow data (i.e., counts of light-duty vehicles (LDVs) and HDVs)
were obtained from the Performance Measurement System (PeMS) website
(http://pems.dot.ca.gov) for different sites located in both freeways (I-110 and I-710) (Figure S2)
during the sampling campaign and are presented in Table 4.2. Average vehicle-km travelled per
hour (VKT) for I-110 (4738±78 vehicle-km hr
-1
) and I-710 (4831±69 vehicle-km hr
-1
) freeways
during traffic rush hours (6 AM-10 AM and 4 PM- 8PM) were higher than between 10 AM-2 PM
(4557±58 vehicle-km hr
-1
for I-110 and 4608±64 vehicle-km hr
-1
for I-710). In addition, as can be
seen in the table, the vehicle-kilometers travelled on each freeway varies by approximately ±10%
during different hours of the day. This allowed us to calculate daily emission rates even though we
had spent a few hours to perform measurements inside each freeway. PN concentrations were
measured using a diffusion size classifier (DiSCmini, Matter Aerosol, Switzerland) calibrated to
measure particles in the 7-500 nm size range. CO2 concentrations were measured with a non-
dispersive infrared (NDIR) analyzer (Licor, model LI-840 CO2/H2O analyzer, Lincoln, NE, USA).
BC concentrations were measured using a MicroAethalometer (AethLabs, model AE51, CA,
USA). PM2.5 concentrations were measured using a DustTrak continuous Aerosol Monitor (TSI
Inc., model 8520, Shoreview, MN, USA). The operational time resolution for all of the
abovementioned instruments was set to 1 second to facilitate the capture of instantaneous
variations in pollutant concentrations
2.2.3. Instrumentation for stationary/mobile measurements at the POLA and POLB

The air pollutant and meteorological measurements designed to capture ships/vessels plumes were
carried out by the Chalmers University of Technology and FluxSense Inc. from October 8 through
17

November 10, 2015, using two different measurement platforms including mobile and stationary
sampling at POLA and POLB. Detailed information on the sampling location and techniques can
be found in Mellqvist et al. (2016;2017). Meteorological conditions during this campaign are
provided in Table 4.1. The prevailing wind direction during the sampling campaign was from the
south, according to Figure 4.2. PN concentrations were measured using a water-based
condensation particle counter (TSI Inc., Model 3007, MN, USA). PM2.5 mass concentrations were
derived from PN size distributions obtained by combining PN size distributions from an engine
exhaust particle sizer, EEPS (TSI Inc., model 3090, MN, USA), which provides size distributions
in the size range of 5.6-560 nm, and an optical particle sizer, OPS (TSI Inc., model 3330, MN,
USA), which provides size distributions in the size range of 0.3-10 µm (only the size channels <
2.5 µm were used in this study). We acknowledge that there might be uncertainties associated with
estimating PM mass concentrations from number size distributions obtained from EEPS and OPS,
but we believe that these uncertainties are likely rather minimal. This is mainly because, even
though measurements taken by the OPS are dependent upon aerosol properties (e.g., the dynamic
shape factor and refractive index), the OPS is factory-calibrated using polystyrene particles that
have a dynamic shape factor of 1 (i.e., spherical particles) and a refractive index of 1.59, which
are very close to those of typical urban aerosols (Strawa et al., 2006; Watson et al., 2002). This
makes the size distributions measured by the OPS quite consistent with the overlapping size ranges
of particle distributions obtained by other instruments (including the EEPS) which rely on the
physical (or geometric) diameter of particles. Several studies in the literature, including in the Los
Angeles Basin, have merged the particle number size distributions obtained from electrical and
optical particle sizers to obtain aerosol volume/mass size distributions, and to calculate the
apparent density of urban aerosols with small uncertainties (Sowlat et al., 2016;Hasheminassab et
18

al., 2014; Chen et al., 2010; Pitz et al., 2003; Khlystov et al., 2004). CO2 concentrations were
measured with a Cavity ring down spectrometer (CRDS) (model G2301, Picarro Inc.). BC
concentrations were measured using a seven-wavelength (370-950 nm) dual spot Aethalometer
(Magee Scientific, model AE33, USA). According to a recent study by Cheng and Lin (2013), a
high correlation was found between BC concentrations measured by AE51 and AE31 instruments
with R
2
value of 0.99 and a slope of 1.05. Therefore, we expect negligible variance in the data due
to the difference in the type of Aethalometer used in this campaign compared to the on-road and
locomotive measurements campaigns. A total of 571 plumes from 132 individual ships, including
cargo, tanker, passenger, and harbor crafts, were measured during the campaign to calculate the
PM2.5, PN, and BC emission factors for the ships at POLA and POLB.
19

Table 2.1 Meteorological conditions during the sampling campaigns of the study.

Sampling campaign Temperature

(°C)
Humidity

(%)
Wind speed/direction
(m.s
-1
)
On-road 21.5 ± 3.2 57% ± 6% 2.87 ± 1.1/south
Ship mobile/stationary 18.7 ± 3.1 65% ± 5% 3.2± 1.5/south
Locomotive emissions
measurements
20.4 ± 2.2 71% ± 9% 2.7 ± 1.1/south-south west

Table 2.2 Vehicle kilometers traveled per hour (VKT) and traffic flow data during the on-road
sampling campaign for I-110 and I -710 freeways. Errors represent one standard deviation.

Time of the day
Station,
freeway
Average VMT (vehicle-km hr
-1
)
Average SD

6 AM-10 AM
1, I-110*
4708 77
2, I-110 4649 65
3, I-710 4775 65
20

4, I-710 4730 70
5, 105 2062 48
6, 405 4179 62

10 AM-2 PM
1, I-110 4498 72
2, I-110 4615 53
3, I-710 4581 61
4, I-710 4634 47
5, 105 1941 65
6, 405 4079 54

4 PM-8 PM
1, I-110 4847 72
2, I-110 4748 53
3, I-710 4932 73
4, I-710 4883 60
5, 105 2102 52
6, 405 4160 68
* Reported parameters are the average of values extracted for each station on the corresponding hours.

* Traffic flow data are obtained from the Performance Measuring System (PeMS) website, operated by
the California Department of Transportation (CalTrans).
21

2.2.4. Instrumentation for Locomotive emissions

The emission factors for locomotives were obtained from a recent study conducted by our groups
at the Alameda Corridor, located near POLA and POLB, on a large number (i.e., N = 88) of plumes
from in-use line-haul freight locomotives (Krasowsky et al., 2015). The instruments used for PN,
BC, PM2.5, and CO2 measurements were identical to that used in our on-road freeway
measurements discussed in section 4.2.2.
Summary information on the instruments used at each sampling campaign to measure PN, PM,
BC, and CO2 concentrations is presented in Table 4.3. Summary information on meteorological
conditions are also presented in Table 4.1 and Figure 4.2. As can be seen in the figure and the
table, although the sampling campaigns in this study took place over different seasons and in
different years, the meteorological conditions during these campaigns were quite similar. This is
mainly due to the small variability of meteorological conditions in Los Angeles during the warmer
period of the year (i.e., April through October). Therefore, we expect that the impact of
meteorological conditions due to the occurrence of sampling campaign in different seasons and
years to be minimal.
22

2.2.5. Calculation of emission factors and rates

2.2.5.1. Emission factors

Emission factors for freeway vehicles (both HDVs and LDVs) were calculated using
the following equation (Kirchstetter et al., 1999):

(2.1)

Where, EFC is the emission factor for species C in units of g kg-fuel
-1
for PM2.5 and BC, and
particles kg-fuel
-1
for PN; [C]m is the measured concentration of species C at the microenvironment
(particles cm
-3
for PN, µg m
-3
for PM2.5 and BC); and [C]bg is the background concentration of
species C; [CO2]m is the measured mass concentration of CO2 at the microenvironment (mg C m
-

3
); [CO2]bg is the is the background mass concentration of CO2 at the microenvironment; ωc is the
weight fraction of carbon in the relevant fuel (0.85 for gasoline and 0.87 for diesel) (W.
Kirchstetter et al., 1999;Graham et al., 2008); and α is a unit correction factor, with a value of 10
3

for PM2.5 and BC, and a value of 10
15
for PN. For on-road measurements, background
concentrations were assumed to be the 5
th
percentile of the total measured concentrations (Riley et
al., 2016).

Ship emission factors for PM2.5, PN, and BC were obtained from Mellqvist et al. (2017), in which
emission factors were calculated from the individual ship plume data, as described in Section 4.2.3.
More detailed information about the methodology can be found in Beecken et al. (2014) and
Beecken et al. (2015). The emission factors for locomotives were obtained from the study by
Krasowsky et al., (2015) as noted earlier. In case of HDVs operating at the POLA and POLB,
emission factors (EFHDVs) were calculated from I-710 and I-110-specific emission factors using
23

equations (2) and (3) (Ning et al., 2008), with the assumption that the emission factors for HDVs
operating at the ports and those on nearby freeways are the same:
EFC = φ * EFC, HDVs + (1-φ) * EFC, LDVs
(2.2)

Where, φ is the fraction of fuel carbon emitted by HDVs; and EFC,HDV and EFC,HDV are
corresponding emission factors for LDVs and HDVs. Due to the very small contribution of HDVs
on the I-110 freeway (about 3%) in the above equation, the emission factor for the I-110 freeway
was assumed to be equal to that of LDVs (i.e., EFC, LDVs). The value of φ was calculated using
the following equation (Allen et al., 2001):
(2.3)
Where, nk is the number of vehicles of type k; Uk is the fuel consumption rate of vehicles of type
k in units of L km
-1
; ρk refers to the fuel density of vehicles of type k in units of kg L
-1
; and Wk
denotes the weight fraction of carbon in the fuel. Emission factors for CHE were also assumed to
be same as those of HDVs, according to Browning and Bailey (2006). Due to the skewness in the
distributions of measurements taken at each microenvironment, we used median emission factors
to minimize the impact of outliers in our calculations, as the median is less affected by outliers
than the arithmetic mean.

2.2.5.2. Emission rates

Having calculated/obtained emission factors for freeways, emission rates were then derived
using the following equation:
ERC = EF
C
× FC × VKT × ρ (2.4)
Where, ERC is the average daily emission rate for species C in units of kg day
-1
for PM2.5 and BC,
and particles day
-1
for PN; FC is the average vehicle fuel consumption (L km
-1
) (0.12 L km
-1
and
24

0.47 L km
-1
for LDVs and HDVs, respectively) (Kirchstetter et al., 1999); ρ is the fuel density (kg
L
-1
) (0.74 kg L
-1
for gasoline and 0.84 kg L
-1
for diesel, respectively) (Ban-Weiss et al., 2008); and
VKT is the vehicle-km traveled per day. Using traffic flow data obtained from the PeMS website,
we calculated freeway-specific values for the above parameters by factoring in the fraction of
LDVs and HDVs in each freeway. For example, in case of the freeway I-710, the overall fuel
density was calculated as follows:
(2.5)
During our on-road measurements, HDVs accounted for 3.12 ± 0.3%, and 20.12 ± 2.2% of total
traffic on I-110 and I-710, respectively. In addition, the total (mean±SD) traffic count (including
both LDVs and HDVs) during the measurements was 6256 ± 682 vehicles hr
-1
and 6893±
349vehicles hr
-1
on I-110 and I-710, respectively. The corresponding VKT values were 113,078±
1647 vehicles-km.day
-1
and 116,493 ± 1999 vehicles-km.day
-1
for I-110 and I-710, respectively.
Since sections of the I-105 and I-405 freeways were also within the zone of impact of POLA and
POLB, emission rates for these freeways were calculated using the freeway-specific emission
factors reported by Shirmohammadi et al. (2017), and the traffic flow and composition on these
freeways during the on-road sampling campaign.
Emission rates for ships were calculated in different engine modes, including maneuvering and
hoteling, according to the CARB’s emission estimation methodology for OGVs (CARB, 2008),
using the following equations:
(ERmaneuvering)C = EF × P × LF × SFC × FCF × n × t (2.6)

(ERhoteling)C = EF × P × LF × SFC × FCF × N (2.7)

Where, P is the maximum power output of the auxiliary engine in different engine modes (kW);
LF is the load factor of the engine based on the engine mode (hoteling, maneuvering); SFC is the
specific fuel consumption of the engine regardless of the engine mode (kg-fuel kW
-1
day
-1
); FCF
25

is the fuel correction factor based on the fuel type (% sulfur content); n is the number of the
departure/arrivals to the port or the number of shifts between the ports (POLA and POLB) on a
daily basis; N is the number of ships in the berth; t is the time that a vessel operates in maneuvering
mode (day).
According to CARB (2008) and POLA (2015), the average of the maximum power output of the
auxiliary engine (P) at POLA and POLB is 901 kW and 2053 kW for hoteling and maneuvering
modes, respectively. The load factor (LF) is proportional to the cube of the vessel’s actual-speed
to maximum-speed ratio. Vessels operating at speeds higher than 9 knots are considered fast, with
an LF of 0.7, and vessels operating at speeds slower than 9 knots are considered slow, with an LF
of 0.3 (POLB, 2015). The weighted average LF was calculated for each engine mode based on the
number of low- and high-speed ships reported in the air emission inventory of POLB (2015) and
POLA (2015), with corresponding values of 0.37 and 0.51 for hoteling and maneuvering modes,
respectively. According to CARB (2008), SFC values of 0.227 kg-fuel kW
-1
hr
-1
and 0.217 kg-fuel
kW
-1
hr
-1
are reported for the auxiliary engine of ships that use heavy fuel oil (92% of population)
and distillate oil (8% of population), respectively. These values were assumed to be the same for
different engine modes. A weighted average of 0.224 kg-fuel kW
-1
hr
-1
was calculated and used
for emission rate calculations, taking into account the proportional composition of ships that use
different types of fuel. In case of FCF, which is a correction factor for SFC, we used the mean
value (0.4) of the range (0.3-0.5) reported in the 2007 air emission inventory released by the POLB
(POLB, 2007a). In addition, according to POLB’s emission inventory, the number of ships in berth
at the POLB in 2015 ranged from 33 to 52 (POLB, 2015), therefore we used the mean value of 42
for each port (i.e., a total of 84) in calculating ERHoteling. The number of the departure/arrivals to
the POLB and POLA was estimated to be 15 per day (i.e., 5378 vessels per year) and 13 per day
26

(i.e., 4631 vessels per year), respectively, based on the annually averaged number of
arriving/departing vessels reported in the POLB’s and POLA’s emission inventory (POLB, 2015;
POLA, 2015). The operating time of vessels in the maneuvering mode was also estimated based
on the values reported in CARB (2008), ranging from 1-3 hours in the ports. The mean value of 2
hr was used in our ERManeuvering calculations.
The following equation was used to calculate emission rates for CHE at the ports (POLB, 2015):
ERCHE = EF × TEC per Year × SFCCHE (2.8)
(2.9)

Where, TEC is the annual total energy consumption of CHE in units of kW hr year
-1
, which is
calculated from equation (8) (POLB, 2015); and SFCCHE is the weighted average of diesel-
equivalent specific fuel consumption of the CHE operating at the port terminal in units of kg-fuel
kW
-1
hr
-1
(with a value of 0.2 kg-fuel kW
-1
hr
-1
according to POLB (2006)); Pi is the average engine
power for specific handling equipment (ranging from 50-500 kW) each working a certain number
of hours per year (ranging between 38-2200 hours per year); Ni is the number of equiupment in
each specific type at the POLA and POLB (ranging from 3-1020). All of the above infromation is
obtained from POLB (2015) and POLA (2015) and provided in Table S6. According to the above
information, a total TEC value of 2.5×10
7
kW hr year
-1
was used in this study. The following
equation was used to calculate emission rates for locomotives at the impact zone of POLA and
POLB (Bergin et al., 2012) :
ERLocomotive= (2.9)
Where, MMGT is the millions of gross tons hauled for all of the locomotive links at the port
terminal per year; N is the number of railroad lines at the port terminal; L is the length of each link
at the port terminal in units of km; and RFCI is the railroad fuel consumption index in units of
27

gross ton-km kg-fuel
-1
. In this study, we used a value of 803 MMGT per year for the total in-port
line activity (POLB, 2015;POLA, 2015). In addition, POLA and POLB have 3 (the value for N in
the above equation) major railroad lines, including Union Pacific Railroad (UP), the BNSF
Railway (BNSF), and the Pacific Harbor Lines (PHL), overall making up to 180.80 km (the value
for L in the above equation) of track lines in the port area (POLB, 2007b; Caltrans, 2012). In
addition, the section of the Alameda Corridor that lies within the impact zone of the port (with
total length of 24 km railroad) was included in our calculations. RFCI was calculated using the
approach suggested by Bergin et al. (2012), leading to a value of 6.76×10
3
gross ton-km kg-fuel
-1

used in our calculations.
For HDVs operating at the ports, PN and BC emission rates were calculated using equation (4).
As mentioned in section 2.5.1., emission factors for HDVs were derived based on I-110 and I-710
freeway emissions. A total VKT value of 14,032,895 vehicle-km per year was used for calculating
the emission rate of HDVs operating at the POLA and POLB, according to the air emission
inventory report of POLB (2015) and POLA (2015).
Since the value of each of the input parameters used in calculating the emission rate for each
mobile source was associated with a range/uncertainty, the total uncertainty associated with the
emission rate calculation for each source was calculated according to the following equation
(Farrance and Frenkel, 2012):
Total Uncertainty= ×ER (2.10)
Where, Pi is the value for parameter i used in the calculation of emission rate for specific source;
Unci is the associated uncertainty for parameter i; ER is the emission rate value calculated for each
specific source.

28

2.3. Results and discussion

2.3.1. PM2.5, PN, and BC emission factors

The calculated/reported PM2.5, PN, and BC emission factors for different sources evaluated in this
study are reported in Table 4.4. Locomotives have the highest PM2.5, PN, and BC emission factors
among all sources, with corresponding values of 1.6±0.34 g kg-fuel
-1
for PM2.5, 2.1±0.78×10
16

particles kg-fuel
-1
for PN, and 0.9±0.45 g kg-fuel
-1
for BC, as reported by Krasowsky et al. (2015).
These values are well within the range of locomotive emission factors reported in the literature
(Bergin et al., 2012, Johnson et al., 2013). For locomotives, PN emission factors are directly
proportional to the sulfur content of the fuel, whereas for PM mass emission factors, in addition to
the sulfur content, other factors such as fuel combustion efficiency and lubricating oil combustion
rate are also important (Johnson et al., 2013). Although locomotive emissions have been subject
to increasingly more stringent regulations over the past few decades (EPA, 2009), their PM
emission factors are still substantially high, especially as compared to other combustion sources
(Table 4.4).
The second highest emission factors (shown in Table 4.4) were observed for ships and vessels,
with corresponding median values of 0.61±0.84 g kg-fuel
-1
for PM2.5, 0.47±0.67×10
16
particles kg-
fuel
-1
for PN, and 0.49±1.30 g kg-fuel
-1
for BC, as reported by Mellqvist et al. (2017). The median
PN emission factor of ships used in this study is in agreement with the values reported by Hobbs
et al. (2000) for distilled (0.4-1.3×10
16
particles kg-fuel
-1
) and residual (1.6± 0.5×10
16
particles kg-
fuel
-1
) fuel, as well as those reported by Juwono et al. (2013) for dredging vessels (0.22-1.50×10
16

particles kg-fuel
-1
). The median PM2.5 emission factor for ships used in the current study is lower
than the value reported by Moldanova et al. (2009) (1.49 g kg-fuel
-1
), mainly because in that study
PM emission factors were reported for main engines, while the PM2.5 emission factor used in this
29

study relates to the ships’ auxiliary engine, as ships are only allowed to use their auxiliary engines
when at the ports terminal. In addition, recent studies have indicated that ships PM emission factors
are correlated with the fuel sulfur content, particularly when it exceeds 1% (Sax and Alexis,
2007;Zhang et al., 2018; Agrawal et al., 2008; Alfoldy et al., 2013). Moldanova et al. (2009)
reported a fuel sulfur content of 1.97% for the sampled ships in their study, whereas the maximum
allowable fuel sulfur content for the ships arriving at POLA and POLB is 0.2%. Therefore, this
significant difference (approximately 10-fold) in the fuel sulfur content across these two studies
can be another reason why we observed lower PM emission factors in this study as compared to
values reported by Moldanova et al. (2009).

30

Table 2.4 Median PM2.5, particle number (PN), and black carbon (BC) emission factors (EF) for
different pollution sources near the Port of Los Angeles (POLA) and Long Beach (POLB). Errors
represent one standard deviation (SD).
In order to evaluate the impact of engine mode and regulations imposed on the sulfur content of
fuels on PN and SO2 emission factors, we have gathered data from studies performed between
2000 and 2015 reporting PN and SO2 emission factors from ships and marine vessels, the results
of which are presented in Figure 4.3. As shown in the figure, while the correlation between EFPN
and EFSO2 has been consistent through the years, ships’ emission factors for both PN and SO2 have
decreased drastically over the years, which can be attributed to the reduction in the sulfur content
of ships’ fuels as mandated by regulations. The current maximum allowable sulfur content at the
POLA and POLB is 0.2% by weight (POLB, 2017; POLA, 2017), whereas the estimated sulfur
content of the ships samples in the other three studies indicated in Figure 4.3 was in the range of
1.5-2.4% (Sinha et al., 2003;Chen et al., 2005;Alfoldy et al., 2013). Another possible reason for
the observed difference in the PN and SO2 emission factors across these studies could be that in
the current study, all ships were running on their auxiliary engines, as mandated by the ports,
whereas in the previous studies, the sampled ships were either using their main engine (in case of
Sinha et al. (2003) and Chen et al. (2005)), or a mix of main and auxiliary engines (in case of
Alfoldy et al. (2013).
particles kg-fuel
-1
for PN, and 0.06±0.02 g kg-fuel
-1
for BC) were 1.5-3 times higher than those of
the I-110 freeway (0.14±0.04 g kg-fuel
-1
for PM2.5, 0.03±0.01×10
16
particles kg-fuel
-1
for PN, and
0.02±0.01 g kg-fuel
-1
for BC). This is mainly because of the larger fraction of HDVs on the I-710
(around 20%) than on the I-110 (around 3%) freeway, as HDVs have significantly higher emission
factors than LDVs (Table 1). The emission factors derived for the I-110 freeway in this study are
31

quite comparable to those reported by Shirmohammadi et al. (2017) for the same freeway.
Shirmohammadi et al. (2017) reported I-110 emission factors of 0.17±0.03 g kg-fuel
-1
for PM2.5,
0.05±0.02×10
16
particles kg-fuel
-1
for PN, and 0.05±0.01 g kg-fuel
-1
for BC, all of which are well
within the range of values obtained in this study.

2.3.2. PN, PM2.5, and BC concentrations on freeways and at the San Pedro Bay ports

Figure 4.4 shows the mean PM2.5, PN, and BC concentrations measured at the POLB and on I-110
and I-710 using box-and-whisker plots. PM2.5 concentrations displayed significant variability
during the sampling time, particularly the levels measured on the I-710 freeway. In addition, I-710
had the highest PM2.5 concentrations among the three microenvironments, with a mean (±SD)
value of 32.81±6.92 µg m
-3
, followed by I-110 with a mean PM2.5 concentration of 25.17±3.85 µg
m
-3
. Lower PM2.5 concentrations were observed at the POLB, with a mean value of 21.34 ±4.72
µg m
-3
, which can be attributed to the fact that measurements at the POLB took place in an area
that was solely downwind of ship and cargo-transport emissions, and not affected by other sources
of PM, including freeway emissions. As shown in Fig 4.4(b), a similar trend was also observed for
the mean PN concentrations, with I-710 exhibiting the highest PN concentrations
(3.99×10
4
±4.81×10
3
particles cm
-3
), followed by I-110 (3.24×10
4
±2.05×10
3
particles cm
-3
) and
POLB (3.23×10
4
±2.01×10
3
particles cm
-3
), which had quite similar PN concentrations. This trend
is consistent with the results of the study of Hu et al. (2008), in which lower PM 2.5 levels were
reported at the oceanfront site, as compared to the sites scattered across the Long Beach
community, which were impacted by a mixture of sources, including emissions from POLA and
POLB as well as traffic-related emissions. The lower concentrations observed at the POLB site
can be partly attributed to the lack of major industrial sources in the harbor area, and to some extent
32

to the prevailing wind direction bringing cleaner marine air into the sampling area and reducing
the ambient concentrations of PM2.5 and PN. The higher concentrations of PM2.5 and PN on the I-
710 than those measure on the I-110 freeway can be due to the higher emission factors of HDVs
as compared to the LDVs (as can be observed in Table 4.2). The PM 2.5 and PN concentrations
measured on I-110 and I-710 freeways are also within the range of values reported in previous on-
road studies in the Los Angeles area (Ning et al., 2008;Shirmohammadi et al., 2017;Liacos et al.,
2012). Fig 4.4(c) indicates the levels of BC measured at the POLB and on I-110 and I-710
freeways. Unlike PM2.5 and PN, BC concentrations were significantly higher at the POLB
(3.34±1.01 µg m
-3
) than the levels measured on the two freeways. This can be likely attributed
to the higher BC emission factors (by an order of magnitude for ships and locomotives operating at
the port terminal, see Table 4.2). In addition, higher BC concentrations were observed on I-710
freeway (2.23±0.91 µg m
-3
) as compared to the I-110 freeway (1.54±0.66 µg m
-3
). This can also
be attributed to the higher BC emission factor for HDVs than for LDVs (Table 1), which has also
been reported in many studies in the literature (Ning et al., 2008;Allen et al., 2001;Geller et al.,
2005;Ban-Weiss et al., 2008).
Figure 2.4 Box-and-whisker plots of variations in the concentrations of the target pollutants on-
road of freeways and at the Port of Long Beach (POLB) for: a) PM2.5; b) particle number (PN); c)
black carbon (BC); and d) CO2. The (×) sign represents the arithmetic mean. Lines inside the box
represent the median values, and the lower and higher lines of the box show the 25
th
and 75
th

percentiles, respectively. Whiskers represent the 5
th
and 95
th
percentiles.
a)

33

b)

I-110 I-710 POLB
I-110 I-710 POLB
34

c)

d)

I-110 I-710 POLB
I-110 I-710 POLB
35

2.3.3. Impact of POLB and POLA emissions at the local and regional scales

POLA and POLB are potentially important sources of air pollutant emissions in the Long Beach
community and can influence the air quality at both local (within the zone of impact of POLA and
POLB, as defined in Section 4.2.1) and regional (across the Los Angeles County) scales. In this
section, we report the absolute and relative contributions of important categories of sources,
including freeways, ships, CHE, locomotives, and HDVs at the ports, to PM2.5, PN, and BC
emissions at these two spatial scales (Table 4.5 and Figure 4.5). At the local scale, emissions from
these important sources were categorized into three main groups: 1) emissions from sections of
the freeways within the impact zone of POLA and POLB (including sections of I-110, I-710, I-
405, and I-105); 2) emissions from ships and CHE, together named as the “internal” goods
movement sources of POLA and POLB; and 3) emissions from locomotives and HDVs operating
at the ports, together named as the “external” mobile sources of POLA and POLB.
36

Table 2.5 Daily emission rates of PM2.5, particle number (PN), and black carbon (BC) for different
sources at the local and regional spatial scales. Errors represent the uncertainty associated with
each value.

