University of Southern California Dissertations and Theses

From tugboats to trees: investigating the coupled systems of urban air pollution and meteorology

(USC Thesis Other)

From tugboats to trees: investigating the coupled systems of urban air pollution and meteorology

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

From tugboats to trees: investigating the coupled systems of urban air
pollution and meteorology

by

Hannah Schlaerth

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ENVIRONMENTAL ENGINEERING)

August 2023

ii
Acknowledgements

I first want to thank my partner, Christian Beneze for cheering me on over the past five
years. I’m so incredibly lucky to have someone supporting me like you do. I’d also like to thank
my family, especially my brother Jake Schlaerth and his partner Lindsay Cherry. Their phone
calls and visits always kept me connected to family while I lived in LA. Thank you, Daisy,
Bmo, and Picasso for all the licks and chirps that have been the backdrop to every zoom call, no
matter how important.
I’ve had the great fortune of having three incredible advisors during my time at USC who
were excellent mentors and outstanding people. I’d like to thank my late advisor, George Ban-
Weiss (1981 – 2021) for hiring me even after reading my twitter during the job interview. I am
so thankful that you were my advisor during the most formative years of my PhD. Thank you for
teaching me how to think and write like a scientist. Thank you to Kelly Sanders for the support
and guidance that made it possible to continue my work after losing George. Finally, thank you
Sam for all of your help in my final year. Your mentorship and perspective were pivotal to
gaining self-confidence and deciding on a career path. I couldn’t have finished this without you!
I can’t believe how many amazing scientists I’ve had the opportunity to work with during
my time at USC. I want to especially thank Yun Li for patiently answering every question I’ve
ever had about WRF. I also want to thank Joe Ko for always being a blast to work with. Thank
you to the rest of the GBW group, Mo Chen, Jiachen Zhang, Arash Mohegh, Trevor Krasowski,
and McKenna Peplinski. Thank you to the Sanders Sustainable Systems group: Stepp Mayes,
Andrew Jin, Diego Ramos Aguilera, and Zoia Comarova. Finally, it has been such a pleasure to
be part of the Atmospheric Composition and Earth Data Science group over the past year with
Kaylee Butler, Obin Sturm, William Yik, and Sahithi Nandyala.
iii
I’d also like to thank my qualifying and defense committee members: Dr. Erika Garcia,
Dr. Felipe de Barros, Dr. Constantinos Sioutas, and Dr. William Berelson. Thank you for the
time you dedicated to reviewing this work and the valuable feedback on my research.
It would not have been possible to do so much computational work without the patient
and kind staff at CARC who were always there to answer my questions and resolve technical
issues. I want to especially thank Derek Strong for helping ensure I had the computing resources
I needed to finish this work after George passed.
Thank you to my 11
th
grade English teacher, Ms. Wiseman, who taught me how to read
and write analytically. Your class was the most important and fun of my entire education.
Finally, thank you Professor Holm for taking the time to answer a very panicked email my first
semester at USC and convincing me not to drop out.

iv
Table of Contents
Acknowledgements ......................................................................................................................... ii
List of Tables ................................................................................................................................. vi
List of Figures ............................................................................................................................... vii
Chapter 1. Introduction ................................................................................................................... 1
1.1 Background ......................................................................................................................................... 1
1.2 Significance of research ...................................................................................................................... 3
1.2.1 Black carbon aerosols .................................................................................................................................. 3
1.2.2 Atmospheric effects of urban greening ........................................................................................................ 4
Chapter 2. Determining black carbon emissions and activity from in-use harbor craft in
Southern California ......................................................................................................................... 8
2.1 Introduction ......................................................................................................................................... 8
2.2 Materials and methods ...................................................................................................................... 11
2.2.1 Activity analysis ........................................................................................................................................ 11
2.2.2 Plume capture measurements .................................................................................................................... 13
2.3 Results and discussion ...................................................................................................................... 16
2.3.1 Vessel activity analysis .............................................................................................................................. 16
2.3.2 Onshore and onboard measurements ......................................................................................................... 19
2.4 Conclusions ....................................................................................................................................... 29
2.5 Funding and support ......................................................................................................................... 30
Chapter 3. Albedo as a competing warming effect of urban greening ......................................... 31
3.1 Introduction ....................................................................................................................................... 31
3.2 Material and methods ........................................................................................................................ 34
3.2.1 Model description and configuration ......................................................................................................... 34
3.2.2 Simulation domains ................................................................................................................................... 35
3.2.3 Land cover data .......................................................................................................................................... 35
3.2.4 Single Layer Urban Canopy Model ........................................................................................................... 36
3.2.5 Noah Land Surface Model ......................................................................................................................... 37
3.2.6 Simulation design ...................................................................................................................................... 39
3.3 Results ............................................................................................................................................... 41
3.3.1 Model validation ........................................................................................................................................ 41
3.3.2 Changes to physical surface properties from urban greening .................................................................... 41
3.3.3 Changes in land surface and near surface air temperatures ....................................................................... 45
3.3.4 Changes in energy fluxes ........................................................................................................................... 49
3.4 Discussion ......................................................................................................................................... 54
v
3.5 Conclusions ....................................................................................................................................... 56
3.6 Funding and support: ........................................................................................................................ 57
Chapter 4. Characterizing ozone sensitivity to urban greening in Los Angeles under current
day and future anthropogenic emissions scenarios ....................................................................... 58
4.1 Introduction ....................................................................................................................................... 58
4.2 Material and methods ........................................................................................................................ 61
4.2.1 Model description and configuration ......................................................................................................... 61
4.2.2 Physical representation of urban vegetation .............................................................................................. 64
4.2.3 Simulation design ...................................................................................................................................... 65
4.2.4 Model validation ........................................................................................................................................ 71
4.3 Results and Discussion ..................................................................................................................... 71
4.3.1 Temperature Changes ................................................................................................................................ 71
4.3.2 Changes in Biogenic Emissions ................................................................................................................. 73
4.3.3 Daily Maximum 8-Hour Ozone Concentrations ........................................................................................ 77
4.3.4 Changes in hourly O3 concentrations ........................................................................................................ 80
4.3.5 Population weighted ozone exposure ........................................................................................................ 84
4.4. Conclusions ...................................................................................................................................... 87
4.5 Funding and support: ........................................................................................................................ 89
Chapter 5: Conclusions ................................................................................................................. 90
References ..................................................................................................................................... 93
Appendices .................................................................................................................................. 108
Appendix for Chapter 2 ........................................................................................................................ 108
Appendix for Chapter 3 ........................................................................................................................ 137
Appendix for Chapter 4 ........................................................................................................................ 143

vi
List of Tables
Table 2-1: Summary of onshore vessel sampling ......................................................................... 19
Table 2-2: Average tugboat EFBC by operating mode .................................................................. 27
Table 3-1: Vegetation parameters used for urban grid cells in the Noah LSM ............................ 38
Table 3- 2: Summary of simulations ............................................................................................. 40
Table 4- 1: Summary of model simulations .................................................................................. 65

vii
List of Figures

Figure 2-1: Summary of the activity analysis using AIS data for 2017. Panel (A) shows the
activity analysis buffers and spatial extent of data coverage. Panel (B) shows the number
of operating hours spent within 24 nautical miles of California’s southern coast for each
vessel type and the proportion of those hours spent in each buffer zone, respectively.
Panel (C) is similar to (B) but shows the relative time spent in each buffer zone. Panel
(D) shows the number of mobile hours spent in each buffer zone. Panel (E) shows the
number of unique MMSI numbers (vessel identifiers) observed in UTM Zone 11 in
2017....................................................................................................................................... 18
Figure 2-2: Summary of onshore measurement results. Panel (A) shows the distribution of all
onshore measurements. Panel (B) shows the distribution of EFBC by vessel type. Boxes
represent the 25
th
and 75
th
percentiles and whiskers extend to the most extreme values
that are not considered outliers. High EFBC plumes (shown as plus signs) are statistical
outliers defined as exceeding the 75
th
percentile plus 1.5 times the interquartile range of
EFBC measured for that vessel type. ...................................................................................... 21
Figure 2-3: Average rbc (A) number and (B) mass-based size distributions before (blue) and
after (black) subtracting the average baseline rbc concentration. ......................................... 23
Figure 2-4: EFBC by operating mode for tugboats (A) and onboard measurements (B). In panel
(B), PB denotes observations made onboard the passenger boat and FB denotes
observations made onboard the fishing boat. ........................................................................ 26
Figure 2- 5: The cumulative EFBC distributions of onshore measurements for all harbor craft.
The fraction of cumulative BC emissions is shown on the y-axis and the fraction of
plume intercepts, from dirtiest to cleanest, is shown on the x-axis. ..................................... 29
viii
Figure 3-1: Surface properties for baseline land cover (left column) and changes in those
properties from increasing urban vegetation (right column). Note that Panel (c) shows
the albedo calculated by WRF for the Baseline_Albedo simulation and Panel (d) shows
the change in albedo between the GVF50_Low_Albedo and the Baseline_Albedo
simulations. The change in albedo between GVF50_High_Albedo and Baseline_Albedo
is nearly identical to that shown in Panel (d) and is shown is Figure S2 in the Appendix
for Chapter 3. ........................................................................................................................ 44
Figure 3-2: Panel (a) shows average diurnal cycle of surface temperature (TS) in urban grid
cells for the baseline simulations and panel (b) shows the change in surface temperature
for the increased GVF simulations compared to their respective baselines. Panels (c) and
(d) are the same as panels (a) and (b) but for 2m air temperature (T2). Shading denotes
the standard deviation. .......................................................................................................... 46
Figure 3-3: The average daily change in surface temperature (a-b) and the average daily
change in 2m air temperature (c-d). Results for GVF50_High_Albedo are shown in
Figure S4 in the Appendix for Chapter 3. ............................................................................. 48
Figure 3- 4: The spatially averaged diurnal cycles of the incoming (a-d) and outgoing (e-l)
energy terms from Equation 3-1 for urban grid cells. Panels on the left represent diurnal
cycles of energy fluxes for baseline land cover and panels on the right are changes in
energy fluxes modeled in the urban greening simulations compared to their respective
baselines. Shading denotes the standard deviation. .............................................................. 51
Figure 4- 1: The nested domains used in our model configuration are shown in panel (a). The
urban fraction of the innermost domain is shown in panel (b) along with the location of
major cities. ........................................................................................................................... 62
ix

Figure 4- 2: The baseline GVF of the innermost domain (a) and the change in GVF used in the
urban greening simulations (b). ............................................................................................ 68
Figure 4- 3: Panel (a) shows the mean daily change in 2m air temperature (T2) between
GVF50 and Baseline and Panel (b) shows the average diurnal cycle of 2m air
temperature in urban grid cells for the baseline and urban greening scenarios. Note that
all of the urban greening simulations had nearly identical 2m air temperature results
since aerosol effects were not included in the model configuration. Similarly, the
modeled 2m air temperature was the same for Baseline and Baseline_Albedo. .................. 72
Figure 4- 4: Panel (a) depicts the mean increase in total daily isoprene emissions for
simulations that assumed new vegetation would be low isoprene emitting (i.e.,
BVOC_Low and BVOC_Low_LA100) and Panel (b) shows the same thing but for the
high isoprene emitting simulations (BVOC_Med, BVOC_Med_LA100, BVOC_High,
and BVOC_High_La100). Panel (c) shows the mean increase in total daily monoterpene
emissions for simulations where new vegetation was assumed to be broadleaf trees (i.e.,
BVOC_Low, BVOC_Low_LA100, BVOC_Med, BVOC_Med_LA100) and Panel (d)
shows the same thing but for the needleleaf tree simulations (i.e., BVOC_High and
BVOC_High_LA100). .......................................................................................................... 75
Figure 4- 5: The average diurnal cycle of (a) isoprene emissions and (b) monoterpene
emissions in urban grid cells for each simulation. Note that biogenic emissions were the
same between the current day and future anthropogenic emissions scenarios (i.e.,
between BVOC_Low and BVOC_Low_LA100, etc). ......................................................... 76
x
Figure 4- 6: Mean changes in the DM8HO3 for each simulation. The left hand column shows
results for simulations that were run with current day anthropogenic emissions and the
right hand column show results for simulations that used future anthropogenic emissions
in the city of Los Angeles. .................................................................................................... 79
Figure 4- 7: The mean diurnal cycle of urban O3 concentrations is shown for the baseline
simulations in Panel (a) and changes in the diurnal cycle for the urban greening
simulations are shown in Panel (b). ...................................................................................... 81
Figure 4- 8: The mean diurnal cycle of O3 concentrations in nonurban grid cells is shown for
the baseline simulations in Panel (a) and changes in the diurnal cycle for the urban
greening simulations are shown in Panel (b). ....................................................................... 84
Figure 4- 9: Population weighted distribution of changes in DM8HO3. Dashed lines indicate
median changes in DM8HO3 for each simulation. Panel (a) shows the population
weighted distribution for the simulations that used current day anthropogenic emissions
and Panel (b) shows the distribution for simulations that used the aggressive
electrification anthropogenic emissions scenario. ................................................................ 86
xi
Abstract
Urban areas are warmer than their rural surroundings due to the Urban Heat Island (UHI)
effect and experience elevated air pollutant concentrations. As anthropogenic climate change
progresses and urban populations continue to grow, the UHI effect will be exacerbated and
human exposure to both extreme heat and air pollution will be affected. Thus, it is more
important now than ever before to find synergistic solutions that mitigate both urban air pollution
and urban heat. Through field measurements and atmospheric modeling, this dissertation aims to
inform effective, optimized policy that reduces heat stress and improves urban air quality. First, I
describe an extensive measurement campaign that characterizes black carbon emissions from
harbor craft, a category of vessels that operate close to shore, contributing to particulate matter
pollution in coastal communities. Black carbon is not only harmful to human health, but also
contributes to global and regional warming through direct and indirect effects. In Chapter 2, I
shift focus to UHI mitigation and use atmospheric modeling to quantify the competing warming
effects of urban greening. Here, I reveal the critical role that albedo plays in determining the
temperature effects of urban greening. Finally, in Chapter 3, I investigate O3 sensitivity to the
chemical and physical effects of urban greening under a current day and reduced anthropogenic
emissions scenario. This dissertation underscores the intricate interplay between urban
meteorology and atmospheric chemistry, revealing mitigation pathways to equitably reduce
urban heat and air pollution exposure while avoiding unintended trade-offs of individual
mitigation strategies.
1

Chapter 1. Introduction
1.1 Background
Urban areas contain more than half of the global population and simultaneously
experience poor air quality and higher temperatures than their rural surroundings, known as the
urban heat island (UHI) effect (Ulpiani, 2021; United Nations, 2018; Rosenfeld, 1998; Oke,
1982). As global mean temperatures continues to rise due to anthropogenic climate change, the
UHI effect will be exacerbated and human exposure to both extreme heat and air pollution will
be affected. Recent research suggests that combined exposure to heat and air pollution increases
premature mortality more than exposure to either heat or air pollution alone (Rahman et al.,
2022). Thus, there is an urgent need to find synergistic solutions that mitigate both urban air
pollution and urban heat.
Air pollution is intricately coupled to meteorology. For example, temperature dependent
emissions increase with rising temperature, which can increase primary and secondary air
pollutant concentrations (Halberstadt, 1989; Tingey, 1981). Additionally, increased air
temperatures can enhance ozone (O3) formation by increasing the rate of atmospheric reactions
(Carter et al., 1979). Changes in air temperature also affect the partitioning of semi-volatile
species between the gas and particle phase (Seinfeld and Pandis, 2016). Changes to ventilation
also have complex air quality effects. For example, higher near-surface air temperatures can
increase the planetary boundary layer height, thus increasing vertical mixing. Changes in wind
speed can affect turbulent mixing and influence where emissions are advected (Seinfeld and
Pandis, 2016).
2
Climate and meteorology are similarly affected by air pollution, which alters the radiative
balance on local to global scales. For example, secondary particulate matter formed from
sulphate emissions can scatter incoming shortwave radiation. Some organic aerosols like black
or brown carbon absorb incoming shortwave radiation, warming the atmosphere directly and
indirectly by affecting the formation of clouds (Bond, 2013). Finally, gaseous pollutants like
ozone can even act as short-lived greenhouse gases.
Given the intricate and implicit connection between air pollution and meteorology,
policies aimed at one have the potential to help or hinder the other. In the body of work
presented here, I address novel and policy-relevant research questions at the intersection of
regional meteorology and air pollution. In Chapter 2, a novel off-road source of BC emissions is
characterized to support policy development in California that aims to reduce human exposure to
port emissions. In Chapters 3 and 4, I explore the efficacy and potential air quality trade-offs of
increasing urban vegetation for UHI mitigation.
The overarching goal of my research is to generate quantitative insights to inform policy
that reduces urban heat and air pollution exposure. In doing so, these studies utilize a variety of
methodologies, ranging from real world air pollutant measurements to numerical weather and
atmospheric chemistry modeling. Measurements were an ideal methodology for the study
presented in Chapter 2, which aimed to characterized baseline emissions so that policy
recommendations could target the most significant emissions sources. The atmospheric
modelling used in Chapters 3 and 4 offers insight into hypothetical scenarios to bound potential
atmospheric effects of UHI mitigation. Idealized model scenarios like these are useful for
assessing the potential unintended consequences of future implementation of environmental
3
policies. My work synthesizes these distinct perspectives of real-world observations and physics-
based modeling to inform effective, optimized policy for urban heat and air pollution mitigation.

1.2 Significance of research
1.2.1 Black carbon aerosols
Black carbon (BC) is a component of particulate matter (PM) pollution that negatively
impacts air quality and human health (Sofia et al., 2020; Pope et al., 2004). BC influences global
and regional climate by absorbing shortwave radiation and through indirect effects on clouds
(Bond, 2013). Off-road emissions sources have historically contributed less to total PM than on-
road vehicles, leaving off-road emissions sources poorly characterized in the scientific literature
and less strictly regulated than on-road vehicles. In recent decades, on-road emissions sources of
PM have steadily declined following widespread adoption of emissions control technologies
(e.g., catalytic converters, selective catalytic reduction, and diesel particulate filters) (Preble et
al., 2015; Dallmann et al., 2011, 2012; Ban-Weiss et al., 2009;). It is now increasingly important
to characterize the off-road emissions sources that have not kept pace with on-road emissions
reductions.
Marine vessel emissions are relatively uncharacterized despite being significant sources
of PM emissions. Globally, marine vessel emissions are predicted to cause ~250,000
cardiovascular and lung cancer deaths and 6.4 million cases of childhood asthma annually
(Sofiev et al., 2018). Harbor craft emissions are of particular interest from a policy standpoint
since these vessels operate locally and spend most of their operating time close to shore, emitting
in proximity to populated areas.
4
In our work presented in Chapter 2, harbor craft were characterized by analyzing spatial
trends in vessel activity and black carbon emission factors were measured from a large sample of
unique vessels in the San Pedro Port Complex. Our results quantified sources of operational
variability in BC emissions from harbor craft. We found that BC emissions from harbor craft
were skewed and that vessels emitted significantly more BC when mobile than when idling. Our
results also indicated that tugboats emitted more BC than other kinds of harbor craft and spent
the most time operating in near shore areas. Finally, our work established a baseline for BC
emissions from harbor craft that was used to inform new emissions regulations for harbor craft
operating in California waters (CARB, 2021).

1.2.2 Atmospheric effects of urban greening
Urban development and environmental policies are increasingly multifaceted as cities not
only need to address emissions sources of emerging importance, like those discussed in section
1.2.1, but also grapple with the growing climate crisis. In particular, mitigating the UHI effect is
crucially important for reducing heat stress in densely populated urban areas.
A prevalent UHI mitigation strategy is called urban greening, where cities pursue
initiatives to increase urban vegetation and tree cover. Urban greening is often proposed because
vegetation has local cooling benefits of providing shade and increasing evapotranspiration
(Rosenfeld, 1998). However, vegetation also has competing warming effects such as reducing
surface albedo (i.e., increasing the fraction of incoming solar radiation that is absorbed at earth’s
surface). Prior work has focused on quantifying the cooling effects of replacing urban landcover
with vegetation rather than holistically assessing the temperature effects of realistic urban
greening implementation (e.g., Fallmann et al., 2016; Li and Norford, 2016; Vahmani and Ban-
5
Weiss, 2016b). Quantifying the competing warming effects of urban greening is critically
important for designing effective urban greening implementation.
In Chapter 3, we use atmospheric modeling to quantify and compare the albedo-induced
effects of urban greening to non-albedo effects under policy relevant urban greening scenarios.
Our results show that accounting for the decrease in albedo from urban greening changes the
daytime surface temperature signal compared to simulations where albedo is ignored. Moreover,
we show that albedo induced effects dampen the nighttime cooling benefits of urban greening.
This work is the first to reveal the critical role that albedo plays in determining the net surface
climate effects of urban greening. Finally, we make the important point that policy makers and
urban planners need to consider albedo reductions from urban greening, especially in areas like
Los Angeles where high albedo roofs and pavements are simultaneously being deployed for UHI
mitigation.
In addition to local cooling benefits, air quality improvements are often cited as co-
benefits of urban greening since leaves act as depositional surfaces for gas and particle phase
pollutants. However, the net air quality effects of vegetation are more complex than considering
depositional effects alone. Vegetation is also an emissions source of biogenic volatile organic
compounds (BVOCs), which lead to the formation of ozone (O3) in areas with high emissions of
nitrogen oxides (NOx) and secondary particulate matter (Seinfeld and Pandis, 2016; Calfapietra et
al., 2013; Guenther et al., 2006). The meteorological effects of vegetation also have the potential
to affect ventilation and temperature dependent chemical reactions that form pollutants, which
could either impair or improve air quality (Halberstadt, 1989; Carter et al., 1979). Prior work has
yet to provide a clear consensus on the potential of urban greening for air pollution abatement
(Eisenman et al., 2019; Nowak et al., 2013; Whitlow et al., 2014, Rao et al., 2014). Notably,
6
there are no findings in the public health literature that indicate that urban greening decreases
cardiovascular illness or childhood asthma (Eisenman et al., 2019).
The sensitivity of O3 concentrations to urban greening remains particularly ambiguous
given the highly nonlinear response of O3 to changes in anthropogenic NOx and biogenic VOC
emissions. Previous work has predicted that NOx reductions following renewable energy
adoption will have local air quality benefits that include reducing O3 concentrations (West et al.,
2013; Bell et al., 2008; Nemet et al., 2010). Several other studies have predicted that urban O3
concentrations will increase with increased BVOC emissions (Taha 1996; Gu et al., 2021; Ma et
al., 2019; Churkina et al., 2017). However, these topics remain largely siloed with no prior work
investigating the potential synergies or competing air quality effects of increasing biogenic
emissions while decreasing anthropogenic emissions. This is a critical knowledge gap as cities
are reducing anthropogenic emissions to address climate change while simultaneously deploying
urban greening for UHI mitigation.
We address these research gaps in Chapter 4, where we use atmospheric chemistry
modeling to investigate O3 sensitivity to urban greening in the Los Angeles Basin. This region
has a long history of proposing urban greening initiatives and has been steadily reducing
anthropogenic emissions for several decades (Pincetl et al., 2013; Yu et al., 2019; Warneke et al.,
2012; Hasheminassab et al., 2014). Our work evaluates the regional response of O3 to changes in
meteorology and BVOC emissions following a large increase in urban vegetation. We evaluate
the response of O3 concentration to urban greening under both a current day and future
anthropogenic emissions scenario, the latter of which reflects the city of Los Angeles’ aggressive
electrification and renewable energy adoption goals. Our results suggest that O3 penalties from
urban greening will persist even as anthropogenic NOx emissions are drastically reduced over the
7
coming decades following climate change mitigation. This work also sheds light on potential
environmental justice concerns as model results indicate that urban greening may increase O3
concentrations in nonurban communities that do not receive any of the cooling benefits of that
vegetation. Based on the results of this work, we recommend that policy makers and city
planners who are implementing urban greening initiatives should adopt low emitting vegetation
to avoid significant air quality trade-offs that will make it more difficult to achieve national O3
standards.

8
Chapter 2. Determining black carbon emissions and activity from
in-use harbor craft in Southern California
Published in Atmospheric Environment (Hannah Schlaerth, Joseph Ko, Rebecca Sugrue,
Chelsea Preble, George Ban-Weiss, 2021)

2.1 Introduction
Black carbon (BC) emitted from diesel fuel combustion contributes significantly to total
particulate matter (PM) emissions that negatively impact air quality and human health (Pope et
al., 2004, Campbell et al., 2005, Zanobetti and Schwartz, 2005, Kennedy, 2007, Hoek et al.,
2013, Manisalidis et al., 2020, Sofia et al., 2020). BC emissions also influence global and
regional climate by absorbing shortwave radiation and through indirect effects on clouds (Bond
et al., 2013). Off-road emissions sources adhere to less stringent emissions regulations than on-
road vehicles because off-road emissions sources have historically contributed less to total PM
emissions in urban areas than on-road emissions (Sawant et al., 2007, Lurmann et al., 2014).
Consequently, off-road emissions sources have been left relatively uncharacterized in the
scientific literature (Dallmann and Harley, 2010, Lurmann et al., 2014). In recent decades, on-
road sources of PM have been greatly reduced with the wide implementation of emission control
technologies, such as three-way catalytic converters for gasoline vehicles and selective catalytic
reduction and diesel particulate filters for diesel vehicles (Ban-Weiss et al., 2009, Dallmann et
al., 2011, Dallmann et al., 2012, Preble et al., 2015). Hence, it is increasingly important to
characterize off-road emissions, such as those from marine vessels.
9
Marine vessels are significant sources of PM emissions. A majority of global marine
vessel emissions (~70%) occur within 400 km of coastlines and ports (Corbett et al., 2007,
Fuglestvedt et al., 2009). A study by Sofiev et al. (2018) predicted that marine vessel emissions
will cause ~250,000 cardiovascular and lung cancer deaths and 6.4 million cases of childhood
asthma annually, even after the global adoption of low sulfur marine fuels (<0.5% sulfur by
weight) required by the International Maritime Organization (IMO) in 2020. PM emissions from
marine vessels also influence global and regional climate through direct and indirect effects that
are estimated to result in net negative radiative forcing (Liu et al., 2016, Turner et al., 2017).
Unlike ocean-going vessels (OGVs), harbor craft (e.g., tugboats, pilot boats, passenger
boats, fishing vessels, and crew and supply ships) operate locally out of ports and spend most of
their operating time close to shore, contributing to air pollution in populated, coastal areas.
Harbor craft emissions are of particular interest in Southern California because it is home to four
major port facilities (Figure 2-1A). This includes the Ports of Los Angeles and Long Beach,
which collectively form the busiest port in North America and are hereafter referred to as the San
Pedro Bay Port Complex (Port of Los Angeles Facts and Figures). Recent measurement
campaigns have indicated that emissions from the San Pedro Bay Port Complex significantly
impact air quality in Los Angeles County by contributing an estimated 33 ± 5% of BC emissions
from mobile sources (Mousavi et al., 2018). Port emissions may also contribute to the toxicity of
particulate matter pollution in the neighborhoods surrounding the San Pedro Bay Port Complex
by contributing 16.3% of oxidative potential of PM0.25 (Mousavi et al., 2019). Harbor craft are
therefore of regulatory interest to agencies looking to reduce port emissions.
Prior work has reported BC emission factors for marine vessels and harbor craft,
including the 2006 TexAQS and 2010 California Air Nexus (CalNex) measurement campaigns
10
(Lack et al., 2008, Buffaloe et al., 2014). In the former, BC measurement techniques were used
to measure vessel emissions near Galveston Bay and the Houston shipping channel in 2006
(Lack et al., 2008). They found that tugboats had the highest geometric mean EFBC when
compared to passenger boats and even cargo vessels (Lack et al., 2008). Following that work, the
CalNex campaign measured emissions from multiple harbor craft and likewise found that
tugboats had higher fleet average EFBC than the other harbor craft sampled (Buffaloe et al.,
2014). When compared to TexAQS, the CalNex study observed lower average EFBC, which was
largely attributed to the differences in fuel sulfur content measured in Texas in 2006 (0.4 ± 0.4
sulfur by weight for medium speed diesel vessels) compared to the low sulfur fuel (<0.1% sulfur
by weight) required in California waters (Buffaloe et al., 2014). Several other studies have
similarly found that BC emissions are lower when marine vessels burn low sulfur fuels instead of
heavy fuel oil (e.g., Khan et al., 2012, Lack et al., 2011, Lack et al., 2012) though this
relationship lacks consensus (Lack et al., 2009, Jiang et al., 2018, Yu et al., 2020). Some effort
has also been made to elucidate potential relationships between engine load and BC emissions.
Lack and Corbett (2012) suggested that BC emissions from marine vessels increase when vessels
operate outside of their engine’s rated load and a case study by Khan et al. (2012)

found that
emissions of elemental carbon (assumed approximately equal to BC) decrease as engine load
approaches rated load. Several studies used vessel speed to estimate engine load and found no
relationship between estimated load and BC emissions, indicating that there is still a need for
consensus on the operational variability of BC emissions from marine vessels (Lack et al., 2008,
Buffaloe et al., 2014, Yu et al., 2020).
The goals of this study were to characterize in-use harbor craft by evaluating spatial
trends in vessel activity and conducting onshore and onboard emissions measurements in
11
Southern California. To investigate fleet-wide and operational variability in emissions,
measured EFBC were characterized by vessel type and operating mode. By better characterizing
the sources of variability in BC emissions from harbor craft, the distribution of emissions within
the fleet, and vessel activity patterns, this work supports emissions inventory development and
provides a baseline prior to future policy change, enabling future studies to measure policy
efficacy (EPA, 2014). Moreover, this study characterizes BC emissions from a unique fleet of
harbor craft since California phased in low sulfur marine diesel starting in 2008 with fuel sulfur
limits of 0.5% sulfur by weight in 2008 and 0.1% sulfur by weight in 2012 (CARB, 2008),
whereas global fuel sulfur limits of 0.5% sulfur by weight have only recently been imposed by
the IMO in 2020 (IMO, 2019). Hence, this study may be relevant in predicting future BC
emissions from harbor craft in other regions of the world as more stringent fuel sulfur limits are
adopted.

2.2 Materials and methods
2.2.1 Activity analysis
The U.S. Coast Guard requires Automatic Identification Systems (AIS) on all self-
propelled vessels >1600 gross tons (Navigation Safety Regulations). Though developed to
improve navigational safety, AIS data can provide insight into the environmental impacts of
marine vessel traffic, as shown by Robards et al. (2016). For the present study, AIS reports were
retrieved from a NOAA database for harbor craft (tugboats, fishing boats, and passenger vessels)
and large OGV (cruise ships, cargo ships, and tankers) activity in the study area over the period
of January 1 to December 31, 2017 (see Table A.1; Marine Cadastre Data Registry). These data
were the most recently available from the archive at the time of analysis and the geographic
12
distribution of vessel activity is not expected to have changed significantly since then. Vessel
activity was quantified first by estimating the amount of time each vessel was powered on within
California waters (see Appendix A in the Appendix for Chapter 2). Activity was then divided
into time spent mobile or stationary using vessel speed and rate of turn. Mobile time included
observations where vessels were likely cruising and maneuvering whereas stationary time
included observations of vessels that were likely idle or at berth. Since the study area includes
three major port facilities (Port Hueneme, the San Pedro Bay Port Complex, and the Port of San
Diego; Figure 2-1A), stationary time for tugboats and large OGVs also includes observations of
vessels that were plugged into shore power facilities (i.e., AIS was reporting while vessels were
not emitting pollutants) (Vaishnav et al., 2016; EPA, 2017). Three buffers were constructed
following established U.S. maritime boundaries with specific legislative implications to increase
the policy relevance of the activity analysis (Figure 2-1A). State-owned tidelands extend from
the coastline to an area offshore known as the baseline, a low water line along the coast that is
approximately 3 nautical miles from shore (U.S. Maritime Limits and Boundaries). The
territorial sea extends 12 nautical miles beyond the baseline and the contiguous zone is the area
extending 12 nautical miles beyond the territorial sea. Total operating time spent in the mobile
and stationary operating modes in each buffer zone was computed for each vessel type. Details
on how operating modes and buffer zones were determined can be found in Appendix A of the
Appendix for Chapter 2. AIS data were also used to identify an on-shore sampling location with
frequent, close-to-shore harbor craft traffic at Pier F in the Port of Long Beach (see Figure B.1 in
the Appendix for Chapter 2).

13
2.2.2 Plume capture measurements
BC mass concentrations were measured at 1 Hz using an Aerosol Black Carbon Detector
(ABCD), a custom-built, filter-based light absorption photometer developed by the University of
California, Berkeley (Caubel et al., 2018, Caubel et al., 2019). BC was drawn through a sample
line made of antistatic silicon rubber tubing using an external pump with a flow rate of 110
cm
3
/min. Carbon dioxide (CO2) was measured in parallel at 1 Hz using a LI-COR LI-840A
analyzer that employs non-dispersive infrared spectroscopy (LI-840A Instruction Manual). An
external pump with a flow rate of 110 cm
3
/min drew CO2 through the sample line. The CO2
sample line was made of Teflon tubing to prevent absorption by the tubing walls (Timko et al.,
2014). During three of the ten days of onshore measurements, a single particle soot photometer
(SP2) was used in parallel with the ABCD and LI-840A to measure size-resolved refractory
black carbon (rBC) mass and number concentrations. The internal pump of the SP2 had a flow
rate of 120 cm
3
/min and the rBC sample line used antistatic silicon rubber tubing. See Appendix
D in the Appendix for Chapter 2 for additional instrumentation details.
Onshore measurements took place in the San Pedro Bay Port Complex on the following
dates: June 19, 26, and 27, 2019; July 22, 2019; August 16 and 21, 2019; November 22, 2019;
January 17 and 27, 2020 (see Table B.1 for measurement locations by date). Emissions from
passing vessels were measured from shore by suspending the sample lines ~3m high using a
pole. Vessels passed within an estimated 50 meters of the sampling site and plumes were
captured tens of seconds to ~3 minutes after the vessel passed the site. Plumes were identified in
the field using real-time CO2 measurements. For each plume, the timestamp of the observed rise
in CO2 was noted along with the vessel type and name identified through real-time AIS data
using the smart-phone application Marine Traffic. Vessel operating modes were noted as mobile
14
when vessels were cruising (i.e., moving at a steady speed), accelerating, and maneuvering. The
stationary operating mode consisted of exhaust plumes from tugboats that were docked in the
port complex with auxiliary and main engines turned on during two days of onshore sampling.
Tugboats were further designated as operating with-load when pushing barges and when tow
lines were attached to ships.
Onboard measurements were conducted onboard a commercial passenger boat used for
sight-seeing tours and onboard a charter fishing boat (i.e., a boat that transports patrons offshore
for recreational fishing) throughout typical operating conditions on July 2 and July 15, 2019,
respectively. Measurements of fresh emissions were made by holding the sample lines within 3
meters of the boats’ exhaust pipe for approximately 1 to 3 seconds and then pulling the sample
line out of the exhaust to capture a baseline concentration (see Appendix C). Observations of the
mobile operating mode included times the vessels were cruising, accelerating, and maneuvering.
Observations of the stationary operating mode included times the vessels were idle with either
auxiliary engines turned on or both auxiliary and main engines turned on. Readers should be
cognizant that the onboard measurements presented here may not be representative of other
passenger and fishing boats and did not cover all engine working conditions. Instead, the
presented data are meant to serve as case studies on the operational variability observed onboard
these two vessels.
Sampled exhaust plumes—hereby referred to as intercepts—from the onshore and
onboard measurements were analyzed using the plume capture method (Ban-Weiss et al., 2009,
Dallmann et al., 2011, Dallmann et al., 2012, Preble et al., 2015). An example plume intercept
from onshore measurements is shown in Appendix D Figure D.1. Fuel-based emission factors
15
(EFBC) were calculated by carbon balance, with units of g BC emitted per kg of fuel burned
(Equation 2-1):
EF
!"
=
∫ ([𝐵𝐶]
#
−[𝐵𝐶]
$%
)𝑑𝑡
#
!
#
"
∫ ([𝐶𝑂
&
]
#
−[𝐶𝑂
&
]
$%
)𝑑𝑡
#
!
#
"
44
12
𝑤
'

Equation 2-1

Here, [BC] is the concentration of BC in μg/m
3
, t is time, BG denotes the baseline concentration,
[CO2] is the concentration of CO2 in mg/m
3
, the factor 44/12 converts CO2 to carbon mass, and
wc is the mass fraction of carbon in diesel fuel (0.87; Ban-Weiss et al., 2009). The baseline
concentrations of CO2 and BC were determined before and after each plume by taking a 10-
second average for onshore data and a 5-second average for onboard data (since plume intercepts
occurred more frequently) and fitting a line between the two baseline points (i.e., before and after
the plume) to determine the baseline throughout the duration of each plume. The 10-second and
5-second averages for onshore and onboard data, respectively, were selected through visual
inspection of the data and were assessed to be long enough to have representative pre- and post-
plume average concentrations but short enough to not interfere with adjacent plumes. This same
method was used to baseline-subtract the rBC mass and number concentrations from the SP2,
using the 10-second average size distribution before and after each plume.
Since onshore measurements were taken downwind of passing vessels, plume dilution
caused some BC concentration measurements to be low compared to background concentrations
and relative to instrument noise. To increase the signal to noise ratio, CO2 and BC data from
onshore measurements were smoothed by taking a 40-second moving average prior to
calculating EFBC. This averaging window was selected based on analysis included in Appendix E
16
in the Appendix for Chapter 2. BC concentrations were significantly higher during onboard
measurements given the proximity of the sample inlet to the engine exhaust outlet, so a 5-second
moving average was used to smooth these data.
The LI-840A has a known sampling artifact where CO2 is overestimated when measuring
plumes with high peak CO2 concentrations, which can result in underestimated emission factors.
Thus, a correction factor of 1.24 was applied to EFBC measured onboard vessels (Appendix F).
The ABCD also has a known filter-loading sampling artifact; BC concentrations were adjusted
during post-processing prior to calculating emission factors, as described in Appendix G.

2.3 Results and discussion
2.3.1 Vessel activity analysis
Vessel activity differs geographically by vessel type, as is shown in Figures 2-1B-C.
Passenger boats spent the most time operating within 24 nautical miles of the southern coast of
California in 2017 relative to all other vessel types (418,124 hours). Tugboats had the next
highest operating time within 24 nautical miles (264,280 hours). Each category of harbor craft
(i.e., passenger, tug, and fishing boats) spent a larger proportion of their total operating time in
state-owned tidelands relative to each category of large OGVs (i.e., cruise ships, cargo ships, and
tankers) (Figure 2-1C). These results are consistent with expected activity patterns, since harbor
craft typically operate locally whereas OGVs travel large distances to make port calls.
17

18
Figure 2-1: Summary of the activity analysis using AIS data for 2017. Panel (A) shows the
activity analysis buffers and spatial extent of data coverage. Panel (B) shows the number of
operating hours spent within 24 nautical miles of California’s southern coast for each vessel type
and the proportion of those hours spent in each buffer zone, respectively. Panel (C) is similar to
(B) but shows the relative time spent in each buffer zone. Panel (D) shows the number of mobile
hours spent in each buffer zone. Panel (E) shows the number of unique MMSI numbers (vessel
identifiers) observed in UTM Zone 11 in 2017.

Figure 2-1D depicts the total time spent in the mobile operating mode, which differed
both geographically and by vessel type. Every vessel type spent the largest amount of stationary
time within state-owned tidelands, the area closest to the shore, whereas more distant buffer
zones were characterized by high proportions of time spent in the mobile operating mode. These
results reflect that vessels have logistical reasons to remain stationary while making port calls
close to shore. Tugboats spent the most mobile time in state-owned tidelands (107,540 hours)
followed by passenger boats (58,679 hours) and cargo barges (21,210 hours). Importantly,
tugboats and passenger boats spent more mobile time in state-owned tidelands than all large
OGVs combined (29,362 mobile hours). These results further establish that harbor craft are
worthwhile targets for emissions reductions given their proximity to populated areas during
mobile operating time, even when compared to large OGVs.
It should be noted that in addition to vessel operating time, fuel consumption plays an
important role in determining total BC emissions, as is shown in the methodology used by Nunes
et al. (2017). Hence, mobile operating time should not be interpreted as directly proportional to
total BC emissions from each vessel type, though previous state emissions inventories have
19
estimated near equal contributions of diesel PM10 from harbor craft and ocean-going vessels
within 100 nautical miles of California’s coast (CARB, 2018). The activity results as presented
instead indicate the geospatial distribution of BC emissions at their source (i.e., prior to
advection), which is important to consider when assessing health risk for localized pollutants like
BC.