Source category Specific source PM2.5 ER

(kg day
-1
)
PN ER

(particles day
-1
)

*10
19

BC ER

(kg day
-1
)

Nearby
freeways at the
local scale*
Total 8.4±3.8 2.4±1.3 2.0±1.6
I-110 1.7±0.7 0.4±0.1 0.2±0.1
I-710 3.9±1.0 1.3±0.4 1.1±0.3
405 1.67±0.8 0.5±0.3 0.5±0.1
105 1.2±0.7 0.2±0.1 0.3±0.1
Internal Total 48.1 ±16.3 32.8±10.3 33.9±8.8
goods movement
Ship hoteling mode 36.73±13.7 28.3±10.6 29.5±8.1
sources

Ship Maneuvering mode 3.2±2.1 2.4±0.9 2.6±0.9
of POLA and

POLB
Cargo handling equipment 8.02±2.7 1.6±0.4 1.8±0.2
External mobile
source
emissions of
POLA and
POLB
Total 49.7±6.3 58.8±5.4 26.4±5.3
Railway locomotive 43.9±9.8 57.6±2.8 24.7±5.6

HDV at harbor

5.7±0.9

1.4±0.5

1.3±0.4
Total freeways in Los Angeles county 506.9±20.3 191.0±40.3 119.4±10.2
37

Figure 2.5 Relative contributions of major sources at the Ports of Los Angeles (POLA) and Long
Beach (POLB) to PM2.5, particle number (PN), and black carbon (BC) emissions: a) at the local
scale; and b) at the regional scale.

38

As can be seen in Table 4.5, at the local scale, the external mobile sources of POLA and POLB
had the largest contributions to PM2.5 and PN emissions compared to all other sources, with
corresponding emission rates of 49.7±6.3 kg day
-1
for PM2.5 and 58.8±5.4×10
19
particles day
-1
for
PN. This was mainly due to the significant emissions from port-related railway locomotives,
emitting as much as 43.9±9.8 kg day
-1
of PM2.5 and 57.6±5.4×10
19
particles day
-1
of PN. The next
major category contributing to PM2.5 and PN emissions was the internal goods movement sources
of POLB, emitting 48.1±16.3 kg day
-1
of PM2.5 and 32.8±10.3×10
19
particles day
-1
of PN.
Emissions from ships in the hoteling mode were the major contributor to total emissions of this
source category, with emission rates of 36.7±13.7 kg day
-1
for PM2.5 and 28.3±10.6×10
19
particles
day
-1
for PN, i.e. 5-20 times higher than those of other sources in this category (i.e., emissions
from ships in maneuvering mode, and those from CHE). In terms of BC emissions, internal goods
movement was the dominant source category, with a total emission rate of 33.9±8.8 kg day
-1
, of
which approximately 90% (i.e., 29.5±8.1 kg day
-1
) was from ship emissions during hoteling mode.
The second major contributor to BC emissions was the external mobile sources at the POLA and
39

POLB, with a total BC emission rate of 26.4±5.3 kg day
-1
, of which around 95% (i.e., 24.7±5.6 kg
day
-1
) was from locomotives.
Nearby freeways (i.e., I-110, I-710, I-105, and I-405) were the third major contributor, with total
emission rates of 8.4±3.8 kg day
-1
for PM2.5, 2.4±1.3×10
19
particles day
-1
for PN, and 2.0±1.6 kg
day
-1
for BC. Among the freeways, I-710 was the main contributor, with corresponding emission
rates of 3.9±1.0 kg day
-1
for PM2.5, 1.3±0.4×10
19
particles day
-1
for PN, and 1.1±0.3 kg day
-1
for
BC, which are 3-6 times higher than those of the other freeway transects, mainly due to the large
percentage of diesel trucks on this freeway.
HDVs operating at the ports are not a major contributor to PM2.5, PN, and BC emissions among
other sources at the ports (5.7±0.9 kg day
-1
for PM2.5, 1.4±0.2×10
19
particles day
-1
for PN, and
1.3±0.4 kg day
-1
for BC), which might be due to the implementation of the Clean Truck Program.
The main objective of this program which started in 2010 was to substantially reduce PM
emissions of diesel trucks that operate at the ports (POLB, 2016). According to the POLB’s
emission inventory (POLB, 2016), there has been a 70% reduction in the HDVs emissions at the
port area due to the implementation of the abovementioned program. In addition, the distinct
difference in the ship emission rates between the hoteling mode and the maneuvering mode is
probably due to the fact that in the hoteling mode ships are near the berth and their auxiliary engine
is operating 24 hr a day; however, in the maneuvering mode, the auxiliary engine is on only for 2-
4 hr per day, depending on the ship type and arriving/departure schedule. To reduce the ship
emissions in the hoteling mode, POLB is planning to require vessels to use green technologies
near the berth (i.e., electrical charging) starting in 2018 (POLB, 2011).
Figure 4.5(a) illustrates the relative contributions of these three categories of sources to PM2.5, PN,
and BC emissions at the local scale. As shown in the figure, for PM2.5, a total of 90% (47±9% from
external mobile sources, and 45±8% from internal goods movement sources) of emissions were
40

from the sources at the POLA and POLB, while nearby freeways only contributed to 8±2% of total
PM2.5 emissions. The same trend was also observed for BC and PN emissions, being dominated
by sources at POLA and POLB. These sources together accounted for 97% of BC emissions
(55±10% from external mobile sources and 42±8% from internal goods movement sources), while
freeways contributed to 3±1% of total BC emissions within the impact zone of POLA and POLB.
In case of PN emissions, the contribution of freeways accounted for 3±0.5% of total PN emissions
within the impact zone of POLA and POLB, while port-related activities contributed to 97±14%
of total PN emissions. These results indicate that within the zone of impact of POLA and POLB,
PM2.5, PN, and BC emissions from the ports are nearly 10-40 times higher than those from nearby
freeway emissions, underscoring the significant contribution and overall impact of internal goods
movement and mobile sources at the ports terminal to total emissions of PM2.5, PN, and BC within
the impact zone of POLA and POLB.
In order to put these results into perspective, contributions from sources at POLA and POLB and
total freeway emissions were also compared at a regional scale, the results of which are illustrated
in Figure 4.5(b). In this scenario, emissions from all sources at the ports (i.e., ships, CHE,
locomotives, and HDVs) are grouped together as the “total POLA and POLB emissions”. The total
freeway emissions of the Los Angeles County were calculated using the average VKT reported by
the CalTrans for the transects of the I-710, I-110, I-105, and I-405 freeways during the on-road
measurements multiplied by the total length of all freeways in Los Angeles County (1500 km).
The total freeway emission rates were estimated using the information above, and the emission
factors calculated (for I-110 and I-710) or reported in the literature (for I-105 and I-405
(Shirmohammadi et al., 2017)) for these transects of freeways. For the purposes of this study, we
assumed that the VKT values as well as the emission factors used for these sections of the
abovementioned freeways adequately represent those of other freeways within the Los Angeles
41

County. In addition, we also used the PM2.5, PN, and BC emission rates calculated for the Los
Angeles Airport (i.e., LAX) by Shirmohammadi et al. (2017) to take into account another
important source of PM emission in the Los Angeles County.
As can be seen in Figure 4.5(b), at the regional scale, emissions from total freeways across the Los
Angeles County far exceed those from sources at the POLA and POLB. The PM2.5, PN, and BC
emissions rates for the total freeways in the Los Angeles County were 506.9±20.3 kg day
-1
,
1.9±0.4×10
21
particles day
-1
, and 119.4±10.2 kg day
-1
, respectively. These correspond to relative
contributions of 81±4% to PM2.5 mass, 61±4% to PN, and 64±6% to BC emissions for traffic,
while the sources at POLA and POLB together contributed 16±7%, 29±9%, and 33±5% to PM2.5,
PN, and BC emissions, respectively. It can also be seen in Figure 4(b) that contributions from LAX
to total PM2.5, PN, and BC emissions were quite minimal compared to those from freeways and
sources at the POLA and POLB, making up to only 3±1% of PM2.5 emissions, 10±1% of PN
emissions, and 3±1.8% of BC emissions across the Los Angeles County.
According to these results, it can be concluded that although sources at the POLA and POLB were
the major contributors to PM2.5, PN, and BC emissions within the impact zone of the ports, at the
regional scale covering the entire Los Angeles County, traffic is the dominant source of PM2.5, PN,
and BC, with emission rates 2-5 times higher than those of the sources at POLA and POLB. Our
results are in agreement with those of Shirmohammadi et al. (2017) who evaluated the impact of
emissions from LAX and freeways at the local (i.e., within the zone of impact of LAX) and
regional (i.e., the whole Los Angeles County) scales. The authors of that study also found that at
the local scale, LAX was the major contributor to PM2.5, PN, and BC emissions, whereas in the
Los Angeles County as a whole, traffic on freeways became the single most important source of
PM2.5, PN, and BC emissions.
42

2.4. Summary and conclusions

In this study, emission rates of PM2.5 mass, PN, and BC from sources at the POLA and POLB,
including ships, locomotives and HDVs operating at the ports, and CHE, we well as those from
the nearby freeway transects were evaluated, and the relative impact of emissions from these
sources were compared on local air quality (i.e., within the “impact zone” of POLA and POLB),
and on the air quality in the Los Angeles County. Results from the present study indicated that
within POLA’s and POLB’s zone of impact, port-related sources have much larger contributions
to PM2.5, PN, and BC emissions; the emission rates from these sources were estimated to be 10-40
times higher than those of the nearby freeways for PM2.5, PN and BC. However, at the regional
scale, that is the whole Los Angeles County, freeway emissions are the dominant contributor
source category to PM2.5, PN, and BC levels, with emission rates of 2-5 times higher than those
from sources at POLA and POLB. In addition to ship emissions, results from the present study
indicated the importance of other important sources, such as locomotives, CHE, and HDVs
operating at the ports, to the total emissions from POLA and POLB, which can negatively impact
the air quality of the communities adjacent to the ports. Interestingly, the PM2.5 mass and PN
emission rates from these seemingly less important sources were estimated to be 1.5-2 times higher
than those coming from ships, and in case of BC, emissions from these sources were as much as
those from ships and vessels. Findings of this research can provide useful information for policy
makers about the importance of sources of PM in the communities adjacent to POLA and POLB
and can be used by epidemiological studies aiming to assess the potential health impacts of
exposure to specific sources of PM.
43

2.5. Acknowledgments

Authors of this paper would like to acknowledge the support from the National Institute of Health
(NIH) and National Institute on Aging (NIA) (grant numbers: 1RF1AG051521-01 and
1R21AG050201-01A1), and USC Viterbi's Ph.D. Fellowship award. We also would like to thank
Chalmers University of Technology and FluxSense Inc. for providing us with the emission factor
data for the ships and vessels at the POLA and POLB.

44

Chapter 3:

Impact of urban activity source emissions on the

ambient PM0.25 oxidative potential across the Los Angeles County.

3.1. Introduction

Air pollution has become one of the most critical challenges all over the world, due to the rapid
rate of urbanization and industrialization in the past several decades. In the most recent global
burden of disease (GBD) study (Gakidou et al., 2017), it was found that exposure to air pollution
causes an estimated 6.1 million premature deaths in 2016 around the world, of which 4.1 million
deaths are attributed to exposure to ambient particulate matter (PM). Based on numerous
epidemiologic and toxicological studies, exposure to ambient PM has been associated with a large
number of adverse health effects, including cardiovascular and respiratory diseases as well as
premature death (Atkinson et al., 2001; Delfino et al., 2005; Dockery et al., 1993; Gauderman et
al., 2015; Oberdörsteret et al., 2005). More recent studies have shown that ultrafine PM (typically
defined as particles with an aerodynamic diameter, Dp < 0.1 – 0.2 µm) have more detrimental
impacts on human health, possibly due to their smaller size and their greater potential to deposit
deeper into the lung (Knol et al., 2009; Moller et al., 2008; Terzanoet et al., 2010; Yorifuji et al.,
2013).
The underlying mechanisms of the health effects associated with exposure to PM remain less
understood; however, it has been speculated that many of these health effects are likely due to the
oxidative stress caused by the interaction of PM with epithelial cells and macrophages (Michael et
al., 2013; Oh et al., 2011). Consequently, a variety of in vitro assays have been developed that are
45

capable of estimating the oxidative potential of ambient PM (or PM extracts), including the cell-
based alveolar macrophage (AM) assay used in this study, which has been found to be associated
with airway and systemic inflammation biomarkers measured in elderly populations in the Los
Angeles basin (Delfino et al., 2010, Zhang et al., 2016). In addition to the PM mass, many of the
PM chemical components have been linked to oxidative potential, including, but not limited to,
polycyclic aromatic hydrocarbons (PAHs) (Cheung et al., 2010; Cho et al., 2005; Janssen et al.,
2014), elemental carbon (EC) and organic carbon (OC), water-soluble organic carbon (WSOC),
and transition metals (Decesari et al., 2017; Saffari et al., 2013; Shirmohammadi et al., 2018).
Since each of these categories of PM components come from a variety of sources, linking the
oxidative potential of PM to its sources and formation mechanisms becomes critical in order to
effectively mitigate sources that are causing most of the toxicity. Therefore, several studies have
evaluated the oxidative potential of different size fractions of ambient PM using the AM assay,
and attempted to explore the association between particle toxicity and individual chemical
components as well as sources in different urban and rural areas around the globe, including the
North America, Europe, Middle East, and China (Cheung et al., 2010; Cho et al., 2005; Verma et
al., 2009; Verma et al., 2010; Wang et al., 2013, Daher et al., 2014, Hamad et al., 2015; Secrest et
al., 2016; Shafer et al., 2016; Shafer et al., 2010; Shuster-Meiseles et al., 2016; Vreeland et al.,
2016; Zhang et al., 2008).
The Los Angeles County is the second largest metropolitan area in the United States. A number of
diverse stationary and mobile sources of PM are located within this region, one of the most
important of which being the Ports of Los Angeles (POLA) and Long Beach (POLB). Several
sources of PM emissions are associated with POLB and POLA, the busiest ports in the United
States (Ault et al., 2009), including marine vessel and heavy-duty vehicle (HDVs) emissions, as
46

well as light-duty vehicle (LDVs) emissions in the port terminals. Concerns about the public health
impact of port-related PM emissions in the highly-populated residential areas surrounding both
POLA and POLB led to the promulgation of the San Pedro Bay Ports Clean Air Action Program
(CAAP) in 2006 to reduce the emissions from POLA and POLB (San Pedro Bay Ports, 2017).
The main objective of this study was to characterize the chemical composition of ambient PM0.25,
and to identify the major sources contributing to the oxidative potential of ambient PM0.25 in three
contrasting locations in the Los Angeles basin. We only considered PM in the quasi-ultrafine
(PM0.25) size range to minimize the impact of regional sources and formation mechanisms and
focus on PM that predominantly originates from primary combustion sources. The oxidative
potential of the collected samples was estimated using the AM in vitro assay (Landreman et al.,
2008). We used Spearman tank-order correlation analysis followed by multiple linear regression
(MLR) analysis to identify the major sources contributing to the oxidative potential of the ambient
PM0.25 samples across the sites.

3.2. Methodology

3.2.1. Sampling sites and meteorology

Samples were concurrently collected at three sites exposed to the mixture of different PM 0.25
sources. Figure 3.1 (a) illustrates a map of the study area and the location of the sampling sites.
The site at the port (PRT), operated by POLB, is located in an open area at the coastline of POLB,
immediately downwind of emissions from arriving/departing ships’ auxiliary engines while at
berth or maneuvering in the port (POLB, 2015; Mousavi et al., 2018c). This site is also affected
by emissions from port-related activities, i.e. emissions from combustion sources that are
responsible for the transportation of goods from/to the pier. These sources include diesel cargo-
47

handling equipment (CHE), heavy-duty vehicles (HDVs), and locomotives operating at the port
terminals (POLB, 2015; Mousavi et al., 2018c). The North Long Beach site (NLB), operated by
the south coast air quality management district (SCAQMD), is located 500 m downwind of the
major interstate freeway I-405, and about 1 km to the east (i.e., downwind) of the I-710 freeway,
which has the highest percentage of HDVs (around 30% during this sampling campaign) in the
Los Angeles freeway network (Lee et al. 2009; Liacos et al., 2012; Ntziachristos et al., 2007). The
above makes NLB a mixed site impacted by both freeway emissions and port-originated PM0.25.
In contrast, the site at USC (operated by USC’s aerosol lab) is approximately 60 km north of the
POLB and is located 150 m downwind of the major freeway I-110, at the University of Southern
California’s Particle Instrumentation Unit (PIU). This site is considered a typical urban
background site impacted by a mixture of fresh and aged traffic emissions as well as PM formed
by secondary photochemical reactions during the warmer months of the year (Mousavi et al.,
2018a; Shirmohammadi et al., 2018; Sowlat et al., 2016). Table 3.1 summarizes the average
metrological conditions (i.e., temperature, relative humidity (RH), and wind speed) during the
sampling period at all sites. In addition, Figures 3.1(b-d) illustrate the wind roses at each of the
sampling sites during the sampling period. According to Table 3.1(a), temperature and relative
humidity were comparable at all sites. Moreover, as shown in Figure 3.1(b-d), the southwesterly
predominant wind direction in the studied area corroborates the impact of upwind port and freeway
emissions on the receptor sites.
48

Figure 3.1 (a) Map of the study area with respect to the location of the three sampling sites. Wind
roses during the sampling period at: (b) PRT; (c) NLB; and (d) USC.
(a)

49

(b)

(c)

50

(d)

Table 3.1 Summary of the meteorological data (average ± standard deviation) at the three
sampling locations for: (a) current study; (b) Hu et al. (2008). Data were obtained from the
website of the California Air Resources Board (CARB) for the sites that were closest to our
sampling locations.
(a)

Site Temperature

(°C)
RH (%) Wind speed (m/s)
PRT 17.3 ± 1.6 65 ± 19 2.5 ± 0.8
NLB 18.1 ± 1.9 53 ± 15 2.3 ± 0.6
USC 18.7 ± 2.1 32 ± 13 2.1 ± 0.8
51

(b)

Site Temperature

(°C)
RH (%) Wind speed (m/s)
PRT 16.1 ± 1.2 54 ± 12 3.1 ± 0.7
NLB 17.2 ± 1.3 50 ± 11 2.9 ± 0.8
USC 17.1 ± 1.7 40 ± 20 2.4 ± 0.4

3.2.2. Collection schedule and method

Five weekly PM0.25 samples were collected concurrently during June and July of 2017 at the three
sites. Twenty-four-hour time-integrated samples were collected on a weekly basis during both
weekdays and weekends. At each site, four parallel Sioutas
TM
personal cascade impactor samplers
(PCISs) (SKC, Inc., Eighty Four, PA, USA) (Misra et al., 2002), operating at a flow rate of 9 lpm,
were installed for size-segregated sampling of PM0.25 and accumulation mode (PM0.25-2.5) PM
fractions. However, as discussed above, the focus of the current study is on the PM0.25 size fraction.
Based on the chemical analysis requirements, three of the PCISs were loaded with 37 mm PTFE
(Teflon) filters (Pall Life Sciences, 3-μm pore, Ann Arbor, MI), while the fourth PCIS was loaded
with 37 mm glass fiber filters (TC-GFF) (Whatman International Ltd., Maidstone, England).

3.2.3. Gravimetric and chemical analysis

PM0.25 mass was calculated through gravimetric analysis via pre- and post-weighing of the filters
with high precision (± 0.001 mg) at a room with controlled temperature and relative humidity. The
electrostatic charges of the filters were diminished using a charge neutralizer before each weighing
52

set. Due to the space and power restrictions in the POLB- and SCAQMD-operated sampling sites,
samples were collected using the light-weight, low-noise-level PCIS samplers that have a limited
flow rate of 9 lpm. In order to be consistent in terms of collection method across the sites, we also
used the PCISs at USC. Five weekly samples collected on the Teflon filters at each site (a total 15
of samples at all sites) were prepared to quantify the elemental composition of the samples by a
magnetic sector inductively coupled plasma mass spectrometer (HR-ICPMS, Finnigan Element
2). Remaining sections of Teflon filters were used for toxicity measurements. Water-soluble
organic carbon (WSOC) content of the composited filters was determined with a Sievers 900 Total
Organic Carbon Analyzer.
The use of the PCIS in each sampling location (for the reasons stated in the previous paragraph)
restricted the total PM0.25 mass collected each week to less than 1 mg; therefore, in order to achieve
the mass loadings necessary for organic tracer chemical analyses, the five weekly samples
collected at each site on TC-GFF filters were combined into two composites. A total of six
composited samples (two composites at each site) were used for determination of the organic
compounds, including (PAHs), n-alkanes, hopanes, steranes, and organic acids (including
secondary organic aerosol (SOA) tracers) using Gas Chromatography-Mass Spectrometry (GC-
MS).

3.2.4. Assessment of PM0.25 oxidative potential by means of the alveolar macrophage (AM) assay
The oxidative potential of PM is defined as the ability of particles to generate oxidative species or
to consume anti-oxidant compounds, and has been widely used as a metric to estimate the toxicity
of ambient PM (Daher et al., 2012; Landreman et al., 2008). In this study, the AM assay was used
to evaluate the oxidative potential of the collected PM0.25 samples, similarly to our previous studies
in the Los Angeles basin (Hu et al., 2008; Saffari et al., 2013; Saffari et al., 2015; Shirmohammadi
53

et al., 2018). According to the protocol, Teflon filters were extracted in 1 ml of sterilized Milli-Q
water followed by a 16-h agitation and 30-min sonication. Rat alveolar cells (NR8383 cell line,
American Type Culture Collection) were then exposed to the extracted aqueous samples and 2,7-
dichlorodihydrofluorescein was used as a fluorescent probe to quantify the oxidative potential of the PM
samples. In this assay, the non-fluorescent 2-,7-dichlorodihydrofluorescein diacetate (DCFH-DA) is first
added to the exposed cell culture, whereupon it enters the cells and is de- acetylated by intracellular
enzymes to form dichlorodihydrofluorescein (DCFH). DCFH is then oxidized by reactive species to form
the highly fluorescent 2,7-dichlorofluorescein that is monitored using a fluorescent microplate reader in
fluorescence units per mass of PM (FU.µg
-1
PM). This assay is performed on dilution series of each sample
extract to establish a linear dose- response region. In addition, Zymosan is used in this assay as a positive
control, mainly because the Toll-like receptors (TLR-2) of macrophage cells can recognize it and initialize
a strong response. After blank correction, the fluorescence data in units of FU.µg
-1
PM is then normalized
to the response of a unit of Zymosan, and reported in units of µg Zymosan.µg
-1
PM. This is done to facilitate
comparisons across datasets, and to minimize the variations in the sensitivity of the methods across different
batches. The values of the above per- PM mass units are further multiplied by the ambient PM concentration
of the pertinent sample to convert them into per unit of air volume, i.e., µg Zymosan.m
-3
air (Landreman et
al., 2008), which might be more relevant from the standpoint of population exposure.

3.2.5. Source apportionment of PM0.25 related oxidative potential

We first performed Spearman rank-order correlation analysis between individual chemical species
and the measured PM0.25 oxidative potential of the samples to identify chemical species that were
statistically significantly correlated with the measured oxidative potential, serving as a guiding
tool to select the species to be included in the MLR analysis. Then, to determine the source factors
contributing to the oxidative potential, we performed an MLR analysis between the chemical
species that showed significant correlations with oxidative potential in the correlation analysis (as
54

independent variables representatives of different source factors) and oxidative potential (as the
dependent variable) using a combined dataset comprising all of the samples from the three
sampling sites, following an approach used in previous studies (Argyropoulos et al., 2016;
Shirmohammadi et al., 2018). Using a sequential regression entry method, MLR identified species
that significantly contribute to the oxidative potential associated with PM0.25. The model was then
used to determine the optimum combination of non co-linear species (variance inflation factor
(VIF) < 2.5) with statistically significant (p < 0.05) correlations with oxidative potential that led
to the highest R
2
regression value. As will be presented in the Results section, the optimum MLR
run, determined based on the R
2
value of the model and statistical significance of the predictors
(i.e., independent variables), included the following marker species: total PAHs, as representative
of combustion– generated PM (Cheung et al., 2010; Cho et al., 2005; Janssen et al., 2014), and
organic SOA tracers, representative of SOA formed by photochemical reactions (Carlton et al.,
2009; Heo et al., 2013; Hu and Yu, 2013; Ding et al., 2012; Shirmohammadi et al., 2016). Since
PAHs are emitted from a variety of combustion sources, we performed another MLR analysis,
using individual PAHs, as tracers of specific combustion sources (including vehicular and port-
related emissions), to identify the contribution of those specific combustion sources to total PAHs
concentrations and, in turn, to the oxidative potential of ambient PM0.25. Using standardized
regression coefficients for each marker species from the two separate MLR runs, we then estimated
the contribution of each source to the oxidative potential of ambient PM0.25.

3.3. Results and discussion

3.3.1. Mass concentration and chemical composition of PM0.25

3.3.1.1. Concentrations of PM0.25 mass and carbonaceous compounds

55

Figure 3.2 illustrates the average PM0.25 mass concentrations at all sites during the sampling period
of the current study as well as the concentrations reported by Hu et al. (2008) in 2007. In both
sampling campaigns, the highest PM0.25 concentrations were observed at USC, followed by NLB
and PRT. The upward trend of PM0.25 levels from PRT towards inland sites could be attributed to
the small but increased contribution of SOA formation at the receptor sites, as demonstrated in
previous studies in the area (Saffari et al., 2015). This observation is further corroborated by the
increasing trend in the concentrations of WSOC (as a surrogate of SOA (Fine et al., 2014b)) as
well as other organic SOA tracers (such as a-5 (3-hydroxyglutaric acid) and a-3 (2-hydroxy-4-
isopropyladipic acid)) (Carlton et al., 2009; Ding et al., 2008) from PRT to USC, as shown in
Figure 3.3. As will be discussed in the following sections, the increased contribution of traffic
emissions from port to central Los Angeles is another likely reason for observing such an
increasing trend in the mass concentrations of ambient PM0.25. As illustrated in Figure 3.2, a
comparison of the measured PM0.25 levels with an earlier study performed at the same sampling
sites and during the same season of the year with comparable meteorological conditions (according
to Table 3.1) indicated a clear decrease in PM0.25 mass concentrations (by about 35% overall
reduction) from 2007 to 2017 at all study sites.
3.3.1.2. Elemental content of PM0.25

Figure 3.4 shows the average concentrations of trace elements and metals in PM0.25 at all sites.
Metals and trace elements in the PM0.25 size range are emitted from a wide range of sources (Thorpe
et al., 2009). For example, elements such as Al, Ca, and Ti mainly originate from the re- suspension
of road dust in the sub-micron size (Harrison, et al., 2012; Yin et al., 2010); these species had
higher concentrations at the urban sites of USC and NLB that are impacted by traffic. Moreover,
Zn, Fe, Ba, Cu, and Pb that are abundant in tire wear and abrasion of the brakes (Harrison et al.,
56

2012; Sanderson et al., 2014) had higher concentrations at NLB and USC. In contrast, V and La
that have been used as tracers of ship and oil refinery emissions (Isakson et al., 2001; Moreno et
al., 2010; Pandolfi et al., 2011; Viana et al., 2009; Viana et al., 2014; Healy et al., 2009) were more
abundant at PRT, which is in closer proximity to the Long Beach oil refineries and is heavily
impacted by fresh ships/vessels emissions. Therefore, the higher concentration of V and La at PRT
can be attributed to contributions from ships/vessels emissions as well as those from the nearby oil
refineries. Previous studies have defined and used the Ce/La ratio to apportion the contribution of
ship emissions to the PM mass, with values below 2 indicating the presence of ship and oil refinery
emissions in the area (Viana et al., 2009; Viana et al., 2014; Pandolfi et al., 2011; Moreno et al.,
2010). In order to compare our results with those in the literature, we calculated the Ce/La ratios
for all three sites, and found values of 0.2, 0.7, and 0.9 at PRT, NLB, and USC sites, respectively.
These below unity values corroborate the impact of oil refineries and ship emissions in the area.
Additionally, Na, a tracer of sea salt particles (Tang et al., 1997), had higher concentrations at PRT
due to its vicinity to the shoreline.
57

3.3.1.3. Organic species

Concentrations of individual organic compounds, including PAHs, hopanes, steranes, and n-
alkanes in the PM0.25 size range are demonstrated for all of the sites in Figure 3.5 (a-c). PAHs are
comprised of several toxic and carcinogenic components (El-Alawi et al., 2013; Lee et al., 2003),
most of which are primarily emitted from the incomplete combustion of diesel and gasoline
engines (Galarneau 2008; Manchester-Neesvig et al., 2003). Higher total PAHs concentrations
were measured at NLB and USC, which are near major freeways, I-710 and I-110, respectively.
PRT exhibited lower but comparable PAHs concentrations, which can be explained by
contributions from ship emissions, as it has been shown that ship emissions contain a very rich
mixture of PAHs (Moldanová et al., 2009).
Hopanes and steranes are known as tracers of vehicular emissions, primarily emitted from the
combustion of the lubricant oil additives in both diesel and gasoline engines (Schauer et al., 1996).
Concentration of hopanes and steranes were comparable across all sites, with slightly higher
concentrations at PRT. These results are in agreement with the levels reported by Hasheminassab
et al. (2013), in which the authors observed higher levels of hopanes and steranes at the site closest
to the POLB and POLA than at the receptor site (i.e., USC), especially in the summer and spring.
Cumulative n-alkanes concentrations were comparable at all sampling sites with the highest levels
at USC, followed by NLB and PRT. In order to attribute n-alkanes to certain sources, we calculated
the carbon preference index (CPI), specified as the concentration ratio of the odd-to-even
numbered carbon n-alkanes (Simoneit 1986). According to the definition, a CPI of around 1
indicates the dominance of anthropogenic sources, while a CPI greater than 2 suggests the
prevalence of the biogenic sources. The calculated CPIs for each of the sites ranged from 1.1 to
1.3, suggesting the dominance of anthropogenic sources including vehicular and ocean-going
58

vessel (OGVs) emissions (Simoneit 1986). As the current study was conducted in the early
summer, the impact of wood burning on n-alkane emissions is expected to be minimal.