2.3.2 Onshore and onboard measurements
2.3.2.1 Fleet average results
A total of 78 exhaust plume intercepts were captured from 42–46 unique vessels
operating in the San Pedro Bay Port Complex, as summarized in Table 2-1. Note that we report a
range of unique vessels because four tugboats were not identifiable through field observations
and were potential duplicates. The majority of measured plume intercepts (64%) were from 25–
29 unique tugboats, corresponding to more than one-third of tugboats operating out of the San
Pedro Bay Port Complex according to 2018 port vessel inventories (Port of Long Beach
Emissions Inventory, Port of Los Angeles Air Emissions Inventory).
Table 2-1: Summary of onshore vessel sampling

20
a
Tugboat average EFBC include only non-idle observations in order to be directly comparable to
the other vessel types.
b
Range of unique vessels is given because four tugboats were not identifiable.

c
The category “other” includes supply boats, work boats, and government vessels.
d
From the Port of Los Angeles Air Emissions Inventory (2018) and Port of Long Beach
Emissions Inventory (2018).

Onshore measurements were approximately lognormally distributed, as shown in Figure
2-2, and hence are reported using the geometric mean. EFBC from all onshore measurements (n=
78 intercepts) had a geometric mean ± standard deviation of 0.49 ± 0.62 g/kg. The arithmetic
mean of onshore measurements (0.69 ± 0.62 g/kg) is notably higher than that reported for the
2010 heavy-duty diesel truck fleet (arithmetic mean ± 95% confidence interval= 0.54 ± 0.07
g/kg) and lower than that reported for locomotives (0.9 ± 0.5 g/kg) (Dallmann et al., 2012,
Krasowsowky et al., 2015).
21

Figure 2-2: Summary of onshore measurement results. Panel (A) shows the distribution of all
onshore measurements. Panel (B) shows the distribution of EFBC by vessel type. Boxes represent
the 25
th
and 75
th
percentiles and whiskers extend to the most extreme values that are not
considered outliers. High EFBC plumes (shown as plus signs) are statistical outliers defined as
exceeding the 75
th
percentile plus 1.5 times the interquartile range of EFBC measured for that
vessel type.
22
The average mass and number based rBC size distributions measured for fourteen plume
intercepts are shown in Figure 2-3. A baseline-subtraction was performed on these data, but the
rBC mode diameters (Dp) observed in the number and mass size distributions were not
considerably affected, as can be observed by comparing the blue and black lines in Figure 2-3.
For the baseline-subtracted number distribution, the rBC mode Dp of our measurements ranged
from 72–118 nm, with an average of 87 nm. A similar average mode Dp of ~90 nm was
observed for rBC number for the shipping fleet measured by Buffaloe et al. (2014) prior to
applying a size-dependent correction for detection efficiency (see our Appendix H for further
discussion). For the base-line subtracted mass distribution, the rBC mode diameter of our
measurements ranged from 111–315 nm, with an average of 125 nm. The mode Dp for the rBC
mass distribution measured by Buffaloe et al. (2014) was ~180 nm prior to applying a size-
dependent correction factor.
Ko et al. (2020) measured rBC size distributions from ambient measurements on Catalina
Island (~35 km from the coast of Los Angeles ). Ko et al. (2020) reported bimodal mass and
number rBC size distributions for rBC attributed to fossil fuel combustion. The smaller of the
two modes was observed at Dp= 160 nm and Dp= 100 nm for the mass and number distributions,
respectively. The larger of the rBC mass mode (Dp= 153 nm) and number mode (Dp= 109 nm)
fall within the ranges observed in our own rBC measurements (Ko et al., 2020).

23

Figure 2-3: Average rBC (A) number and (B) mass-based size distributions before (blue) and
after (black) subtracting the average baseline rBC concentration.

2.3.2.2 EFBC by vessel type
The distribution of BC emission factors from all onshore measurements are characterized
by each vessel type in Figure 2-2B. Mean EFBC for each vessel type and across all measurements
are also reported in Table 2-1. Passenger boats (n=7 intercepts) had the highest geometric mean
EFBC of 0.56 ± 0.86 g/kg. This is higher than the geometric mean EFBC for passenger boats
measured during CalNex (0.30 ± 0.41 g/kg ; Buffaloe et al., 2014) and TexAQS (0.32 ± 0.20
24
g/kg; Lack et al., 2008). Median EFBC, which are less sensitive to outlying data points than the
geometric mean, agree well for the current study versus CalNex, which might suggest that the
current study contains more observations of high emitters. In the present study, high emitters are
defined as EFBC greater than the 75
th
percentile of all EFBC measured for that vessel type plus 1.5
times the interquartile range. Differences between the current study and TexAQS are more likely
explained by fleet composition differences, such as vessel ages, engine type, age, and rated load,
and the use of emission control technologies (e.g., selective catalytic reduction or diesel
particulate filters). Additionally, the TexAQS and CalNex studies both used photoacoustic
methods to measure BC (Lack et al., 2008, Buffaloe et al., 2014). Though Buffaloe et al. (2014)
found that photoacoustic methods showed reasonable agreement with a particle soot absorption
photometer (an instrument that operates on principles similar to those of the ABCD), differences
in instrumentation may contribute to the EFBC differences observed between the current study
and previous studies.
Tugboats had the next highest EFBC (after passenger boats) of 0.54 ± 0.56 g/kg. The
geometric mean EFBC for tugboats in TexAQS (0.74 ± 0.66; Lack et al., 2008) was considerably
higher than that of non-idling tugboat observations in the current study. This difference is
potentially due to the differences in sulfur content between fuel used in Texas at the time of
TexAQS and California, 0.4 ± 0.6% and ≤ 0.1% sulfur by weight, respectively (Lack et al., 2008;
Buffaloe et al., 2014). This assumes tugboats observed in the current study are based in the San
Pedro Bay Port Complex and therefore, burn low sulfur (<0.1% sulfur by weight) marine diesel.
This finding is consistent with previous studies, suggesting that lower fuel sulfur content is
associated with decreased BC emissions (Khan et al., 2012, Lack et al., 2011, Lack and Corbett.,
2012).
25
Vessels were required to use low sulfur (<0.1% sulfur by weight) fuels at the time of
CalNex (Buffaloe et al., 2014). The geometric mean EFBC for tugboats was somewhat lower
during CalNex (0.43 ± 0.46; Buffaloe et al., 2014) than those of non-idling tug and tow
observations from the current study. This is likely due to the small proportion of vessels moving
while actively tugging/pushing container ships and barges during CalNex (Buffaloe et al., 2014).
Indeed, the geometric mean EFBC from the current study of tugboats moving without load (0.48
g/kg; discussed in the next section) better agrees with the CalNex tugboat average (0.43 g/kg;
Buffaloe et al., 2014).
Pilot boats and fishing vessels had similar geometric mean EFBC values of 0.38 ± 0.33
and 0.36 ± 1.2 g/kg, respectively. The EFBC for pilot boats agrees well with observations made
during the CalNex study (0.33 ± 0 .28 g/kg), though fishing vessels had higher average EFBC and
more variability in the current study than in CalNex (0.22 ± 0.18 g/kg; Buffaloe et al., 2014).
Onshore observations of passenger boats and fishing vessels had higher standard
deviations than observations made onboard the passenger boat and fishing boat during the
mobile operating mode (standard deviation= 0.12 and 1.14 g/kg, respectively). This finding
suggests that the vessel-to-vessel variability observed in the onshore measurements is greater
than the operational variability observed in the onboard measurements.

2.3.2.3 EFBC by operating mode
As shown in Figure 2-4A and reported in Table 2-2, EFBC from tugboats were
categorized based on the three operating modes observed: (1) without load, when tugboats were
travelling without pushing a barge or when no tow line was attached to a ship; (2) with load,
when tugboats were pushing barges or had tow lines attached to ships; and (3) stationary, when
26
both auxiliary and main engines were on but the tugboats were stationary. Stationary
measurements were taken during the two days of dock measurements described above.

Figure 2-4: EFBC by operating mode for tugboats (A) and onboard measurements (B). In panel
(B), PB denotes observations made onboard the passenger boat and FB denotes observations
made onboard the fishing boat.
27
Table 2-2: Average tugboat EFBC by operating mode
Operating Mode
Geometric Mean BC
Emissions Factor (g/kg)
Arithmetic Mean BC
Emission Factor (g/kg)
Standard Deviation
(g/kg)
Without Load 0.48 0.71 0.67
With Load 0.64 0.70 0.29
Stationary 0.19 0.19 0.06
All Tugboat
Measurements
0.50 0.67 0.56

Geometric mean EFBC were highest when tugboats were operating with load (0.64 ± 0.29
g/kg), second highest without load (0.48 ± 0.67 g/kg), and lowest while stationary (0.19 ± 0.06
g/kg). The difference between with load and without load observations was not statistically
significant given the limited sample size of with load observations (n=16); however,
observations of the stationary operating mode were statistically significantly lower than with
load and without load observations at a 95% confidence interval despite the small sample size of
stationary observations (n=3). The largest amount of variability was observed in the without load
operating mode (n=31). This variability likely reflects both vessel-to-vessel and operating
variability, such as observations made while vessels were accelerating versus cruising.
Onboard measurements showed a similar trend. Emission factors observed for the mobile
operating mode were slightly higher than the stationary operating mode for the passenger boat.
For the fishing boat, EFBC values from the mobile operating mode were statistically significantly
higher than those of the stationary operating mode (Figure 2-4B). Moreover, the onboard
observations of the mobile operating modes for both the passenger boat and fishing boat had
28
higher standard deviations than their corresponding stationary operating modes, reflecting the
wider range of operational variability included in the former.

2.3.2.4 Skewness of EFBC from onshore measurements
Figure 2-5 illustrates the skewness of the measured fleet by ranking each plume intercept
by descending EFBC and plotting against cumulative BC emissions. EFBC from all non-idle,
onshore measurements are skewed, with the dirtiest 20% of plume intercepts contributing ~47%
of BC emissions. EFBC from fishing boats are the most skewed with the single highest EFBC
observed accounting for over 52% of BC emissions from this vessel category. For tugboat EFBC,
skewness was highest while vessels were moving without load, with the dirtiest 20% of plume
intercepts contributing nearly 50% of total BC emissions from tugboats moving without load
(see Figure I.1 in the Appendix for Chapter 2).
29

Figure 2- 5: The cumulative EFBC distributions of onshore measurements for all harbor craft.
The fraction of cumulative BC emissions is shown on the y-axis and the fraction of plume
intercepts, from dirtiest to cleanest, is shown on the x-axis.

2.4 Conclusions
The activity analysis revealed clear trends between vessel type and geographic
distribution of operating time. In general, harbor craft spent more total and mobile operating time
close to shore compared to large OGVs. This is important given that (a) EFBC values were
typically found to be higher during mobile versus stationary operation, and (b) increased
operating time near the shore is expected to result in increased population exposures to BC. The
activity analysis and reported EFBC values are two critical inputs for estimating total BC
emissions for these marine vessels, but further information about engine specifications and
differences in duty cycles is needed to estimate fuel consumption.
30
The results from this study suggest that reducing BC emissions from tugboats may have
the greatest impact on harbor craft fleet emissions with the least number of engine replacements
and/or retrofits for equipping vessels with emission control technologies. Tugboats spent the
largest amount of mobile operating time in state-owned tidelands and there were far fewer
unique tugboats operating in California waters in 2017 compared to large OGVs and passenger
boats (Figure 2-1E). Yet, tugboats had the highest median EFBC compared to all other harbor
craft observed in this study (Figure 2-2B, 0.32 ± 0.20 g/kg). This suggests that a small fraction of
tugboats contribute a relatively large amount of BC and in turn, that reducing tugboat emissions
may considerably reduce total BC emissions from harbor craft. This work supports emissions
inventory development and policy building with better characterization of BC emission
variability, fleet-wide emission distributions, and vessel activity patterns.

2.5 Funding and support
The project was funding by the California Air Resource Board under contract number
18TTD005. This material is based upon work supported by the National Science Foundation
Graduate Research Fellowship under Grant No. DGE-1752814 and DGE-1842487. Any
opinion, findings, and conclusions or recommendations expressed in this material are those of
the authors(s) and do not necessarily reflect the views of the sponsors. An extension of gratitude
to Trevor Krasowsky for helping conceptualize the project, Thomas Kirchstetter for advising this
work, and David Quiros and Wei Liu for project management.

31
Chapter 3. Albedo as a competing warming effect of urban greening
Submitted to the Journal of Geophysical Research: Atmospheres in March 2023 (Hannah
Schlaerth, Sam J. Silva, Yun Li, and Dan Li, in review)

3.1 Introduction
The urban heat island (UHI) effect describes the warmer temperatures observed in urban
areas compared to their rural surroundings (Rosenfeld et al., 1998). Anthropogenic climate
change will exacerbate the UHI effect and urbanization is rapidly increasing, making UHI
mitigation of great public health interest to reduce heat exposure in densely populated urban
areas (IPCC, 2021; United Nations, 2018).
The UHI effect is caused by the unique surface properties of urban areas, as is illustrated
by comparing the surface energy balance of urban and rural land cover. The surface energy
balance for an idealized control volume with no horizontal advection of heat can be expressed
with Equation 3-1(Wang and Li, 2021; Oke et al., 2017).

𝐴𝐻+(1−𝛼)𝑆𝑊
()
+𝜀𝐿𝑊
()
=𝑆𝐻+𝐿𝐻+𝐺+𝜀𝜎𝑇
+
,

Equation 3-1
where AH is the anthropogenic heat flux (W/m
2
), 𝛼 is albedo of the surface, SWin is incoming
shortwave radiation (W/m
2
), 𝜀 is emissivity of the surface, LWin is the incoming longwave
radiation (W/m
2
), SH is the sensible heat flux (W/m
2
), LH is the latent heat flux (W/m
2
), G is the
ground heat flux (W/m
2
), 𝜎 is the Stefan-Boltzmann constant (W/m
2
K
4
) and Ts is the land
surface temperature (K; Wang and Li, 2021). The left-hand side of Equation 3-1 represents
incoming energy to the control volume whereas the right-hand side represents different pathways
32
for energy transfer away from the control volume. In cities, the contribution of anthropogenic
heat from buildings and cars as well as the presence of low albedo, thermally massive surfaces
(i.e., low 𝛼 and high 𝜀 values) results in a large flux of incoming energy. The abundance of
impervious surface cover and a lack of vegetation makes water availability low in urban areas,
shifting outgoing energy towards higher sensible heat fluxes and lower latent heat fluxes (Oke,
1982). As a result of these physical surface properties as well as the difference in the heat
capacity of built versus natural land cover, urban areas experience warmer surface and near
surface air temperatures relative to their rural surroundings (Oke, 1982).
UHI mitigation strategies aim to favorably shift the surface energy balance by changing
the surface properties of urban areas. Urban greening and tree planting goals are increasingly
proposed for UHI mitigation because plants increase evapotranspiration and provide shade,
thereby increasing the latent heat flux and, in the case of shade, reducing incoming energy fluxes
below the plant canopy (Rosenfeld et al., 1998). Urban greening also has the co-benefits of
providing ecosystem services and reducing building energy use (Rosenfeld et al., 1998; Akbari,
2002; Akbari et al., 2001). While urban greening may increase evaporative fluxes, vegetation
also has lower albedo and higher emissivity than the bare soil it often replaces, leading to larger
fluxes of incoming shortwave and longwave radiation at the surface (Betts, 2000; Bonan, 2008).
Moreover, vegetation also has enhanced surface roughness compared to bare land cover, which
may reduce the cooling effects of wind on near surface air temperatures. Quantifying the
competing warming and cooling effects of vegetation is critically important for optimizing urban
greening for UHI mitigation.
Observational studies have offered clear consensus that areas shaded by vegetation are
cooler than areas without shade (Pincetl et al., 2013; Souch and Souch, 1993; Mcginn, 1982), but
33
few have attempted to compare the relative contributions of shading and evapotranspiration from
vegetation to the overall cooling observed

(e.g., Pincetl et al., 2013) and none consider the
competing warming effects of vegetation outlined above. Similarly, atmospheric modeling
studies have focused on quantifying the cooling effects of urban greening, such as the effects of
shading
(
Wang et al., 2018; Morakinyo et al., 2017) and evapotranspiration (Li and Liu, 2021),
without considering the potential warming effects of changes to surface properties. Though
studies like these provide valuable insight into the physical processes that contribute to cooling
from vegetation, the competing warming effects of urban greening remain largely unquantified.
Another limiting assumption of several previous works is that new vegetation would replace
existing infrastructure such as urban land cover or lawns in urban greening scenarios, which
leaves model results difficult to generalize to real world urban greening implementation (e.g.,
Fallmann et al., 2016; Li and Norford, 2016; Vahmani and Ban-Weiss, 2016b).
The albedo-induced effects of urban greening have yet to be considered in any prior
modeling studies, but recent attention has been paid to the albedo-induced effects of global
changes in vegetative land cover. Rohatyn et al (2022) quantified the carbon sequestration
potential of planting forests in drylands and compared it to the radiative forcing from
accompanying changes in albedo in those areas (Rohatyn et al., 2022). Their model results
showed that the climate mitigation potential of forestation was low after accounting for albedo
effects, only offsetting ~1% of carbon emissions under a business as usual emissions scenario
(Rohatyn et al., 2022).
In the work presented here, we aim to address the literature gaps outlined above by
quantifying and comparing the albedo-induced and non-albedo effects (e.g., emissivity, surface
roughness, and changes to evaporative fluxes) of urban greening on surface climate using
34
realistic, policy-driven scenarios. We used the Weather Research and Forecasting model (WRF)
V3.7 with chemistry to simulate a 50% increase in urban vegetation in the Los Angeles Basin, an
area with a long history of proposing urban greening initiatives (Pincetl et al., 2013). We
quantified the response of albedo, evaporative fluxes, emissivity, and surface roughness to
increased urban vegetation across the Los Angeles Basin in this realistic policy scenario. This
work offers valuable insight for policy makers, urban planners, and regional modelers by
demonstrating the critical role of albedo in determining the regional surface climate effects of
urban greening.

3.2 Material and methods
3.2.1 Model description and configuration
We simulate the response of the urban atmosphere to increasing vegetation using WRF
V3.7 coupled to Chemistry (WRF-Chem) and the Single Layer Urban Canopy Model (SLUCM).
Since the focus of this manuscript is on regional meteorology, information on chemistry
schemes, chemical boundary conditions, and the emissions datasets used can be found in the
Appendix for Chapter 3. The physics schemes used in the model were as follows: the Yonsei
University Planetary Boundary Layer Scheme (Dyer and Hicks, 1970), the MM5 surface layer
scheme (Hong et al., 2006; Paulson, 1970), the Lin et al scheme for cloud microphysics (Lin et
al., 1983), the rapid radiative transfer model longwave radiation scheme (Mlawer et al., 1997),
the Goddard shortwave radiation scheme (Chou and Suarez, 1999), and the Grell 3D convective
parameterization (Grell and Dévényi, 2002).

35
3.2.2 Simulation domains
WRF uses two-way nested domains so that coarser, parent domains can be used as
boundary conditions for the higher resolution, child domains that they contain. For the
simulations presented in this work, three two-way nested domains were used with horizontal
resolutions of 18, 6, and 2 km that were centered at 33.9 N, 118.14 W, as shown in Figure S1 in
the Supporting Information. Each domain had 29 unequally spaced, terrain-following vertical
levels from the surface to 100 hPA. The North American Regional Reanalysis (NARR) dataset
was used for initial and boundary meteorology conditions for all three domains (Mesinger et al.,
2006).