Figure 3.5 Average concentrations of organic species: (a) polycyclic aromatic hydrocarbons
(PAHs); (b) hopanes and steranes; and (c) n-alkanes. Error bars represent total standard deviation
(SD) of compounds.
(a)

(b)
59

(c)

3.3.2. PM0.25 Oxidative Potential

Figure 3.6 illustrates the oxidative potential of PM0.25 at each site. There are two different ways of
reporting the oxidative potential of PM, either on per PM mass collected, which reflects the
intrinsic PM redox activity, or on per m
3
of air volume reflecting the extrinsic PM oxidative
activity that represents its actual airborne concentration, therefore making it more relevant to
population exposure. In this section, we focus and discuss the per-volume oxidative potential of
the collected samples.
The highest extrinsic oxidative potential of PM0.25 was observed at USC (50.2 ±3.4 ng.m
-3
)
followed by NLB (45.5 ±2.3 ng.m
-3
) and PRT (35.3 ±4.2 ng.m
-3
) (Figure 4). In order to compare
the oxidative potential levels determined in the current study with those of recent studies conducted
using same sampling sites and to assess the consistency across studies, the data reported by Saffari,
et al. (2013) and the current work are compared in Table 3.2. According to the data in this table,
the annual average PM0.25 oxidative potential levels reported for the USC (58.6±18.3 µg
60

Zymosan.m
-3
) and NLB (52.8±16.7 µg Zymosan.m
-3
) sites in the earlier study by Saffari et al.
(2013) are quite consistent with the range of values measured in the current study (51.2±7.5
µgZymosan.m
-3
and 45.1± 6.7 µgZymosan.m
-3
for USC and NLB sites, respectively).

Figure 3.6 Oxidative potential of PM0.25 at the three locations normalized per m
3
of air volume.
Error bars represent one standard deviation (SD).

70
60
50
40
30
20
10
0
PRT NLB USC

3.3.3. PM0.25 source apportionment and associated oxidative potential of dominant sources

3.3.3.1. Correlation between individual species and oxidative potential

Spearman correlation coefficients between oxidative potential and individual PM0.25 components
are reported in Table 3.3. Due to the limited sample size at each site, we combined data points
from all sites to increase the statistical power of the correlation analysis. The number of data points
for metals and trace elements was 15 (5 per site), while that of WSOC and GC-MS organics
(including PAHs, hopanes, steranes, SOA tracers, n-alkanes) was 6 (2 per site). Moderate-to-strong
Oxidative potential
(µg Zymosan.m
-3
)
61

associations were observed between the PM0.25 oxidative potential and different metals and trace
elements; for instance, Cu, Ni, and Al (tracers of traffic emissions (Pakbin et al., 2011; Querol et
al., 2007; Shirmohammadi et al., 2015; Shirmohammadi et al., 2017) as well as V and La (tracers
of ship and oil refinery emissions (Moreno et al., 2009)) showed relatively high (R> 0.6)
correlations with PM0.25 oxidative potential. All of these associations were statistically significant,
with p values lower than 0.05, as indicated in Table 3.3.

62

Table 3.2 Spearman rank correlation coefficients (R) between the concentrations of the chemical
species (ng.m
-3
) and the oxidative potential (µg of Zymosan.m
-3
) of PM0.25 in the study area: (a)
elements and metals (N = 15); and (b) WSOC and organic compounds (N = 6). R values above 0.5
are bolded.

a)

Species Correlation

coefficient
Species Correlation

coefficient
Ba 0.45
*
Cu 0.60*
Ti 0.08 Ca 0.15
V 0.59* K 0.08
Cr -0.06 Mn 0.36
Fe 0.354 S -0.03
Ni 0.67* Al 0.50
*

Zn 0.32 Mg -0.15
As 0.23 Na -0.61
Cd -0.49 La 0.72
**

63

b)

Species Correlation

coefficient
WSOC 0.77
**

N-alkanes 0.48
Hopanes and stranes -0.31
SOA tracers 0.77
**

Total PAHs 0.82
*

Phenanthrene 0.71
**

Fluoranthene 0.31
Pyrene -0.20
Benz(a)anthracene -0.15
Chrysene 0.55
Benzo(b)fluoranthene 0.72
Benzo(e)pyrene 0.63
Indeno(1,2,3-cd)pyrene 0.94
**

Benzo(g,h,i)perylene 0.74
*

Coronene 0.75
*
Represents statistical significance at P < 0.05.

**
Represents statistical significance at P < 0.01.

Furthermore, our results indicated moderate -to -high correlations (R > 0.6) between the oxidative
potential of PM0.25 and several organic components, including species that are products of primary
64

combustion emissions, such as PAHs (Cheung et al., 2010; Cho et al., 2005; Janssen et al., 2014),
as well as tracers of secondary organic aerosols, such as WSOC, a-5 (3-hydroxyglutaric acid), and
a-3 (2-hydroxy-4-isopropyladipic acid) (Carlton, et al., 2009; Ding et al., 2008; Fine et al., 2014b).
Since PAHs are emitted from a variety of combustion sources, we also evaluated the correlation
between the oxidative potential of PM0.25 and individual PAHs as tracers of specific combustion
sources. As shown in Table 3.3(b), strong and statistically significant correlations were found
between oxidative potential of PM0.25 and three PAHs, including phenanthrene, indeno(1,2,3-
cd)pyrene, and benzo(g,h,i)perylene. Indeno(1,2,3-cd)pyrene and benzo(g,h,i)perylene are high
molecular weight PAHs that are associated with vehicular emissions, particularly emissions from
gasoline vehicles (Ning et al., 2008; Phuleria et al., 2007; Polidori et al., 2008; Schauer et al.,
1996). Therefore, they can be used as tracers of vehicular emissions in the study area. Figure 3.7(a)
indicates an increasing trend for the concentrations of these two PAHs from the port to central Los
Angeles, which is in line with our earlier observations in the previous sections indicating higher
concentrations of inorganic and organic tracers of vehicular emissions in the sites further inland.
Phenanthrene, however, is a lower molecular weight PAH that has been associated with diesel
engine emissions (Ning et al., 2008; Phuleria et al., 2007; Polidori et al., 2008). Moreover, some
earlier studies also indicated that phenanthrene is one of the most abundant PAHs found in ship
emissions, therefore it could be used as a tracer for ship emissions in areas near ports as well
(Donateo et al., 2014; Streibel et al., 2017). Emission sources at the Ports of Los Angeles and Long
Beach include not only ships, but also diesel cargo-handling equipment (CHE), locomotives, and
heavy-duty vehicles (HDVs) operating at the port (POLB, 2015; Mousavi et al., 2018);
phenanthrene could therefore be used as a tracer for all of these emissions from combustion sources
related to port activities. The spatial trend for the concentration of phenanthrene (Figure 3.7(b))
65

also indicates slightly higher levels at the port compared to the NLB and USC sites. This in contrast
to the trend observed for tracers of vehicular emissions all of these correlations corroborate the
significant impacts of primary combustion emissions (including vehicular and port-related
emissions) and SOA formation on the oxidative potential of ambient PM0.25 in the area, and were
used as a guiding tool for the selection of marker species to be included in the subsequent MLR
analysis to apportion the contribution of sources to the oxidative potential of PM0.25.
66

3.3.3.2. MLR analysis and oxidative potential source apportionment

We performed the MLR analysis to apportion the sources of PM0.25 oxidative potential using the
15-sample dataset (including metals and trace elements) and the 6-sample dataset (including
organic compounds), the results of which are presented in Table 3.4(a). The optimum model
included total PAHs, as a tracer of products of primary combustion emissions (Cheung et al., 2010;
Cho et al., 2005; Janssen et al., 2014), and organic tracers of SOA (Carlton et al., 2009; Heo et al.,
2013; Hu and Yu, 2013; Ding et al., 2012; Shirmohammadi et al., 2016), as the major contributors
to the oxidative potential of PM0.25. The model explained 97% of variability in the data, based on
the R
2
value of 0.97. The higher standardized regression coefficient of total PAHs (0.62) and SOA
tracers (0.46) also indicated the larger contribution of combustion sources to PM 0.25 oxidative
potential.
As mentioned in Section 3.2.5., since multiple combustion sources can release PAHs into the
atmosphere, we performed a second MLR analysis to apportion the sources of PAHs in the study
area using individual PAHs as tracers of specific combustion sources, the results of which are
presented in Table 3.4(b). Two sources, including vehicular emissions, represented by
indeno(1,2,3-cd)pyrene (Ning et al., 2008; Phuleria et al., 2007; Polidori et al., 2008), and port-
related emissions, represented by phenanthrene (Donateo et al., 2014; Streibel et al., 2017; Ning
et al., 2008) were the major contributors to the total PAHs concentrations in the study area,
explaining 96% of variability in the data. In addition, based on the standardized regression
coefficients of indeno(1,2,3-cd)pyrene (1.0) and phenanthrene (0.42), vehicular emissions appear
to be the major contributor to total PAHs concentrations than port-related emissions.. Our MLR
analysis also indicated negligible contributions from other combustion sources, including biomass
and natural gas burning, to total PAHs concentrations, in line with the results from several studies
67

performed earlier in this area (Hu et al., 2008; Mousavi et al., 2018b; Hasheminassab et al., 2013;
Minguillón et al., 2008).
Table 3.4 Results of the multiple linear regression (MLR) analysis performed between: (a) PM0.25
associated oxidative potential as dependent variables, and PM0.25 organic chemical species as
independent variables; (b) total PAH concentrations as dependent variable and individual PAHs
concentrations as independent variables; using the combined dataset (N=6) including WSOC and
organic species.
(a)

PM0.25 as dependent variable
Species Source Standardized

coefficient
Standard

error
Partial

R
P-

value
R
2
Constant 34.10 0.97
Total PAHs Combustion emissions 0.62 4.60 0.81 0.02

SOA tracers SOA formation 0.46 3.10 0.68 0.05

(b)

PM0.25 as dependent variable
Species Source Standardized

coefficient
Standard

error
Partial

R
P-

value
R
2
Constant -3.27 0.96
Indeno(1,2,3-
cd)pyrene
Vehicular emissions 1.00 8.04 0.78 0.01

Phenanthrene Port-related
emissions
0.42 3.33 0.69 0.04

68

Combining the results from these two separate MLR analyses, and using the standardized
coefficients obtained in each MLR analysis, we were able to quantify the contributions of the
vehicular emissions, SOA, and port-related emissions, to the overall oxidative potential of PM0.25,
the results of which are presented in Figure 3.8. The PM0.25 oxidative potential is mostly driven by
vehicular emissions and SOA formation, each contributing to 39±2% and 40±5%, respectively, to
the overall oxidative potential. Emissions related to combustion sources at the port were the next
major contributor, accounting for 16±3% of the PM0.25 oxidative potential. In addition, the higher
concentrations of indeno(1,2,3-cd)pyrene and benzo(g,h,i)perylene as well as those of WSOC and
other organic tracers of SOA formation at USC compared to NLB and PRT (Figures 3.3 and 3.7)
indicate the larger contributions of vehicular emissions and SOA formation to the oxidative
potential of ambient PM0.25 in central Los Angeles. In contrast, the higher concentrations of V and
phenanthrene at PRT compared to NLB and USC (Figures 3.4 and 3.7) suggest higher
contributions of port-related emissions to the oxidative potential of PM0.25 at the port area.
69

The main limitation of our study was the duration of our sampling campaign, which was conducted
for 7 weeks during June-July of 2017. This period was selected firstly to make our results
comparable to an earlier study conducted in the same area and time period in 2007, before the
implementation of CAAP regulations. Moreover, to apportion the ship and refinery emissions from
POLB at the NLB site and resolve their contributions to PM0.25 at sites far from the port area, we
believe that sampling should occur under meteorological conditions favoring atmospheric
dispersion that is required to transport the particles from the port area to areas further north.
Therefore, we had to exclude the late fall and winter months that are characterized by stagnation
induced by the meteorological stability (Hasheminassab et al., 2013; Hasheminassab et al., 2014).
To further corroborate the representativeness of our sampling campaign, we have compared the
annual average oxidative potential levels of PM0.25 found by Saffari et al. (2013) at the same or
very close to the same sampling sites of the Los Angeles County and the oxidative potential levels
measured in the current study, and provide the results in Table 1. The results of this comparison
confirm that our measurements were conducted during a time period that captures quite adequately
the overall annual average of the oxidative potential of PM0.25 reported in the earlier study by
Saffari et al. (2013), and should therefore be considered a reasonably robust “average” fingerprint.
The main objective of Saffari et al. (2013) study was to investigate the effect of atmospheric aging
of primary emissions on the toxicity of ultrafine particulate matter at different sampling sites
ranging from source (north Long Beach) to receptor (one in central Los Angeles and one is
Riverside, which is downwind of primary emissions from Los Angeles). In contrast, the main goal
of our current study was to evaluate the impact and contribution of ship emissions on the oxidative
potential of ambient ultrafine PM. To address this question, in addition to our sites in north Long
Beach and central Los Angeles, we chose a remote site at the port (as shown in Figure S1) that
70

was only impacted by port activities, and was located upwind of the emissions of nearby freeways
surrounding the port area, based on the prevailing wind directions shown in Figure S1 (b).
measurements performed at this site were critical, because they allowed us to estimate the impact
of ship emissions on the oxidative potential of ambient PM0.25 in the area by using it as a reference
site to which the chemical composition and oxidative potential of ambient PM0.25 at the other two
sampling sites were compared and contrasted. This is a key distinction between this study and the
study by Saffari et al. (2013).

Another potential limitation of the current study is the impact of sampling artifacts on the collected
samples, mainly caused by the volatilization of labile organic and inorganic species. However,
several studies in the literature have indicated that these artifacts are substantially reduced with
increased PM mass loadings (Chang et al., 2000; Wang and John, 1988). Cheng and Tsai
(1997) compared experimental losses of pure ammonium nitrate particles during filter sampling

with the losses predicted by Zhang and McMurry (1992). Cheng and Tsai (1997) showed that
severe evaporation loss occurs during the initial particle collection stage, when only few particles
are collected on the filter. The formation of a “particle cake” with prolonged sampling decreases
the rate of evaporation. Particle mass loading on the collection substrate is one of the most
significant factors affecting the evaporation loss of semi-volatile species (Zhang and McMurry
1992). The sooner the formation of a particle cake, the lower the evaporative losses. Considering
that our PM mass loadings were in the range of 300-500 micrograms, we therefore expect these
losses to be rather small. Furthermore, previous studies have indicated that volatilization of organic
and inorganic labile species can be minimized by using samplers with lower pressure drops, and
by reducing filter velocities (Yu et al., 2005). The Personal Cascade Impactor Samplers (PCISs)
71

that we used in this study accommodate both of these conditions, as they are extremely low
pressure drop samplers, and have low filter velocities because of the low flow rate of the sampler
(Misra et al., 2002). This helps minimize the evaporative loss of nitrate and labile organics during
samples collection. Lastly, given that the focus of our study was to link source factors to oxidative
potential of ambient PM0.25, and considering the fact that previous studies have indicated that
inorganic ions, such as ammonium nitrate, are rather innocuous species (Decesari et al., 2017;
Verma et al., 2012), we did not measure these species in this study. Therefore, any discussion
pertaining to the impact of artifacts on the concentrations of ammonium nitrate would be irrelevant.

3.4. Summary and conclusions

The main objective of this study was to explore the impact of emissions from the ports of Los
Angeles and Long Beach on oxidative potential of ambient PM0.25 measured across the Los
Angeles area in three contrasting locations. These included the PRT, as an upwind site, heavily
impacted by emissions from the port, the NLB site impacted by both port-related emissions as well
as traffic-related emissions, and the downwind USC site, which is mainly impacted by traffic-
related emissions and to a moderate to lesser degree secondary organic aerosols (SOA). Higher
concentrations of PM0.25 mass and its associated oxidative potential were observed in central Los
Angeles (i.e., USC) in comparison to the sites that were closer to the port (i.e., PRT and NLB).
Comparison of our results with an earlier study performed at the same sampling sites and during
the same season indicated a 35% reduction in PM0.25 mass concentrations from 2007 to 2017.
Overall, vehicular emissions and SOA emissions were found to be the dominant sources
contributing 39±2% and 40±5% of PM0.25 oxidative potential, respectively, underscoring the major
impact of vehicles on freeways and SOA formation on PM0.25 toxicity at the receptor sites. The
72

contribution of port-related emissions, including emissions from ships, CHE, HDVs, and
locomotives operating at the port terminals, to the PM0.25 oxidative potential was also found to be
16 ± 5% across the sampling sites. The spatial trends for the concentrations of marker species
indicated relatively higher contributions of port-related emissions to the oxidative potential of
PM0.25 at PRT and NLB, while only negligibly contributing to PM0.25 oxidative potential in central
Los Angeles, highlighting the impact of Los Angeles-Long Beach port emissions on the toxicity
of ambient PM0.25 in the communities near the ports of Los Angeles and Long Beach.

3.5. Acknowledgements

This study was supported by the National Institutes of Health (NIH) (grants numbers:
1R21AG050201-01A1 and 1RF1AG051521-01). The authors wish to thank the South Coast
AQMD staff (Dr. Olga Pikelnaya), Port of Long Beach staff (Mr. Hayden Backman), and Leidos
Inc. air quality project manager (Mr. Joel Tarcolini) for their help with the sampling at PRT and
NLB. We also would like to acknowledge support of the PhD fellowship award from the USC
Viterbi School of Engineering.
73

Chapter 4:

Source contributions and temporal trends of Redox-Active Metals in
central Los Angeles

4.1. Introduction

Air pollution is becoming an increasingly severe problem in urban areas around the globe and can
cause human health impacts (Naddafi et al., 2012). Particularly, exposure to ambient particulate
matter (PM) has been known to cause human health impacts ranging from respiratory and
cardiovascular diseases to neurodegenerative effects (Davis et al., 2013; Delfino et al., 2010;
Gauderman et al., 2015). Historically, the focus of earlier epidemiological studies has been on total
mass of PM, while the findings from more recent studies suggest that, other than the total PM
mass, physical characteristics (including number concentration, particle size, and surface area) as
well as chemical composition of PM are also critical in driving the health end-points (Davis et al.,
2013; Delfino et al., 2010).
Iron (Fe), Chromium (Cr), Cupper (Cu), and Manganese (Mn) are among the most important
redox-active transition metals that have been linked to human health effects mainly because of
their toxicological properties. This is believed to be due to the ability of these metals to induce
oxidative stress through generating reactive oxygen species (ROS), which eventually leads to
inflammation of target cells and organs (Li et al., 2009; Tao et al., 2003). Therefore, the study of
sources, transport, and spatio-temporal characteristics of these redox-active metals becomes of
paramount importance, as this information will help policy makers to more effectively regulate
and mitigate their sources and to bring down exposure to these toxic species.
74

Positive matrix factorization (PMF) has been globally used as a reliable tool for source
apportionment of PM mass (Sowlat et al., 2013; Tao et al., 2016; Waked et al., 2014) and PM
number concentrations (Kasumba et al., 2009; Pey et al., 2009; Sowlat et al., 2016a; Zhou et al.,
2004). A major drawback of source apportionment studies on chemically-speciated PM data is that
they typically rely on 24-hr samples, mainly because of the limitation of filter-based analytical
methods, which require substantial PM mass loadings and need to be performed in the laboratory.
However, sources and atmospheric chemical processes involved in producing PM have much finer
time scales (i.e., few hours as opposed to days), making it essential to capture temporal profiles of
these PM formation mechanisms. Only a few recent studies have applied PMF source
apportionment of ambient PM combined with chemical speciation measurements at higher time-
resolutions in metropolitan areas of the US and Europe (Dall'Osto et al., 2013; Fang et al., 2015;
Gao et al., 2016; Morishita et al., 2011; Pancras et al., 2013; Pancras et al., 2006; Richard et al.,
2011; Shi et al., 2017). To the best of our knowledge, such time-resolved PMF apportionment
studies have not been performed in the megacity of Los Angeles, which is one of the most polluted
air sheds in the US (Jerrett et al., 2005).
In this study, we measured the concentrations and determined the diurnal variations (with a time
resolution of 2 hrs) of four redox-active metals, i.e. Fe, Cr, Cu, and Mn, using previously developed
on-line metal monitors (Sowlat et al., 2016b; Wang et al., 2014; Wang et al., 2016). Our
measurements were conducted over a relatively long time-period (i.e., June-August 2016 as the
warm season and December 2016 through February 2017 as the cold season) in central Los
Angeles, and we subsequently employed PMF to apportion the sources of these metals in the study
area. To further characterize the sources, a variety of other species were also measured and
included in the model run. These included elemental and organic carbon (EC and OC); PM size
75

distributions data; PM2.5 mass concentrations, gaseous pollutants data, including NO2 and
O3;meteorological parameters, including temperature and relative humidity (RH); and traffic data
(for heavy- (HDVs) and light-duty vehicles (LDVs)).
4.2. Methods

4.2.1. Sampling site and study period

Samples were continuously collected in warm (i.e., June-August 2016) and colder seasons
(December 2016 through February 2017) at the University of Southern California’s Particle
Instrumentation Unit (PIU), which is located in the central part of Los Angeles immediately
(approximately 150 m) downwind of the I-110 freeway. Figure 2.1 illustrates the location of the
sampling site in central Los Angeles. According to the previous studies carried out at this site, PIU
is considered to be a mixed urban site with significant impact from traffic-related sources (Sowlat
et al., 2016a).
Figure 4.1 Map of the study area.

76

4.2.2. Instrumentation

The PM2.5 mass concentrations of the four redox-active metals were semi-continuously (i.e., every
2 hrs) measured at the PIU using the techniques previously developed for Cu (Wang et al., 2014)
and Fe, Mn, and Cr (Wang et al., 2016). Briefly, these techniques employ a two-module system:
1) PM collection module; and 2) chemical analysis module. The PM collection module is the same
for all the aforementioned metals, in which air is drawn at a flow rate of 200 lpm first through an
inlet (an inertial impactor) with a cut-point at 2.5 µm, then into a saturation tank (kept at 30-32
°C), which creates a mixture of ambient PM and saturated ultrapure water vapor. The airflow then
passes through a chiller reducing the aerosol temperature to about 21-22 °C. The resulting
supersaturation through this cooling process leads to condensation of water vapor onto the existing
particles, growing them to 3-4 µm droplets, which are subsequently collected as slurry samples
using inertial impaction. More details pertaining to the collection module can be found in (Wang
et al., 2013). For Cu measurement, our technique utilizes a cupric combination ion selective
electrode (ISE) coupled to a high impedance millivolt meter. The concentration of Cu in the
ambient air is proportional to the potential difference between the membrane and the reference
measured by the ISE and the millivolt meter (Wang et al., 2014). The Fe, Mn, and Cr monitor, on
the other hand, relies on the spectrophotometric measurement of the colored complexes formed as
a result of addition of pertinent reagents (e.g., Ferrozine for Fe measurement, Formaldoxime
(FAD) for Mn measurement, and diphenycarbazide (DPC) for Cr measurement) to the slurry
sample. In this system, the ambient concentration of the target metal is proportional to the intensity
of color formed, which is measured in a micro volume flow cell (MVFC) using a portable
spectrophotometry (Wang et al., 2016).
We measured particle number (PN) concentration and size distribution in the range of 14 nm–10
77

µm using a combination of an Optical Particle Sizer (OPS™, Model 3330, TSI Inc., USA) (for
0.3–10 μm particles) and a scanning mobility particle sizer (SMPS™, TSI Model 3081) (for 14
760 nm particles) connected to a condensation particle counter (CPC, model 3020, TSI Inc., USA).
The time-resolution for PNSD measurements was 5 min. In order to form the complete size
distribution from 14 nm to 10 µm, size channels from SMPS in the range of 13.6–710 nm were
merged with size channels from OPS in the range of 0.809-9.01 μm. More detailed information on
the merging process and its validity can be found in Sowlat et al. (2016). EC and OC concentrations
were measured with a time-resolution of 1 hr with Sunset Laboratory (Inc., USA) EC/OC monitor
(Model 4) using the National Institute of Occupational Safety and Health (NIOSH) 5040 protocol.

4.2.3. Auxiliary variables

Concentrations of gaseous pollutants as well as meteorological parameters data measured in the
North Main Street sampling site (about 3 km away from the PIU) by the California Air Resources
Board (CARB) were collected from the CARB website. We also collected traffic data for the traffic
station closest to the PIU from the freeway performance measurement system (PeMS) website.
The time-resolution for all of the auxiliary variables was 1 hr. Table 2.1 presents a summary of all
the variables that were included in the input matrix along with their time resolution and
measurement method/collection source. Since the time-resolution for all of the species that are put
into the PMF model should be the same, considering that the time-resolution of metal
measurements was 2 hrs, we calculated 2-hr averages for all of other auxiliary data to fully match
those of metal concentrations. Figure 2.2 illustrates the diurnal variation of temperature, RH, and
WS in central Los Angeles by season. Two-hour averages were calculated for all parameters with
a different time resolution to achieve uniformity across all data before inserting them into the
78

model.
Table 4.1 Inputs of the PMF model along with their time resolution and measurement
method/collection source.

Parameter Measurement method/collection source Time resolution
(h)
Redox-active metals (Fe, Cr, Cu,

and Mn)
Metal monitor 2

Size distribution (13-760 nm)

SMPS

5 min

Size distribution (0.3-10 µm)

OPS

5 min

EC and OC

Sunset Laboratory monitor

1 h

PM2.5

BAM monitor

1 h

Gaseous pollutants

CARB

1 h

Meteorological parameters

CARB

1 h

Traffic data

PeMS

1 h
79

4.2.4. PMF run

We employed the USEPA PMF version 5.0 to apportion the sources of the four redox-active metals
in central Los Angeles. The uncertainties associated with the model outputs were estimated using
the Bootstraps (BS) and Displacement (DISP) methods. Since we included variables of different
nature in the PMF analysis, the approach used to calculate the uncertainty values for each type of
input variables was also different. For EC/OC data, uncertainty values are reported by the
instrument for every single data point, so they were directly used and added to the uncertainty
matrix. For the metal species, the uncertainty was estimated based on the detection limit for each
species, using the following formula (Reff et al., 2007):
Sij = (0.05 x Xij) + DLj (4.1)

4.3. Results and discussion

4.3.1. Overview of the data

Table 2.2 presents the key statistics for the input parameters of the PMF model. The mean
concentrations of the redox-active metals were higher in the cold season than in the warm season.
Similarly, the mean PM2.5 mass concentration was higher in the cold season than in the warm
season. This can be attributed to the impact of meteorological conditions on the atmospheric
stability and mixing height, limiting pollutants dispersion in the atmosphere in the cold season.
80

Table 4.2 Summary statistics for the parameters inserted into the model by season (N=181).