3.2.3 Land cover data
Predicting the exchange of heat, momentum, and moisture between the land and
atmosphere necessitates real world representation of land surface properties. Hence, we replaced
WRF default values for several physical properties with real time remote sensing data (Vahmani
and Ban-Weiss, 2016b). Input data for the green vegetation fraction (GVF), leaf area index, and
albedo were derived from real-time satellite observations made by the MODerate resolution
Imaging Spectroradiometer (MODIS) for the innermost domain. Raw data were retrieved from
the United States Geological Society’s Earth Explorer website and regridded to the innermost
domain following Vahmani and Ban-Weiss (2016a).
Landcover was categorized using the National Land Cover Database (NLCD) from 2006
for all three domains. This 33-category land classification dataset categorizes urban areas as low-
intensity residential, high-intensity residential, and industrial/commercial (Fry et al., 2011).
Whereas the default version of the SLUCM uses predefined urban fraction values for each urban
36
category, we replace default values with gridded urban fraction data for the innermost domain
using the 2006 NLCD imperviousness dataset (Wickham et al., 2013) following Vahmani and
Ban-Weiss (2016a). When available, a gridded dataset of urban morphology parameters (e.g.,
building heights, road widths, and roof widths) was created with the National Urban Database
and Access Portal Tool (NUDAPTS; Ching et al., 2009). Where NUDAPTS data were
unavailable, average building and road morphology for each of the aforementioned NLCD urban
categories was used from Los Angeles Region Imagery Acquisition Consortium (Wickham et al.,
2013), as was done in Zhang et al (2018).

Lastly, we use an irrigation module tuned for Southern
California that assumes that irrigation occurs three times a week at 23:00 PST in the pervious
fraction of urban grid cell (Vahmani and Hogue, 2014).

3.2.4 Single Layer Urban Canopy Model
In WRF, urban grid cells are divided into pervious and impervious land cover, the latter
of which is referred to as the urban fraction. For the urban fraction of grid cells, WRF uses the
SLUCM (Kusaka et al., 2012; Chen et al., 2011; Yang et al., 2014) to calculate the surface
energy balance between urban surfaces and the atmosphere. The SLUCM accounts for the
radiative effects of the geometry of urban areas (i.e., shading from buildings, reflection off
canyon walls, and trapped radiation) and includes anthropogenic heat fluxes. The remaining
pervious fraction of urban grid cells is composed of bare soil and vegetation. Our configuration
of WRF includes an added land cover category called Urban Vegetation for vegetative land
cover in the pervious fraction of urban grid cells.

37
3.2.5 Noah Land Surface Model
The Noah Land Surface Model (LSM) is used to calculate the surface energy balance for
pervious land cover in urban grid cells and for all non-urban grid cells. GVF, defined here as the
fraction of pervious land cover that is vegetation, is a key parameter in the Noah LSM. Our
model code is modified to convert pixel-level GVF to the GVF of the pervious fraction of urban
grid cells, hereafter referred to as the pervious-level GVF and described in Vahmani and Ban-
Weiss (2016b). GVF is used in the Noah LSM to partition total evaporation between direct
evaporation from bare soil and evaporation from the plant canopy (e.g., through transpiration or
evaporation of intercepted precipitation; Chen and Dudhia, 2001). Seasonal variability in GVF is
used to scale several other land surface parameters that include roughness length, emissivity, and
albedo. Roughness length and emissivity are scaled proportionally with increasing GVF using
Equation 3-2:

𝑉𝐸𝐺𝑃𝐴𝑅𝑀 = (1−
%-.
%-.
#$%
/%-.
#&'
)𝑉𝐸𝐺𝑃𝐴𝑅𝑀
0()
+
%-.
%-.
#$%
/%-.
#&'
𝑉𝐸𝐺𝑃𝐴𝑅𝑀
012

Equation 3-2

where VEGPARM is the calculated parameter value used in WRF, GVF is the current GVF,
GVFmin and GVFmax are gridded datasets of the yearly minimum and maximum GVF, and
VEGPARMmin and VEGPARMmax are minimum and maximum values of roughness length or
emissivity for each land cover category. The minimum and maximum parameter values for urban
land cover are shown in Table 3-1. In the default Noah LSM, albedo is scaled inversely with
increasing GVF using Equation 3-3:

38
𝐴𝑙𝑏𝑒𝑑𝑜 = (1−
%-.
%-.
#$%
/%-.
#&'
)𝑎𝑙𝑏𝑒𝑑𝑜
012
+
%-.
%-.
#$%
/%-.
#&'
𝑎𝑙𝑏𝑒𝑑𝑜
0()

Equation 3-3

Where Albedo is the calculated albedo used in WRF in the absence of gridded, real-time albedo
data and albedomin and albedomax are the yearly minimum and maximum albedo values for each
land cover category, respectively.

Table 3-1: Vegetation parameters used for urban grid cells in the Noah LSM
Vegetation
Category
Min.
Emissivity
Max.
Emissivity
Min.
Roughness
Length (m)
Max.
Roughness
Length (m)
Urban
vegetation
0.93 0.97 0.14 0.34
Urban and built-
up land
0.88 0.88 0.50 0.50

The Noah LSM has a few shortcomings that should be noted when interpreting the
results of this study. First, the Noah LSM does not include the effects of shade from trees when
calculating surface temperatures of pervious landcover and only includes effects from shade
when calculating evaporative fluxes from bare soil. Thus, the diagnostic 2m air and surface
temperatures discussed in the results reflect the temperature above the plant canopy and the
temperature of the vegetated surface, respectively, rather than temperatures below the plant
canopy (e.g., Vahmani and Banweiss, 2016a).

Additionally, by treating the urban and pervious
39
fractions separately, potential interactions between urban landcover and vegetation are not
resolved. In particular, the effect of increased vegetation on roughness length may be
overestimated since buildings would dominate surface friction in urban areas. Moreover, this
modeling framework does not allow us to explore potential effects of adding vegetation to the
urban fraction of model grid cells. Our experiments also do not investigate potential effects of
converting urban landcover to pervious landcover following urban greening. Additionally, pixel-
level albedo values are used to calculate the surface energy balance for sub pixel-level land
cover, which may not accurately reflect the individual albedo of pervious and impervious
landcover. Future work is needed to improve the representation of heterogeneous land cover
in regional atmospheric models.

3.2.6 Simulation design
To quantify and compare the albedo-induced and non-albedo effects of urban greening,
we designed two sets of simulations that are summarized in Table3-2 and described here. The
first set of simulations are the GVF50 and Baseline simulations, which we hereafter refer to as
the non-albedo simulations. The Baseline scenario used the MODIS-derived GVF described
above to represent the baseline land cover. For the urban greening scenario, GVF50, the existing
pervious-level GVF in urban grid cells was increased by a relative change of 50%. In the rare
instance where increasing GVF by 50% would surpass the total pervious fraction of grid cells,
the pervious-level GVF was set to 100%, meaning new vegetation never replaced existing urban
land cover. In GVF50 and Baseline, we turned off the option in the Noah LSM that scales albedo
with the seasonal variation in GVF and instead used the MODIS-derived albedo described above.
Hence, changes between GVF50 and Baseline reflect only the non-albedo effects (i.e., changes
to emissivity, roughness length, and evaporative fluxes) of urban greening.

40
Table 3- 2: Summary of simulations
Simulation Name GVF Albedo
Baseline MODIS MODIS
GVF50 +50% MODIS
Baseline_Albedo MODIS Calculated by the Noah LSM
GVF50_Low_Albedo +50%
Min. albedo: 0.18
Max. albedo: 0.24
GVF50_High_Albedo +50%
Min. albedo: 0.23
Max. albedo: 0.24

The second simulation set includes Baseline_Albedo, GVF50_Low_Albedo, and
GVF50_High_Albedo, which are referred to as the albedo simulations. For the albedo
simulations, we turned on the option in the Noah LSM that adjusts albedo proportionally with
seasonal variations in GVF. Accordingly, the albedo simulations include albedo-induced effects
in addition to the non-albedo effects of urban greening. The Baseline_Albedo simulation used
the MODIS-derived GVF to establish a baseline for the Noah-calculated albedo against which
the GVF50_Low_Albedo and GVF50_High_Albedo simulations could be directly compared. In
the GVF50_Low_Albedo and GVF50_High_Albedo simulations, pervious-level GVF was
increased by 50% in urban grid cells, as was done in GVF50. We tested the sensitivity of our
model results to the Albedomin value used in Equation 3-3 by increasing the look-up table value
from 0.18 to 0.23 between the GVF50_Low_Albedo and GVF50_High_Albedo simulations,
41
respectively. For both the albedo and non-albedo simulations, we assumed no additional
irrigation would be needed to sustain the newly planted vegetation given the already water
stressed region. We ran our simulations from July 1, 2012 00:00 through July 31, 2012 23:00
LST to represent a typical summer month and employed a five-day model spin-up.

3.3 Results
3.3.1 Model validation
Hourly 2m air temperature from the Baseline simulation was validated against hourly
measurements from the U.S. EPA Air Quality System network (US EPA Air Quality Systems
Data Mart). The 2m air temperature from the Baseline simulation was strongly correlated with
observations (r= 0.91; see Figure S5 in the Appendix for Chapter 3). The model tends to
underestimate 2m air temperature relative to observations, with a mean error of 2.1 K. This
model performance is consistent with previous studies that have used similar model
configurations in the same study area (Vahmani and Ban-Weiss 2016a; Vahmani and Ban-Weiss
2016b, Zhang et al., 2018; Li et al., 2019).

3.3.2 Changes to physical surface properties from urban greening
Figure 3-1 shows surface properties for the baseline landcover and the changes in those
properties from increasing GVF in the urban greening scenarios, GVF50, GVF50_Low_Albedo,
and GVF50_High_Albedo. The mean ± standard deviation of the GVF in urban grid cells was
33.4 ± 12.7% for Baseline and Baseline_Albedo (Figure 3-1a). Since the urban greening
scenarios employed a relative increase in pervious-level GVF, the increase in GVF was spatially
42
heterogeneous. Following a 50% relative increase in GVF, the mean GVF increased by 16.6 ±
6.23% in the urban greening scenarios, shown in Figure 3-1b.
43

44
Figure 3-1: Surface properties for baseline land cover (left column) and changes in those
properties from increasing urban vegetation (right column). Note that Panel (c) shows the albedo
calculated by WRF for the Baseline_Albedo simulation and Panel (d) shows the change in
albedo between the GVF50_Low_Albedo and the Baseline_Albedo simulations. The change in
albedo between GVF50_High_Albedo and Baseline_Albedo is nearly identical to that shown in
Panel (d) and is shown is Figure S2 in the Appendix for Chapter 3.

Increasing GVF considerably decreased albedo in both the GVF50_Low_Albedo and
GVF50_High_Albedo simulations relative to the Baseline_Albedo simulation, as shown in
Figures 3-1c and 3-1d. For urban grid cells in the Baseline_Albedo simulation, the mean ±
standard deviation of albedo calculated by the Noah LSM was 0.18 ± 0.05. Albedo decreased on
average by 0.06 ± 0.03 and 0.06 ± 0.04 in the GVF50_Low_Albedo and GVF50_High_Albedo
simulations, respectively. Our estimated decrease in albedo from urban greening is within the
range estimated by Rose and Levinson (2013), a remote sensing study that quantified changes in
albedo following increased vegetation in neighborhoods in Sacramento and Los Angeles, CA.
Though the albedomin value for urban vegetation increased considerably in the
GVF50_High_Albedo simulation, there was practically no change compared to the albedo
calculated for the GVF50_Low_Albedo simulation (Figure S2 in the Appendix for Chapter 3),
indicating that the calculated albedo is more sensitive to the large increase in GVF as opposed to
the lookup table albedo values.
Figures 3-1e and 3-1f show the roughness length for baseline landcover and the change in
roughness length for the urban greening scenarios. Increasing GVF by 50% increased roughness
length considerably in urban grid cells. The mean increase was 0.11 ± 0.05 m in the urban
45
greening simulations, which corresponded to a relative increase of 42.3% compared to the
roughness length of the baseline land cover. Increasing GVF did not cause as large of a change to
emissivity, as shown in Figures 3-1g and 3-1h. The mean increase in emissivity for urban grid
cells was 0.03 ± 0.01 in the urban greening simulations, which was a relative increase of 2.8%.

3.3.3 Changes in land surface and near surface air temperatures
The baseline and changes in diurnal cycles of surface and near surface air temperature in
urban grid cells are shown in Figure 3-2. Accounting for the albedo effects of urban greening
changed the direction of the daytime surface temperature signal and reduced the magnitude of
the nighttime cooling signal, as shown in Figures 3-2a and 3-2b. When albedo was left
unchanged in GVF50, mean daytime (06:00 – 19:00 LST) surface temperature cooled by 0.27 ±
0.72 K in urban grid cells. Including the albedo-induced effects of increasing GVF led to warmer
mean daytime surface temperatures of 0.70 ± 0.89 and 0.61 ± 0.93 K in urban grid cells in
GVF50_Low_Albedo and GVF50_High_Albedo, respectively. Nighttime surface temperature
cooled in urban grid cells by 0.52 ± 0.94 K in GVF50. In comparison, accounting for albedo-
induced effects reduced the nighttime cooling signal by more than 50% with mean nighttime
surface cooling of 0.23 ± 0.91 K in GVF50_Low_Albedo and cooling of 0.27 ± 0.90 K in
GVF50_High_Albedo. The large standard deviations of these surface temperature results are
partly due to the spatial heterogeneity of changes in GVF in the urban greening simulations.
46

Figure 3-2: Panel (a) shows average diurnal cycle of surface temperature (TS) in urban grid cells
for the baseline simulations and panel (b) shows the change in surface temperature for the
increased GVF simulations compared to their respective baselines. Panels (c) and (d) are the
same as panels (a) and (b) but for 2m air temperature (T2). Shading denotes the standard
deviation.

Figures 3-2c and 3-2d illustrate the average diurnal cycle of 2m air temperature in urban
grid cells for the suite of simulations. Including albedo-induced effects resulted in daytime 2m
air temperature warming of 0.33 ± 0.41 K and 0.29 ± 0.38 K in urban grid cells in
47
GVF50_Low_Albedo and GVF50_High_Albedo, respectively. Although surface temperature
slightly cooled when albedo-induced effects were neglected, daytime 2m air temperature had a
slight warming signal during the day of 0.04 ± 0.32 K in urban grid cells in GVF50. This
daytime warming is consistent with the effect of increased surface roughness reducing wind
speeds (Figure S3 in the Appendix for Chapter 3), resulting in less turbulent transport of warm,
near surface air compared to the baseline scenario. At night, accounting for the albedo-induced
effects led to a slight nighttime cooling signal in urban areas of 0.05 ± 0.46 K and 0.08 ± 0.42 K
for GVF50_Low_Albedo and GVF50_High_Albedo whereas a stronger cooling signal of 0.21 ±
0.47 K was modeled when albedo-induced effects were neglected in GVF50.
Figure 3-3 shows the spatial distribution of changes in the daily mean surface
temperature and 2m air temperature for the urban greening scenarios compared to their baselines.
Changes in surface temperature, shown in Figures 3-3a and 3-3b, were constrained to urban areas
and visually comparable in magnitude to the spatial distribution of increased GVF shown in
Figure 1a for both GVF50 and GVF50_Low_Albedo. Surface temperature changes
corresponded to small 2m air temperature changes in the albedo and non-albedo simulations, as
shown in Figures 3-3c and 3-3d. While changes in 2m air temperature followed roughly the same
spatial pattern as changes in surface temperature for the GVF50 simulation, the 2m air
temperature warming modelled in the GVF50_Low_Albedo simulation extended to some areas
outside of the urban grid cells. Notably, including the albedo-induced effects of urban greening
in GVF50_Low_Albedo reversed the mean daily surface and near surface air temperature signals
in most urban grid cells.
48

Figure 3-3: The average daily change in surface temperature (a-b) and the average daily change
in 2m air temperature (c-d). Results for GVF50_High_Albedo are shown in Figure S4 in the
Appendix for Chapter 3.

49
3.3.4 Changes in energy fluxes
Figure 3-4 shows the spatially averaged diurnal cycles of the incoming and outgoing
energy terms from Equation 1 for urban grid cells with the diurnal cycles for the energetic
contribution of incoming shortwave and longwave fluxes shown in Figures 3-4a-d. The incoming
shortwave fluxes shown in Figure 3-4a are different for Baseline and Baseline_Albedo since the
former used gridded observational albedo data and the latter calculated albedo values using GVF.
In the albedo simulations, the decreased albedo from urban greening caused more incoming
shortwave radiation to be absorbed by the surface. The mean ± standard deviation of the daytime
change in the incoming shortwave flux for GVF50_Low_Albedo and GVF50_High_Albedo was
41.01 ± 42.88 and 38.5 ± 44.82 W/m
2
respectively. Note that the wide standard deviation reflects
the spatial heterogeneity of the albedo-induced effects since GVF was increased by a relative
percent. In contrast, GVF50 had almost no change in the shortwave flux compared to the
baseline simulation. Small variations modelled during daytime hours are consistent with slight
changes in scattered shortwave radiation by aerosols between GVF50 and Baseline. Incoming
longwave fluxes changed slightly between the baseline and urban greening simulations, as
shown in Figures 4c and 4d. During the day, incoming longwave fluxes increased by 9.28 ± 7.88,
9.90 ± 8.07, and 9.58 ± 7.95 W/m
2
in the GVF50, GVF50_Low_Albedo and
GVF50_High_Albedo simulations compared to their respective baselines (Figure 3-4d). At night,
incoming longwave fluxes similarly increased by 8.94 ± 9.04, 8.99 ± 9.13, and 8.86 ± 8.87 W/m
2

for each respective simulation (Figure 3-4d). The incoming longwave fluxes are similar for the
albedo and non-albedo simulations because the incoming longwave flux is more sensitive to
changes in emissivity rather than changes in albedo, which is consistent with Equation 3-1.

50
51

Figure 3- 4: The spatially averaged diurnal cycles of the incoming (a-d) and outgoing (e-
l) energy terms from Equation 3-1 for urban grid cells. Panels on the left represent diurnal
cycles of energy fluxes for baseline land cover and panels on the right are changes in
energy fluxes modeled in the urban greening simulations compared to their respective
baselines. Shading denotes the standard deviation.

52

Figures 3-4e through 3-4h depict the spatially averaged diurnal cycles of sensible and
latent heat fluxes, which are the first two outgoing energy terms on the right-hand side of
Equation 3-1. The sensible heat flux increased only slightly during the day in GVF50 compared
to Baseline with a mean daytime increase in urban grid cells of 2.96 ± 16.55 W/m
2
. In
comparison, the sensible heat flux increased by nearly an order of magnitude more in
GVF50_Low_Albedo and GVF50_High_Albedo, with mean daytime changes of 25.62 ± 27.44
and 23.46 ± 28.35 W/m
2
, respectively. Note that the standard deviation can be large given the
spatial heterogeneity of increased GVF and the month-long simulation length. At night, the
albedo and non-albedo simulations had slightly lower sensible heat fluxes compared to their
respective baselines with average nighttime change for urban grid cells of -3.61 ± 5.71, -2.02 ±
5.06, and -2.10 ± 4.99 W/m
2
, for GVF50, GVF50_Low_Albedo, and GVF50_High Albedo,
respectively.
Increasing GVF changed evaporative fluxes in urban grid cells and is reflected by
changes in the latent heat flux as shown in Figures 3-4g and 3-4h. During the day, latent heat
fluxes increased slightly in the GVF50 simulation compared to the Baseline simulation with a
mean change of 4.62 ± 10.20 W/m
2
. Daytime latent heat fluxes increased by approximately twice
that modeled in GVF50, with GVF50_Low_Albedo increasing by 9.63 ± 11.63 W/m
2
and
GVF50_High_Albedo by 9.09 ± 11.38 W/m
2
compared to Baseline_Albedo. The increased
daytime latent heat fluxes in the albedo and non-albedo simulations reflect increased
evapotranspiration from added vegetation. The albedo simulations had higher increases in the
latent heat fluxes due to the additional effect of increased near surface air temperatures, which
promote evapotranspiration and direct evaporation from bare soil. At night, total evaporation is
53
dominated by evaporation from bare soil since vegetation is treated as photosynthetically
inactive with closed stomata. Since the urban greening simulations had less landcover made up
of bare soil, nighttime latent heat fluxes decreased. The average nighttime changes were -1.87 ±
5.71 W/m
2
in GVF50, -1.59 ± 5.06 W/m
2
in GVF50_Low_Albedo, and -1.65 ± 4.99 W/m
2
in
GVF50_High_Albedo.
Figures 3-4i through 3-4l show the baseline and changes to the diurnal cycles of the
ground heat flux and the outgoing longwave radiation flux from the right-hand side of Equation
3-1. The GVF50 simulation had minimal changes in surface temperature during the day and only
a small increase in emissivity, resulting in a slight daytime increase of 10.76 ± 5.78 W/m
2
in
outgoing longwave radiation compared to Baseline, as shown in Figures 3-4k and 3-4l. The
albedo simulations experienced increased surface temperatures and correspondingly had higher
increases in outgoing longwave radiation during the day, with average daytime increases of
16.27 ± 9.37 and 15.73 ± 9.16 W/m
2
, respectively. Though surface temperature cooled at night in
GVF50, the increased emissivity from increasing GVF led to slightly increased outgoing
longwave radiation with an average increase of 5.78 ± 6.22 W/m
2
in urban grid cells. In the
albedo simulations, nighttime surface temperatures decreased less than in the GVF50 simulation,
resulting in larger increases in outgoing longwave radiation of 8.22 ± 6.37 W/m
2
in
GVF50_Low_Albedo and 8.31 ± 6.32 w/m
2
in GVF50_High_Albedo.
During the day, energy is transferred from the atmosphere to ground storage through
conduction, shown as a positive ground flux in Figure 3-4i. Less energy was transferred to
ground storage in GVF50 compared to Baseline during daytime hours, with an average daytime
change of -8.55 ± 16.73 W/m
2
in urban grid cells. Smaller decreases were modeled in the albedo
simulations, with average changes of -4.90 ± 13.50 and -5.19 ± 8.31 W/m
2
in the
54
GVF50_Low_Albedo and GVF50_High_Albedo, respectively (Figure 3-4j). Again, note that the
large variability can be attributed in part to the spatial heterogeneity of increased GVF in the
region. At night, energy is released to the atmosphere from ground storage which is represented
as a negative flux in Figure 3-4i. The change in the nighttime ground storage terms for the albedo
and non-albedo simulations were approximately equal in magnitude and opposite in direction to
their corresponding daytime changes described above (Figure 3-4j).