Species Season Mean SE Minimum Maximum
Fe (ng/m
3
) Warm 43.88 2.49 9.00 139.00

Cold 53.27 2.30 18.00 94.00
Cr (ng/m
3
) Warm 2.24 0.11 0.70 7.30

Cold 4.05 0.21 1.10 7.90
Cu (ng/m
3
) Warm 17.23 0.94 1.50 65.00

Cold 19.03 0.69 6.00 36.00
Mn (ng/m
3
) Warm 3.74 0.20 0.50 11.00

Cold 4.68 0.22 1.90 10.10
OC (µg/m
3
) Warm 8.18 0.35 3.76 19.26

Cold 7.37 0.29 4.22 14.46
EC (µg/m
3
) Warm 0.74 0.08 0.00 4.37

Cold 2.34 0.10 0.58 4.37
PM 2.5

(µg/m
3
)
Warm 16.15 0.57 5.25 34.50
Cold 19.39 0.73 10.25 32.88
O 3 (ppb) Warm 30.88 1.40 3.88 64.50

Cold 12.57 1.43 1.20 38.11
NO 2 (ppb) Warm 12.76 0.68 0.15 30.63

Cold 26.18 0.70 15.23 36.31
RH (%) Warm 63.17 1.35 22.25 84.13

Cold 62.05 2.24 19.00 83.94
T (°C) Warm 23.27 0.34 18.06 33.13

Cold 14.12 0.44 6.43 21.96
HDV

(vehicles/h)
Warm 232 13 23 898
Cold 391 9 183 485
LDV

(vehicles/h)
Warm 4880 409 834 5081
Cold 5839 231 1941 7819

81

Figure 4.3 presents the diurnal variations of PM2.5 mass, Fe, Cr, Cu, and Mn concentrations during
the entire campaign. As shown in the figure, in the warm season, the diurnal variations of PM 2.5
mass concentrations exhibited a major peak during morning rush hours, but no peak was observed
during late afternoon/early evening hours. However, in the cold season, the average concentrations
are higher in the late afternoon/early evening traffic rush hours compared to those in the warm
season. The observed increase in PM2.5 concentrations could be primarily because of semi-volatile
species partitioning in the particle phase due to lower temperatures and the depression in mixing
height that prevail in the wintertime. The diurnal variations for Fe and Cu exhibited a major peak
during morning rush hours, and also another minor peak during late afternoon/early evening hours.
However, Mn concentrations only displayed a major peak during morning rush hours in both
seasons. Cr concentrations showed a similar trend as Mn in the warm season, with a peak during
morning rush hours. In the cold phase, however, although Cr concentrations increased sharply in
the morning, they remained relatively unchanged until the late evening hours.
Figure 4.4 presents the diurnal variations of mass fractions of Fe, Cr, Cu, and Mn concentrations.
As can be seen in the figure, Cu concentrations indicated peaks during rush hours, whereas for Mn
and Cr, this peak was much less pronounced. Fe concentrations showed a more stationary phase
with non-significant peak variations during the day. On the other hand, the mass fractions of Cr
remained high during the day, and Mn also had high concentrations at night, in addition to
relatively high concentrations during the day. This indicates that for all species, except Cr, the
seasonal/diurnal difference is driven mostly by the PM mass variations
Figure 4.3 Diurnal variations of: (a) PM2.5 mass, (b) Fe, (c) Cr, (d) Cu, and (e) Mn concentrations
during the entire campaign. Error bars correspond to one standard error (SE).

82

4.3.2. Number of factors

We followed the approach of Masiol etl al. (2017), in which the authors performed model runs
separately for chemical species and size-distribution data. For each of the datasets, PMF was run
with varying input uncertainties and number of factors. The final solutions were chosen based on
several criteria, including profiles of the chemical species and auxiliary variables as well as profiles
of number and volume size distributions for each factor, diurnal variations of factors resolved by
the model, relative and absolute contribution of each factor to PM2.5 concentrations, diurnal
variations of the ambient concentrations of the redox-active metals, and uncertainty estimates. In
order to facilitate factor identification and to better associate the factors resolved in the two
separate runs (i.e., factors resolved from the metal data matrix and those obtained from size
distribution matrix), we also used the results of correlation analysis between factor contributions
from the two runs, the results of which are presented in Table 2.3 The most plausible solution for
the chemical species data was with 4 factors, while the most plausible solution for the size
distribution data was with 5 factors.
Table 4.3 Correlation matrix between the factor contributions from the two separate runs for the
chemical species and size distribution data
Figure 4.5 illustrates the factor profiles resolved by the model for redox-active metals and auxiliary
variables. Given the wide range of units and levels for this input matrix, the factor loadings were
normalized by the mean concentration of each species, and the results are presented as explained
variation of that species in each factor (%). Figure 2.6 indicates the factor profiles resolved by the
model for the volume size distribution data. For the factor profiles of the size distribution data,
mass/volume concentration in each size channel is represented by solid lines (primary Y axis),
while the explained variation is represented by triangles (secondary Y axis). Figure 2.7 depicts the
83

pie chart of relative contributions of each factor to PM2.5 mass, Fe, Cr, Cu, and Mn, concentrations.
Figure 2.8 illustrates absolute contributions of each source factor to the PM2.5 mass concentrations
in the warm and cold seasons. Figure 2.9 presents the diurnal variations for the contribution of
each of the factors to the PM2.5 concentrations in the warm and cold seasons. All of these figures
are going to be discussed in the next section, i.e. factor identification.

Figure 4.5 Factor profiles resolved by the PMF model for redox-active metals and auxiliary
variables: (a) Fresh traffic; (b) Urban background aerosol; (c) Secondary aerosol; and (d) soil/road
dust.
Figure 4.6 Factor profiles resolved by the model for the PM10 volume size distribution data: (a)
Nucleation; (b) Fresh traffic; (c) Urban background aerosol; (d) Secondary aerosol; and (e)
Soil/Road dust. Volume/mass concentrations in each size channel are represented by solid lines
(primary Y axis), while the explained variation is represented by triangles (secondary Y axis).
Figure 4.7 Relative contributions of each factor to: (a) Fe, (b), Cr, (c) Cu, (d), Mn, and (e) PM2.5
mass concentrations.

84

Figure 4.8 Absolute contributions (by cold or warm phases) of each factor to the total PM2.5 mass
concentrations

Figure 4.9 Diurnal variations for the contribution of each of the factors to the total PM2.5
concentrations in the warm and cold seasons: (a) Fresh traffic; (b) Urban background aerosol; (c)
Secondary aerosol; and (d) soil/road dust.

Contribution (μg/m
3
)
85

4.3.3. Identification of factors

Factor 1: Fresh traffic

This factor contributes 27% to total PM2.5 concentrations, and is the major contributor to the
concentrations of Fe and Cu, making up to more than 50% of Fe and Cu concentrations (Figures
2.5 and 2.7), both of which are associated with vehicular emissions in the PM2.5 range (Megido et
al., 2016). In addition, a large portion of the EC concentration (around 90%), a well-known tracer
of vehicular emissions (as observed in many studies, such as Argyropoulos et al. (2016) and
Shirmohammadi et al. (2017)), comes from this factor (Figure 4). Judging by the loadings of
auxiliary variables, this factor also indicates significant associations with gaseous pollutants
emitted by traffic, including NO2, as well as with counts of HDVs and LDVs, all of which suggest
the impact of vehicular emissions.
This factor was highly correlated with the fresh traffic factor obtained from the size distribution
model run, indicating a volume peak at 60-100 nm, i.e., the Aitken mode (Figure 2.6). According
to Figure 2.9, this factor indicated a major peak in the early morning traffic rush hours (around 8-
10 am), whereas in the cold season another peak was also observed in the late afternoon/early
evening traffic rush hours. The diurnal trend of the contribution of this factor is quite similar to the
diurnal variations of Fe and Cu concentrations, indicating a major peak during morning traffic rush
hours (Figures 2.2 and 2.3 ), again corroborating the major impact of traffic-related sources.
Moreover, a minor peak was also observed in the diurnal variation chart of Fe concentration in the
evening rush hours (6-8pm). Relatively lower concentrations of Fe in evening rush hours in
comparison with morning rush hours could be attributed to the higher mixing height and
atmospheric instability in the late afternoon/early evening hours in summer, caused by higher
86

temperatures, wind speeds and incoming solar radiation during the day (Hasheminassab et al.,
2014b). Finally, as can be seen in Figure 2.8, the fresh traffic factor had a significantly larger
contribution to PM2.5 concentrations in the cold season, in line with PM2.5 source apportionment
studies previously performed in central Los Angeles, observing similar characteristics and
contribution from traffic to PM mass (Hasheminassab et al., 2014b; Sowlat et al., 2016a).
Interestingly, the fresh traffic factor was also correlated with the nucleation factor obtained from
the size-distribution model run that exhibited a peak in the PM volume size distribution at < 20
nm, which is typical of particles formed via nucleation (Kumar et al., 2014; Ogulei et al., 2007).
This association could be mainly due to the influence of traffic-related emissions (as precursors)
to form particles via nucleation during morning rush hours, during which lower temperatures
facilitate the partitioning of semi-volatile gasses (coming from the tailpipe) into the particle phase
(Harrison et al., 2011; Ntziachristos et al., 2007). This could also explain the minor loadings of
OC, relative humidity, and temperature in this factor, which are all typical characteristics of the
nucleation factor (Sowlat et al., 2016a).
Factor 2: Urban background aerosol

Factor 2 contributes roughly 20% to the PM2.5 mass concentrations and is associated with the urban
background aerosol obtained from the size distribution model run which indicated a volume mode
in the accumulation mode (200-300 nm) (Figures 2.5 and 2.6). It also has large contributions to
the redox-active metal concentrations, making up to 58% of Cr concentrations (i.e., the major
contributor to Cr concentrations), and around 33% of Fe and 51% of Mn (Figure 2.7). This factor
had high loadings of HDV and LDV, T, O3, which is an indication that this factor is a mixture of
urban aerosol sources that include also aged traffic emissions. Judging by the volume mode
diameter (Figures 2.6 ) in the accumulation mode, the particles in this factor are considerably more
87

aged than freshly emitted particles (for example, compared to those observed for traffic sources),
implying that these are particles likely associated with regional background and aged traffic
emissions. This is also corroborated by the diurnal and seasonal variation of this factor, indicating
a peak in the early afternoon hours and being higher in the warm season (Figures 2.5 and 2.8) during
which atmosphere is more unstable and the mixing height is higher (Figures 2.9 and 2.1), which
facilitates the vertical as well as horizontal dispersion of particles in the metropolitan area of Los
Angeles. This factor could also include a multitude of smaller local sources that could not be
specifically distinguished by the model, but apparently exist in the area; for example, the very high
loading of Cr in this factor may be due to several small plating facilities distributed across the Los
Angeles Basin that are known sources of Cr in the area (Ospital et al., 2008; Propper et al., 2015).

Factor 3: Secondary aerosol

This factor contributes the largest to the total PM2.5 mass concentration (around 20%) (Figures 2.5
and 2.7), and was the major contributor to OC concentrations (around 40%). However, this factor
has a very minor contribution to the concentrations of the four redox-active metals as well as EC.
Additionally, loadings of RH and temperature suggest the impact of temperature and water content
of the atmosphere on the formation of particles in this factor. This factor was also well correlated
with the Secondary Aerosol factor obtained from the model run on size distribution data, with a
peak in the PM volume size distribution in the larger end of the accumulation mode (around 500
nm) (Figure 2.6), which has been attributed to particles of secondary nature by previous studies
((Aiken et al., 2008; Cesari et al., 2016; Kim et al., 2003) .
The diurnal variation of the PM2.5 concentrations also indicated a major peak during early
88

afternoon hours in the warm season (Figure 2.9), which could be a result of peak photochemical
activity in the middle of the day, leading to the formation of SOA and secondary ammonium sulfate
(Hasheminassab et al., 2014b; Saffari et al., 2015). In the cold season, according to Figure 2.9, this
factor also had higher concentrations at nighttime, which could be attributed to the formation of
particle-phase ammonium nitrate, caused by the significant drop in atmospheric temperature and
increased RH at night (Hasheminassab et al., 2014b; Ying, 2011). Additionally, Saffari et al.
(2016) indicated that in Los Angeles, the enhanced liquid water content of ambient PM in the cold
seasons facilitates the formation of aqueous-phase Secondary Organic Aerosol (SOA) during
nighttime. Decesari et al. (2017) also reported the nighttime formation of SOA during fog episodes
in the cold seasons in the Po Valley, Italy. The seasonal trend chart (Figure 2.8) also indicated
larger contribution of this factor to PM2.5 concentrations in the cold season, in line with the above
observations and the prior literature. Based on the above discussion, we refer to this factor as
"secondary aerosol" that include secondary ammonium sulfate and nitrate and as well as SOA
(formed both by daytime photochemical as well as nighttime aqueous phase reactions, as discussed
earlier). This is also in concert with other studies conducted in central Los Angeles
(Hasheminassab et al., 2014b; Saffari et al., 2016; Saffari et al., 2015).
It should be noted that the distinction between the urban background aerosol and secondary aerosol
is that the former reflects a mixture of aged and well-mixed particles that are mostly primary and
emitted from a multitude of point and line (i.e., freeway) sources dispersed in the Los Angeles
basin, whereas the latter consists of particles formed by distinct daytime and nighttime
chemical/photochemical reactions in the atmosphere. In other words, both factors are associated
with aerosols that have a relatively homogeneous spatial distribution, but they are formed via
distinctly different processes, the details of which are presented above.
89

Factor 4: Soil/Road dust

Factor 4 is also a major contributor to PM2.5 mass concentrations, making up to more than 30% of
PM2.5 concentrations (Figure 2.7). This factor had a high correlation with the soil/road dust factor
obtained from the model run on size distribution data, and indicated a major peak in the volume
size distribution at 4-5 µm that has been attributed to soil/road dust particles in previous studies
(Figure 2.6) (Sowlat et al., 2016a). This factor also contributes to about 32% of Mn concentrations,
which is often used as a reliable tracer of particles originating from dust (Świetlik et al., 2015).
Additionally, major contributions of Mn might be attributed to wear debris from vehicular traffic
as stated in previous works (Birmili et al., 2006). This factor also contributes around 39% to Cu
concentrations, which is also associated with traffic-related sources (Gietl et al., 2010; Hassanvand
et al., 2015; Sowlat et al., 2012) as suggested by their substantial loadings in traffic factor, as
discussed earlier. The high loading of Cu can be attributed to particles from the brake wear, which
is used as an indicator of non-tailpipe emissions (Cheung et al., 2011). This also matches the
number and volume mode diameter for this factor observed at 3-4 µm (Figure 2.6), which is the
typical size range for brake wear particles (Harrison et al., 2012).

According to Figure 4.5, this factor is also influenced by temperature and RH. The counterintuitive
high loading of O3 observed in this factor is probably due to the association of temperature and O3
concentrations, both peaking in the middle of the day and having higher values in summertime. A
considerable portion of OC was attributed to the soil/road dust factor, due to the appreciable OC
content of road dust, as reported by previous sources apportionment studies performed in central
Los Angeles (Hasheminassab et al., 2014a; Hasheminassab et al., 2014b) and elsewhere (Kim and
Hopke, 2004; Zheng et al., 2005). The contribution of this factor to PM2.5 mass concentrations was
90

higher in the warm season (Figure 2.8), in line with other studies carried out in the same location
(Sowlat et al., 2016a). Additionally, this factor had higher contributions during nighttime (Figure
2.9), which could most likely be attributed to the impact of vehicles passing at higher speeds during
the night, leading to elevated concentrations of resuspended non-tailpipe particles, as observed
previously in a study at the same location (Cheung et al., 2011; Cheung et al., 2012; Sowlat et al.,
2016a; Sowlat et al., 2016b). Hence, "soil/road dust" is the most likely label for this factor.

4.4. Summary and conclusions

In this study, we implemented a novel semi-continuous metal monitor to measure the
concentrations of Fe, Cr, Cu, and Mn in central Los Angeles at a very fine time resolution (i.e., 2
hrs) and performed a time-resolved source apportionment using PMF. Results indicated that fresh
traffic was the major source of Fe and Cu, whereas soil/road dust contributed considerably to Mn
and Cu concentrations. Cr, on the other hand, mostly came from a multitude of small local sources.
The factor profiles of size distribution data also indicated that the metals emitted by traffic-related
sources (i.e., Fe and partly Cu) were mostly in the ultrafine size range, while those coming from
urban background and soil/road dust (i.e., Cr, Mn, and Cu) were mostly bound to the accumulation-
range and super-micron PM. The diurnal trends in the contribution of the sources indicated a major
impact of traffic on all of the redox-active metals. Seasonal trends also indicated higher
contributions of fresh traffic to metals and PM2.5 concentrations in the cold season, whereas
soil/road dust had higher contributions in the warm season. Urban background aerosol had
comparable contributions in both seasons, while secondary aerosol contributed only negligibly to
the concentrations of these metals. Using this online metal monitor, we were able to obtain the
diurnal variations of the sources that contribute to the metals and PM2.5 concentrations; currently
91

available techniques, which rely on filter-based measurements and subsequent chemical analysis
in laboratory, do not offer this advantage, as the time-resolution of measurements for most of these
techniques is on the order of 24 hrs, which is inadequate to capture the diurnal profiles of the
sources producing these species, which vary on much finer time scales than 24 hrs. This highlights
the major advantage of using such novel measurement techniques, as the processes/reactions in
the atmosphere occur at rather fine time resolutions than a day.

4.5. Acknowledgements

The present work was financially supported by the United States National Institute of Allergy and
Infectious Diseases (award number: 5R01AI065617-15). The authors are grateful to Giulia
Simonetti of the Sapienza University of Roma for her help with the sampling.
92

Chapter 5:

Spatio-temporal trends and source apportionment of black carbon
(BC) in the Los Angeles Basin: fossil fuel and biomass burning

5.1. Introduction

Particulate matter (PM) has significant influences on the variation of chemical composition and
radiative properties of the atmosphere (Forster et al., 2007). For instance, the radiative budget of
the earth could vary according to the absorbing/scattering ratio of specific compounds that
contribute to the light absorption of PM (Jacobson, 2001). Carbonaceous matter (CM), which is
composed of organic carbon (OC) and black carbon (BC), is a major constituent of atmospheric
aerosols and one of the most important components that alter the chemical and radiative properties
of atmospheric compounds (Ramanathan and Carmichael, 2008). BC, as the light absorbing
component of the carbonaceous material, has more impact on unbalancing the net absorption by
the earth’s atmosphere of the radiative solar energy (Chýlek et al., 1995;Ning et al., 2013).
Generally, BC consists of carbonaceous matter resulting from the incomplete combustion of both
fossil and non-fossil fuel (Cooke et al., 1999). In most parts of the world, such as Europe, Asia and
America, BC concentrations are majorly impacted by transportation sources, industrial emissions,
especially power plants, and biomass burning particularly during winter time (Lanz et al., 2007).
However, it has been estimated that decreasing trends of traffic and industrial emissions in the
developed world would make wood-burning contributions more impactful in the near future
(Crippa et al., 2013).
In addition to its impact on climate change, BC has been associated with adverse health effects
93

(Atkinson et al., 2015;WHO, 2012;Grahame and Schlesinger, 2010) and is also used in cancer risk
assessment as a surrogate of diesel particulate exhaust (Baan et al., 2006;Ramanakumar et al.,
2008;Straif et al., 2000). Moreover, fine BC particles can penetrate the alveolar –blood barrier of
the respiratory system of the human body and cause cardiovascular diseases (Koelmans et al.,
2006;Janssen et al., 2011;Straif et al., 2000). The above considerations have been the impetus of
several recent studies globally aiming to apportion the various sources of BC in urban and rural
areas (Herich et al., 2011;Favez et al., 2010;Sandradewi et al., 2008a;Healy et al., 2017;Zotter et
al., 2017;Liu et al., 2014;Ning et al., 2013;Petit et al., 2015).
The Aethalometer model is one of the most commonly used techniques for source apportionment
of total BC concentrations. This model has been used by several studies to quantitatively determine
the contributions of fossil fuel combustion and biomass burning to BC concentrations, with the
assumption of minimal contribution from other sources. For instance, Healy et al. (2017) used the
Aethalometer model for source apportionment of fossil and non-fossil originated BC in several
locations in Ontario, Canada, and reported significant seasonal variations in the fossil fuel
contributions to total BC concentrations in that area. In a long-term exposure assessment study,
Herich et al. (2011) used a 2.5-year dataset to apportion fossil fuel and wood-burning-originated
BC in two different urban and suburban locations in Switzerland. Results from that study indicated
that the total BC concentrations as well as the contributions of wood burning to total BC
concentrations are expectedly higher in the cold season compared to the warm season. In another
study, Favez et al. (2009) used the Aethalometer model along with two commonly used source
apportionment models, i.e. chemical mass balance (CMB) and positive matrix factorization (PMF),
to quantify BC sources in Grenoble, France.
The Los Angeles Basin is one of the largest metropolitan areas in the U.S. with a population of
94

over 10.2 milion residents in 2016 and, historically, it has been considered as one of the most
polluted urban areas in terms of particulate air pollution (Jerrett et al., 2005). Earlier studies in the
Los Angeles metropolitan area indicated that the major sources of BC are traffic-related emissions,
with other stationary sources, such as power plants and local industries also making observable
but rather small contributions (Pratsinis et al., 1984). In addition, wintertime biomass burning as
well as summertime wildfire events also contribute to a certain degree to the total BC
concentrations (Cass et al., 1982). More recently, the California’s Diesel Control Program
established in 2013 as well as the regulations implemented in 2007 to reduce BC emissions by
90% by 2030, have resulted in substantial reductions of direct traffic emissions to the total PM 2.5
mass (Hasheminassab et al., 2014a;Shirmohammadi et al., 2016b). Therefore, understanding the
current BC sources as well as their spatial, seasonal, and diurnal variations would help decision-
makers to implement more effective regulatory programs to further reduce the emissions of BC in
the LA Basin.
The main goal of this study was to determine the spatial and temporal variability of fossil and non-
fossil-originated BC contributions in the Los Angeles Basin using multiwavelength Aethalometers
and the Aethalometer model described above. Additionally, we explored the correlation between
BCff and fossil fuel tracers (including NOx and CO (Miller et al., 2012;Gamnitzer et al.,
2006;Lopez et al., 2013)) as well as between BCbb and biomass burning tracers (including K
+
/K
and levoglucosan (Simoneit, 2002;Echalar et al., 1995;Andreae et al., 1998)) to support and
validate the results of the source apportionment analysis using the Aethalometer model. We also
investigated the effect of regulations imposed for BC emission reductions in the past decade on
the relative contribution of fossil fuel combustion and biomass burning on the total BC
concentrations in the studied areas within the LA Basin.
95

5.2. Methodology

5.2.1. Sampling sites and period

Sampling was conducted by the South Coast Air Quality Management District (SCAQMD) at four
sites within the LA Basin, including central Los Angeles (CELA), Anaheim, Fontana, and
Riverside. Figure 5.1 and Table 5.1 illustrate the locations and characteristics of the four sampling
sites, respectively. CELA site is located in the central part of Los Angles, near major freeways and
urban activities. The Anaheim site is also located to the southeast of Los Angeles, downwind of
the major freeways (e.g., I-710 and southern section of I-110) with high percentages of heavy-duty
vehicles (HDVs). Fontana is a sub-urban sampling site located to the northeast of Los Angeles,
and is exposed to freeway emissions to a lower extent than the other sites. Riverside is a semi-rural
site with lower levels of urban and traffic emissions, but located downwind of major urban sources
in the area, due to the prevailing westerly and southwesterly wind direction in the Los Angeles
Basin.

96

Table 5.1 Geographical coordinates and characteristics of the four sampling sites along with the
parameters measured/collected in each site.

Sampling site
Site Coordinate
Site type

Measurements
Latitude Longitude
CELA

34° 03' 59"N

118° 13' 36"W

Urban
BC
1
, EC
2
, CO
3
,
NO x, K
+
/K,
levoglucosan
4,

temperature,
wind speed
5

Anaheim
33° 49' 09"N

117° 55' 18"W

Urban
BC, EC,
levoglucosan,

temperature,
wind speed
Riverside
33° 59' 58"N

117° 24' 57"W

Rural/agricultural
BC, EC, CO,
NOx, K
+
/K,

temperature,
wind speed
Fontana
34° 06' 0"N

117° 29' 31"W

Sub-urban
BC, EC,
temperature,

wind speed
5

1
BC data were collected using AE33 during the 2012-2013 and 2016-2017 sampling periods.

2
EC data were measured using the EC-OC Sunset Laboratory Carbon Analyzer during the 2012-2013 sampling
period.
3
Gaseous pollutants (CO and NO x) as well as K
+
/K data collected from the CSN network for the 2012-2013
sampling period.
4
Levoglucosan concentrations measured over 2012-2013 were obtained from Shirmohammadi et al.
(2016a).

5
Temperature and wind speed data were collected from California Air Resources Board (CARB) website for the
2012- 2013 and 2016-2017 sampling period.

97

Figure 5.1 Map of the Los Angeles Basin along with the location of sampling sites.

Our dataset included data collected from March 2012 through February 2013, and from March
2016 through February 2017. The four sampling sites were also selected based on the availability
of data during these time periods, with 2012-2013 as the baseline for comparisons, and 2016-2017
as the most recent data available at the time of performing this study. The entire sampling period
for each year was divided into three different periods, including warm, transition, and cold periods.
The warm period spanned from April through September, and the cold period spanned from
November through February. October and March were considered as the transition period, due to
the transient metrological conditions during these two months. AFmbient temperatures were quite
higher in the warm period than in the cold period for all of the sites, although this seasonal trend
was more pronounced for CELA and Anaheim. In addition, lower temperatures are observed in
different seasons in Fontana. Wind speeds were also higher in the warm period, increasing the
N
W E
S
98

atmospheric dilution of primary emissions.

5.2.2. Instrumentation

Light absorption measurements were performed using 7-wavelength Aethalometers (Model AE33,
Magee scientific, USA), an online measurement instrument that determines the optical fraction of
BC based on the aerosol light absorption at different wavelengths ranging from near-ultraviolet
(N-UV) to near-infrared (NIR). Absorption coefficients (babs) for each of the wavelengths, that are
a main component of the Aethalometer model, can be derived using light attenuation through the
filter (ATN). Continuous sampling was done with 2.5 µm cut-off inlets at a flow rate of 5.0±0.1
lpm. The original time resolution for the measurements was 1 min, which allows us to evaluate the
diurnal variations of the results. The limit of detection (LOD) was 0.002 µg m
-3
. BC concentrations
are derived through light absorption processes which correlate optical evaluations to amount BC
particles deposited on the filter. Due to its limitations, this method is not a direct approach of
measuring carbon concentrations in the atmosphere. On the other hand, elemental carbon (EC)
represents thermally refractory portion of the carbon which is measured by oxidation and
quantifying the output gas (Gray and Cass, 1998). While BC is mostly the result of incomplete
combustion of fossil and bio-fossil fuels, major EC sources in urban areas are diesel engine
emissions (Horvath, 1993). Therefore, EC and BC are not necessarily considered as measures of
the same species, but they are generally highly correlated (Jeong et al., 2004; Lack et al., 2014).
In order to apportion BC originating from fossil and non-fossil fuel using the Aethalometer model,
thermal EC concentrations need to be determined. For EC measurements, 24-hour time integrated
samples were collected once every sixth day on quartz filters over 2012-2013. The sampling
schedule of these measurements was synchronized with the schedule used by the national air-
99

monitoring network. The sampler’s inlet was equipped with a 2.5 μm cut-off point impactor which
was installed at a height ranging from 3 m to 15 m above the ground (according to the USEPA’s
guidelines). After collection, filter samples were taken back to the lab and the EC and OC
concentrations were measured with a Desert Research Institute (Reno, Nevada) thermal/optical
carbon analyzer, employing the thermal/optical transmittance measurement protocol implemented
recommended by the National Institute of Occupational Safety and Health (NIOSH 5040). EC data
are crucial in correcting the Aethalometer-measured BC data and calculating mass absorption cross
section (MAC) values in the Aethalometer model. Figure 5.2 shows the BC vs. EC linear
regression line using the combined data obtained at all sites in 2012-2013 sampling campaign. As
can be seen in the figure, the EC and BC data are highly correlated, with a high R
2
value of 0.71
and a slope of 0.57. Using the EC-BC regression line equation obtained for the 2012-2013 data,
EC data were estimated for the 2016-2017 campaign as well, which were then used for calculating
MAC values using the equations mentioned in section 5.2.3.
Figure 5.2 BC vs. EC linear regression line using data from all of the sites in 2012-2013 sampling
campaign.

100

5.2.3. BC source apportionment principles

The Aethalometer source apportionment model using dual spot multi-wavelength Aethalometer
(AE33) quantifies the fossil fuel burning and the biomass burning fractions of BC due to the
difference between the light absorption of particles originating from fossil fuel sources and
biomass combustion in the N-IR (880nm) and N-UV (370nm) wavelengths (Drinovec et al., 2015).
The new dual spot technology allows determination of attenuation at each wavelength using two
spots with different flow rates, allowing the calculation of black carbon concentrations and MAC
values (Drinovec et al., 2015). Briefly, equation (1) shows the relation between incoming and
outgoing light intensity through the filter media and the change in light attenuation (ATN), where
I0 is the incoming light intensity and I is the remaining light intensity:
ATN ≡100 Ln(I
0
/I)

The aerosol attenuation coefficient (bATN) is derived using equation (2), where ∆ATN is the light
intensity attenuation in the time period of ∆t, Q is the sampling air flow rate, set at 5 lpm in this
study, and A is the area of the filter spot of the Aethalometer (1.67 cm
2
for AE33 model).
The correction of the b ATN to calculate the absolute aerosol absorption coefficient babs is described
in equation (3).

In this equation, C and R(ATN) are correction factors for multiple scattering and shadowing
effects, respectively, which are related to the Aethalometer sampling properties. C is a constant
with a value of 2.14, which is determined empirically by comparing different soot particles
originating from diesel and diesel mixed gaseous species (Weingartner et al., 2003). Equation (4)
101

demonstrates that R(ATN) is a linear function of Ln(ATN) and varies based on optical
specifications of the aerosol as the weight of particles on the collection filter changes. In this
equation, f is a parameter that accounts for the instrumental error (i.e., the shadowing effect of
particles on the filter) with values of 1.14 and 1.09 at 370 nm and 880 nm, respectively. All of
these parameters are described in detail elsewhere (Sandradewi et al., 2008a).

MAC value at a specific wavelength λ (MACλ) is derived using equation (5), in which babs,λ
represents the total absolute aerosol absorption in wavelength of λ, and EC represents the estimated
EC thermal concentrations via BC-EC regression line (Knox et al., 2009):
Then, source apportionment of total BC measured by the Aethalometer can be derived using the
following equations (Healy et al. 2016):
Where, babs, λ represents the absolute aerosol absorption at wavelength λ, babs,ff,370 and babs,ff,880
represent the absolute aerosol absorption originating from fossil fuel sources at 370 nm and 880
nm, respectively. Similarly, babs,bb,370 and babs,bb,880 represent aerosol absorption from biomass
burning sources at 370 nm and 880 nm, respectively. BCff denotes the concentration of total black
carbon apportioned to fossil fuel combustion, and BCbb denotes the concentration of total black
carbon apportioned to biomass burning. MAC880 refers to the mass absorption cross-section of BC
at 880 nm. Lastly, absorption Ångstrom exponent (AAE) values are tagged for aerosols from each
source; αff and αbb for fossil fuel and biomass, respectively. Based on the available literature, in
this study we used αff and αbb values of 0.90 and 2.09, respectively (Zotter et al., 2017;Kirchstetter
et al., 2004;Schnaiter et al., 2005).
102

5.2.4. Tracer analysis

As mentioned in previous sections, in order to further corroborate the BC ff and BCbb source
apportionment results, we collected and used data for tracers of vehicular emissions in the Los
Angeles basin (i.e., CO and NOx) and biomass burning (and K
+
/K ratio and levoglucosan) to
evaluate how their temporal profiles compare with those of BC ff and BCbb. Although biomass
burning generally contributes to CO concentrations, in Los Angeles, due to predominant traffic
emissions, CO concentrations are mostly driven by fossil fuel combustion (Shirmohammadi et al.,
2017; Sowlat et al., 2016: Mousavi et al., 2018). This was supported by the weak correlations
between CO and BCbb for both seasons (R
2
values ranging from 0 to 0.11). For CELA and
Riverside, hourly data for NOx and CO along with weekly data for potassium ion (K
+
) and mineral
potassium (K) were collected from the Chemical Speciation Network (CSN) of the Environmental
Protection Agency (EPA) for the 2012-2013 and 2016-2017 sampling campaigns. It is noteworthy
that the concentrations of K
+
and K were measured using ion chromatography (IC) and energy-
dispersive X-ray fluorescence (EDXRF), respectively. Levoglucosan weekly concentrations were
obtained from the study of Shirmohammadi et al. (2016a) in which levoglucosan was measured as
a tracer for biomass burning in central LA and Anaheim over 2012-2013 using gas
chromatography-mass spectrometry (GC-MS). The tracer analysis included comparison of the
temporal trends (i.e., monthly and diurnal variations) of BCff and BCbb with those of the tracers, as
well as the regression analysis between BCff and BCbb and tracer concentrations in different seasons
and sites.

5.3. Results and discussion

5.3.1. Data overview and average BC concentrations

103

Seasonal variations of BC concentrations for all of the four sampling sites in 2012-2013 and 2016-
2017 are illustrated in Figure 5.3. As shown in the figure, BC levels decreased from 2012-2013 to
2016-2017 in almost all of the sites by 10-20%. For example, the cold-period average BC
concentration at CELA site decreased from a value of 1.17 ± 0.20 µg m
-3
in 2012-2013 to 0.84±
0.10 µg m
-3
in 2016-2017. This reduction could be attributed to the impact of stringent regulations
posed by the State of California on vehicular emissions, especially emissions from diesel trucks,
which are considered as one of the main sources of BC in the area (California Code of Regulations,
2008; San Pedro Bay Ports, 2017). Earlier studies performed in the LA Basin show results similar
to those presented in this study, indicating reductions in vehicular emissions and contributions to
total BC concentrations (Shirmohammadi et al., 2016b;Krasowsky et al., 2016). It can also be seen
in Figure 5.3 that in all of the sampling sites (except for Fontana), the average BC concentrations
are almost twice as high in the cold period compared to the warm period. For instance, in 2012-
2013, the average BC concentrations in the CELA and Anaheim sampling sites were 1.17 ± 0.20
µg m
-3
and 2.02 ± 0.20 µg m
-3
in the cold period, compared to warm period values of 0.60 ± 0.09 µg
m
-3
and 0.73 ± 0.20 µg m
-3
, respectively. Since the magnitude of traffic count and emissions in
summer and winter are almost the same in Los Angeles (with winter levels being only about 10%,
based on California Department of Transportation’s Performance Measurement System), the
roughly doubling of BC concentrations in the cold period could be attributed to the more stable
atmospheres in the winter caused by lower mixing heights (i.e., limited vertical dispersion) (Ware
et al., 2016) and lower wind speeds (i.e., limited horizontal dispersion). Similar impacts of
meteorology on ambient levels of PM have been observed in previous studies performed in the LA
Basin (Sowlat et al., 2016;Daher et al., 2013;Shirmohammadi et al., 2016b). In addition, in
Fontana, the meteorological conditions were quite similar in the cold and warm periods, which
104

could explain the consistent BC concentrations in different seasons. The average BC
concentrations in the transition period were higher than those in the warm period, but quite lower
than those observed in the cold period, which is also consistent with meteorological conditions
observed for each period. Figure 5.3 also illustrates that the average BC levels in the warm period
are highest in Fontana, followed by Riverside, Anaheim, and CELA. This suggests a clear pattern
of increasing BC concentrations from ocean to inland, which is primarily due to the accumulation
of PM during transport from source to receptor areas of the basin (Hasheminassab et al., 2014b;Fine
et al., 2004). Figure 5.3 also indicates that in the cold period, BC concentrations in Anaheim were
considerably higher than the other sites. Stable meteorological conditions in the winter in addition
to local BC sources nearby (e.g., Ports of Los Angeles and Long Beach as well as Long Beach
power plant station) are major sources of BC in this area. Further, Anaheim is located downwind
of the three major freeways (i.e., I-110, I-710, and I-405 freeways), with high fraction and number
of heavy duty vehicles (HDV), and the prevailing westerly and southwesterly winds blow the
majority of these emissions to this site, as observed in previous studies (Hasheminassab et al.,
2014b;Saffari et al., 2013;Shirmohammadi et al., 2016a).
Figure 5.3 Seasonal variations of total BC concentrations at the 4 sampling sites in 2012-2013 and
2016-2017. Error bars represent one standard deviation (SD).

105

5.3.2. MAC values and seasonal trends of sources

Figure 5.4 depicts the seasonal variation of MAC values for all of the sites in 2012-2013 and 2016-
2017. It should be noted that MAC values directly depend on BC concentrations and could vary
based on the location and the model of the Aethalometer used (Hansen et al. 1984). Derived MAC
values in this study were on average 16.5-20 m
2
g
-1
for MAC 370 nm and 5.3-8.1 m
2
g
-1
for MAC
880 nm. Using the same source apportionment model to apportion BC fossil fuel contribution in
Ontario, Healy et al. (2017) reported average values of 11.45 m
2
g
-1
and 10.70 m
2
g
-1
for MAC 880
nm in two different near-road sites after making corrections using thermally measured EC data.
Our results are also consistent with those observed in other studies conducted in multiple locations
around the world (Healy et al., 2017;Herich et al., 2011;Sandradewi et al., 2008a).
As shown in Figure 5.4, consistent MAC 370 (17.6 ± 2.3 m
2
g
-1
) and MAC 880 (6.7 ± 1.4 m
2
g
-1
)
values were observed among different sites and seasons. However, we observed a decrease in
average MAC values for all of the sites from warm period to cold period (18.8 to 16.3 m
2
g
-1
for
MAC 370 and 7.7 to 6.1 m
2
g
-1
for MAC 880), which could be attributed to the higher BC values
observed in the cold period as discussed in section 5.3.1. The results of the t-test indicated that all
of MAC values were statistically significantly lower in the cold period compared to the warm
period, with P <0.05 in all sites except of CELA, where they approached statistical significance
(P <0.1).
As expected, MAC values did not significantly change over the years for any of the sites. As
indicated above, the inter-study variations of MAC values are more pronounced than the seasonal
variations within the same study, which implies that MAC value is more dependent on the
characteristics of the area than seasonal variations (Zotter et al., 2017). To further evaluate the
spatial and seasonal variations in the calculated MAC values, we performed intraclass correlation
106

analysis on the MAC values calculated for different seasons and years at each of the sites. The
intraclass correlation coefficients (ICC) calculated for the seasonal and inter-year variability of
MAC values at each site indicated ICC values as high as 0.70-0.87, suggesting small temporal
variability in the MAC 370 and MAC 880 values in the studied sites.
Additionally, as a result of particle aging, optical properties of black carbon may slightly change
from the source to the receptor areas, causing a small increase in the MAC value, a process referred
to as mass absorption coefficient enhancement (MACE) (Krasowsky et al., 2016;Cappa et al.,
2012). These studies, which were conducted in the LA Basin, have reported MACE values ranging
from 1.03 to 1.06 for MAC 370 from Central Los Angeles to Riverside, quite consistent with
values calculated in the current study. In order to further explore the spatial variation of MAC
values across the sites, we calculated ICC values for the annually averaged MAC values of each
wavelength, i.e. MAC 370 and MAC 880. The calculated ICC values of 0.68-0.70 indicated that
the spatial variations of the calculated MAC values are also small across the studied site.

Figure 5.4 Seasonal variations of Mass Absorption Cross-section (MAC) in the 2012-2013 and
2016-2017 campaigns at different sites: a) MAC 370 nm and b) MAC 880 nm. Error bars represent
one standard deviation (SD).
a)
107

b)

Figures 5.5 and 5.6 illustrate the seasonal variations of absolute concentrations of BCff and BCbb
and their contributions to total BC concentrations over the entire sampling period for all of the
sites. As shown in the figures, in 2012-2013, average BCff concentrations are higher in the cold
period (1.10 µg m
-3
) for all of the sites in comparison with the warm period (0.65 µg m
-3
), while
the contribution of BCff to total BC concentrations decreases from warm period (95±5%) to cold
period (86±6%) in all of the sites, due to the widespread biomass burning activities that frequently
occur in the cold period (Hasheminassab et al., 2013;Saffari et al., 2013). In addition, CELA
exhibited the highest contribution of BCff to the total BC concentrations (97±2% in warm period,
89±2% in transition period, and 87±3% in the cold period) in both campaigns. This may be due to
the highest number of traffic sources in central LA. On the other hand, Riverside, which is
considered as the most rural/semi-rural of all the sites, exhibited the lowest contribution of BCff to
the total BC concentrations; the corresponding BC ff contributions were 91±2% in warm period,
84±2% in transition period, and 73±3% in cold period, respectively. Anaheim, located downwind
of three major freeways, exhibited significant contributions of BCff to total BC concentrations in
warm period (97±1% and 95±1% in 2012-2013 and 2016-2017, respectively); in the cold period,
however, quite smaller contributions of BCff and higher contributions of BCbb were observed, most
108

likely due to the increased biomass burning activities in the nearby residential region. Furthermore,
Fontana indicated a similar seasonal trend of BCff% variation (93±2% in the warm period, 85±2%
in the transition period, and 75±3% in the cold period) to Riverside in both 2012-2013 and 2016-
2017 campaigns. Additionally, the relative contributions of BCff to total BC concentrations have
decreased over the years, which could be attributed to the state regulation implemented in 2007 to
reduce BC and PM emissions form vehicular sources. For example, in the warm period the average
of BCff% for all of the sites was 92±6% in 2016-2017 in comparison with 86±5% in 2012-2013.
Results of the independent sample t test indicated that this decrease was statistically significant (P
<0.001).
Figure 5.5 Seasonal variation of BCff and BCbb concentrations at the 4 sampling sites in the 2012-
2013 and 2016-2017 campaigns: a) BCff, and b) BCbb. Error bars represent one standard
deviation (SD).

a)

109

b)

Figure 5.6 Seasonal variation in the BC fossil fuel (BCff) and BC biomass burning (BCbb)
contributions at the 4 sampling sites in the 2012-2013 and 2016-2017 campaigns: a) BCff % b)
BCbb %. Error bars represent one standard deviation (SD).

a)

110

b)

111

5.3.3. Diurnal variation of BCff and BCbb

The average diurnal variations of fossil fuel and biomass burning contributions to BC
concentrations for 2012-2013 and 2016-2017 campaigns are illustrated for each site and different
periods in Figure 5.7. Additionally, diurnal trends based on the absolute concentrations are
presented in Figure 5.8 and Figure 5.9 for BCff and BCbb, respectively. In the warm period, we
observed a minor peak in the %BCff in the early morning (6-9 AM) traffic rush hours with the
maximum amount of emissions from vehicles for the sites that are close to the freeways, including
CELA and Anaheim (Shirmohammadi et al., 2016b;Ruprecht et al., 2011). However, such a peak
was not observed in Riverside, because, as mentioned earlier, Riverside County has semi-rural
characteristics (Daher et al. 2013). In addition, the %BCbb did not show any significant peaks
during the day in the warm period, and its contributions are negligible (less than 10% on average
for all of the sites).
In the cold period, on the other hand, clear major peaks were observed for %BC ff in both early
morning and early evening rush hours (4-7 PM) in CELA; this is quite consistent with the CELA
site being majorly impacted by fresh nearby traffic BC emissions (Shirmohammadi et al., 2018).
In addition, the diurnal variations of BCff contributions in Anaheim displayed an increase during
morning traffic rush hours and remained relatively high until late afternoon hours. We also
observed minimum BCff contributions during nighttime in almost all of the sites, when urban traffic
activities are minimal (Gariazzo et al., 2007). In Fontana, BCff contributions were relatively
constant during the day in the cold period, most likely because this site is far from traffic zones in
the Los Angeles Basin. In case of BCbb (Figure 5.9), it can be seen from the figure that
concentrations were higher during the nighttime for the urban sites during the cold period. In semi-
rural sites, however, BCbb concentrations remained high even during the day, as biomass burning
112

emissions are more pronounced in such areas and persist during the day (Yan et al., 2015).

Figure 5.7 Diurnal variations of BCff% and BCbb% at all sites during the entire study period in: a)
warm period, b) cold period. Error bars represent one standard deviation (SD).

a)

113

b)

114

Figure 5.8 Diurnal variations of BCff (ng.m
-3
) at all sites during the entire study period in: a)
warm period, b) cold period. Error bars represent one standard deviation (SD).

a)

b)

115

Figure 5.9 Diurnal variations of BCbb (ng.m
-3
) at all sites during the entire study period in: a)
warm period, b) cold period. Error bars represent one standard deviation (SD).

a)

b)

116

5.3.4. BC vs fossil fuel/biomass burning tracers

Monthly variations of BCff, NOx, and CO for both CELA and Riverside sites for the 2012-2013
and 2016-2017 campaigns are indicated in Figure 5.10 and Figure 5.11, respectively. As can be
seen in the figure, in CELA and Riverside the concentrations of all of the tracer species are higher
in the cold period compared to the warm period. The higher concentrations of NOx and CO in
CELA compared to Riverside are consistent with the fact that CELA is more impacted by vehicular
emissions (Saffari et al. 2013). The BCff monthly variations in both sites show similar trends with
those of NOx and CO concentrations. Figures 5.12 and 5.13 illustrates the diurnal variations of
BCff, NOx, and CO in the cold and warm periods in CELA and Riverside for the 2012-2013 and
2016-2017 sampling campaigns, respectively. As shown in the figure, major peaks were observed
for both NOx and CO concentrations during the morning rush hours (6-8 am) in both periods and
both sites. Additionally, the BCff diurnal variation trend was also consistent with those of NOx and
CO in the both sites, which demonstrates that all of these species come from the same source, i.e.
traffic (Lopez et al. 2013). The results of the regression analysis for 2012-2013 (Figure 5.14) and
2016-2017 (Figure 5.15) also exhibited positive associations between BC ff and NOx and CO in
both seasons and both sites.
117

Figure 5.10 Monthly variations of BCff, NOx, and CO concentrations in: a) CELA, and b)
Riverside at 2012-2013. Error bars represent one standard deviation (SD).

a)

b)

118

Figure 5.11 Monthly variations of BCff, NOx, and CO concentrations during the at 2016-2017
campaign in: a) CELA, and b) Riverside. Error bars represent one standard deviation (SD).
a)

b)

119

Figure 5.12 Diurnal variations of a) BCff, b) NOx, and c) CO in the cold and warm phases in
CELA and Riverside sites in the 2012-2103 campaign. Error bars represent one standard
deviation (SD).
a)

b)

c)

120

Figure 5.13 Diurnal variations of a) BCff, b) NOx, and c) CO in the cold and warm phases in
CELA and Riverside sites in the 2016-2107 campaign. Error bars represent one standard
deviation (SD).

Figure 5.14 Linear regression between BCff and fossil fuel tracers (NOx and CO) in the cold and
warm periods in: a) CELA and b) Riverside at 2012-2013 sampling campaign. The regression
lines for NOx are plotted based on seasonally-averaged data points for diurnal variations.
a)

121

b)

122

Figure 5.15 Linear regression between BCff and fossil fuel tracers (NOx and CO) in the cold and
warm periods of the 2016-2017 campaign in: a) CELA, and b) Riverside.

Regression lines between BCbb concentration and K
+
/K ratio (as a tracer of biomass burning) for
CELA and Riverside over the 2012-2013 and 2016-2017 campaigns are illustrated in Figure 5.16
and Figure 5.17, respectively. According to the figure, there was a high correlation between BCbb
and K
+
/K values in both seasons and both sites, corroborating that they most likely come from the
same source, i.e. biomass burning. As can be seen in the figures, R
2
values for CELA are 0.83 and
0.60 in the cold and warm periods of the 2012-2013 campaign, and 0.76 and 0.77 in the cold and
warm periods of the 2016-2017 campaign, respectively; the corresponding values for Riverside
are 0.80 and 0.82 in the cold and warm periods of the 2012-2013 campaign, and 0.66 and 0.73 in
the cold and warm periods of the 2016-2017 campaign, respectively. Similarly, we also observed
a strong positive association between BCbb and levoglucosan (Figure 5.18) in both CELA (R
2

=0.71) and Anaheim (R
2
=0.95) during the 2012-2013 campaign, again corroborating our BC
123

source apportionment results.
Figure 5.16 Regression lines of BCbb and K
+
/K (weekly-averaged data) in the cold and warm
phases for: a) CELA 2012-2013, b) Riverside at 2012-2013.

a)

b)
124

Figure 5.17 Regression lines of BCbb and K
+
/K (weekly-averaged data) in the cold and warm
periods of the 2016-2017 sampling campaign in: a) CELA, and b) Riverside.

a)

125

b)

126

Figure 5.18 Regression lines of BCbb and levoglucosan (weekly samples) sites in the cold periods
of the 2012-2013 campaign: a) CELA and b) Anaheim.

a)

b)

127

5.3.5. Sensitivity analysis of the Aethalometer model for biomass burning fraction of BC
Although the tracer analysis performed on the model-derived BCbb and BCff contributions
corroborates the validity of fossil fuel and biomass burning contributions in the region, we have
performed a sensitivity analysis on the αff and αbb variations to investigate the impact of brown
carbon on the BCbb absorption levels, following the approach suggested by Fuller et al. (2014). In
the first step, we evaluated the impact of variations in αff and αbb values on Babs370 absorption levels,
the results of which are demonstrated in Figure 5.19(a). According to the figure, B abs370 values
ranged from 70 M m
-1
at αff =0.8 (roughly +10% higher than 65 M m
-1
at αff =0.9) to 60 M m
-1
at
αff =1 (-8% lower than the 65 M m
-1
value at αff =0.9) when αbb was set at 2.09. Similarly, Babs370
changed by approximately 5% when αbb varied from 1.8-2.2 at αff =0.9. These results indicate that
Babs370 is almost twice more sensitive to the changes in αff than αbb.
The strong correlation between levoglucosan concentrations and the model-derived BCbb
concentrations in CELA (R
2
=0.71) and Anaheim (R
2
=0.95) are shown in Figures 5.18(a-b). Using
original values of αff =0.9 and αbb =2.09 led to very small intercepts in the regression lines between
levoglucosan concentrations and the model-derived BCbb concentrations (CELA= 4.30 ng m
-3
and
Anaheim=1.48 ng m
-3
), which may suggest a slight influence of brown carbon on BC bb
contributions. To achieve a zero intercept in the regression line, α ff and αbb had to be adjusted to
result in an optimal BCbb contributions at which brown carbon effect is minimal. Since we used
the suggested α values by Zotter et al. (2014) (αff =0.9 and αbb =2.09) with the lowest associated
uncertainty in the Aethalometer model output, we investigated the sensitivity of B abs370 and
levoglucosan regression intercepts by changing αff and αbb around the abovementioned values.
Figure 5.19(b) demonstrates the intercept of levoglucosan and Babs370 as a function of αff with αbb
varying from 1.8-2.2 at the Anaheim site with the highest BCbb contributions. As shown in Figure
128

5.19(b), the levoglucosan intercept was zero at α ff = 0.93, which is a very small deviation (2%)
from our assumed value of 0.9. Based on Figure 5.19(a), a 2% increase in αff from 0.9 (used value
in the current work) at αbb =2.09 would lead to a 7% decrease in Babs370, which is within the range
of the model uncertainty value of 18% used in this work to account for AEE variations. This further
corroborates the validity of the αff and αbb values used in the Aethalometer model to minimize the
brown carbon effect on the BCbb absorption enhancement.

Figure 5.19 (a) Babs370 and (b) Levoglucosan reduced major axis (RMA) intercept as a function of
αff when αbb varies between 1.8-2.2.
(a)

(b)
129

5.4. Summary and conclusions

In this study, we evaluated the spatial and temporal variations of BC concentrations and the sources
(fossil fuel and biomass burning) that contribute to these levels at four distinct sites in the Los
Angeles Basin over the years 2012-2013 and 2016-2017. Our results indicated higher BC
concentrations in the cold period compared to transition and warm periods, due to more stable
atmospheres and lower mixing heights (Ware et al., 2016). The highest BC levels were also
observed in Anaheim, which is in close proximity to/downwind of numerous sources, including
major freeways, ports of Los Angeles and Long Beach, and Long Beach power plant station.
Results from the present study indicated a higher overall contribution of fossil fuel burning (which
mostly comes from traffic-related sources in the Los Angeles Basin) than biomass burning to the
total BC concentrations in the studied areas. The analysis of temporal trends also revealed elevated
BCff contributions in the warm season, while the opposite was observed for BCbb. BCbb
contributions as high as 25-30% were observed during the cold period at all sites, suggesting the
noticeable contribution of this increasingly important source to total BC concentrations in the LA
Basin, especially in the cold season. We also observed a clear decrease in BC concentration as
well as fossil fuel contributions from 2012-2013 to 2016-2017, most likely due to the
implementation of stringent regulations in California to reduce transportation-related PM and BC
emissions. The decrease in traffic-related BC emissions has resulted in an overall increase in the
contributions of other sources, such as biomass burning, to total BC concentrations in the LA basin.
Therefore, more attention should now be paid to these sources when designing and implementing
effective control measures in future aiming to reduce BC concentrations in the area.