3.4 Discussion
Urban greening is one of the most often proposed methods of UHI mitigation, yet little
work has been done to quantify the competing warming effects of increased vegetative cover on
surface climate. In particular, no prior work has considered the albedo-induced effects of urban
greening on surface climate. Understanding the competing warming effect of urban greening,
like that of reduced albedo, is critically important for optimizing urban greening implementation
for UHI mitigation.
In this study, we used WRF-Chem to simulate realistic, policy-driven urban greening
scenarios and quantified the albedo-induced and non-albedo effects (e.g., changes in surface
roughness, emissivity, and evaporative fluxes) of urban greening in the Los Angeles Basin. Our
model results showed that accounting for the albedo-induced effects of urban greening changed
the direction of the daytime surface temperature signal and suppressed nighttime cooling.
Interestingly, in both the albedo and non-albedo simulations, changes in surface temperature
corresponded to only slight changes in 2m air temperature despite the large increase in GVF.
The unintuitive daytime surface temperature warming modeled in the albedo simulations
is made clear when compared to the changes in the incoming and outgoing energy flux terms
55
from Equation 3-1. In the albedo simulations, the reduced albedo from urban greening led to
large increases in the incoming shortwave flux and increases in the sensible heat flux that
outweighed increases in the latent heat flux, resulting in daytime surface temperature warming
up to 0.70 ± 0.89 K in GVF50_Low_Albedo. In contrast, neglecting the albedo effects in the
non-albedo simulation led to increased latent heat fluxes that outweighed increases in the
sensible heat flux, resulting in cooler mean daytime surface temperatures in urban areas by 0.27
± 0.72 K.
Our temperature results are consistent with findings from Pincetl et al (2013), who
analyzed remote sensing data and found that irrigated grass lawns in Los Angeles, CA did not
reduce surface temperatures, suggesting that evapotranspiration did not contribute significantly
to cooling from vegetation (Pincetl et al., 2013). However, prior regional modelling studies on
urban greening predict more cooling than was modelled in both the albedo and non-albedo
simulations. Beyond differing model configurations, one likely cause for the discrepancy is the
different urban greening implementation scenarios between our studies. Fallmann et al (2016)
and Li and Norford (2016) used urban greening scenarios where vegetation replaced urban land
cover and report more near surface air cooling than was modeled in our non-albedo simulation.
In contrast, our urban greening scenarios assumed that new urban vegetation would only replace
bare soil and hence, we made no changes to the impervious fraction of urban grid cells. Our
results therefore isolate the role of increased vegetation in a manner that better aligns with how
cities will realistically increase vegetated land.
Our model results point to important directions for future work. First, our temperature
results showed that the albedo-induced effects of urban greening outweighed cooling from
increased evapotranspiration during the day, but fully capturing the net temperature effects of
56
urban greening requires a better representation of heterogenous land cover in urban areas. As
described in the methodology, WRF calculates the surface energy balance separately for
pervious and impervious land cover, which means that WRF does not capture several important
radiative effects of urban greening, such as shading of pavement from street trees or the trapping
of outgoing longwave radiation from increased canopy cover. Future work is needed to better
represent temperatures below the plant canopy and within urban canyons. Our model results also
show that large regional increases in urban vegetation have the potential to reduce albedo, which
needs to be considered as cities are simultaneously implementing high albedo roofs and
pavements for UHI mitigation. As an illustrative example, Ko et al (2022) measured the albedo
of a pilot scale implementation of high albedo pavement in a neighborhood in Covina, CA and
measured an increase in albedo compared to pre installation ranging from 0.08 to 0.26. In
comparison, the 50% increase in vegetation in our study corresponded to a mean decrease in
albedo of 0.06 ± 0.03 in GVF50_Low_Albedo. The extent to which decreases in albedo from
urban greening may counteract increases in albedo from high albedo roofs and pavements needs
to be quantified in future work.

3.5 Conclusions
In the work presented here, we used WRF-Chem to quantify the response of albedo,
evaporative fluxes, emissivity, and surface roughness to a 50% increase in urban vegetation
across the Los Angeles Basin under a realistic urban greening scenario. We analyzed our model
results in the context of the surface energy balance by looking at changes in incoming and
outgoing energy fluxes from the albedo-induced and non-albedo effects of urban greening. This
work is the first to consider changes in albedo from urban greening.
57
The contrasting daytime temperature results between the albedo and non-albedo
simulations and the reduced nighttime cooling signal in the albedo simulations highlight the
critical importance of constraining the role of albedo when assessing the net effects of urban
greening on the surface energy balance. The daytime surface warming signal in the albedo
simulations suggests that more irrigation may be needed to outweigh the competing warming
effects of urban greening, which may not be realistic in water stressed regions like Southern
California. Moreover, the dampened nighttime cooling signal in the albedo simulations may
indicate that previous modeling studies overestimated cooling from urban greening. These
albedo-induced effects of urban greening need to be carefully considered in future modeling
studies and more work is needed to investigate temperature effects below the plant canopy.
Reductions in albedo from urban greening should also be carefully considered by policy makers
and urban planners, especially as cool surfaces like high albedo roofs and pavements are
simultaneously being deployed for UHI mitigation in many cities.

3.6 Funding and support:
This work was supported by the National Science Foundation Graduate Research
Fellowship under Grant No. DGE-1842487. Any opinion, findings, and conclusions or
recommendations expressed in this material are those of the authors(s) and do not necessarily
reflect the views of the sponsors. An extension of gratitude to the late George Ban-Weiss for
advising this project in its early stages.

58
Chapter 4. Characterizing ozone sensitivity to urban greening in
Los Angeles under current day and future anthropogenic emissions
scenarios
In preparation (Hannah L. Schlaerth, Sam J. Silva, Yun Li)

4.1 Introduction
Sustainable urban development is increasingly multifaceted as cities aim to address the
growing climate crisis and locally adapt to rising temperatures. In particular, urban areas tend to
be warmer than the rural landcover surrounding them due to the urban heat island (UHI) effect,
which is expected to be exacerbated as global temperatures continue to rise due to anthropogenic
climate change (Rosenfeld et al., 1998; IPCC, 2021). Hence, cities are reducing greenhouse gas
(GHG) emissions to address climate change while simultaneously deploying UHI mitigation
strategies.
Urban greening is a popular UHI mitigation strategy, where cities propose to increase
vegetation in urban areas through landscaping and tree planting. This increased vegetation can
provide local cooling benefits by increasing shade and evaporative cooling but has ambiguous
net effects on air quality. Air quality co-benefits are often cited for urban greening since leaves
act as depositional surfaces for pollutants (e.g., Nowak et al., 2013); however, urban greening
also has additional complex effects on air quality. Vegetation emits biogenic volatile organic
compounds (BVOCs), which may increase ozone (O3) formation in areas with high
concentrations of nitrogen oxides (NOx) and lead to the formation of secondary particulate
59
matter (Seinfeld and Pandis, 2016; Calfapietra et al., 2013; Guenther et al., 2006, Sillman, 1999).
Additionally, the local cooling benefits of urban greening may reduce ozone concentrations by
slowing down temperature-dependent reactions (Carter et al., 1979). Local cooling may also
reduce temperature dependent emissions, thus reducing secondary particulate matter and O3
concentrations (Halberstadt, 1989). Another effect of cooler temperatures is a lower planetary
boundary layer height which can reduce ventilation of pollutants (Seinfeld and Pandis, 2016).
Urban tree planting can further reduce ventilation by increasing surface roughness, thus
decreasing wind speed and correspondingly, turbulent mixing and advection of pollutants.
Hence, whether urban tree planting for UHI mitigation would result in air quality co-benefits or
unintended consequences depends on highly coupled interactions between the biosphere,
meteorology, and atmospheric chemistry.
At the same time, climate change mitigation strategies like renewable energy adoption
and electrification not only reduce GHG emissions but also affect regional atmospheric
chemistry by reducing co-emitted pollutants and precursors (Bistline et al., 2022; Requia et al.,
2018; Gallagher and Holloway, 2020). Studies that characterize the air quality effects of climate
change mitigation represent a large body of literature (e.g., West et al., 2013; Bell et al., 2008;
Nemet et al., 2010 ; Cifuentes et al., 2001) and characterizing air quality impacts from urban
greening is an area of increasing research interest (e.g., Gu et al., 2021, Arghavani et al., 2019;
Eisenman et al., 2019). However, these research topics have remained largely siloed with no
studies investigating the interactions of climate change mitigation and urban greening, despite
their implementations being planned on concurrent timelines. Addressing this research gap is
critically important because tropospheric O3 forms through complex, nonlinear mechanisms that
vary with NOx and VOC concentrations (Sillman, 1999). Moreover, nighttime O3 depletion is
60
determined by NOx concentrations through O3 titration (Sillman, 1999). Thus, the magnitude and
direction of the response of O3 concentrations to climate change mitigation and urban greening
likely depends on the relative proportions of decreases in anthropogenic NOx emissions and
increases in BVOC emissions following each mitigation strategy.
There are several prior works that investigated the nonlinear response of urban O3
concentrations to increased BVOC emissions with particular emphasis on isoprene emissions.
Isoprene represents over half of global BVOC emissions followed by monoterpenes (𝛼-pinene +
β-pinene), which represent 8% of BVOC emissions (Guenther et al., 2012). Taha (1996)
modeled a 30% increase in tree cover in Los Angeles and found that the population weighted
exposure to O3 concentrations would increase if new vegetation was medium or high isoprene
emitting. Gu et al. (2021) compared BVOC emissions from urban greening scenarios to
anthropogenic VOC emissions and found that urban greening could counteract air quality
benefits from anthropogenic VOC reductions. Several other studies have found that BVOCs
contribute significantly to urban O3 during heat waves (e.g., Ma et al., 2019; Churkina et al.,
2017). On the other hand, previous studies predict O3 reductions following reduced NOx
emissions as renewable energy and electrification are adopted (West et al., 2013; Bell et al.,
2008; Nemet et al., 2010 ; Cifuentes et al., 2001) with some regional increases in urban O3
concentrations following reduced O3 titration at night (Gallagher and Holloway, 2020). These
prior works have characterized O3 responses to either increases in biogenic emissions or
decreases in anthropogenic emissions, but none investigate potential synergies or competing air
quality effects of increasing biogenic emissions while decreasing anthropogenic emissions.
In this study, we investigate O3 sensitivity to urban greening in the Los Angeles Basin, a
region with a long history of proposing urban greening initiatives that has been steadily reducing
61
anthropogenic emissions for several decades (Pincetl et al., 2013; Yu et al., 2019; Warneke et al.,
2012; Hasheminassab et al., 2014). We evaluate the regional response of O3 to changes in
meteorology and BVOC emissions from a 50% increase in urban vegetation under both a current
day and future anthropogenic emissions scenario. The future anthropogenic emissions scenario
reflects the city of Los Angeles, California’s goal of achieving 100% renewable electricity
generation with aggressive electrification by the year 2045. Our results offer critical insight for
policy makers and urban planners to help avoid O3 penalties from urban greening that may
counteract the air quality benefits of anthropogenic emissions reductions.

4.2 Material and methods

4.2.1 Model description and configuration
We simulated atmospheric chemistry on-line with meteorology using the Weather
Research and Forecasting (WRF) model V3.7 coupled to Chemistry (WRF-Chem) and the Single
Layer Urban Canopy Model (SLUCM). The physics schemes used in the model included the
Yonsei University Planetary Boundary Layer Scheme (Dyer and Hicks, 1970), the MM5 surface
layer scheme (Hong et al., 2006; Paulson, 1970), the Lin et al. scheme for cloud microphysics
(Lin et al., 1983), the rapid radiative transfer model longwave radiation scheme (Mlawer et al.,
1997), the Goddard shortwave radiation scheme (Chou and Suarez, 1999), and the Grell 3D
convective parameterization (Grell and Dévényi, 2002). The chemistry schemes used were the
Madronich TUV photolysis scheme (Madronich, 1987), the RACM-ESRL scheme for gas phase
chemistry (Kim et al., 2009), and the MADE/VBS aerosol scheme (Ackermann et al., 1998;
Ahmadov et al., 2012). Lastly, a modified version of the Wesley dry deposition scheme was used
62
for compatibility with the SLUCM (Fallmann et al., 2016). This model configuration is similar to
previous work investigating urban atmospheric chemistry in the same study area (Li et al., 2019,
Zhang et al., 2019).
WRF uses two-way nested domains such that coarser domains act as boundary conditions
for the higher resolution domains that they contain. In the presented work, we used three two-
way nested domains centered at 33.9 N, 118.14 W with horizontal resolutions of 18, 6, 2 km, as
shown in Figure 4-1a. Each domain contained 29 unequally spaced terrain following levels from
the surface to 100 hPA. The innermost domain contains the cities of Los Angeles, San Diego,
and Riverside, which are labeled in Figure 4-1b.

Figure 4- 1: The nested domains used in our model configuration are shown in panel (a). The
urban fraction of the innermost domain is shown in panel (b) along with the location of major
cities.
63

In our model configuration, WRF-Chem requires initial and boundary conditions for
meteorology and atmospheric chemistry. For all three domains, the North American Regional
Reanalysis (NARR) dataset was used for initial boundary meteorology conditions (Mesinger et
al., 2006). The Model for ozone and related chemical tracers was used to generate chemistry
boundary conditions for the outermost domain and chemistry initial conditions for all three
domains (Emmons et al., 2010).
To better represent real world land surface properties of the innermost domain, we
replaced WRF default values for green vegetation fraction (GVF), leaf area index, and albedo
with real time satellite observation made by the Moderate Resolution Imaging Spectroradiometer
(MODIS) as was done in Vahmani and Ban-Weiss (2016a). MODIS data were obtained from the
United States Geological Society’s Earth Explorer website and regridded to the innermost
domain (Vahmani and Ban-Weiss, 2016a). In all three domains, landcover was categorized using
the National Land Cover Database (NLCD). The USGS 33-category land cover classification
scheme was used, which categorizes urban areas as low-intensity residential, high-intensity
residential, and industrial/commercial (Fry et al., 2011). We replace the predefined urban
fraction values for each urban category with gridded urban fraction data for the innermost
domain using the 2006 NLCD imperviousness dataset (Wickham et al., 2013) following
Vahmani and Ban-Weiss (2016a) .We also used a gridded dataset of urban morphology
parameters that included building heights, road widths, and roof widths that was created with the
National Urban Database and Access Portal Tool (NUDAPTS; Ching et al., 2009). In areas
where NUDAPTS data were unavailable, we used the average building and road morphology
from the Los Angeles Region Imagery Acquisition Consortium (Wickham et al., 2013) for each
64
of the urban categories, as was done in Zhang et al (2018). Finally, we used the irrigation module
from Vahmani and Hogue (2014) that was tuned for Southern California and assumes irrigation
occurs 3 times weekly at 23:00 LST in the pervious fraction of urban grid cells.

4.2.2 Physical representation of urban vegetation
In our configuration of WRF, urban grid cells are divided into pervious and impervious
land cover. WRF uses the SLUCM (Kusaka et al., 2012; Chen et al., 2011; Yang et al., 2014) to
calculate the surface energy balance between impervious land cover, referred to as the urban
fraction, and the atmosphere. The SLUCM includes urban heat fluxes and radiative effects from
urban morphology, such as shading from building, reflection off canyon walls, and trapped
radiation within the urban canyon.
The surface energy balance for pervious landcover, which includes the pervious fraction
of urban grid cells and all nonurban grid cells, is calculated using the Noah Land Surface Model
(LSM). The GVF, which we define here as the fraction of pervious land cover that is vegetated,
is a key parameter in the Noah LSM that is used to partition total evaporation between direct
evaporation from bare soil and evaporation from the plant canopy (e.g., plant transpiration and
evaporation of precipitation that has been intercepted by the plant canopy; Chen and Dudhia,
2001). Seasonal variability in GVF is also used to scale land surface parameters including
roughness length and emissivity in our model configuration. It should be noted that current
regional atmospheric models do not include effects from shading of pavement by vegetation or
the trapping of outgoing radiation by the plant canopy. Hence, the surface and 2m air
temperature results presented should be interpreted as the temperature of the vegetated surface
and the temperature above the plant canopy, respectively (e.g., Vahmani and Ban-Weiss 2016b).
65
Additionally, the effect of increased vegetation on roughness length may be somewhat
overestimated since urban landcover like buildings would likely dominate surface friction and
shield the effects of added vegetation. Future work is needed to better represent heterogenous
land cover in urban areas so that potential interactions between built and natural landcover can
be resolved.

4.2.3 Simulation design
We designed a suite of simulations to quantify ozone sensitivity to physical meteorology
and BVOC emissions under current and future anthropogenic emissions scenarios, summarized
in Table 4-1. The simulations called Baseline and Baseline_LA100 used MODIS-derived GVF
to represent current day land cover and default biogenic emissions. GVF50 and GVF50_LA100
were used to isolate O3 sensitivity to the physical meteorological effects of urban greening and
are described in Section 4.2.3.1. Section 4.2.3.2 describes the BVOC simulations that were used
to test sensitivity to isoprene and monoterpene emissions. Finally, Section 4.2.3.3 describes the
anthropogenic emissions datasets that were used to represent current day emissions and those of
the future scenario with renewable energy and aggressive electrification.

Table 4- 1: Summary of model simulations
Simulation Name Landcover
Biogenic
Isoprene
Emissions
Biogenic
Monoterpene
Emissions
Anthropogenic Emissions
66
Baseline Baseline
MEGAN
defaults
MEGAN
defaults
2012 emissions inventory
(SCAQMD, 2017)
Baseline_LA100 Baseline
MEGAN
defaults
MEGAN
defaults
LA100 emissions dataset
(Heath et al., 2021)
GVF50 +50% GVF
MEGAN
defaults
MEGAN
defaults
2012 emissions inventory
(SCAQMD, 2017)
GVF50_LA100 +50% GVF
MEGAN
defaults
MEGAN
defaults
LA100 emissions dataset
(Heath et al., 2021)
BVOC_Low +50% GVF Low Low
2012 emissions inventory
(SCAQMD, 2017)
BVOC_Low_LA100 +50% GVF Low Low
LA100 emissions dataset
(Heath et al., 2021)
BVOC_Med +50% GVF High Low
2012 emissions inventory
(SCAQMD, 2017)
BVOC_Med_LA100 +50% GVF High Low
LA100 emissions dataset
(Heath et al., 2021)
BVOC_High +50% GVF High High
2012 emissions inventory
(SCAQMD, 2017)
BVOC_High_LA100 +50% GVF High High
LA100 emissions dataset
(Heath et al., 2021)

67
4.2.3.1 Urban greening scenarios
We isolated the physical meteorological effects of urban greening in GVF50 and
GVF50_LA100. In our model scenarios, we simulated urban greening as a 50% increase in GVF
in the pervious portion of urban grid cells, as shown in Figure 4-2. This means that new
vegetation replaced the bare soil fraction of urban grid cells rather than replacing existing urban
land cover. For a small number of grid cells, increasing the existing GVF by 50% would have
surpassed the total pervious fraction of grid cells, so the GVF for the pervious fraction of these
grid cells was set to 100%. We also made no changes to irrigation given the water stressed
region. Hence, the physical effects included in our simulations included changes to emissivity,
roughness length, and evaporative fluxes from urban greening. Collectively, we refer to the
simulations with increased GVF as the urban greening scenarios (i.e., GFV50, GVF50_LA100,
BVOC_Low, BVOC_Low_LA100, BVOC_Med, BVOC_Med_LA100, BVOC_High, and
BVOC_High_LA100).