130

5.5. Acknowledgments
Authors of this paper would like to acknowledge the USC Viterbi's PhD fellowship award and
South Coast Air Quality Management District for providing us with the collected dataset.
131

Chapter 6:

Source apportionment of black carbon (BC)

from fossil fuel and biomass burning in metropolitan Milan, Italy

6.1. Introduction

Airborne particulate matter (PM) in urban and suburban areas throughout the world is largely
composed of residual chemical components of incomplete fossil fuel combustion, industrial
emissions, and products of biomass burning and natural wildfires (Cooke et al., 1999). PM
resulting from the combustion of both fossil fuel and non-fossil fuel (e.g., biomass burning)
sources consists largely of carbonaceous material (CM) at its core that consists of black carbon
(BC) and organic carbon (OC). Several additional compounds adsorb onto the CM core surface,
including metals, polycyclic aromatic hydrocarbons (PAHs), semi-volatile organic aerosols, and
primary and secondary ions. BC is a PM species of great importance, due mainly to its impacts on
human health, as well as on the optical properties and radiative forcing of the atmosphere, which
can contribute to global warming (Ning et al., 2013;Atkinson et al., 2015;WHO et al.,
2012;Grahame and Schlesinger, 2010;Chýlek et al., 1995).
Several studies have reported severe adverse health impacts resulting from exposure to BC,
including respiratory and cardiovascular mortality and morbidity due to asthma, ischemic heart
disease (IHD), chronic obstructive pulmonary disease (COPD) (Atkinson et al., 2015;Grahame
and Schlesinger, 2010; Laeremans et al., 2018a;Laeremans et al., 2018b; Magalhaesa et al., 2018).
Additionally, cancer risk assessment studies often use BC as a surrogate for diesel exhaust
particulate (DEP) which is considered carcinogenic (Baan et al., 2006;Ramanakumar et al.,
2008;Lovett et al., 2018). BC has also been found to be a better indicator than PM2.5 or PM10 mass
132

when estimating human health impacts (Atkinson et al., 2015). Studies have also indicated that
exposure to products of biomass combustion, as a major source of BC, can be associated with
excessive risk of developing lung cancer (Stabile et al., 2018; Pacitto et al., 2018; Sarigiannis et
al., 2015).
In addition to its health effects, BC is also a known contributor to global warming. Although BC
is a short-lived pollutant, the absorption of incoming solar radiation by aerosols containing black
carbon significantly affects the atmosphere’s radiative properties, resulting in a substantial positive
forcing effect (Forster et al., 2007; Solomon et al., 2007; Stocker et al., 2013). The magnitude of
this forcing depends on the specific chemical composition of carbonaceous aerosols and their
sources of origin (e.g., fossil fuel combustion or biomass burning), as well as the subsequent
interactions between surface-bound particulate components and the black carbon at its core, which
determine the BC mixing state (Mohr et al., 2013;Saleh et al., 2013). The increase in atmospheric
radiative forcing due to the absorption and re-radiation of solar energy by BC in its various mixing
states has a significant impact on global warming (Jacobson, 2001). Several studies worldwide
have investigated BC levels in urban and semi-rural areas of the world, with a focus on
apportioning BC concentration to its various sources, including fossil fuel combustion and biomass
burning (Sandradewi et al., 2008; Herich et al., 2011; Zotter et al., 2017; Healy et al., 2017).
The Aethalometer model is a commonly used technique to apportion sources of BC in different
areas around the globe. The Aethalometer model relies on the light-attenuation (ATN) data
provided by an Aethalometer at two wavelengths (i.e., 370 nm and 880 nm) to calculate absorption
coefficients (babs) and mass absorption cross-section (MAC) values, based on which fractions of
BC can be apportioned to either fossil fuel combustion (BCff) or biomass burning (BCbb), with the
assumption that other sources have minimal contributions to BC levels (Sandradewi et al., 2008).
133

While the optical properties of BC can also be affected by the presence of any “brown carbon”
species, which are primarily the organic products of biomass burning that strongly absorb light in
the ultraviolet range (Massabò et al., 2015;Laskin et al., 2015;Andreae and Gelencsér, 2006),
radiocarbon analysis of the
14
C present in collected PM samples (e.g. as described in Zotter et al.,
2017) allows us to distinguish between the two common sources of carbonaceous species (i.e.,
biomass burning and fossil fuel combustion) based on the relative age of carbon in the samples,
and thus account for the effect of brown carbon in the Aethalometer model. Additionally, since
the Aethalometer provides continuous data with a very fine time-resolution, it is possible to explore
the diurnal variations of BCff and BCbb contributions to total BC concentrations as its composition
changes as a result of atmospheric aging and processing.
The Aethalometer model has been used in various studies to determine concentrations and sources
of BC in several areas of Europe and North America. Sandradewi et al. (2008) and Herich et al.
(2011) both compared traffic (fossil fuel) and wood-burning sources of BC in the Alpine Valley
region of Switzerland. Healy et al. (2017) examined BC contributions of traffic and biomass
burning at urban and background sites during cold and warm seasons in Ontario, Canada. Titos et
al. (2017) looked at the contributions of fossil fuel versus biomass burning in urban as compared
to suburban regions of Granada, Spain. Most recently, Mousavi et al. (2018a) evaluated the
contributions of BCff and BCbb to BC concentrations across Los Angeles, California. One general
finding across all of these studies is that the biomass burning has larger contributions to ambient
BC concentrations in rural and suburban areas than in urban centers. In addition, all of the
abovementioned studies have reported that biomass burning makes a larger contribution to BC
emissions during the cold season, as compared to warmer periods, due to the increased wood
burning for residential heating.
134

The Po Valley in the Lombardy region of northern Italy is an area where several PM studies have
been conducted, due to its unique meteorological and geographical properties. The Po Valley
extends over 650 km, from the Western Alps to the Adriatic Sea, and is characterized by industrial,
traffic, and biomass burning emissions throughout its urban, suburban, and rural areas. Within the
valley, concentrations of PM and other pollutants are enhanced by fog and low-level clouds, which
are transient phenomena in the atmosphere during the wintertime fog season (November to
March). This enclosed valley topography of this region is also subject to the formation of inversion
layers, resulting in trapped, stagnant air causing pollutant concentrations to increase (Decesari et
al., 2017;Cermak et al., 2009;Fuzzi et al., 1992;Decesari et al., 2001).
Milan, the capital of Lombardy, is the second largest city in Italy with a population of 2.4 million.
As a high-density urban center within the Po Valley, fossil fuel derived PM emissions from traffic
and other sources dominate in the atmosphere above central Milan, thus making it an ideal location
to study sources of BC in the region (Decesari et al., 2017;Sandrini et al., 2014). Recent studies
have reported high concentrations of BC in Milan, ranging from 2.1-5.5 µg/m
3
across various areas
of the city, which are attributed mostly to traffic emissions along with some contributions from
wood smoke in the wintertime (Daher et al., 2012;Invernizzi et al., 2011). Bareggio, a suburb
located 14 km (9 miles) to the west of central Milan, is much less densely populated, with only
17,000 residents. In this suburban location, traffic emissions are significantly reduced in
comparison to the urban core of Milan, and a relatively larger fraction of the particulate matter
comes from biomass burning, particularly during the colder, winter months when wood-burning
for household heating is common.
In this study, we aimed at determining the temporal variability of ambient BC concentrations and
apportioning its sources to biomass burning and fossil fuel combustion in the city of Milan
135

compared to surrounding suburban areas within the Po Valley. Data collected from single- and
multi-wavelength Aethalometers were used in the Aethalometer model to apportion BC to these
two sources. Using
14
C radiocarbon analysis of PM samples, we were also able to account for any
influence of brown carbon on the optical properties of BC.
136

6.2. Methodology

6.2.1. Sampling sites and collection periods

BC concentrations were measured using two Aethalometers at an urban site located in metropolitan
Milan, as well as at a suburban site located approximately 14 km to the west, in the municipality
of Bareggio. Figure 1(a-b) shows the locations of the two sampling sites. BC concentrations were
monitored during three time-periods of a year-long sampling campaign: from July to August 2017
(summer/warm phase), from September to October 2017 (intermediate phase), and from January
to March 2018 (winter/cold phase). PM2.5 filter samples were also collected during these phases
for EC/OC and
14
C analyses. Meteorological data, including average daily temperatures, wind
speed, and wind direction during the three sampling periods were obtained from an international
meteorological website (http://www.wunderground.com) for the sites closest to our sampling
locations (i.e., Roveda di Sedriani (MI) site in Bareggio and Milano Francesco Sforza site in
Milan). The seasonally averaged wind speed and temperature for each site are presented in Table
1. At both sites, we experienced lower ambient temperatures during the cold period than during
the warm period, with this trend being more pronounced at the suburban site of Bareggio (25.2 ±
2.8 °C in summer and 5.9 ± 1.1 °C in winter). In addition, as can be seen in the table, during the
warm period wind speeds were greater, resulting in increased dispersion of atmospheric aerosols.
Figure S1 shows the wind rose plots for Bareggio and Milan during the three sampling periods.
The dominant wind direction during all of the three sampling periods is south and southwest for
Milan and Bareggio, respectively.
137

Figure 6.1 a) Map of the metropolitan areas within the Po valley (star sign indicates Milan); and

b) locations of the two sampling sites within the metropolitan Milan area.

138

6.2.2. Instrumentation

BC concentrations were measured using Aethalometers at each sampling location (Hansen et al.,
1984; Sandradewi et al., 2008). At the central Milan site, we measured BC concentrations using a
7-wavelength Aethalometer (Model AE31, Magee Scientific, USA), with a measurement time-
resolution of 2 minutes and flow rate of 5 lpm. At the suburban Bareggio site, we used 2 single-
wavelength Aethalometers (Model AE51, AethLabs, USA) in parallel to measure BC
concentrations at 370 nm and 880 nm, with a time resolution of 5 minutes and a flow rate of 0.05
lpm. The Limit of Detection (LOD) for both devices is 0.002 µg/m
3
. A recent study by Cheng and
Lin (2013) indicated a high correlation between BC concentrations measured by AE51 and AE31
instruments with an R
2
value of 0.99, therefore we expect negligible variance in the data due to
the different Aethalometers used in this study at the two sites. In fact, prior to initiating the
sampling campaign, our AE31 and AE51 instruments were run in parallel, and values measured
by the instruments were found to be highly correlated, with Pearson correlation coefficients
between AE31 and AE51 measurements of 0.89 and 0.94 at wavelengths of 880 nm and 370 nm,
respectively.
In addition to collecting BC concentrations data derived from Aethalometer measurements, time-
integrated PM2.5 samples were also collected at each site for EC/OC and
14
C analyses using
Personal Cascade Impactor Samplers (PCISs, SKC Inc., Eighty-Four, PA, USA; Misra, et al.,
2002; Singh, et al., 2003), downstream of a 2.5 µm cut-point impactor. The PCISs were fitted with
37 mm quartz filters pre-baked for 12 hours at 200 °C and stored in baked aluminum foil. The
PM2.5 samples were analyzed for elemental carbon (EC) concentrations by the thermal
evolution/optical transmittance method, using the National Institute for Occupational Safety and
Health (NIOSH) Thermal Optical Transmission (TOT) protocol (Gray and Cass, 1998;Birch and
139

Cary, 1996). During the summer/warm phase, two time-integrated PM2.5 samples were collected
at each site over 2-7 days of continuous sampling. During the winter/cold phase, three time-
integrated PM2.5 samples were collected at each site over 4-7 days of continuous sampling. All
PCISs were operated at 10 lpm flow rate.
In the Aethalometer model, we need EC data to correct the BC data measured by the Aethalometers
and ultimately calculate MAC values. A detailed discussion on estimating uncertainties associated
with the EC and BC concentrations is included in the supplementary information (SI). Figure S2
shows the correlation line between BC and EC obtained by lumping the data from both seasons
and sampling sites together. As shown in the figure, there is a relatively high correlation between
EC and BC data, with an R
2
value of 0.60 and a slope of 1.35. The average (BC/EC) ratio was also
found to be 1.72 ± 0.85. These values are quite consistent with those reported in a recent BC source
apportionment study in Los Angeles Basin. In that study, Mousavi et al. (2018a) reported a
(BC/EC) ratio of 1.65 ± 0.87 and a BC-EC correlation slope of 1.23, respectively. This regression
line was used to determine the corresponding EC concentration for a given BC concentration when
calculating MAV values, as described in section 2.3.

Table 6.1 Average meteorological data during the sampling campaigns in Milan and Bareggio.

6.2.3. Source apportionment of BC

To determine the relative contributions of biomass burning emissions and fossil fuel combustion
140

products to the ambient BC concentrations at both sampling locations, the Aethalometer source
apportionment model was used. This model uses light absorption data measured by multi-
wavelength Aethalometers in the near-infrared (N-IR, 880 nm) and near-ultraviolet (N-UV, 370
nm) wavelength regions to quantify differences in light absorption between fossil fuel and biomass
particulate sources, and thus determine the relative contributions of each source. The Aethalometer
“dual-spot” technology allows us to determine attenuation at each wavelength of interest (370 nm
and 880 nm), and subsequently calculate black carbon concentrations based on these data
(Drinovec et al., 2015). In order to fully resolve the potential impact of the loading effect on the
measurements from both instruments (i.e., AE31 and AE51), we followed the methodology
developed by Virkkula et al. (2007) to correct the ATN values recorded by the Aethalometers. The
correction for the loading effect can be done using the following equation (Virkkula et al., 2007):

It should be noted that the value of ATN after filter change in AE51 or after filter advancement in
AE31 is almost never zero. Therefore, it is suggested to take this fact into consideration by using
the following equation instead (Virkkula et al., 2007) :

Where, ATNini is the ATN value of the first measurement after filter change/advancement; and k
is an empirically derived constant, which was found to be 0.0051 for our dataset. This value is
consistent with the values reported by Virkkula et al. (2007) for urban sites in Europe. Finally, the
EC data provided by PCIS-collected PM samples are then used in the Aethalometer model to
correct the Aethalometer-measured BC data and calculate MAC values.
Light attenuation (ATN) due to a given PM sample is based on the proportion of incoming to
141

outgoing light intensities as measured through the filter media by the Aethalometer. ATN values
are calculated using equation (3) below, where Io is the incoming light intensity and I is the
intensity of light passing through the filter:
From ATN values provided by the Aethalometer, the aerosol attenuation coefficient (bATN) is
calculated for each time point using equation (4) where, ∆ATN is the change in the attenuation of
light over the measurement time-interval ∆t, Q is the flow rate of the instrument, and A is the filter
spot surface area.
To calculate the absorption coefficient of the aerosol (babs), we use the attenuation coefficient bATN,
as well as correction factors C and R(ATN), in Equation (5) These correction factors, C and
R(ATN), are values specific to the Aethalometer that compensate for multiple scattering and the
shadowing effects, respectively, caused by increased filter load. The value of C for the
Aethalometers used in this study was 2.14, as empirically determined by comparing the light
scattering properties of diesel soot particles and diesel soot mixed with gaseous species
(Weingartner et al., 2003).
The R(ATN) correction factor is a linear function of ln(ATN) and its value changes according to
the aerosols’ optical properties collected on the filter. R(ATN) is calculated using Equation (6):

where, f is an empirically determined parameter that takes into account the shadowing effect of
particles collected on the filter. The f values for the Aethalometers used in this study are 1.14 at
370 nm and 1.09 at 880 nm. These parameters of R(ATN) are described in further detail by
Sandradewi et al. (2008).
In the next step, Equation (7) was used to calculate the MAC value at a specific wavelength λ
(MACλ):

142

where, babs,λ represents the absorption coefficient of the aerosol at a specific wavelength λ, while
the thermal EC concentration is determined by the thermal evolution/optical transmittance method
(Knox et al., 2009).
The calculated absolute absorption coefficients at the two wavelengths, 370 nm and 880 nm, can
then be used to determine the source apportionment of total BC, as described in Zotter et al. (2014)
and Healy et al. (2017), using equations (8) through (13) below. Of aerosol at wavelength λ, with
babs,ff,880 and babs,ff,370 representing the absolute aerosol absorption values of BC from fossil fuel
combustion at 880 nm and 370 nm, respectively. Similarly, babs,bb,880 and babs,bb,370 denote aerosol
absorption values for BC from biomass burning at 880 nm and 370 nm, respectively. BCff
represents the fraction of BC concentration from fossil fuel combustion, and BCbb represents the
fraction of BC concentration from biomass burning. MAC880 represents the mass absorption cross-
section of BC at 880 nm. Lastly, αbb and αff represent the absorption Ångstrom exponent
(AAE) values for biomass burning and fossil fuel combustion, respectively. According to the
literature, αff values ithe range of 0.90-1.10 and αbb values in the range of 1.8-2.2 have typically
been used in the Aethalometer model (Massabò et al., 2015;Healy et al., 2017;Favez et al.,
2010;Herich et al., 2011). However, we empirically determined alpha (α) values specific to our
collected BC samples based on
14
C radiocarbon analysis data to provide more accurate source
apportionment results and to exclude the impact of brown carbon.

6.2.4.
14
C analysis

Radiocarbon analysis relies on the fact that
14
C is completely extinct in fossil fuels, whereas it is
on the contemporary level in modern biomass (Reddy et al., 2002;Zotter et al., 2014;Minguillón
et al., 2011). Radiocarbon is determined as F
14
C value, i.e. the isotopic ratio
14
C/
12
C present in
each sample normalized to the
14
C/
12
C ratio of the reference year 1950.
143

EC was isolated from four PM samples collected (2 per site) in summer and winter seasons as
described by Zhang et al. (2012). In brief, the filters are water-extracted and treated in the thermo-
optical OC/EC Analyzer (Model 4L, Sunset Laboratory Inc.) using the Swiss_4S protocol, which
includes three thermal steps to remove OC, two of which applying oxygen followed by an
additional helium step. Finally, EC is transformed to CO2 by combustion in pure oxygen at 850°C.
During the whole process, filter transmittance is monitored by the 660-nm tuned-diode laser of the
instrument and, thus, the formation of pyrolyzed carbon from OC, so-called charring, and
premature losses of EC are quantified. We used the MIni radioCArbon DAting System
(MICADAS) (Synal et al., 2007) at the LARA laboratory at the University of Bern (Szidat et al.,
2014) to determine the F
14
C values of the EC fractions (Szidat et al., 2014;Zhang et al.,
2012;Agrios et al., 2015). The OC/EC Analyzer is directly coupled to the interface of the gas ion
source of the MICADAS (Ruff et al., 2007; Wacker et al., 2013), which allows for direct online
14
C analysis (Agrios et al., 2015). The F
14
C values are then corrected for charring and premature
losses of EC (Zhang et al., 2012). Due to the nuclear weapon tests in the 1950s and 1960s, the
atmospheric
14
C content increased and F
14
C exhibits values > 1. Therefore, non-fossil fractions
(fNF) of EC were estimated using a reference value of F
14
C = 1.06±0.05, representing pure biomass-
burning EC based on a tree-growth model as described in Mohn et al. (2008). fNF(EC) values for
the filters are shown in Table S1.
We used Equation (14) to calculate αbb based on the results of the
14
C isotope analysis (i.e., 1-fNF
which equals to ECff/ECtotal) and the wavelength-specific babs values obtained from the
Aethalometer model:

144

As can be seen in Equation (14), the ratio of MACff,880/MACbb,880 is also needed to derive the
correlation between αbb and αff for each of the samples. Typically, the first iteration starts
assuming a MACff,880/MACbb,880 ratio equal to unity (Sandradewi et al., 2008;Herich et al.,
2011; Zotter et al., 2017). Then, a least-square fitting technique is used to find the best fit for the
αbb and αff values for each of the samples. Using the best fit, the impact of deviations in the
MACff,880/MACbb,880 ratio from unity is assessed by comparing the ECff/EC value from the
14
C analysis with the BCff/BC value from the Aethalometer model.

6.3. Results and discussion

6.3.1. Data overview

Average BC concentrations during the sampling campaign for Milan and Bareggio are presented
in Figure 2. Overall, the average BC concentration at the Bareggio site (2763 ± 567 ng.m
-3
) was
higher than at the Milan site (1921 ± 345 ng.m
-3
). In addition, as can be seen in the figure, BC
concentrations were significantly higher during the winter season as compared to the summer
season, which could in part be attributed to the more stable meteorological conditions due to lower
temperatures and wind speeds, as shown in Table 1 but more importantly, to the increased wood
burning for residential heating both in Milan and Bareggio, as previously has been reported in
other areas of Europe (Puxbaum et al., 2007;Saarikoski et al., 2008;Chakrabarty et al.,
2012;Perrone et al., 2012).
Monthly variations of BC concentrations at both sites are also demonstrated in Figure S3. The
AE31-measured BC concentrations by the Regional Environmental Protection Agency (Agenzia
Regionale per la Protezione Ambientale (ARPA) Lombardy) in metropolitan Milan are also
illustrated in Figure 2 for the SENATO and PASCAL ARPA sampling sites. The PASCAL station
145

is the closest measurement unit to our Bareggio site, and SENATO is the urban station closest to
our site in central Milan. As can be seen in the figure, the values reported by ARPA are quite
consistent and within the range of the values obtained from our campaign during the three periods,
and the small discrepancies observed between ours and ARPA’s measurements are well within the
range of errors reported for the seasonal averages.

Figure 6.2 Seasonal variations of BC concentrations in Milan and Bareggio. Error bars represent
standard error (SE). SENATO and PASCAL stations are the closest ARPA stations near the
Milan and Bareggio sites, respectively.

146

6.3.2. MAC values and the temporal trends in the source contributions

6.3.2.1. Overview of the Aethalometer model result and MAC values

In order to reduce the uncertainty associated with the MAC value determination and BCff and BCbb
calculation using the Aethalometer model (which is approximately 39% based on Mousavi et al.
(2018a)), we used the EC ff/EC ratios derived from the
14
C analysis in Eq. 14 to derive the best
fitted (MACff,880/MACbb,880) ratio by changing the values of αff, αbb, and independent (ECff/EC)
measurements. Our analysis indicates that best fitted (MACff,880/MACbb,880) value is 0.95. Then
using the (MACff,880/MACbb,880) ratio of 0.95 in Equation 14, by fitting Eq (14) against
independently measured (ECff/ EC) ratios, we derived the best fitted αff and αbb values for the area
(0.9 for αff and 1.82 for αbb, respectively) (Figure 3). The derived α values in this study are well
within the range of those reported previously by Zotter et al. (2017), who used a similar technique
to derive the best fitted α values to be used in the Aethalometer model.
Figure 4 illustrates the seasonal variations of calculated MAC values in Milan and Bareggio during
all three study periods. The annually averaged values for MAC 370 were 16.5 ± 0.6 m
2
.g
-1
and
16.2 ± 1.2 m
2
.g
-1
in Milan and Bareggio, respectively. The corresponding values for MAC 880 are

9.5 ± 1.3 m
2
.g
-1
and 8.5 ± 1.2 m
2
.g
-1
, respectively. These values are again consistent with those
reported by earlier studies conducted in other European cities (Herich et al., 2011;Sandradewi et
al., 2008;Zotter et al., 2017). For instance, Herich et al. (2011) reported an average MAC 880 value
of 9.9-13.2 m
2
.g
-1
for sites ranging from urban to suburban areas in Switzerland. Similarly, Zotter
et al. (2017) observed MAC 880 values in the range of 9.2–15.1 m
2
.g
-1
. However, it should be
noted that MAC values can vary based on the location (i.e., urban, suburban, and rural) (Favez et
al., 2010; Healy et al., 2017; Sandradewi et al., 2008; Hansen et al., 1984), so the slight differences
147

in the values across different studies can be attributed to this effect.
Based on Figure 4, comparable MAC 370 and MAC 880 values are obtained for both sites in
different seasons with a decreasing trend from summer to the winter season. For example, at the
Milan site, the average MAC 370 and MAC 880 values ranged from 17.1 m
2
.g
-1
and 10.6 m
2
.g
-1
in
the summer to 15.8 m
2
.g
-1
and 8.2 m
2
.g
-1
in the winter, respectively. This could be attributed in
part to the different geographical characteristics of the sampling sites, each being impacted by a
different mixture of combustion sources (Favez et al., 2010; Healy et al., 2017; Sandradewi et al.,
2008; Zotter et al., 2017). The seasonal variations in the MAC values were further evaluated by
performing intra-class correlation analysis on the MAC values derived for different seasons and in
different sites. The intraclass correlation coefficient (ICC) is a measure of the significance of
difference between the values in two different datasets. Any value higher than 0.6-0.7 would
suggest that the values in the two datasets are significantly different from each other (Bartko, 1966;
Donner and Wells, 1986). The calculated ICC were as high as 0.74-0.90 at the two sites, indicating
small seasonal variability in the MAC 880 and MAC 370 values in Milan and Bareggio. Based on
the above discussion, the spatial variability of MAC values at different geographic locations are
more significant than the seasonal variations at the same site, which corroborates the findings of
previous studies that MAC values are more sensitive to site characteristics and location than to
seasonal variation (Zotter et al., 2017).
148

Figure 6.3 Fig. 3. Residuals of BCff/BC compared to ECff/EC (ΔBCff=BC) as a function of
measured ECff/EC for the 4 analyzed samples using different combinations of αbb when αff=0.9.
Figure 6.4 Seasonal variations of mass absorption cross-section (MAC) in Milan andBareggio:
a) MAC 370 nm and b) MAC 880 nm. Error bars represent one standard error (SE).

149

6.3.2.2. Spatial and temporal variation of BCff and BCbb

Results of the Aethalometer source apportionment model, i.e. the BC ff and BCbb contributions to
total BC concentrations, are illustrated in Figures 5 and S4. The annually averaged BC ff
concentrations in Milan and Bareggio were 1378 ± 100 ng.m
-3
and 1331 ± 130 ng.m
-3
, respectively.
Despite the comparable BCff concentrations across the two sites, the average BCbb concentration
in Bareggio (1432 ± 340 ng.m
-3
) was more than two times higher than at the Milan site (547± 87
ng.m
-3
) (Figure 5(b)). A similar trend was observed during the winter season, with BCbb values as
high as 3284 ± 421 ng.m
-3
observed in Bareggio, compared to a wintertime average BC bb
concentration of 1154 ± 89 ng.m
-3
in Milan. This in part can be attributed to the enhanced biomass
burning activities in the suburban area of Bareggio in the wintertime (Herich et al., 2011;Gilardoni
et al., 2016;Favez et al., 2010;Szidat et al., 2007) compared to the Milan site, which is mostly
exposed to traffic emissions as well as minor black carbon emissions associated with other
combustion sources, including natural gas combustion, residental heating and cooking (Perrone et
al., 2012; Invernizzi et al., 2011). This finding is quite consistent with the results from previous
studies performed in suburban/rural parts of Europe. For example, one recent study by Gilardoni
et al. (2016) reported that more than 60- 90% of submicron organic aerosols in the rural Po Valley
in wintertime can be attributed to biomass burning sources. Additionally, Szidat et al. (2007)
investigated the impact of wood burning on particulate matter in the Alpine Valley during winter
and reported a 65-88% contribution of biomass burning to the total carbonaceous material in the
rural area. The comparable but higher annual average BCff concentrations at the Milan site (1378
± 100 ng.m
-3
) in comparison to the Bareggio site (1331 ± 130 ng.m
-3
ng.m
-3
), indicate the higher
impact of traffic-related emissions on BC concentrations in urban areas. The Aethalometer-model-
resolved average BCff and BCbb concentrations reported in this study for different seasons and at
150

different sites are in general agreement with the range of values derived for various areas of
Europe, Canada, and the United States (Herich et al., 2011;Zotter et al., 2017;Mousavi et al.,
2018a;Healy et al., 2017), with slight variations based on the location and specific geographical
and meteorological characteristics of each area. For example, Herich et al. (2011) performed a 2.5-
year (2008-2010) study in Switzerland. This study was performed at three sites, two of which were
located in a rural area and one located in the city center of Zurich. The highest BCbb concentrations
were found in the wintertime at the rural site located south of the Alps (MAG), with a BCbb
concentration of 2290 ± 140 ng.m
−3
. On the other hand, the summertime BCff concentration at the
urban (ZUR) site, located in central Zurich, was 830 ± 50 ng.m
-3
.
In addition, Healy et al. (2017) determined the BC fossil fuel and biomass burning concentrations
at nine sites in Ontario, Canada, using the Aethalometer model, and reported annual BCff and BCbb
concentrations of 1500 ng.m
-3
and 240 ng.m
-3
, respectively, for a site near a highway.
Corresponding annual average BCff and BCbb concentrations for the Milan site, a high-traffic urban
area, were 1378 ± 100 ng.m
-3
and 547 ± 87 ng.m
-3
, respectively. Most recently, Mousavi et al.
(2018a) utilized the Aethalometer source apportionment model in the Los Angeles air basin at four
sites ranging from urban to suburban background. The annual average BCff concentration across
all four sites was 1003 ± 589 ng.m
-3
during the 2016-2017 sampling period, slightly lower than
those measured in Milan and Bareggio in the current study (1354 ± 388 ng.m
-3
). The lower BCff in
the Los Angeles Basin is evidence of the effectiveness of regulations enacted in recent years in
California to reduce traffic-related PM, especially black carbon, emissions (California Code of
Regulations, 2008; San Pedro Bay Ports, 2017). Mousavi et al. (2018a) observed very low BCbb
concentrations at all of the sites, even in the wintertime (150-400 ng.m
-3
), due mainly to the
minimal impact of biomass burning on the PM and BC levels in the study area, whereas in the
151

current study BCbb concentrations at the Milan and Bareggio sites reached values as high as 1154
± 89 ng.m
-3
and 3284 ± 321 ng.m
-3
, respectively, due to the increased wood burning activities in
the wintertime.
Figure 6.5 Seasonal variations of: a) BCff; and b) BCbb concentrations in Milan and Bareggio. Error
bars represent one standard error (SE).