68
Figure 4- 2: The baseline GVF of the innermost domain (a) and the change in GVF used in the
urban greening simulations (b).

4.2.3.2 Biogenic emissions scenarios
Biogenic emissions make up a significant portion of total VOC emissions which play an
important role in tropospheric O3 chemistry (Guenther et al., 2006). For all simulations, biogenic
emissions were calculated online with meteorology using the Model of Emissions of Gases and
Aerosols from Nature (MEGAN) version 2.06 ( Guenther et al., 2006). MEGAN uses a gridded
emissions dataset for isoprene emissions and calculates emissions of non-isoprene species of
BVOCs online. For non-isoprene species of BVOCs, MEGAN uses a look-up table of emission
factors based on the plant functional type. Then, BVOC emissions are calculated by adjusting
base emission factors to account for deviations from standard temperature, pressure, and relative
humidity. Additional adjustments are made to account for canopy loss, canopy production, and
activity (Guenther et al., 2006).
We designed three BVOC emissions scenario to investigate O3 sensitivity to isoprene and
monoterpene (here defined as a-pinene and b-pinene) emissions. Isoprene and monoterpenes are
particularly important species of BVOCs because they represent 53% and 9% of global BVOC
emissions, respectively (Guenther et al., 2012). To simulate increased BVOC emissions
following a policy relevant urban greening scenario, we changed the gridded isoprene emission
factors and plant functional type distributions that are used as inputs by MEGAN. Low and high
isoprene emission factors were estimated by cross referencing species-specific isoprene emission
factors that are used in the MEGAN 3.0 with a dataset of common California-native tree species
from CALSCAPE (CALSCAPE; MEGAN3 Data and Code, 2019). Note that our configuration
69
of WRF-Chem uses MEGAN 2.06 to calculate biogenic emissions, but here we used an input
dataset for the newer MEGAN 3.0 to calculate our isoprene emissions scenarios. The low
isoprene emission factor selected was the mean from this analysis (52.4 mol/km
2
/hr) and the high
isoprene emission factor was the mean plus one standard deviation (147.3 mol/km
2
/hr). In
comparison, the mean isoprene emission factor of the MEGAN default dataset was 26.2
mol/km
2
/hr for the urban grid cells in our innermost model domain. New grid-level isoprene
emission factors were calculated by taking a weighted average between the old grid-level
isoprene emission factor and the portion of each grid cell with the new high or low isoprene
emission factor. In other words, we adjusted the grid-level isoprene emission factors by the
added emissions from the portion of the grid cell occupied by new vegetation.
The distribution of plant functional types that are used in the calculation of non-isoprene
species of BVOCs was scaled in the same way as was described for isoprene emissions. In
general, the broadleaf tree (BT) plant functional type used in MEGAN uses low monoterpene
(i.e., ⍺ -pinene and β -pinene) emission factors and the needleleaf tree (NT) vegetation type used
in MEGAN uses high monoterpene emission factors. The low isoprene emission factor and the
BT vegetation type was used for the low BVOC emissions scenario (BVOC_Low and
BVOC_Low_LA100), the high isoprene emission factor and the BT vegetation type was used
for the moderate BVOC emissions scenario (BVOC_Med and BVOC_Med_LA100), and the
high isoprene emission factor and NT vegetation type was used for the most extreme BVOC
emission scenario (BVOC_High and BVOC_High_LA100). The gridded isoprene emission
factors and plant functional type distributions for each BVOC scenario are shown in Figure S1
and Figure S2 in the Appendix for Chapter 4.

70
4.2.3.3 Anthropogenic emissions scenarios
In our configuration of WRF-Chem, gridded anthropogenic emissions datasets are
required input files for each domain. For the outer two domains of all simulations, an emissions
inventory from the California Air Resource Board (CARB) was used for areas within California
and the National Emissions Inventory (NEI) from the U.S. Environmental Protection Agency
(EPA) was used for areas outside of California that were still within the simulation domains
(CARB; US EPA, 2014). The CARB emissions inventory contained hourly emissions for the
year 2012 at 4-km resolution and represents the most up-to-date emissions data set available at
the time of analysis.
In the innermost domain for the current day anthropogenic emissions scenario, we used a
gridded emissions dataset from the South Coast Air Quality Management District (SCAQMD)
that represents anthropogenic emissions for the year 2012 (SCAQMD, 2017). For the future
anthropogenic emissions scenario, we used an emissions inventory generated during the Los
Angeles 100% Renewable Energy Study (Cochran and Denholm, 2021). Simulations that used
the future anthropogenic emissions dataset are denoted with “LA100” at the end of the
simulation name. Details on the methodology used to generate the emissions inventory can be
found in Chapter 9 of the study report (Heath et al., 2021). The future anthropogenic emissions
scenario represents emissions projections for the year 2045 where the city of LA has achieved
100% renewable electricity generation with aggressive electrification of light-duty vehicles,
buses, commercial and residential buildings, and the ports of Los Angeles and Long Beach. In
the future anthropogenic emissions scenario, daily average NOx emissions in the city of LA
decreased by 63% relative to those in the current day anthropogenic emissions, as described in
Heath et al (2021). The emissions inventories described above were regridded to match the
71
model domains and chemical speciation was converted to align with the RACM-ESRL and
MADE/VBS mechanisms following Li et al. (2019).

4.2.4 Model validation
Hourly 2m air temperature, hourly O3 concentrations, and daily maximum 8-hour O3
(DM8HO3) concentrations from the Baseline simulation were validated against measurements
from the US EPA Air Quality System network (US EPA Air Quality Systems Data Mart). The
2m air temperature from the Baseline simulation was strongly correlated (r = 0.91) with
observations and had a mean error of 2.1 K, indicating that the model tends to underestimate 2m
air temperature (see Figure S3 in the Appendix for Chapter 4). The model somewhat
underestimated hourly O3 concentrations (see Figure S4 in the Appendix for Chapter 4) and
DM8HO3. Hourly O3 concentrations had a correlation coefficient of r = 0.75 and DM8HO3 had a
correlation coefficient of r = 0.65, which are both within the range of acceptable model
benchmarks (Emery et al., 2017; See Table S1 in the Appendix for Chapter 4). These model
performance results are consistent with prior work in the same study area that used similar model
configurations (e.g., Vahmani and Ban-Weiss 2016a; Vahmani and Ban-Weiss 2016b, Zhang et
al., 2018; Li et al., 2019, Zhang et al., 2019).

4.3 Results and Discussion
4.3.1 Temperature Changes
Temperature changes in the urban greening simulations were modest despite the large
increase in vegetative cover. Daily mean 2m air temperature cooled by 0.08 ± 0.42 K in urban
grid cells in GVF50 compared to the baseline simulation, as shown in Figure 4-3a. The modeled
72
temperature changes were the same up to three significant figures for the other urban greening
simulations, which are shown in Fig S5 in the Appendix for Chapter 4. Figure 4-3b shows the
average diurnal cycle of 2m air temperature in urban grid cells for each simulation, which were
also approximately the same for the other urban greening simulations compared to their
respective baselines. During the day, 2m air temperature slightly warmed in the urban greening
simulations with a mean increase of 0.04 ± 0.32 K in urban grid cells. This unintuitive daytime
warming is consistent with the effect of increased surface roughness reducing wind speeds
during the day (see Figure S6 in the Appendix for Chapter 4), thus resulting in less transport of
warm, near surface air. At night, urban areas cooled with a mean decrease in 2m air temperature
of 0.21 ± 0.47 K. Changes in 2m air temperature were largely constrained to urban areas and
visually consistent with the increased GVF (see Figure 4-2). Changes in near surface air
temperature were near zero for nonurban grid cells.

Figure 4- 3: Panel (a) shows the mean daily change in 2m air temperature (T2) between GVF50
and Baseline and Panel (b) shows the average diurnal cycle of 2m air temperature in urban grid
cells for the baseline and urban greening scenarios. Note that all of the urban greening
simulations had nearly identical 2m air temperature results since aerosol effects were not
73
included in the model configuration. Similarly, the modeled 2m air temperature was the same for
Baseline and Baseline_Albedo.

4.3.2 Changes in Biogenic Emissions
The changes in total daily isoprene emissions that were modeled in the urban greening
scenarios are summarized in Figure 4-4a and 4-4b and discussed in this section. When the
meteorological effects of urban greening were isolated (i.e., in GVF50 and GVF50_LA100)
daily total isoprene emissions had a near zero change compared to their respective baseline
simulations (i.e., Baseline and Baseline_LA100). This is consistent with expectations because
the temperature changes from urban greening were small, as discussed in Section 4.3.1.
Increasing the gridded isoprene emission factors, as described in Section 4.2.6.3, resulted in large
increases in total daily isoprene emissions in urban grid cells. When the low isoprene emission
factor was used for new vegetation (i.e., in BVOC_Low and BVOC_Low_LA100), mean total
daily isoprene emissions increased by 40.16 ± 29.6 mol/km
2
, which corresponds to a relative
increase of 35%. When the high isoprene emission factor was used for the new vegetation (i.e.,
in BVOC_Med, BVOC_Med_LA100, BVOC_High, and BVOC_High_LA100), total daily
isoprene emissions increased by 140.0 ± 102.2mol/km
2
(122% relative to baseline isoprene
emissions). Considering the small temperature changes modeled in the urban greening scenarios,
the increased isoprene emissions in the BVOC scenarios were driven by the scaled emission
factors rather than the meteorological effects of urban greening.
The mean change in total daily monoterpene emissions are shown in Figure 4-4c and 4-d.
Matching the results for isoprene emissions, the modest meteorological effects of urban greening
caused a near zero change in monoterpene emissions compared to the baseline simulations.
74
Whereas isoprene emissions increased proportionally to the increased GVF, total daily
monoterpene emissions increased proportionally with changes to the plant functional type
distribution, as was described in the Section 4.2.3.3. Total daily monoterpene emissions
increased by 1.9 ± 1.4 mol/km
2
(45% relative to baseline biogenic emissions) when new
vegetation was assumed to be broadleaf (i.e., in the BVOC_Low and BVOC_Med simulations)
and by 7.9 ± 5.0 mol/km
2
(183% relative to baseline biogenic emissions) when new vegetation
was assumed to be needleleaf trees (i.e., in the BVOC_High simulations). Again, the small
temperature changes modeled in the urban greening scenarios had little effect on monoterpene
emissions.

75

Figure 4- 4: Panel (a) depicts the mean increase in total daily isoprene emissions for simulations
that assumed new vegetation would be low isoprene emitting (i.e., BVOC_Low and
BVOC_Low_LA100) and Panel (b) shows the same thing but for the high isoprene emitting
simulations (BVOC_Med, BVOC_Med_LA100, BVOC_High, and BVOC_High_La100). Panel
(c) shows the mean increase in total daily monoterpene emissions for simulations where new
vegetation was assumed to be broadleaf trees (i.e., BVOC_Low, BVOC_Low_LA100,
BVOC_Med, BVOC_Med_LA100) and Panel (d) shows the same thing but for the needleleaf
tree simulations (i.e., BVOC_High and BVOC_High_LA100).
76

Figure 4-5 shows the mean diurnal cycle of isoprene and monoterpene emissions in urban
grid cells for each simulation. For both isoprene and monoterpenes, the largest increases in
emissions took take place during daytime hours when vegetation is treated as photosynthetically
active in the model configuration. Whereas biogenic isoprene emissions are zero during
nighttime hours (21:00 – 05:00), vegetation still emits small amounts of monoterpene emissions
throughout the night. Increases in monoterpene emissions were greatest in magnitude during
daytime hours and relative increases were approximately constant throughout the diurnal cycle.

Figure 4- 5: The average diurnal cycle of (a) isoprene emissions and (b) monoterpene emissions
in urban grid cells for each simulation. Note that biogenic emissions were the same between the
current day and future anthropogenic emissions scenarios (i.e., between BVOC_Low and
BVOC_Low_LA100, etc).
77

4.3.3 Daily Maximum 8-Hour Ozone Concentrations
The mean changes in the DM8HO3 are shown for each simulation in Figure 4-6. The
reduced ventilation and slight daytime warming from urban greening led to small increases in
DM8HO3 throughout the Los Angeles basin in the simulations that isolated the meteorological
effects of urban greening (i.e., in GVF50 and GVF50_LA100). In the GVF50 simulation, the
mean increase in DM8HO3 was 0.07 ± 0.16 ppb in both urban and non-urban grid cells. In
GVF50_LA100, the meteorological effects of urban greening led to still smaller increases in
DM8HO3 with mean changes of 0.05 ± 0.13 and 0.05 ± 0.16 ppb in urban and non-urban grid
cells, respectively.

78

79

Figure 4- 6: Mean changes in the DM8HO3 for each simulation. The left hand column shows
results for simulations that were run with current day anthropogenic emissions and the right hand
column show results for simulations that used future anthropogenic emissions in the city of Los
Angeles.

Under both the current and future anthropogenic emissions scenarios, increasing biogenic
emissions increased DM8HO3 with larger changes modeled under the current anthropogenic
emissions scenario. Notably, in all of the biogenic emissions scenarios, the increase in BVOC
emissions in urban grid cells resulted in regional increases in DM8HO3 that extended beyond the
urban areas with increased vegetation. Spatially, increases in DM8HO3 were greatest in inland
urban areas downwind of Los Angeles. DM8HO3 concentrations were sensitive to isoprene
emissions, with more sensitivity modeled under the current anthropogenic emissions scenario
than was modeled in the aggressive electrification scenario with reduced NOx emissions. In
BVOC_Low, where the new vegetation was assumed to be low isoprene emitting, DM8HO3
increased by 0.47 ± 0.44 ppb in urban grid cells and by 0.33 ± 0.29 ppb in nonurban grid cells. In
contrast, DM8HO3 increased by 0.32 ± 0.30 and by 0.18 ± 0.23 ppb in urban and nonurban grid
cells under the future anthropogenic emissions scenario (BVOC_Low_LA100). When new
vegetation was high isoprene emitting, the increase in DM8HO3 more than doubled in urban and
nonurban areas, with average increases of 1.25 ± 1.11 and 0.84 ± 0.29 ppb for urban and
nonurban grid cells respectively in BVOC_Med. This sensitivity to isoprene emissions was also
modeled under the future anthropogenic emissions scenario with mean increases of 0.84 ± 0.74
80
ppb in urban grid cells and increases of 0.42 ± 0.48 ppb in nonurban grid cells in
BVOC_Med_LA100.
DM8HO3 increased slightly with increased monoterpene emissions, which is shown by
comparing the medium and high BVOC emissions scenarios in Figure 4-6 (i.e., BVOC_Med and
BVOC_High). In the current day anthropogenic emissions scenario (BVOC_High), the mean
increase in DM8HO3 was 1.37 ± 1.22 and 0.92 ± 0.76 ppb in urban and nonurban areas,
respectively. Increases in DM8HO3 from increasing monoterpene emissions were smaller under
the aggressive electrification scenario than in the current day anthropogenic emissions scenario.
In BVOC_High_LA100, DM8HO3 increased by 0.92 ± 0.82 ppb in urban grid cells and by 0.45
± 0.52 ppb in nonurban grid cells.

4.3.4 Changes in hourly O3 concentrations
Figure 4-7 illustrates the diurnal cycle of urban O3 concentrations for each of the baseline
simulations and the changes in urban O3 modeled for the different BVOC emissions scenarios.
During the day, the meteorological effects of urban greening led to slight increases in urban O3
concentrations with mean increases of 0.10 ± 0.17 and 0.13 ± 0.17 ppb modeled in GVF50 and
GVF50_LA100, respectively. As biogenic emissions increased, daytime urban O3 concentrations
also increased with the most sensitivity modeled in response to increased isoprene emissions
(i.e., comparing the blue and yellow lines in Figure 4-7b) as opposed to monoterpene emissions
(i.e., comparing the yellow and red lines in Figure 4-7b). Comparing the dashed lines to the solid
lines in Figure 7b illustrates the flattened O3 response to increased BVOC emissions in the
aggressive electrification scenario. Daytime urban O3 concentrations were highest in the urban
greening simulations with current day anthropogenic emissions. In urban grid cells, the mean
81
daytime increase in O3 concentrations was 0.36 ± 0.35, 0.86 ± 0.75 and 0.95 ± 0.82 ppb for the
low, moderate, and high BVOC emissions scenarios (i.e., BVOC_Low, BVOC_Med, and
BVOC_High). In comparison, increases in daytime urban O3 concentrations were 0.32 ± 0.30,
0.67 ± 0.57, and 0.73 ± 0.63 ppb for the low, moderate, and high BVOC emissions scenarios
with future anthropogenic emissions (i.e., BVOC_Low_LA100, BVOC_Med_LA100,
BVOC_High_LA100). These results suggest that urban areas will transition away from VOC-
limited O3 formation following reduced NOx emissions in the city of Los Angeles in the
aggressive electrification scenario.

Figure 4- 7: The mean diurnal cycle of urban O3 concentrations is shown for the baseline
simulations in Panel (a) and changes in the diurnal cycle for the urban greening simulations are
shown in Panel (b).

82
At night, urban greening somewhat reduced urban O3 concentrations in both the current
day and future anthropogenic emissions scenarios. The spatial distribution of changes in
nighttime O3 concentrations are shown in the Appendix for Chapter 4 in Figure S7. Nighttime
reductions in urban O3 concentrations were likely driven by the meteorological effects of urban
greening with the greatest decreases modeled in GVF50. The mean decrease in nighttime urban
O3 concentrations was 0.41 ± 0.47 ppb in GVF50. Increasing BVOC emissions in BVOC_Low,
BVOC_Med, and BVOC_High somewhat counteracted the O3 improvements from the
meteorological effects of urban greening with respective mean changes of -0.33 ± 0.49, -0.22 ±
0.55, and -0.19 ± 0.57 ppb. Nighttime urban O3 concentrations were higher in the urban greening
simulations that used the aggressive electrification scenario with mean changes of -0.14 ± 0.36, -
0.14 ± 0.36, -0.09 ± 0.37, and 0.00 ± 0.43 ppb, modeled in GVF50_LA100,
BVOC_Low_LA100, BVOC_Med_LA100, and BVOC_High_LA100. Notably, nighttime O3
reductions were smaller than daytime O3 penalties from urban greening for both the current and
future anthropogenic emissions scenarios.
These nighttime urban O3 trends are consistent with increased urban vegetation reducing
ventilation, resulting in enhanced NOx concentrations and hence increased O3 titration in urban
grid cells. Indeed, under the current day anthropogenic emissions scenario, nighttime increases in
NOx concentrations were strongly correlated with decreases in O3 concentrations (see Figure S8
in the Appendix for Chapter 4). While reducing NOx emissions in the aggressive electrification
scenario lessened O3 penalties from urban greening during the day, NOx reductions also curtailed
nighttime O3 benefits of urban greening. Since NOx emissions were drastically reduced in the
aggressive electrification scenario, less O3 titration took place compared to the current day
83
anthropogenic emissions scenario, resulting in smaller changes in nighttime O3 concentrations in
urban areas.
The spatially averaged diurnal cycle of hourly O3 concentrations for non-urban grid cells
is shown in Figure 4-8. Unlike the trends modeled in urban grid cells, nonurban grid cells had
enhanced O3 concentrations throughout the day and night. Daily maximum O3 concentrations
occurred later in the day for nonurban grid cells than for the urban grid cells, reflecting the
transportation of pollutants to inland areas from urban, coastal areas. In GVF50 and
GVF50_LA100, mean daytime O3 concentrations respectively increased by 0.07 ± 0.13 and 0.06
± 0.12 ppb, suggesting that the meteorological effects of urban greening caused slight O3
penalties downwind of new vegetation. Similar to the results modeled in urban grid cells,
daytime O3 concentrations increased in nonurban grid cells following increased BVOC
emissions. Again, daytime O3 concentrations were less sensitive to the increased BVOC
emissions under the aggressive electrification scenario, suggesting a migration away from the
NOx saturated regime following reduced NOx emissions in Los Angeles. At night, O3
concentrations increased with increasing BVOC emissions, which may be due to reduced
advection of NOx to inland areas (see Figure S9 in the Appendix for Chapter 4) causing reduced
nighttime O3 titration. These results indicate that the O3 penalties of urban greening extend
beyond the local scale, impacting the nonurban areas that notably do not receive the local cooling
benefits of that vegetation.