152

6.3.3. Impact of local events on the BCff and BCbb concentrations

On July 24th and 25th, 2017, an extensive local fire event occurred in a municipal warehouse near
the Milan site, which prompted us to investigate the variations in the BCbb concentrations during
the incident. Since the fire event occurred during the summer sampling campaign (July-August)
when biomass burning activities are minimal, a rise in daily BCbb concentrations could be
attributed to this local fire event. The daily variations of the BCbb concentrations during the incident
in July are illustrated in Figure S5. As can be seen in Figure S5(a), a clear peak was observed
on July 24th for BCbb concentrations in Milan, a value as high as 683 ng.m
-3
. However, this
increase was less pronounced at Bareggio (200 µg.m
-3
), mainly because this site was located 14
km to the west of the fire event. These episodic observations further substantiate the validity of the
Aethalometer model results and its ability to capture local black carbon sources.

6.3.4. Diurnal variations of BCff and BCbb

Figure 6 illustrates the average diurnal variations of BCff and BCbb concentrations during the entire
study campaign at each site. At the Milan site, we observed a major peak in the BCff concentrations
during the early morning (6-8 AM) traffic rush hours (Figure 6(a)), when the maximum amount
of emissions from vehicles influence the sites that are close to the freeways (Ruprecht et al., 2011).
The highest BCff concentrations at the Bareggio site were also observed between 6-9 AM (Figure
6(b)), which can be attributed to the higher traffic emissions in suburban areas once the daily
activity starts. The BCff concentrations at both sites also showed a minor peak in the late
afternoon/early evening hours, which was less pronounced than the peak observed in the morning,
due to higher wind speeds and mixing heights in the afternoon causing more atmospheric
instability (Mousavi et al., 2018b;Sowlat et al., 2016;Ware et al., 2016). In addition, the afternoon
153

peak in BCff concentrations was more noticeable in Milan compared to Bareggio, due to the higher
impact of traffic-related emissions at this site. The lowest BC ff concentrations were observed in
the nighttime for both sites, when traffic activities are minimal (Gariazzo et al., 2007).
In the case of BCbb, as can be seen in Figure 6, the concentrations started to increase in the
afternoon hours and reached a maximum during the nighttime when wood burning for residential
heating peaks and boundary layer heights are minimum. This trend was similarly observed for both
sites, even though the levels were generally higher at Bareggio. As expected, the lowest BC bb
concentrations were observed during the day when wood burning activities for residential heating
are minimal (Yan et al., 2015).
Figure 6.6 Diurnal variations of BCff and BCbb during the whole campaign in: a) Milan; and b)
Bareggio. Error bars represent one standard error (SE).

6.4. Summary and conclusions

In this study, we studied the temporal variations in the ambient concentrations of BC and the
154

contributions of their major sources (i.e., fossil fuel and biomass burning) in the greater Milan area
throughout 2016-2017 at two distinct sites: central Milan and the municipality of Bareggio,
representing urban and suburban areas, respectively. Results from the present study indicated that
BC concentrations are higher at both sites during the cold period, as compared to the intermediate
and warm periods, mostly due to the increased biomass burning activities for domestic heating, as
well as highly stable atmospheres during the cold season. The Aethalometer model results
indicated a higher concentration BCff (primarily from traffic-related emissions in the Milan
metropolitan area) than BCbb (a product of wood burning for residential heating) during the
summer season, whereas biomass burning was the dominant source of BC emissions in the
wintertime, especially in the Bareggio area, due to increased domestic wood burning activities.
BCbb concentrations as high as 1154 ± 89 ng.m
-3
and 3284 ± 321 ng.m
-3
were observed during the
cold season at the Milan and Bareggio sites, respectively, suggesting a substantial contribution of
this source to the ambient concentrations of BC in the Milan metropolitan area, especially in the
cold season. These results highlight the significant impact of residential wood burning for heating
purposes on the total BC concentrations, particularly in suburban areas of metropolitan Milan. Our
results also illustrate the significance of the within-city variations in the BC concentrations and
source contributions, which should be taken into consideration when performing comprehensive
exposure assessment studies to be used in effective regulatory efforts to minimize the population
exposure.

6.5 Acknowledgements

This study was supported by the National Institute of Health (NIH) grants 1R21AG050201-01A1
and 1RF1AG051521-01.
155

Chapter 7: Concluding remarks

7.1. Major findings:

The results from these studies provide insight into the spatio-temporal variation and sources of
ambient PM components, along with their associated toxicity, in different urban environments.
The major findings of these studies help us by improving our understanding of the sources
contributing to PM and its components (i.e., BC, redox-active metals, PN and PM) in an area.
Using the results of source apportionment in combination with emission profiles of different
sources in an area, we can determine the major sources of concern, which may then inform future
epidemiology studies and policy decisions.
Results from the first study indicated that fresh traffic PM emissions mostly in the ultrafine size
range and urban background and soil/road dust in the accumulation-range contributed considerably
to the redox-active metals concentrations. Using an online monitor (i.e., time resolution of 2 hrs)
enabled us to identify and quantify the diurnal and seasonal source contributions to the redox active
metals in an urban area, which is almost impossible using the filter-based measurements with time-
resolutions the order of 24 hrs. The diurnal trends in the metal concentrations from contributing
sources indicated a major impact of traffic on all of the redox-active metals.
In Chapter 3 (second study), a 35% reduction in PM0.25 mass concentrations from 2007 to 2017
was observed in central Los Angeles. Overall, vehicular emissions and SOA emissions were found
to be the dominant sources of PM0.25 oxidative potential at the receptor sites. Further, port-related
emissions were determined to be most significant in affecting the oxidative potential of PM0.25 in
the communities near the ports of Los Angeles and Long Beach.
156

Research presented in Chapter 4 (third study) indicated the major impact of port-related sources
to PM2.5, PN, and BC emissions, in comparison to the nearby freeways (10-40 times higher), at the
local scale. However, at the regional scale, i.e. all of Los Angeles County, freeway emissions
showed emission rates 2-5 times higher than those from sources at POLA and POLB. Findings
from Chapters 3 and 4 combined provide useful information for policy makers about the
importance of sources of PM in the communities adjacent to POLA and POLB, which can also be
utilized by epidemiologists aiming to assess the potential health impacts of exposure to specific
sources of PM.
Finally, in Chapter 5, the spatio-temporal variations of BC contributions from fossil fuel (BCff)
and biomass burning (BCbb) sources were investigated in the Los Angeles basin. The results
showed a dominant contribution of BCff and a noticeable contribution of BCbb, which is an
increasingly important source to total BC concentrations in the LA Basin, especially during the
cold season. We also observed a clear decrease in BC concentration as well as fossil fuel
contributions (BCff) over the period from 2012-2013 to 2016-2017, which is likely attributable to
the regulations in California aimed at reducing transportation-related PM and BC emissions. The
decrease in traffic-related BC emissions has resulted in an overall relative increase in the
contributions of other sources, such as biomass burning, to total BC concentrations in the LA basin.
Therefore, more attention should now be paid to these sources when designing and implementing
effective control measures in the future aimed at reducing BC concentrations in the area.

7.2. Recommendations and future research ideas

Most important implications of the findings of this dissertation can be summarized as follows:

157

Advancing future air quality regulations: current PM regulations in the United States aim at
controlling the total PM mass concentration. Over the past few decades since the enactment of the
clean air act, the major trend of changes in PM regulations have been to decrease the NAAQS
mass concentration standards to lower levels. While these progressive actions have been, overall,
successful in improvement of air quality in major urban areas, they might not be the most efficient
and cost-effective method of regulation, given that PM mass is just a surrogate of other more
important PM characteristics, especially the chemical and toxicological properties. Findings of this
dissertation can potentially facilitate the design and implementation of more targeted and cost-
effective regulatory measures that would provide public health protection while possibly lifting
some of the unnecessary burden from the industrial sector.

Improving PM-related epidemiological studies: epidemiological studies typically require a large
spatio-temporal dataset covering a variety of different populations and conditions. Comprehensive
documentation of the temporal and spatial variability of ultrafine particle composition and
oxidative potential as well as major PM toxic species ( i.e., BC and redox-active metals) discussed
in this dissertation can be used as an input for longitudinal cohort investigations that aim to study
the PM health effects over long periods of time and wide geographical regions. This data become
particularly important for cohorts that are designed for chronic health effects of air pollution
exposure (e.g. study of lung cancer or the Alzheimer’s disease), given that such studies are in dire
need of long-term and geographically resolved PM pollution and characterization data.

Statistical approach design for air pollution monitoring: Conclusion of the current research
suggest focusing on the innovative statistical-approach design environmental pollution research to
158

radically increase the spatial and temporal resolution by deploying supplementary remote sensing
dataset. This objective will create unique solutions for different sectors such as public health, smart
transportation and urban planning. For instance, a great addition to already well-established source
apportionment studies in polluted metropolitan cities around the world (i.e., Los Angeles, Beijing,
Delhi), would be the online and accurately processed (i.e., stream analytics) air pollution remote
sensing data (i.e., ground-based sensor networks or NASA satellite) through cloud service that can
be coupled with sophisticated data analytics tools (i.e., Geo coding, R analytics and machine
learning forecasting) resulting in time-resolved variation in source contributions in contrasting
communities. Such data acquisition would be also beneficial for environmental exposure
assessment to effectively monitor sensitive communities (i.e., children, pregnant women or
elderly) in the society. This approach will substantially account for the current cavity of fairly high
uncertainty associated with the population exposure assessment research.
159

Bibliography

Aiken, A.C., DeCarlo, P.F., Kroll, J.H., Worsnop, D.R., Huffman, J.A., Docherty, K.S., Ulbrich,
I.M., Mohr, C., Kimmel, J.R., Sueper, D., Sun, Y., Zhang, Q., Trimborn, A., Northway, M.,
Ziemann, P.J., Canagaratna, M.R., Onasch, T.B., Alfarra, M.R., Prevot, A.S.H., Dommen, J.,
Duplissy, J., Metzger, A., Baltensperger, U., Jimenez, J.L., 2008. O/C and OM/OC Ratios of
Primary, Secondary, and Ambient Organic Aerosols with High-Resolution Time-of-Flight
Aerosol Mass Spectrometry. Environ. Sci. Tech. 42, 4478-4485.
Argyropoulos, G., Besis, A., Voutsa, D., Samara, C., Sowlat, M.H., Hasheminassab, S., Sioutas,
C., 2016. Source apportionment of the redox activity of urban quasi-ultrafine particles (PM
0.49) in Thessaloniki following the increased biomass burning due to the economic crisis in
Greece. Sci. Total Environ. 568, 124-136.
Apeagyei, E., Bank, M. S., and Spengler, J. D., 2011. Distribution of heavy metals in road dust
along an urban-rural gradient in Massachusetts. Atmos. Environ., 45(13), 2310-2323.
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., and Katsouyanni, K., 2011. Acute
effects of particulate air pollution on respiratory admissions: results from APHEA 2 project.
Air Pollution and Health: a European Approach, Am J Respir Crit Care Med, 164, 1860-1866.
Ault, A. P., Moore, M. J., Furutani, H., and Prather, K. A., 2011. Impact of emissions from the
Los Angeles port region on San Diego air quality during regional transport events, Environ Sci
Technol, 43, 3500-3506.
Birmili, W., Allen, A.G., Bary, F., Harrison, R.M., 2006. Trace metal concentrations and water
solubility in size-fractionated atmospheric particles and influence of road traffic. Environ. Sci.
Tech. 40, 1144-1153.
Brown, S.G., Eberly, S., Paatero, P., Norris, G.A., 2015. Methods for estimating uncertainty in PMF
solutions: Examples with ambient air and water quality data and guidance on reporting PMF
results. Sci. Total Environ. 518, 626-635.
Brunekreef, B., Forsberg, B., 2005. Epidemiological evidence of effects of coarse airborne particles
on health. European Respiratory Journal 26, 309-318.
Carlton, A., Wiedinmyer, C., and Kroll, J., 2009. A review of Secondary Organic Aerosol (SOA)
160

formation from isoprene, Atmos. Chem. Phys., 9, 4987-5005.
Cesari, D., Amato, F., Pandolfi, M., Alastuey, A., Querol, X., Contini, D., 2016. An inter-
comparison of PM10 source apportionment using PCA and PMF receptor models in three
European sites. Environ. Sci. Poll. Res. 23, 15133-15148.
Cheung, K., Daher, N., Shafer, M.M., Ning, Z., Schauer, J.J., Sioutas, C., 2011. Diurnal trends in
coarse particulate matter composition in the Los Angeles Basin. J. Environ. Monit. 13, 3277-
3287.
Cheung, K., Shafer, M.M., Schauer, J.J., Sioutas, C., 2012. Diurnal trends in oxidative potential
of coarse particulate matter in the Los Angeles Basin and their relation to sources and chemical
composition. Environ. Sci. Tech. 46, 3779-3787.
Chang, M. C., Sioutas, C., Kim, S., Gong Jr, H., and Linn, W. S., 2000. Reduction of nitrate losses
from filter and impactor samplers by means of concentration enrichment, Atmos. Environ., 34,
85-98.
Cheng, Y.-H., and Tsai, C.-J., 1997. Evaporation loss of ammonium nitrate particles during filter
sampling, J Aerosol Sci., 28, 1553-1567.
Cheung, K. L., Ntziachristos, L., Tzamkiozis, T., Schauer, J. J., Samaras, Z., Moore, K. F., and
Sioutas, C., 2010. Emissions of Particulate Trace Elements, Metals and Organic Species from
Gasoline, Diesel, and Biodiesel Passenger Vehicles and Their Relation to Oxidative Potential,
Aerosol Sci Technol., 44, 500-513.
Cho, A. K., Sioutas, C., Miguel, A. H., Kumagai, Y., Schmitz, D. A., Singh, M., Eiguren-
Fernandez, A., and Froines, J. R., 2005. Redox activity of airborne particulate matter at different
sites in the Los Angeles Basin, Environ. Res., 99, 40-47.
Chow, J. C., Watson, J. G., Kuhns, H., Etyemezian, V., Lowenthal, D. H., Crow, D., Kohl, S. D.,
Engelbrecht, J. P., and Green, M. C., 2004. Source profiles for industrial, mobile, and area
sources in the Big Bend Regional Aerosol Visibility and Observational study, Chemosphere,
54, 185-208.
Daher, N., Ruprecht, A., Invernizzi, G., De Marco, C., Miller-Schulze, J., Heo, J. B., Shafer, M.
M., Shelton, B. R., Schauer, J. J., and Sioutas, C., 2012. Characterization, sources and redox
activity of fine and coarse particulate matter in Milan, Italy, Atmos. Environ., 49, 130-141.
Dall'Osto, M., Querol, X., Amato, F., Karanasiou, A., Lucarelli, F., Nava, S., Calzolai, G., Chiari,
M., 2013. Hourly elemental concentrations in PM 2.5 aerosols sampled simultaneously at urban
161

background and road site during SAPUSS–diurnal variations and PMF receptor modelling.
Atmos. Chem. Phys. 13, 4375-4392.
Davis, D.A., Bortolato, M., Godar, S.C., Sander, T.K., Iwata, N., Pakbin, P., Shih, J.C., Berhane,
K., McConnell, R., Sioutas, C., 2013. Prenatal exposure to urban air nanoparticles in mice
causes altered neuronal differentiation and depression-like responses.
Decesari, S., Sowlat, M.H., Hasheminassab, S., Sandrini, S., Gilardoni, S., Facchini, M.C., Fuzzi,
S., Sioutas, C., 2017. Enhanced toxicity of aerosol in fog conditions in the Po Valley, Italy.
Atmos. Chem. Phys. 17, 7721-7731.
Delfino, R.J., Sioutas, C., Malik, S., 2005. Potential role of ultrafine particles in associations
between airborne particle mass and cardiovascular health. Environmental health perspectives,
934-946.
Delfino, R.J., Tjoa, T., Gillen, D.L., Staimer, N., Polidori, A., Arhami, M., Jamner, L., Sioutas, C.,
Longhurst, J., 2010b. Traffic-related Air Pollution and Blood Pressure in Elderly Subjects With
Coronary Artery Disease. Epidemiology (Cambridge, Mass.) 21,
10.1097/EDE.1090b1013e3181d1095e1019b.
Ding, X., Zheng, M., Yu, L., Zhang, X., Weber, R. J., Yan, B., Russell, A. G., Edgerton, E. S., and
Wang, X., 2008. Spatial and Seasonal Trends in Biogenic Secondary Organic Aerosol Tracers
and Water-Soluble Organic Carbon in the Southeastern United States, Environ Sci Technol.,
42, 5171-5176.
Ding, X., Wang, X.M., Gao, B., Fu, X.X., He, Q.F., Zhao, X.Y., Yu, J.Z. and Zheng, M., 2012.
Tracer‐based estimation of secondary organic carbon in the Pearl River Delta, south China.
Journal of Geophysical Research: Atmospheres, 117(D5).
Dockery , D. W., Pope , C. A., Xu , X., Spengler , J. D., Ware , J. H., Fay , M. E., Ferris , B. G. J.,
and Speizer , F. E., 1993. An Association between Air Pollution and Mortality in Six U.S. Cities,
N Engl J Med, 329, 1753-1759.
Dockery, D.W., Stone, P.H., 2007. Cardiovascular risks from fine particulate air pollution. New
England Journal of Medicine 356, 511-513.
Donaldson, K., Li, X.Y., MacNee, W., 1998. Ultrafine (nanometre) particle mediated lung injury.
Journal of Aerosol Science 29, 553-560.
Donateo A, Gregoris E, Gambaro A, Merico E, Giua R, Nocioni A, 2014. Contribution of harbour
activities and ship traffic to PM2. 5, particle number concentrations and PAHs in a port city of
162

the Mediterranean Sea (Italy). Environ Sci Pollut Res., 21: 9415-9429.
El-Alawi, Y. S., McConkey, B. J., Dixon, D. G., and Greenberg, B. M., 2012. Measurement of
short-and long-term toxicity of polycyclic aromatic hydrocarbons using luminescent bacteria.
Ecotoxicol Environ Saf, 51(1), 12-21.
Fang, T., Guo, H., Verma, V., Peltier, R.E., Weber, R.J., 2015. PM 2.5 water-soluble elements in
the southeastern United States: automated analytical method development, spatiotemporal
distributions, source apportionment, and implications for heath studies. Atmos. Chem. Phys.
15, 11667-11682.
Fine, P. M., Cass, G. R., and Simoneit, B. R., 2004a. Chemical characterization of fine particle
emissions from the fireplace combustion of wood types grown in the Midwestern and Western
United States, J Environ Eng Ecol Sci., 21, 387-409.
Fine, P. M., Chakrabarti, B., Krudysz, M., Schauer, J. J., and Sioutas, C., 2004b. Diurnal Variations
of Individual Organic Compound Constituents of Ultrafine and Accumulation Mode Particulate
Matter in the Los Angeles Basin, Environ Sci Technol., 38, 1296-1304.
Gao, J., Peng, X., Chen, G., Xu, J., Shi, G.L., Zhang, Y.C., Feng, Y.C., 2016. Insights into the
chemical characterization and sources of PM 2.5 in Beijing at a 1-h time resolution. Sci. Total
Environ. 542, 162-171.
Gakidou, E., Afshin, A., Abajobir, A. A., Abate, K. H., Abbafati, C., Abbas, K. M., ... and Abu-
Raddad, L. J., 2017. Global, regional, and national comparative risk assessment of 84
behavioural, environmental and occupational, and metabolic risks or clusters of risks, 1990-
2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet, 390(10100),
1345-1422.
Galarneau, E., 2008. Source specificity and atmospheric processing of airborne PAHs: implications
for source apportionment. Atmos. Environ., 42(35), 8139-8149.
Gauderman, W.J., Urman, R., Avol, E., Berhane, K., McConnell, R., Rappaport, E., Chang, R.,
Lurmann, F., Gilliland, F., 2015. Association of improved air quality with lung development in
children. New England Journal of Medicine 372, 905-913.
Gietl, J.K., Lawrence, R., Thorpe, A.J., Harrison, R.M., 2010. Identification of brake wear particles
and derivation of a quantitative tracer for brake dust at a major road. Atmos. Environ. 44, 141-
146.
Hamad SH, Shafer MM, Kadhim AKH, Al-Omran SM, Schauer JJ. Seasonal trends in the
163

composition and ROS activity of fine particulate matter in Baghdad, Iraq. Atmospheric
Environment 2015; 100: 102-110.
Harrison, R.M., Beddows, D.C.S., Dall’Osto, M., 2011. PMF analysis of wide-range particle size
spectra collected on a major highway. Environ. Sci. Tech. 45, 5522-5528.
Harrison, R.M., Jones, A.M., Gietl, J., Yin, J., Green, D.C., 2012. Estimation of the contributions
of brake dust, tire wear, and resuspension to nonexhaust traffic particles derived from
atmospheric measurements. Environ. Sci. Tech. 46, 6523-6529.
Hasheminassab, S., Daher, N., Schauer, J. J., and 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.
Hasheminassab, S., Daher, N., Ostro, B.D., Sioutas, C., 2014a. 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. Environ. Poll. 193, 54-64.
Hasheminassab, S., Daher, N., Saffari, A., Wang, D., Ostro, B.D., Sioutas, C., 2014b. Spatial and
temporal variability of sources of ambient fine particulate matter (PM2.5) in California. Atmos.
Chem. Phys. 14, 12085-12097.
Hassanvand, M.S., Naddafi, K., Faridi, S., Nabizadeh, R., Sowlat, M.H., Momeniha, F.,
Gholampour, A., Arhami, M., Kashani, H., Zare, A., 2015. Characterization of PAHs and metals
in indoor/outdoor PM 10/PM 2.5/PM 1 in a retirement home and a school dormitory. Sci. Total
Environ. 527, 100-110.
Healy, R. M., O'Connor, I. P., Hellebust, S., Allanic, A., Sodeau, J. R., and Wenger, J. C., 2009.
Characterisation of single particles from in-port ship emissions, Atmospheric Environment, 43,
6408-6414.
Heo, J., Dulger, M., Olson, M.R., McGinnis, J.E., Shelton, B.R., Matsunaga, A., Sioutas, C. and
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.
Atmospheric Environment, 73, 51-61.
Hu, S., Polidori, A., Arhami, M., Shafer, M., Schauer, J., Cho, A., and Sioutas, C., 2008. 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.
Hu, D. and Yu, J.Z., 2013. Secondary organic aerosol tracers and malic acid in Hong Kong:
164

seasonal trends and origins. Environmental Chemistry, 10(5), 381-394.
Isakson J, Persson TA, Selin Lindgren E., 2001. Identification and assessment of ship emissions
and their effects in the harbour of Göteborg, Sweden. Atmos. Environ., 35: 3659-3666.
Janssen, N. A. H., Yang, A., Strak, M., Steenhof, M., Hellack, B., Gerlofs-Nijland, M. E.,
Kuhlbusch, T., Kelly, F., Harrison, R., Brunekreef, B., Hoek, G., and Cassee, F., 2014.
Oxidative potential of particulate matter collected at sites with different source characteristics,
Sci Total Environ., 472, 572-581.
Jerrett, M., Burnett, R.T., Ma, R., Pope Iii, C.A., Krewski, D., Newbold, K.B., Thurston, G., Shi,
Y., Finkelstein, N., Calle, E.E., 2005. Spatial analysis of air pollution and mortality in Los
Angeles. Epidemiology 16, 727-736.
Kasumba, J., Hopke, P.K., Chalupa, D.C., Utell, M.J., 2009. Comparison of sources of submicron
particle number concentrations measured at two sites in Rochester, NY. Sci. Total Environ. 407,
5071-5084.
Kim, E., Hopke, P.K., 2004. Source apportionment of fine particles in Washington, DC, utilizing
temperature-resolved carbon fractions. J. Air Waste Manag. Assoc. 54, 773-785.
Kim, E., Hopke, P.K., Edgerton, E.S., 2003. Source Identification of Atlanta Aerosol by Positive
Matrix Factorization. J. Air Waste Manag. Assoc. 53, 731-739.
Kim, K. H., Jahan, S. A., Kabir, E., and Brown, R. J., 2013. A review of airborne polycyclic
aromatic hydrocarbons (PAHs) and their human health effects. Environ Int., 60, 71-80.
Kleeman, M. J., Riddle, S. G., Robert, M. A., and Jakober, C. A., 2008. Lubricating Oil and Fuel
Contributions To Particulate Matter Emissions from Light-Duty Gasoline and Heavy-Duty
Diesel Vehicles, Environ Sci Technol., 42, 235-242.
Knol, A. B., de Hartog, J. J., Boogaard, H., Slottje, P., van der Sluijs, J. P., Lebret, E., Cassee, F.
R., Wardekker, J. A., Ayres, J. G., Borm, P. J., Brunekreef, B., Donaldson, K., Forastiere, F.,
Holgate, S. T., Kreyling, W. G., Nemery, B., Pekkanen, J., Stone, V., Wichmann, H.-E., and
Hoek, G., 2009. Expert elicitation on ultrafine particles: likelihood of health effects and causal
pathways, Part Fibre Toxicol., 6, 19.
Kulkarni, P., Chellam, S., and Fraser, M. P., 2007. Tracking petroleum refinery emission events
using lanthanum and lanthanides as elemental markers for PM2.5. Environ Sci
Technol., 41(19), 6748-6754.