84

Figure 4- 8: The mean diurnal cycle of O3 concentrations in nonurban grid cells is shown for the
baseline simulations in Panel (a) and changes in the diurnal cycle for the urban greening
simulations are shown in Panel (b).

4.3.5 Population weighted ozone exposure
Since changes in O3 concentrations were spatially heterogeneous throughout the basin,
we investigated the changes in population weighted DM8HO3 using 2021 population data from
the US Census Bureau as a first approximation of changes in exposure to O3 (data.census.gov).
For each simulation, changes in DM8HO3 were aggregated to the census tract level using the
methodology described in S5 of the Supplementary Information.
Population weighted changes in DM8HO3 are shown in Figure 4-9. In general, the
population weighted distributions in changes in DM8HO3 skewed towards higher values under
the current day anthropogenic emissions scenarios. In GVF50 and GVF50_LA100, changes in
85
DM8HO3 were approximately normally distributed around 0 ppb. In GVF50, 66.3% of the
population experienced increases in DM8HO3, with increases greater than or equal to 0.25 ppb
affecting only 12.0% of the total population. Similarly, 69.1% of the population would have
increased DM8HO3 in GVF50_LA100, with only 3.4% of the population experiencing increases
over 0.25 ppb. Interestingly, the results for GVF50 and GVF50_LA100 both suggest that even
when urban greening has minimal effects on biogenic emissions, most of the population would
not experience any change in O3 exposure. As BVOC emissions increased, the population
weighted distribution skewed towards higher increases in DM8HO3. When the new vegetation
was low isoprene emitting, 37.5% and 23.3% of the population would experience increases in
DM8HO3 over 0.5 ppb in BVOC_Low and BVOC_Low_LA100, respectively. Similarly, when
vegetation was assumed high isoprene emitting, DM8HO3 increased by more than 0.5 ppb for
69.5% and 58.3% in BVOC_Med and BVOC_Med_LA100, respectively. The population
weighted distribution increased slightly with monoterpene emissions, with 71.8% and 60.2% of
the population experiences increases over 0.5 ppb for BVOC_High and BVOC_High_LA100.
These results suggest that urban greening would at best have a neutral effect on O3 exposure and
at worse would increase DM8HO3 concentrations for a majority of the population in the study
domain.

86

Figure 4- 9: Population weighted distribution of changes in DM8HO3. Dashed lines indicate
median changes in DM8HO3 for each simulation. Panel (a) shows the population weighted
distribution for the simulations that used current day anthropogenic emissions and Panel (b)
shows the distribution for simulations that used the aggressive electrification anthropogenic
emissions scenario.
87

4.4. Conclusions
In this study, we investigated O3 sensitivity to urban greening under a current day and
future anthropogenic emissions scenario with drastically reduced NOx emissions. We used WRF-
Chem to simulate O3 sensitivity to urban greening scenarios during a typical summer month in
the Los Angeles Basin. We isolated the dynamic effect of urban greening on O3 concentrations
by increasing urban vegetation by 50% and making no direct changes to biogenic emissions in
GVF50 and GVF50_LA100. Then, we tested O3 sensitivity to increased isoprene and
monoterpene emissions using low, moderate, and high BVOC emissions scenarios. We explicitly
evaluated potential interactions between urban greening and anthropogenic emissions reductions
by running our urban greening scenarios with current day anthropogenic emissions and future
anthropogenic emissions that represent the city of Los Angeles’ goal of achieving 100%
renewable energy with aggressive electrification by the year 2045.
Our model results indicate that urban greening will likely make it harder to achieve
national DM8HO3 standards regardless of biogenic and anthropogenic emissions scenarios. The
meteorological effects of urban greening led to small increases in urban and nonurban DM8HO3
concentrations. Increases in biogenic emissions led to increases in DM8HO3 as high as 1.37 ppb
in urban grid cells and 0.92 ppb in nonurban grid cells when new vegetation was high isoprene
and monoterpene emitting. The O3 penalties of urban greening were somewhat mitigated under
the aggressive electrification scenario, suggesting that anthropogenic emissions reductions will
shift O3 production away from the NOx-saturated regime. Our results still suggest however that
urban greening may counteract air quality benefits of anthropogenic emissions reductions by
increasing O3 concentrations. Moreover, the population weighted distribution in DM8HO3
88
concentrations indicates that urban greening would at best have a neutral effect on O3 exposure
and at worst would increase DM8HO3 concentrations for a majority of the population in the
study area irrespective of anthropogenic emissions.
Our model results also show that urban greening will have complex effects on hourly O3
concentrations, with likely O3 penalties throughout the basin during the day and reduced urban
O3 concentrations at night. Nighttime reductions in urban O3 concentrations were strongly
correlated with increases in NOx concentrations, suggesting that reduced ventilation of NOx from
urban greening enhanced O3 titration. On the other hand, nonurban grid cells downwind of urban
areas experienced increased nighttime O3 because of the reduced advection of NOx and
correspondingly, dampened O3 titration. Nighttime O3 concentrations were higher in the
aggressive electrification scenario compared to simulations with current day anthropogenic
emission as a consequence of reduction in anthropogenic NOx emissions. Hence, while NOx
reductions from aggressive electrification somewhat dampened the daytime O3 penalties of urban
greening, they also dampened the nighttime air quality benefits of urban greening.
This study adds to the mounting evidence that urban greening will likely result in
regional O3 penalties, especially if the added vegetation is high isoprene emitting (e.g., Gu et al.,
2021; Ma et al., 2019; Churkina et al., 2017; Taha, 1996). Our model results for the aggressive
electrification scenario further suggest that O3 penalties from urban greening will persist even as
anthropogenic NOx emissions are drastically reduced over the coming decades following climate
change mitigation. Finally, it is worth highlighting that the O3 penalties of urban greening
modeled in our results impacted the nonurban, inland areas that did not receive the cooling
benefits of that vegetation. Policy makers and urban planners should carefully consider these
regional O3 impacts of urban greening to avoid potential environmental justice concerns.
89

4.5 Funding and support:
This work was supported by the National Science Foundation Graduate Research
Fellowship under Grant No. DGE-1842487. Any opinion, findings, and conclusions or
recommendations expressed in this material are those of the authors(s) and do not necessarily
reflect the views of the sponsors. An extension of gratitude to the late George Ban-Weiss for
advising this project in its early stages.

90
Chapter 5: Conclusions

Global mean temperatures are rising due to anthropogenic climate change while urban
populations are rapidly growing, making it more important now than ever to reduce heat stress
and improve urban air quality (United Nations, 2018; IPCC 2021). As this dissertation shows,
regional meteorology and air pollution are intricately linked systems. Their reciprocal nature
presents two paths forward for urban environmental policy: one towards synergistic solutions
that equitably reduce heat and air pollutant exposure and the second towards unintended
consequences of addressing one system without consideration for the other. This dissertation
marches down path one, answering policy relevant research questions at the intersection of urban
air pollution and meteorology while characterizing the potential pitfalls of taking path two.
In Chapter 2, we conducted a measurement campaign to characterize BC emissions from
harbor craft, a class of vessels that operate close to shore, contributing to local air pollution. Our
work in Chapter 2 quantified novel sources of fleet-wide and operational variability in harbor
craft emissions. We found that tugboats emitted more BC than other vessel types and spent the
most time emitting in near shore areas despite having relatively few unique vessels. Our work
identified tugboats as a particularly desirable target for future emissions reductions, representing
a large fraction of total BC emissions from harbor craft while at the same time making up a small
fraction of the total fleet. This work represents an important contribution to emissions inventory
development and was used to inform new harbor craft emissions regulations in the state of
California.
We shifted focus to UHI mitigation in Chapter 3, where we used atmospheric modeling
to quantify the competing warming effects of urban greening, a popular UHI mitigation strategy.
While urban greening can provide local cooling benefits through shade and evaporative cooling,
91
vegetation also has lower albedo than the bare soil it often replaces. Our results in Chapter 3
reveal the critical role that albedo plays in determining the surface climate effects of urban
greening. We found that accounting for decreases in albedo from urban greening resulted in
daytime warming and drastically reduced nighttime cooling compared to simulations that
assumed urban greening caused no change in albedo. Our results suggest that prior modelling
studies that ignore albedo-induced effects likely overestimate cooling from urban greening.
Chapter 3 also makes the important point that regional decreases in albedo from urban greening
have the potential to counteract other UHI mitigation strategies, like the high albedo roofs and
pavements that are simultaneously being deployed in cities like Los Angeles.
This dissertation comes to a head in Chapter 4, where we use atmospheric chemistry
modeling to investigate O3 sensitivity to urban greening under a current day and future
anthropogenic emissions scenario. The future emissions scenario represents Los Angeles City’s
goal of achieving 100% renewable energy generation with aggressive electrification. In prior
work, the air quality effects of urban greening and climate change mitigation were investigated
separately, which is particularly concerning since O3 forms through complex mechanisms that
are controlled by both VOC and NOx concentrations. Here, we address this key literature gap by
testing sensitivity of O3 concentrations to BVOC emissions under a current day and drastically
reduced NOx emissions scenario. Our results show that urban greening has O3 penalties
regardless of anthropogenic emissions reductions. Moreover, nonurban areas did not receive the
cooling benefits of urban greening, yet still experienced O3 tradeoffs, which may pose
environmental justice concerns. This work offers key insight for policymakers at a critical time
when cities simultaneously need to reduce anthropogenic emissions to address the growing
climate crisis and deploy UHI mitigation strategies to reduce heat stress.
92
This dissertation underscores the interconnected nature of urban meteorology and air
pollution while addressing policy relevant research questions. Our work characterizes a novel
source of BC emissions, critically examines underlying assumptions about a UHI mitigation
strategy, and elucidates complex interactions between climate change mitigation and adaptation.
Highlighted in my research is the importance of understanding the interplay between regional
meteorology and air quality so that future policy can take advantage of potential synergies while
avoiding unintended trade-offs. Doing so will ensure that future policy is effective and
optimized, reducing heat stress and improving air quality to better protect human health in urban
areas.