165

Kumar, P., Morawska, L., Birmili, W., Paasonen, P., Hu, M., Kulmala, M., Harrison, R.M.,
Norford, L., Britter, R., 2014. Ultrafine particles in cities. Environ. Int. 66, 1-10.
Landreman, A. P., Shafer, M. M., Hemming, J. C., Hannigan, M. P., and Schauer, J. J. , 2008. A
macrophage-based method for the assessment of the reactive oxygen species (ROS) activity of
atmospheric particulate matter (PM) and application to routine (daily-24 h) aerosol monitoring
studies, Aerosol Sci Technol., 42, 946-957.
Lee, H. J., Villaume, J., Cullen, D. C., Kim, B. C., and Gu, M. B., 2003. Monitoring and
classification of PAH toxicity using an immobilized bioluminescent bacteria, Biosens.
Bioelectron, 18, 571-577.
Lee, G., You, S., Ritchie, S., Saphores, J. D., Sangkapichai, M., and Jayakrishnan, R., 2009.
Environmental impacts of a major freight corridor: a study of I-710 in California. Transportation
Research Record: Journal of the Transportation Research Board, (2123), 119-128.
Li, N., Wang, M., Bramble, L.A., Schmitz, D.A., Schauer, J.J., Sioutas, C., Harkema, J.R., Nel,
A.E., 2009. The adjuvant effect of ambient particulate matter is closely reflected by the
particulate oxidant potential. Environ. Health Perspect. 117, 1116-1123.
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., 2013. 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.
Liacos, J. W., Kam, W., Delfino, R. J., Schauer, J. J., and Sioutas, C., 2012. Characterization of
organic, metal and trace element PM2.5 species and derivation of freeway-based emission rates
in Los Angeles, CA, Sci Total Environ., 435-436, 159-166.
Manchester-Neesvig, J. B., Schauer, J. J., and 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, Air Waste Manag Assoc., 53, 1065-1079.
Masiol, M., Hopke, P.K., Felton, H.D., Frank, B.P., Rattigan, O.V., Wurth, M.J., LaDuke, G.H.,
2017. Source apportionment of PM2.5 chemically speciated mass and particle number
concentrations in New York City. Atmos. Environ. 148, 215-229.
Megido, L., Negral, L., Castrillón, L., Marañón, E., Fernández-Nava, Y., Suárez-Peña, B., 2016.
Traffic tracers in a suburban location in northern Spain: relationship between carbonaceous
166

fraction and metals. Environ. Sci. Poll. Res. 23, 8669-8678.
Michael, S., Montag, M., and Dott, W., 2013. Pro-inflammatory effects and oxidative stress in lung
macrophages and epithelial cells induced by ambient particulate matter, Environ Pollut., 183,
19-29.
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. New England Journal of Medicine 356, 447-458.
Minguillón, M. C., Arhami, M., Schauer, J. J., and 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.
Misra, C., Singh, M., Shen, S., Sioutas, C., and Hall, P. M., 2002. Development and evaluation of
a personal cascade impactor sampler (PCIS), J Aerosol Sci., 33, 1027-1047.
Moldanová, J., Fridell, E., Popovicheva, O., Demirdjian, B., Tishkova, V., Faccinetto, A., and
Focsa, C., 2009. Characterisation of particulate matter and gaseous emissions from a large ship
diesel engine, Atmos. Environ., 43, 2632-2641.
Moller, W., Felten, K., Sommerer, K., Scheuch, G., Meyer, G., Meyer, P., Haussinger, K., and
Kreyling, W. G., 2008. Deposition, retention, and translocation of ultrafine particles from the
central airways and lung periphery, Am J Respir Crit Care Med, 177, 426-432.
Moore, K., Krudysz, M., Pakbin, P., Hudda, N., and 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.
Moreno, N., Viana, M., Pandolfi, M., Alastuey, A., Querol, X., Chinchón, S., Pinto, J. F., Torres,
F., Díez, J. M., and Saéz, J. (2009). Determination of direct and fugitive PM emissions in a
Mediterranean harbour by means of classic and novel tracer methods, J Environ Manage., 91,
133-141.
Moreno T, Querol X, Alastuey A, de la Rosa J, Sánchez de la Campa AM, Minguillón M, 2010.
Variations in vanadium, nickel and lanthanoid element concentrations in urban air. Sci Total
Environ., 408: 4569-4579.
Morishita, M., Keeler, G.J., Kamal, A.S., Wagner, J.G., Harkema, J.R., Rohr, A.C., 2011.
Identification of ambient PM 2.5 sources and analysis of pollution episodes in Detroit, Michigan
using highly time-resolved measurements. Atmos. Environ. 45, 1627-1637.
167

Mousavi, A., Sowlat, M. H., and Sioutas, C., 2018a. Diurnal and seasonal trends and source
apportionment of redox-active metals in Los Angeles using a novel online metal monitor and
Positive Matrix Factorization (PMF), Atmos. Environ., 174, 15-24.
Mousavi A, Sowlat MH, Hasheminassab S, Polidori A, Sioutas C., 2018b. Spatio-temporal trends
and source apportionment of fossil fuel and biomass burning black carbon (BC) in the Los
Angeles Basin. Sci Total Environ., 640-641: 1231-1240.
Mousavi, A., Sowlat, M. H., Hasheminassab, S., Pikelnaya, O., Polidori, A., Ban-Weiss, G., and
Sioutas, C., 2018c. Impact of particulate matter (PM) emissions from ships, locomotives, and
freeways in the communities near the Ports of Los Angeles (POLA) and Long Beach (POLB)
on the air quality in the Los Angeles County, Atmos. Environ.
Naddafi, K., Sowlat, M.H., Safari, M.H., 2012. Integrated assessment of air pollution in Tehran,
over the period from September 2008 to September 2009. Iran. J. Public Health 41, 77.
Nel, A., Xia, T., Mädler, L., Li, N., 2006. Toxic potential of materials at the nanolevel. Science
311, 622-627.
Ning Z, Polidori A, Schauer JJ, Sioutas C., 2008. Emission factors of PM species based on freeway
measurements and comparison with tunnel and dynamometer studies. Atmos. Environ., 42:
3099-3114.
Ntziachristos, L., Froines, J.R., Cho, A.K., Sioutas, C., 2007. Relationship between redox activity
and chemical speciation of size-fractionated particulate matter. Particle Fibre toxicol. 4, 1.
Ntziachristos, L., Ning, Z., Geller, M. D., Sheesley, R. J., Schauer, J. J., and Sioutas, C., 2007. Fine,
ultrafine and nanoparticle trace element compositions near a major freeway with a high heavy-
duty diesel fraction. Atmos. Environ., 41(27), 5684-5696.
Oberdörster, G., Sharp, Z., Atudorei, V., Elder, A., Gelein, R., Lunts, A., Kreyling, W., Cox, C.,
2002. Extrapulmonary translocation of ultrafine carbon particles following whole-body
inhalation exposure of rats. Journal of Toxicology and Environmental Health Part A 65, 1531-
1543.
Oberdörster, G., Oberdörster, E., and Oberdörster, J., 2005. Nanotoxicology: An Emerging
Discipline Evolving from Studies of Ultrafine Particles, Environ Health Perspect, 113, 823-839.
Ogulei, D., Hopke, P.K., Chalupa, D.C., Utell, M.J., 2007. Modeling source contributions to
submicron particle number concentrations measured in Rochester, New York. Aerosol Sci.
Tech. 41, 179-201.
168

Oh, S. M., Kim, H. R., Park, Y. J., Lee, S. Y., and Chung, K. H., 2011. Organic extracts of urban
air pollution particulate matter (PM2. 5)-induced genotoxicity and oxidative stress in human
lung bronchial epithelial cells (BEAS-2B cells), Mutat Res Genet Toxicol Environ Mutagen.,
723, 142-151.
Ospital, J., Cassmassi, J., Chico, T., 2008. Multiple air toxics exposure study in the south coast air
basin. MATES III Draft Final Report, South Coast Air Quality Management District, Diamond
Bar, CA (USA).
Pakbin P, Ning Z, Shafer MM, Schauer JJ, Sioutas C., 2011. Seasonal and spatial coarse particle
elemental concentrations in the Los Angeles area. Aerosol Sci Technol., 45: 949-963.
Pancras, J.P., Landis, M.S., Norris, G.A., Vedantham, R., Dvonch, J.T., 2013. Source
apportionment of ambient fine particulate matter in Dearborn, Michigan, using hourly resolved
PM chemical composition data. Sci. Total Environ. 448, 2-13.
Pancras, J.P., Ondov, J.M., Poor, N., Landis, M.S., Stevens, R.K., 2006. Identification of sources
and estimation of emission profiles from highly time-resolved pollutant measurements in
Tampa, FL. Atmos. Environ. 40, 467-481.
Pandolfi M, Gonzalez-Castanedo Y, Alastuey A, de la Rosa JD, Mantilla E, de la Campa AS, 2011.
Source apportionment of PM10 and PM2.5 at multiple sites in the strait of Gibraltar by PMF:
impact of shipping emissions. Environ Sci Pollut Res., 18: 260-269.
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. Particle and fibre toxicology 3, 1.
Peters, A., Wichmann, H.E., Tuch, T., Heinrich, J., Heyder, J., 1997. Respiratory effects are
associated with the number of ultrafine particles. American journal of respiratory and critical
care medicine 155, 1376-1383.
Pey, J., Querol, X., Alastuey, A., Rodríguez, S., Putaud, J.P., Van Dingenen, R., 2009. Source
apportionment of urban fine and ultra-fine particle number concentration in a Western
Mediterranean city. Atmos. Environ. 43, 4407-4415.
Phuleria HC, Sheesley RJ, Schauer JJ, Fine PM, Sioutas C., 2007. Roadside measurements of size-
segregated particulate organic compounds near gasoline and diesel-dominated freeways in Los
Angeles, CA. Atmos. Environ., 41: 4653-4671.
POLA: air emission inventory, port of Long Beach, July 2016, 2015.
169

Polidori A, Hu S, Biswas S, Delfino R, Sioutas C., 2008. Real-time characterization of particle-
bound polycyclic aromatic hydrocarbons in ambient aerosols and from motor-vehicle exhaust.
Atmos. Chem. Phys., 8: 1277-1291.
Pope, C.A., Burnett, R.T., Thurston, G.D., Thun, M.J., Calle, E.E., Krewski, D., Godleski, J.J.,
2004b. Cardiovascular mortality and long-term exposure to particulate air pollution
epidemiological evidence of general pathophysiological pathways of disease. Circulation 109,
71-77.
Pope Iii, C.A., Burnett, R.T., Thun, M.J., Calle, E.E., Krewski, D., Ito, K., Thurston, G.D., 2002.
Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air
pollution. Jama 287, 1132-1141.
Propper, R., Wong, P., Bui, S., Austin, J., Vance, W., Alvarado, Á., Croes, B., Luo, D., 2015.
Ambient and Emission Trends of Toxic Air Contaminants in California. Environ. Sci. Tech. 49,
11329-11339.
Querol, X., Viana, M., Alastuey, A., Amato, F., Moreno, T., Castillo, S., Pey, J., De la Rosa, J., De
La Campa, A. S., and Artíñano, B., 2007. Source origin of trace elements in PM from regional
background, urban and industrial sites of Spain, Atmos. Environ., 41, 7219-7231.
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.
Richard, A., Gianini, M.F.D., Mohr, C., Furger, M., Bukowiecki, N., Minguillón, M.C., Lienemann,
P., Flechsig, U., Appel, K., DeCarlo, P.F., 2011. Source apportionment of size and time resolved
trace elements and organic aerosols from an urban courtyard site in Switzerland. Atmos. Chem.
Phys. 11, 8945-8963.
Saffari, A., Hasheminassab, S., Shafer, M.M., Schauer, J.J., Chatila, T.A., Sioutas, C., 2016.
Nighttime Formation of Aqueous-Phase Secondary Organic Aerosols in Los Angeles and its
Implication for Fine Particulate Matter Composition and Oxidative Potential. Atmos. Environ.
133, 112-122.
Saffari, A., Hasheminassab, S., Wang, D., Shafer, M.M., Schauer, J.J., Sioutas, C., 2015. Impact of
primary and secondary organic sources on the oxidative potential of quasi-ultrafine particles
(PM0. 25) at three contrasting locations in the Los Angeles Basin. Atmos. Environ. 120, 286-
296.
170

Saffari, A., Daher, N., Shafer, M. M., Schauer, J. J., and Sioutas, C., 2013. Seasonal and spatial
variation in reactive oxygen species activity of quasi-ultrafine particles (PM 0.25) in the Los
Angeles metropolitan area and its association with chemical composition, Atmos Environ, 79,
566-575.
Saffari, A., Hasheminassab, S., Wang, D., Shafer, M. M., Schauer, J. J., and Sioutas, C., 2015.
Impact of primary and secondary organic sources on the oxidative potential of quasi-ultrafine
particles (PM0.25) at three contrasting locations in the Los Angeles Basin, Atmos. Environ.,
120, 286-296.
Sanderson, P., Delgado-Saborit, J. M., and Harrison, R. M., 2014. A review of chemical and
physical characterisation of atmospheric metallic nanoparticles, Atmospheric Environment, 94,
353-365.
San Pedro Bay Ports, 2017. Clean Air Action Plan, 2017 update.
Schaap M, Spindler G, Schulz M, Acker K, Maenhaut W, Berner A, 2004. Artefacts in the sampling
of nitrate studied in the “INTERCOMP” campaigns of EUROTRAC-AEROSOL. Atmos.
Environ.38: 6487-6496.
Schauer, J. J., Rogge, W. F., Hildemann, L. M., Mazurek, M. A., Cass, G. R., and Simoneit, B. R.
T., 1996. Source apportionment of airborne particulate matter using organic compounds as
tracers, Atmos. Environ., 30, 3837-3855.
Secrest MH, Schauer JJ, Carter EM, Lai AM, Wang Y, Shan M, et al. The oxidative potential of
PM2.5 exposures from indoor and outdoor sources in rural China. Science of The Total
Environment 2016; 571: 1477-1489.
Shafer, M. M., Hemming, J. D., Antkiewicz, D. S., and Schauer, J. J., 2016. Oxidative potential of
size-fractionated atmospheric aerosol in urban and rural sites across Europe, Farad. discus., 189,
381-405.
Shafer MM, Perkins DA, Antkiewicz DS, Stone EA, Quraishi TA, Schauer JJ. Reactive oxygen
species activity and chemical speciation of size-fractionated atmospheric particulate matter
from Lahore, Pakistan: an important role for transition metals. Journal of Environmental
Monitoring 2010; 12: 704-715.
Shi, G., Xu, J., Peng, X., Xiao, Z., Chen, K., Tian, Y., Guan, X., Feng, Y., Yu, H., Nenes, A.,
Russell, A.G., 2017. pH of Aerosols in a Polluted Atmosphere: Source Contributions to Highly
Acidic Aerosol. Environ. Sci. Tech. 51(8), 4289-4296.
171

Shirmohammadi F, Hasheminassab S, Wang D, Saffari A, Schauer JJ, Shafer MM, 2015. Oxidative
potential of coarse particulate matter (PM10–2.5) and its relation to water solubility and sources
of trace elements and metals in the Los Angeles Basin. Environ Sci Process Impacts., 17: 2110-
2121.
Shirmohammadi, F., Hasheminassab, S., Saffari, A., Schauer, J. J., Delfino, R. J., and Sioutas, C.,
2016. Fine and ultrafine particulate organic carbon in the Los Angeles basin: trends in sources
and composition, Sci Total Environ., 541, 1083-1096.
Shirmohammadi F, Wang D, Hasheminassab S, Verma V, Schauer JJ, Shafer MM, 2017. Oxidative
potential of on-road fine particulate matter (PM2.5) measured on major freeways of Los
Angeles, CA, and a 10-year comparison with earlier roadside studies. Atmos. Environ., 148:
102-114.
Shirmohammadi, F., Lovett, C., Sowlat, M. H., Mousavi, A., Verma, V., Shafer, M. M., Schauer,
J. J., and Sioutas, C., 2018. Chemical composition and redox activity of PM0.25 near Los
Angeles International Airport and comparisons to an urban traffic site, Sci Total Environ., 610-
611, 1336-1346.
Shirmohammadi, F., Lovett, C., Sowlat, M.H., Mousavi, A., Verma, V., Shafer, M.M., Schauer,
J.J., Sioutas, C., 2017. Chemical composition and redox activity of PM0. 25 near Los Angeles
International Airport and comparisons to an urban traffic site. Sci. Total Environ. 610, 1336.
Shuster-Meiseles T, Shafer MM, Heo J, Pardo M, Antkiewicz DS, Schauer JJ, et al. ROS-
generating/ARE-activating capacity of metals in roadway particulate matter deposited in urban
environment. Environmental Research 2016; 146: 252-262.
Simoneit, B. R., 1986. Characterization of organic constituents in aerosols in relation to their rigin
and transport: a review, Int J Environ Anal Chem., 23, 207-237.
Sioutas, C., Delfino, R.J., Singh, M., 2005a. Exposure assessment for atmospheric ultrafine
particles (UFPs) and implications in epidemiologic research. Environmental Health
Perspectives 113, 947-955.
Sioutas, C., Delfino, R.J., Singh, M., 2005b. Exposure assessment for atmospheric ultrafine
particles (UFPs) and implications in epidemiologic research. Environmental health
perspectives, 947-955.
Sowlat, M.H., Hasheminassab, S., Sioutas, C., 2016a. Source apportionment of ambient particle
number concentrations in central Los Angeles using positive matrix factorization (PMF).
172

Atmos. Chem. Phys. 16, 4849-4866.
Sowlat, M.H., Naddafi, K., Yunesian, M., Jackson, P.L., Lotfi, S., Shahsavani, A., 2013. PM10
source apportionment in Ahvaz, Iran, using positive matrix factorization. CLEAN Soil Air
Water 41, 1143-1151.
Sowlat, M.H., Naddafi, K., Yunesian, M., Jackson, P.L., Shahsavani, A., 2012. Source
apportionment of total suspended particulates in an arid area in southwestern Iran using positive
matrix factorization. Bull. Environ. Contam. Toxicol. 88, 735-740.
Sowlat, M.H., Wang, D., Simonetti, G., Shafer, M.M., Schauer, J.J., Sioutas, C., 2016b.
Development and field evaluation of an online monitor for near-continuous measurement of
iron, manganese, and chromium in coarse airborne particulate matter (PM). Aerosol Sci. Tech.
50, 1306-1319.
Streibel, T., Schnelle-Kreis, J., Czech, H., Harndorf, H., Jakobi, G., Jokiniemi J., 2017. Aerosol
emissions of a ship diesel engine operated with diesel fuel or heavy fuel oil. Environ Sci Pollut
Res., 24: 10976-10991.
Świetlik, R., Trojanowska, M., Strzelecka, M., Bocho-Janiszewska, A., 2015. Fractionation and
mobility of Cu, Fe, Mn, Pb and Zn in the road dust retained on noise barriers along expressway
– A potential tool for determining the effects of driving conditions on speciation of emitted
particulate metals. Environ. Poll. 196, 404-413.
Tao, F., Gonzalez-Flecha, B., Kobzik, L., 2003. Reactive oxygen species in pulmonary
inflammation by ambient particulates. Free Radical Biology and Medicine 35, 327-340.
Tao, J., Zhang, L., Zhang, R., Wu, Y., Zhang, Z., Zhang, X., Tang, Y., Cao, J., Zhang, Y., 2016.
Uncertainty assessment of source attribution of PM2.5 and its water-soluble organic carbon
content using different biomass burning tracers in positive matrix factorization analysis — a
case study in Beijing, China. Sci. Total Environ. 543, Part A, 326-335.
Tang, I. N., Tridico, A., and Fung, K., 1997. Thermodynamic and optical properties of sea salt
aerosols, J Geophys Res Atmos., 102, 23269-23275.
Terzano, C., Di Stefano, F., Conti, V., Graziani, E., and Petroianni, A., 2010. Air pollution ultrafine
particles: toxicity beyond the lung, Eur Rev Med Pharmacol Sci., 14, 809-821.
Thorpe., A., and Harrison, R. M., 2008. Sources and properties of non-exhaust particulate matter
from road traffic: a review, Sci Total Environ., 400, 270-282.
Turpin, B.,J., Saxena, P., Andrews, E., 2000. Measuring and simulating particulate organics in the
173

atmosphere: problems and prospects. Atmos. Environ., 34: 2983-3013.
Vecchi, R., Valli, G., Fermo, P., D'Alessandro, A., Piazzalunga, A., Bernardoni, V., 2009. Organic
and inorganic sampling artefacts assessment. Atmos. Environ., 43: 1713-1720.
Venkataraman, C., 1999. Comparison of particle lung doses from the fine and coarse fractions of
urban PM-10 aerosols. Inhalation toxicology 11, 151-169.
Verma, V., Ning, Z., Cho, A. K., Schauer, J. J., Shafer, M. M., and Sioutas, C., 2009. Redox activity
of urban quasi-ultrafine particles from primary and secondary sources, Atmos. Environ., 43,
6360-6368.
Verma, V., Rico-Martinez, R., Kotra, N., King, L., Liu, J., Snell, T.W., 2012. Contribution of water-
soluble and insoluble components and their hydrophobic/hydrophilic subfractions to the
reactive oxygen species-generating potential of fine ambient aerosols. Environ Sci Technol.,
46: 11384-11392.
Verma V, Shafer MM, Schauer JJ, Sioutas C. Contribution of transition metals in the reactive
oxygen species activity of PM emissions from retrofitted heavy-duty vehicles. Atmospheric
Environment 2010; 44: 5165-5173.
Vreeland H, Schauer JJ, Russell AG, Marshall JD, Fushimi A, Jain G, et al. Chemical
characterization and toxicity of particulate matter emissions from roadside trash combustion in
urban India. Atmospheric Environment 2016; 147: 22-30.
Viana, M., Amato, F., Alastuey, A., Querol, X., Moreno, T., García Dos Santos, S., 2009. Chemical
Tracers of Particulate Emissions from Commercial Shipping. Environ Sci Technol., 43: 7472-
7477.
Viana, M., Hammingh, P., Colette, A., Querol, X., Degraeuwe, B., de Vlieger, I., 2014. Impact of
maritime transport emissions on coastal air quality in Europe. Atmospheric Environment, 90:
96-105.

Waked, A., Favez, O., Alleman, L.Y., Piot, C., Petit, J.E., Delaunay, T., Verlinden, E., Golly, B.,
Besombes, J.L., Jaffrezo, J.L., Leoz-Garziandia, E., 2014. Source apportionment of
PM₁₀ in a north-western Europe regional urban background site (Lens, France)
using positive matrix factorization and including primary biogenic emissions. Atmos. Chem.
Phys. 14, 3325-3346.
Wang, H.-C., and John, W., 1998. Characteristics of the Berner impactor for sampling inorganic
174

ions, Aerosol Science and Technology, 8, 157-172.
Wang D, Pakbin P, Shafer MM, Antkiewicz D, Schauer JJ, Sioutas C. Macrophage reactive oxygen
species activity of water-soluble and water-insoluble fractions of ambient coarse, PM2.5 and
ultrafine particulate matter (PM) in Los Angeles. Atmospheric Environment 2013; 77: 301-310.
Wang, D., Pakbin, P., Saffari, A., Shafer, M.M., Schauer, J.J., Sioutas, C., 2013. Development and
evaluation of a high-volume aerosol-into-liquid collector for fine and ultrafine particulate
matter. Aerosol Sci. Tech. 47, 1226-1238.
Wang, D., Shafer, M.M., Schauer, J.J., Sioutas, C., 2014. Development of a technology for online
measurement of total and water-soluble copper (Cu) in PM2.5. Aerosol Sci. Tech. 48, 864-874.
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.
Wang, H., Naghavi, M., Allen, C., Barber, R., Carter, A., Casey, D., Charlson, F., Chen, A., Coates,
M., Coggeshall, M., 2016. Global, regional, and national life expectancy, all-cause mortality,
and cause-specific mortality for 249 causes of death, 1980â 2015: a systematic analysis for the
Global Burden of Disease Study 2015. The Lancet 388, 1459-1544.
Wichmann, H.-E., Spix, C., Tuch, T., Wölke, G., Peters, A., Heinrich, J., Kreyling, W.G., Heyder,
J., 2000. Daily mortality and fine and ultrafine particles in Erfurt, Germany part I: role of
particle number and particle mass. Research report (Health Effects Institute), 5-86.
Yin, J., Harrison, R. M., Chen, Q., Rutter, A., and Schauer, J. J., 2010. Source apportionment of
fine particles at urban background and rural sites in the UK atmosphere, Atmos. Environ., 44,
841-851.
Ying, Q., 2011. Physical and chemical processes of wintertime secondary nitrate aerosol formation.
Front. Environ. Sci. Eng. China 5, 348-361.
Yorifuji, T., Kashima, S., Tsuda, T., Ishikawa-Takata, K., Ohta, T., Tsuruta, K.-i., and Doi, H.,
2013. Long-term exposure to traffic-related air pollution and the risk of death from
hemorrhagic stroke and lung cancer in Shizuoka, Japan, Sci Total Environ., 443, 397-402.
Zhang, X., and McMurry, P. H., 1992. Evaporative losses of fine particulate nitrates during
sampling, Atmos. Environ., 26, 3305-3312.
Zhang, X., Staimer, N., Gillen, D. L., Tjoa, T., Schauer, J. J., Shafer, M. M., Hasheminassab, S.,
Pakbin, P., Vaziri, N. D., Sioutas, C., and Delfino, R. J., 2016. Associations of oxidative stress
175

and inflammatory biomarkers with chemically-characterized air pollutant exposures in an
elderly cohort, Environ. Res., 150, 306-319.
Zhang Y, Schauer JJ, Shafer MM, Hannigan MP, Dutton SJ. Source Apportionment of in Vitro
Reactive Oxygen Species Bioassay Activity from Atmospheric Particulate Matter.
Environmental Science & Technology 2008; 42: 7502-7509.
Zheng, M., Salmon, L.G., Schauer, J.J., Zeng, L., Kiang, C.S., Zhang, Y., Cass, G.R., 2005.
Seasonal trends in PM2. 5 source contributions in Beijing, China. Atmos. Environ. 39, 3967-
3976.
Zhou, L., Kim, E., Hopke, P.K., Stanier, C.O., Pandis, S., 2004. Advanced Factor Analysis on
Pittsburgh Particle Size-Distribution Data Special Issue of Aerosol Science and Technology
on Findings from the Fine Particulate Matter Supersites Program. Aerosol Sci. Tech. 38, 118-
132.

Abstract (if available)

Abstract There is compelling evidence indicating strong associations between long/short term exposure to airborne particulate matter (PM) aerosol and increased risk of a wide range of adverse health outcomes in humans mostly attributable to PM oxidative potential (i.e., the capacity of PM species to induce cellular oxidative stress in biological systems). Despite the epidemiological evidence of PM-related health effects, the state of knowledge regarding the exact causative components and the extent to which urban aerosol source is contributing to its severity is uncertain. The research gaps in understanding the PM source-specific adverse health effects as well as the insufficient target PM criteria pollutant urge the objective research of the physico-chemical and toxicological characteristics of ambient PM aerosol. This proposed research would provide significant insight in connecting urban activity source emissions, atmospheric aerosol physicochemical characteristic, and its capability to cause detrimental health effects. This aim will essentially be used in the implementation of targeted source-specific mitigation policies to control adverse health impacts on the exposed population around the globe. ❧ The main objective of the presented dissertation is to quantify the relative impact of different urban activity source emissions on physical, chemical and toxicological properties of ambient aerosol in local and regional scales in various urban settings. To fulfil this goal, a series of time-integrated field campaigns were designed including deployment and development of state of art aerosol technologies and air monitoring methods coupled with advanced statistical and mathematical models. Physicochemical and toxicological properties of collected ambient PM aerosol demonstrated the temporal and spatial variation of important unregulated toxic PM components such as particle number (PN), redox-active metals and black carbon (BC) in contrasting complex urban environment impacted by a variety of PM sources ( i.e., Los Angeles, USA, Milan, Italy, Amsterdam, Netherlands and Tehran, Iran). The experimental and modelling approach were used to evaluate the formation mechanism of urban aerosol and their gas-particle phase partitioning in the different microenvironment around the globe and the linkages of these properties to the oxidative potential. Findings of this research advance our knowledge of complex source emission impacts on the urban aerosol toxicity and physicochemical composition in different microenvironments and provide valuable insights for more targeted and cost-effective air pollution control schemes in polluted areas around the globe.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

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...

PDF

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

PDF

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

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...

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

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

PDF

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

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

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

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

PDF

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

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

PDF

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

PDF

Influence of atmospheric processes on the physico-chemical properties of primary particles and their impact on public exposure

PDF

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

PDF

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

PDF

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

PDF

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

PDF

Personal exposure to particulate matter PM2.5 sources during pregnancy and birthweight

PDF

Beyond greenhouse gases and towards urban-scale climate mitigation: understanding the roles of black carbon aerosols and the urban heat island effect as local to regional radiative forcing agents

Asset Metadata

Creator Mousavi Nasabi Shams, Amirhosein (author)

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

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Engineering (Environmental Engineering)

Publication Date 05/03/2020

Defense Date 03/04/2020

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

Tag aerosol science,Air pollution,airborne PM,black carbon,Los Angeles,Milan metropolitan,OAI-PMH Harvest,oxidative potential,particulate matter,PM2.5,Port of Los Angeles,redox-active metals,San Pedro ports,source apportionment,statistical model,toxicity,urban pollution

Language English

Contributor Electronically uploaded by the author (provenance)

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

Creator Email a.mousavinasabi@gmail.com,amousavi@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-295263

Unique identifier UC11663925

Identifier etd-MousaviNas-8407.pdf (filename),usctheses-c89-295263 (legacy record id)

Legacy Identifier etd-MousaviNas-8407.pdf

Dmrecord 295263

Document Type Dissertation

Rights Mousavi Nasabi Shams, Amirhosein

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