93
References

2017 Cruise Ship Schedules for All US Ports. Crew Center, 23 June 2017, crew-
center.com/2017 cruise-ship-schedules-all-us-ports. Last accessed 10 May 2021.
Ackermann, Ingmar J., et al. "Modal aerosol dynamics model for Europe." Atmospheric
Environment, vol. 32, no. 17, 1998, pp. 2981-2999.
Ahmadov, R., et al. "A volatility basis set model for summertime secondary organic aerosols
over the eastern United States in 2006." Journal of Geophysical Research: Atmospheres,
vol. 117, no. D6, 2012, pp. n/a-n/a.
AIS Vessel Type Codes. National Oceanic and Atmospheric Administration, 2018,
coast.noaa.gov/data/marinecadastre/ais/VesselTypeCodes2018.pdf. Last Accessed 2
Nov. 2020.
Akbari, H. "Shade trees reduce building energy use and CO2 emissions from
powerplants." Environmental Pollution, vol. 116, 2002, pp. S119-S126.
Akbari, H., et al. “Cool Surfaces and Shade Trees to Reduce Energy Use and Improve Air
Quality in Urban Areas.” Solar Energy, vol. 70, no. 3, Elsevier Ltd, 2001, pp. 295–310,
https://doi.org/10.1016/S0038-092X(00)00089-X.
Arghavani, Somayeh, et al. "Numerical evaluation of urban green space scenarios effects on
gaseous air pollutants in Tehran Metropolis based on WRF-Chem model." Atmospheric
Environment, vol. 214, 2019, p. 116832.
Ban-Weiss, George A., et al. "Measurement of Black Carbon and Particle Number Emission
Factors from Individual Heavy-Duty Trucks." Environmental Science & Technology,
vol. 43, no. 5, 2009, pp. 1419-1424
Bell, Michelle L., et al. "Ancillary human health benefits of improved air quality resulting from
climate change mitigation." Environmental, Health 7.1 (2008): 1-18.
Betts, Richard A. “Offset of the Potential Carbon Sink from Boreal Forestation by Decreases in
Surface Albedo.” Nature (London), vol. 408, no. 6809, 2000, pp. 187–90,
https://doi.org/10.1038/35041545.
94
Bistline, John ET, et al. "Economy-wide evaluation of CO2 and air quality impacts of
electrification in the United States." Nature Communications, 13.1 (2022): 6693.
Bonan, Gordon B. “Forests and Climate Change: Forcings, Feedbacks, and the Climate Benefits
of Forests.” Science, vol. 320, no. 5882, 2008, pp. 1444–49,
https://doi.org/10.1126/science.1155121.
Bond, T. C., et al. "Bounding the role of black carbon in the climate system: A scientific
assessment." Journal of Geophysical Research: Atmospheres, vol. 118, no. 11, 2013,
pp. 5380-5552.
Buffaloe, G. M., et al. "Black carbon emissions from in-use ships: a California regional
assessment." Atmospheric Chemistry and Physics, vol. 14, no. 4, 2014, pp. 1881-1896.
CA State Boundary. California Open Data, 10 Sept. 2019, data.ca.gov/dataset/ca-geographic-
boundaries/resource/3db1e426-fb51-44f5-82d5-a54d7c6e188b. Last accessed 2 Nov.
2020.
Calfapietra, C.; Fares, S.; Manes, F.; Morani, A.; Sgrigna, G.; Loreto, F. “Role of Biogenic
Volatile Organic Compounds (BVOC) Emitted by Urban Trees on Ozone Concentration
in Cities: A Review.” Environmental Pollution, 2013, 183, 71−80.
California Native Plant Society. "List of native trees to California." Calscape, calscape.org/.
Accessed April 24, 2022.
Campbell, A., et al. "Particulate Matter in Polluted Air May Increase Biomarkers of
Inflammation in Mouse Brain." NeuroToxicology, vol. 26, no. 1, Jan. 2005, pp. 133-140.
CARB. "ARB’s Emission Inventory Activities." California Air Resources Board,
ww2.arb.ca.gov/emission-inventory-
activities#:~:text=An%20emission%20inventory%20is%20an,%2C%20in%20a%20cale
ndar%20year).
CARB. "Draft Technology Assessment: Ocean Going Vessels." California Air Resources
Board, 2018, ww2.arb.ca.gov/resources/documents/technology-and-fuels-assessments.
Last accessed 11 March 2021.
95
CARB. "Commercial Harbor Craft About Page." California Air Resources Board, 2021,
ww2.arb.ca.gov/our-work/programs/commercial-harbor-craft/about. Accessed
24 Apr. 2022.
CARB. "Rulemaking to Consider the Adoption of a Proposed Regulation for Fuel Sulfur and
Other Operational Requirements for Ocean-Going Vessels Within California Waters and
24 Nautical Miles of the California Baseline." California Air Resources Board, 2008,
ww3.arb.ca.gov/regact/2008/fuelogv08/fuelogv08.htm. Last accessed 26 Feb. 2021.
Carter, William P., et al. "Smog chamber studies of temperature effects in photochemical
smog." Environmental Science & Technology, vol. 13, no. 9, 1979, pp. 1094-1100.
Caubel, Julien J., et al. "A Distributed Network of 100 Black Carbon Sensors for 100 Days of
Air Quality Monitoring in West Oakland, California." Environmental Science &
Technology, vol. 53, no. 13, 2019, pp. 7564-7573.
Caubel, Julien, et al. "A New Black Carbon Sensor for Dense Air Quality Monitoring
Networks." Sensors, vol. 18, no. 3, 2018, p. 738.
Chen, Dongsheng, et al. "Estimating ship emissions based on AIS data for port of Tianjin,
China." Atmospheric Environment, vol. 145, 2016, pp. 10-18.
Chen, Fei, and Jimy Dudhia. "Coupling an Advanced Land Surface–Hydrology Model with the
Penn State–NCAR MM5 Modeling System. Part I: Model Implementation and
Sensitivity." Monthly Weather Review, vol. 129, no. 4, 2001, pp. 569-585.
Chen, Fei, et al. "The integrated WRF/urban modelling system: development, evaluation, and
applications to urban environmental problems." International Journal of Climatology,
vol. 31, no. 2, 2011, pp. 273-288.
Chen, Shukai, et al. "Analysis of Tugboat Activities using AIS Data for the Tianjin
Port." Transportation Research Record: Journal of the Transportation Research Board,
vol. 2674, no. 5, 2020, pp. 498-509.
Ching, Jason, et al. "National Urban Database and Access Portal Tool." Bulletin of the American
Meteorological Society, vol. 90, no. 8, 2009, pp. 1157-1168.
96
Chou, M.-D. and Suarez, M. J.: “Technical Report Series on Global Modeling and Data
Assimilation,” Volume 15, in: A Solar Radiation Parameterization for Atmospheric
Studies, Goddard Space Flight Center, Greenbelt, MD, USA, 1–38, 1999.
Churkina, G.; Kuik, F.; Bonn, B.; Lauer, A.; Grote, R.; Tomiak, K.; Butler, T. M. “Effect of
VOC Emissions from Vegetation on Air Quality in Berlin during a Heatwave.”
Environmental Science & Technology, 2017, 51, 6120−6130.
Cifuentes, Luis, et al. "Hidden health benefits of greenhouse gas mitigation." Science, 293.5533
(2001): 1257-1259.
Cochran, Jaquelin, and Paul Denholm, eds. 2021. “The Los Angeles 100% Renewable Energy
Study. Golden, CO: National Renewable Energy Laboratory.” NREL/TP-6A20-79444.
https://maps.nrel.gov/la100/.
Corbett, James J., et al. "Mortality from Ship Emissions: A Global Assessment." Environmental
Science & Technology, vol. 41, no. 24, 2007, pp. 8512-8518.
Dallmann, Timothy R., and Robert A. Harley. "Evaluation of mobile source emission trends in
the United States." Journal of Geophysical Research, vol. 115, no. D14, July 2010.
Dallmann, Timothy R., et al. "Effects of Diesel Particle Filter Retrofits and Accelerated Fleet
Turnover on Drayage Truck Emissions at the Port of Oakland." Environmental Science &
Technology, vol. 45, no. 24, 2011, pp. 10773-10779.
Dallmann, Timothy R., et al. "On-Road Measurement of Gas and Particle Phase Pollutant
Emission Factors for Individual Heavy-Duty Diesel Trucks." Environmental Science &
Technology, vol. 46, no. 15, 2012, pp. 8511-8518.
De Souza, Erico N., et al. "Improving Fishing Pattern Detection from Satellite AIS Using Data
Mining and Machine Learning." PLOS ONE, vol. 11, no. 7, 2016, p. e0158248.
Dyer, A. J., and B. B. Hicks. "Flux-gradient relationships in the constant flux layer." Quarterly
Journal of the Royal Meteorological Society, vol. 96, no. 410, 1970, pp. 715-721.
Eisenman et al. “Urban trees, air quality, and asthma: An interdisciplinary review.” Landscape
and Urban Planning, (2019).
97
Emery, Christopher, et al. "Recommendations on statistics and benchmarks to assess
photochemical model performance." Journal of the Air & Waste Management
Association, 67.5 (2017): 582-598.
Emmons, L. K., et al. "Description and evaluation of the Model for Ozone and Related chemical
Tracers, version 4 (MOZART-4)." Geoscientific Model Development, vol. 3,
no. 1, 2010, pp. 43-67.
EPA. "Shore Power Technology Assessment at U.S. Ports PDF." US EPA, Mar. 2017,
www.epa.gov/ports-initiative/shore-power-technology-assessment-us-ports#assessment.
Last accessed Nov. 2019.
EPA. “Amendments Related to Marine Diesel Engine Emission Standards.” Environmental
Protection Agency Documents and Publications, Federal Information & News Dispatch,
LLC, 2 Oct. 2020.
EPA. “California State Nonroad Engine Pollution Control Standards; Diesel Engines on
Commercial Harbor Craft; Request for Within-the-Scope and Full Authorization;
Opportunity for Public Hearing and Comment.” The Federal Register, vol. 79, no. 226,
Federal Information & News Dispatch, LLC, 24 Nov. 2014, p. 69842–.
Fallmann, Joachim, et al. "Secondary effects of urban heat island mitigation measures on air
quality." Atmospheric Environment, vol. 125, 2016, pp. 199-211.
Fry, J.A., et al. “Completion of the 2006 National Land Cover Database for the Conterminous
United States.” Photogrammetric Engineering and Remote Sensing, vol. 77, no. 9, 2011,
pp. 859–64.
Fuglestvedt, Jan, et al. "Shipping Emissions: From Cooling to Warming of Climate—and
Reducing Impacts on Health." Environmental Science & Technology, vol. 43,
no. 24, 2009, pp. 9057-9062.
Gallagher, Ciaran L., and Tracey Holloway. "Integrating air quality and public health benefits in
US decarbonization strategies." Frontiers inPublic Health, 8 (2020): 563358.
Grell, Georg A., and Dezső Dévényi. "A generalized approach to parameterizing convection
combining ensemble and data assimilation techniques." Geophysical Research Letters,
vol. 29, no. 14, 2002, pp. 38-1-38-4.
98
Gu, Shan, et al. “Effects of Anthropogenic and Biogenic Volatile Organic Compounds on Los
Angeles Air Quality.” Environmental Science & Technology, vol. 55, no. 18, 2021, pp.
12191 201, https://doi.org/10.1021/acs.est.1c01481.
Guenther, A. B., et al. "The Model of Emissions of Gases and Aerosols from Nature version 2.1
(MEGAN2. 1): an extended and updated framework for modeling biogenic
emissions." Geoscientific Model Development, 5.6 (2012): 1471-1492.
Guenther, A., et al. "Estimates of global terrestrial isoprene emissions using MEGAN (Model of
Emissions of Gases and Aerosols from Nature)." Atmospheric Chemistry and Physics,
vol. 6, no. 11, 2006, pp. 3181-3210.
Halberstadt, M.L. "Effective control of highway vehicle hydrocarbon emissions." JAPCA,
International Journal of Air Pollution Control and Waste Management, vol. 39, no. 6,
1 July 1989, www.osti.gov/biblio/5188163.
Hasheminassab, S.; Daher, N.; Ostro, B. D.; Sioutas, C. Long. “Term Source Apportionment of
Ambient Fine Particulate Matter (PM2.5) in the Los Angeles Basin: A Focus on
Emissions Reduction from Vehicular Sources.” Environ. Pollut. 2014, 193, 54−64.
Heath, Garvin, George Ban-Weiss, Vikram Ravi, Yun Li, and Jiachen Zhang. 2021. “Chapter 9:
Air Quality and Public Health.” In The Los Angeles 100% Renewable Energy Study,
edited by Jaquelin Cochran and Paul Denholm. Golden, CO: National Renewable Energy
Laboratory. NREL/TP-6A20-79444-9. https://www.nrel.gov/docs/fy21osti/79444-9.pdf.
Hoek, Gerard, et al. "Long-term air pollution exposure and cardio- respiratory mortality: a
review." Environmental Health, vol. 12, no. 1, 2013.
Hong, Song-You, et al. "A New Vertical Diffusion Package with an Explicit Treatment of
Entrainment Processes." Monthly Weather Review, vol. 134, no. 9, 2006, pp. 2318-2341.
IMO. "Sulphur 2020 – Cutting Sulphur Oxide Emissions." 2019,
www.imo.org/en/MediaCentre/HotTopics/Pages/Sulphur-2020.aspx. Last accessed
26 Feb. 2021.
IMO. “Third IMO Greenhouse Gas Study.” The International Maritime Organization, 2015,
www.imo.org/en/OurWork/Environment/Pages/Greenhouse-Gas-Studies-2014.aspx.
Last accessed 10 March 2021.
99
IPCC, 2021. “Climate Change 2021: The Physical Science Basis. Contribution of Working
Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate
Change.” [Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N.
Caud, Y. Chen, L. Goldfarb, M.I., Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R.
Matthews, T.K. Maycock, T. Waterfield, O., Yelekçi, R. Yu, and B. Zhou (eds.)].
Cambridge University Press. In Press.
Jiang, Yu, et al. "Sources of variance in BC mass measurements from a small marine engine:
Influence of the instruments, fuels and loads." Atmospheric Environment, vol. 182, 2018,
pp. 128-137.
Kennedy, Ian M. "The health effects of combustion-generated aerosols." Proceedings of the
Combustion Institute, vol. 31, no. 2, 2007, pp. 2757-2770.
Khan, M. Y., et al. "Benefits of Two Mitigation Strategies for Container Vessels: Cleaner
Engines and Cleaner Fuels." Environmental Science & Technology, vol. 46, no. 9, 2012,
pp. 5049-5056.
Kim, S.-W., et al. "NO2 columns in the western United States observed from space and
simulated by a regional chemistry model and their implications for NOx
emissions." Journal of Geophysical Research, vol. 114, no. D11, 2009.
Ko, Joseph, et al. "Measuring the impacts of a real-world neighborhood-scale cool pavement
deployment on albedo and temperatures in Los Angeles." Environmental Research
Letters, 17.4 (2022): 044027.
Ko, Joseph, Trevor Krasowsky, and George Ban-Weiss. "Measurements to determine the
mixing state of black carbon emitted from the 2017–2018 California wildfires and urban
Los Angeles." Atmospheric Chemistry and Physics, 20.24 (2020): 15635-15664.
Krasowsky, Trevor, et al. "Measurement of particulate matter emissions from in-use
locomotives." Atmospheric Environment, vol. 113, 2015, pp. 187-196.
Kusaka, Hiroyuki, et al. "Numerical Simulation of Urban Heat Island Effect by the WRF Model
with 4-km Grid Increment: An Inter-Comparison Study between the Urban Canopy
Model and Slab Model." Journal of the Meteorological Society of Japan. Ser. II,
vol. 90B, no. 0, 2012, pp. 33-45.
100
Lack, D. A., and J. J. Corbett. "Black carbon from ships: a review of the effects of ship speed,
fuel quality and exhaust gas scrubbing." Atmospheric Chemistry and Physics, vol. 12,
no. 9, 2012, pp. 3985-4000.
Lack, Daniel A., et al. "Impact of Fuel Quality Regulation and Speed Reductions on Shipping
Emissions: Implications for Climate and Air Quality." Environmental Science &
Technology, vol. 45, no. 20, 2011, pp. 9052-9060.
Lack, Daniel, et al. "Light absorbing carbon emissions from commercial shipping." Geophysical
Research Letters, vol. 35, no. 13, 2008.
Lack, Daniel, et al., "Particulate emissions from commercial shipping: Chemical, physical, and
optical properties." Journal of Geophysical Research, vol. 114, 2009.
LI-840A Instruction Manual. LI-COR Biosciences, June 2016,
www.licor.com/documents/y10gor2jal2p3t8ev4hm. Last accessed 10 March 2021.
Li, Xian-Xiang, and Xuan Liu. "Effect of tree evapotranspiration and hydrological processes on
urban microclimate in a tropical city: A WRF/SLUCM study." Urban Climate, 40
(2021): 101009.
Li, Xian-Xiang, Leslie K. Norford, “Evaluation of cool roof and vegetations in mitigating urban
heat island in a tropical city, Singapore." Urban Climate,Volume 16, 2016, Pages 59-74,
ISSN 2212-0955, https://doi.org/10.1016/j.uclim.2015.12.002.
Li, Yun, et al. "Effects of urbanization on regional meteorology and air quality in Southern
California." Atmospheric Chemistry and Physics, vol. 19, no. 7, 2019, pp. 4439-4457.
Lin, Yuh-Lang, et al. "Bulk Parameterization of the Snow Field in a Cloud Model." Journal of
Climate and Applied Meteorology, vol. 22, no. 6, 1983, pp. 1065-1092.
Liu, Huan, et al. "Health and climate impacts of ocean-going vessels in East Asia." Nature
Climate Change, vol. 6, no. 11, 2016, pp. 1037-1041.
Lurmann, Fred, et al. "Emissions reduction policies and recent trends in Southern California’s
ambient air quality." Journal of the Air & Waste Management Association, vol. 65,
no. 3, 2014, pp. 324-335.
101
Ma, M.; Gao, Y.; Wang, Y.; Zhang, S.; Leung, L. R.; Liu, C.; Wang, S.; Zhao, B.; Chang, X.;
Su, H.; Zhang, T.; Sheng, L.; Yao, X.; Gao, H. “Substantial Ozone Enhancement over
the North China Plain from Increased Biogenic Emissions Due to Heat Waves and Land
Cover in Summer 2017.” Atmospheric Chemistry and Physics, 2019, 19, 12195−12207.
Madronich, Sasha. “Photodissociation in the Atmosphere: 1. Actinic Flux and the Effects of
Ground Reflections and Clouds.” Journal of Geophysical Research,
doi:10.1029/JD092iD08p09740.
Manisalidis, Ioannis, et al. "Environmental and Health Impacts of Air Pollution: A
Review." Frontiers in Public Health, vol. 8, 2020.
Marine Cadastre Data Registry. "Vessel Traffic Data." National Oceanic and Atmospheric
Administration, 2018, marinecadastre.gov/AIS. Last accessed 10 March 2021.
MarineTraffic. AIS Marine Traffic,
www.marinetraffic.com/en/ais/home/centerx:117.903/centery:33.606/zoom:17. Last
accessed 10 March 2021.
Mcginn, Charles Evans. “The microclimate and energy use in suburban tree canopies
(California).” ProQuest Dissertations Publishing, 1982.
MEGAN3 Data and Code. "Data and Code for MEGAN3.2." UCI BAI, Apr. 2019,
bai.ess.uci.edu/megan/data-and-code. Accessed 24 Apr. 2022.
Mesinger, F., et al. “North American Regional Reanalysis,” 2006. Bulletin of the American
Meteorological Society, 87(3), 343-360.
Mlawer, Eli J., et al. "Radiative transfer for inhomogeneous atmospheres: RRTM, a validated
correlated-k model for the longwave." Journal of Geophysical Research: Atmospheres,
vol. 102, no. D14, 1997, pp. 16663-16682.
Moore, Thomas J., et al. "Exploring ship traffic variability off California." Ocean & Coastal
Management, vol. 163, 2018, pp. 515-527.
Morakinyo, Tobi Eniolu, et al. "A study on the impact of shadow-cast and tree species on in
canyon and neighborhood's thermal comfort." Building and Environment, 115 (2017): 1
17.
102
Mousavi, Amirhosein, et al. "Impact of emissions from the Ports of Los Angeles and Long
Beach on the oxidative potential of ambient PM0.25 measured across the Los Angeles
County." Science of The Total Environment, vol. 651, 2019, pp. 638-647.
Mousavi, Amirhosein, et al. "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." Atmospheric
Environment, vol. 195, 2018, pp. 159-169.
Navigation Safety Regulations, 33 Fed. Reg. 164 (July 1, 2003). Electronic Code of Federal
Regulations. Web.
Nemet, Gregory F., Tracey Holloway, and Paul Meier. "Implications of incorporating air-quality
co-benefits into climate change policymaking." Environmental Research Letters, 5.1
(2010): 014007
Nowak, David J., et al. "Modeled PM2. 5 removal by trees in ten US cities and associated health
effects." Environmental Pollution, 178 (2013): 395-402.
Nunes, R.A.O., et al. "The activity-based methodology to assess ship emissions – A
review." Environmental Pollution, vol. 231, 2017, pp. 87-103.
Oke, T. R., et al. Urban Climates. Cambridge University Press, 2017.
Oke, T.R., The energetic basis of the urban heat island. Quarterly Journal of the Royal
Meteorological Society,1982. 108: 124. DOI: https://doi.org/10.1002/qj.49710845502
Paulson, C. A. "The Mathematical Representation of Wind Speed and Temperature Profiles in
the Unstable Atmospheric Surface Layer." Journal of Applied Meteorology, vol. 9,
no. 6, 1970, pp. 857-861.
Pincetl, Stephanie, et al. “Urban Tree Planting Programs, Function or Fashion? Los Angeles and
Urban Tree Planting Campaigns.” GeoJournal, vol. 78, no. 3, Springer, 2013, pp. 475
93, https://doi.org/10.1007/s10708-012-9446-x.
Pope, C. A., et al. "Cardiovascular Mortality and Long-Term Exposure to Particulate Air
Pollution." Circulation, vol. 109, no. 1, Jan. 2004, pp. 71-77.
103
Port of Long Beach Emissions Inventory. The Port of Long Beach, 2018,
https://polb.com/environment/air/#emissions-inventory. Last accessed 10 March 2021.
Port of Los Angeles Air Emissions Inventory. The Port of Los Angeles, 2018,
www.portoflosangeles.org/environment/air-quality/air-emissions-inventory. Last
accessed 10 March 2021.
Port of Los Angeles Facts and Figures. The Port of Los Angeles, 2020,
www.portoflosangeles.org/business/statistics/facts-and-figures. Last accessed 10 March
2021.
Preble, Chelsea V., et al. "Effects of Particle Filters and Selective Catalytic Reduction on Heavy
Duty Diesel Drayage Truck Emissions at the Port of Oakland." Environmental Science
& Technology, vol. 49, no. 14, 2015, pp. 8864-8871.
Rahman, Md Mostafijur, et al. "The effects of coexposure to extremes of heat and particulate air
pollution on mortality in California: implications for climate change." American Journal
of Respiratory and Critical Care Medicine, 206.9 (2022): 1117-1127.
Rao, Meenakshi, et al. "Assessing the relationship among urban trees, nitrogen dioxide, and
respiratory health." Environmental Pollution, vol. 194, 2014, pp. 96-104.
Requia, Weeberb J., et al. "How clean are electric vehicles? Evidence-based review of the
effects of electric mobility on air pollutants, greenhouse gas emissions and human
health." Atmospheric Environment, 185 (2018): 64-77.
Robards, MD, et al. "Conservation science and policy applications of the marine vessel
Automatic Identification System (AIS)—a review." Bulletin of Marine Science, vol. 92,
no. 1, 2016, pp. 75-103.
Rohatyn, Shani, et al. "Limited climate change mitigation potential through forestation of the
vast dryland regions." Science, 377.6613 (2022): 1436-1439.
Rose, L. Shea, and Ronnen Levinson. "Analysis of the effect of vegetation on albedo in
residential areas: Case studies in suburban Sacramento and Los Angeles,
CA." GIScience & Remote Sensing, 50.1 (2013): 64-77.
Rosenfeld, Arthur H., et al. "Cool communities: strategies for heat island mitigation and smog
reduction." Energy and Buildings, vol. 28, no. 1, 1998, pp. 51-62.
104
Sawant, Aniket A., et al. "Regulated and Non-Regulated Emissions from In-Use Diesel-Electric
Switching Locomotives." Environmental Science & Technology, vol. 41,
no. 17, Sept. 2007, p. 6074–83, doi:10.1021/es061672d.
SCAQMD. "Final 2016 Air Quality Management Plan." Air Quality Management
District, Mar. 2017, www.aqmd.gov/home/air-quality/clean-air-plans/air-quality-mgt-
plan/final-2016-aqmp. Accessed 24 Apr. 2022.
Seinfeld, John H., and Spyros N. Pandis. Atmospheric Chemistry and Physics: From Air
Pollution to Climate Change. John Wiley & Sons, 2016.
Sillman, Sanford. "The relation between ozone, NOx and hydrocarbons in urban and polluted
rural environments." Atmospheric Environment, 33.12 (1999): 1821-1845.
Single Particle Soot Photometer User Manual. Droplet Measurement Technologies, 2017.
ra9r13nh313ew0s1pxuptw7p-wpengine.netdna-ssl.com/wp-
content/uploads/2020/02/DOC-0171-G-9-SP2-Operator-Manual.pdf. Last accessed
2 Nov. 2020.
Sofia, Daniele, et al. "Mitigation strategies for reducing air pollution." Environmental Science
and Pollution Research, vol. 27, no. 16, 2020, pp. 19226-19235.
Sofiev, Mikhail, et al. "Cleaner fuels for ships provide public health benefits with climate
tradeoffs." Nature Communications, vol. 9, no. 1, 2018.
Souch, C.A., and C. Souch. “The Effect of Trees on Summertime Below Canopy Urban
Climates: a Case Study in Bloomington, Indiana.” Journal of Arboriculture, vol. 19, no.
5, International society of arboriculture, 1993, pp. 303–12.
Taha, Haider. "Modeling impacts of increased urban vegetation on ozone air quality in the
South Coast Air Basin." Atmospheric Environment, vol. 30, no. 20, 1996, pp. 3423-
3430.
Tingey, David T., Rosemary Evans, and Marcia Gumpertz. "Effects of environmental conditions
on isoprene emission from live oak." Planta 152 (1981): 565-570.
Timko, Michael T., et al. "Sampling Artifacts from Conductive Silicone Tubing." Aerosol
Science and Technology, vol. 43, no. 9, 2009, pp. 855-865.
105
Turner, David R., et al. "Shipping and the environment: Smokestack emissions, scrubbers and
unregulated oceanic consequences." Elem Sci Anth, vol. 5, no. 0, 2017, p. 45.
Ulpiani, Giulia. "On the linkage between urban heat island and urban pollution island: Three-
decade literature review towards a conceptual framework." Science of The Total
Environment, vol. 751, 2021, p. 141727.
U.S. EPA. Air Quality System Data Mart. Available at http://www.epa.gov/ttn/airs/aqsdatamart.
Last accessed 27 March 2023.
U.S. Maritime Limits and Boundaries. National Oceanic and Atmospheric Administration,
Office of Coast Survey, 8 June 2020, nauticalcharts.noaa.gov/data/us-maritime-limits-
and-boundaries.html. Last accessed 10 March 2021.
United Nations. "World Urbanization Prospects: The 2018 Revision: key facts." 2018,
population.un.org/wup/Publications/Files/WUP2018-KeyFacts.pdf.
US EPA. "Previous NEI Reports." US EPA, Apr. 2014, www.epa.gov/air-emissions-
inventories/previous-nei-reports. Last accessed 24 Apr. 2022.
Vahmani, P., and G. A. Ban-Weiss. "Impact of remotely sensed albedo and vegetation fraction
on simulation of urban climate in WRF-urban canopy model: A case study of the urban
heat island in Los Angeles." Journal of Geophysical Research: Atmospheres, vol. 121,
no. 4, 2016b, pp. 1511-1531.
Vahmani, P., and G. Ban‐Weiss. "Climatic consequences of adopting drought‐tolerant
vegetation over Los Angeles as a response to California drought." Geophysical Research
Letters 43.15 (2016a): 8240-8249.
Vahmani, Pouya, and Terri S. Hogue. "Incorporating an Urban Irrigation Module into the Noah
Land Surface Model Coupled with an Urban Canopy Model." Journal of
Hydrometeorology, vol. 15, no. 4, 2014, pp. 1440-1456.
Vaishnav, Parth, et al. "Shore Power for Vessels Calling at U.S. Ports: Benefits and
Costs." Environmental Science & Technology, vol. 50, no. 3, 2016, pp. 1102-1110.
Wang, Chenghao, et al. “Cooling Effect of Urban Trees on the Built Environment of Contiguous
United States.” Earth’s Future, vol. 6, no. 8, 2018, pp. 1066–81,
https://doi.org/10.1029/2018EF000891.
106
Wang, Liang, and Dan Li. “Urban Heat Islands During Heat Waves: a Comparative Study
Between Boston and Phoenix.” Journal of Applied Meteorology and Climatology, 2021,
https://doi.org/10.1175/JAMC-D-20-0132.1.
Warneke, C.; de Gouw, J. A.; Holloway, J. S.; Peischl, J.; Ryerson, T. B.; Atlas, E.; Blake, D.;
Trainer, M.; Parrish, D. D. “Multiyear Trends in Volatile Organic Compounds in Los
Angeles, California: Five decades of decreasing emissions.” Journal of Geophysical
Research, 2012, 117, 1.
West, J. Jason, et al. "Co-benefits of mitigating global greenhouse gas emissions for future air
quality and human health." Nature Climate Change, 3.10, 2013: 885-889.
Whitlow, Thomas H., et al. "Comments on ‘Modeled PM2.5 removal by trees in ten U.S. cities
and associated health effects’ by Nowak et al. (2013)." Environmental Pollution,
vol. 191, 2014, p. 256.
Wickham, James D., et al. "Accuracy assessment of NLCD 2006 land cover and impervious
surface." Remote Sensing of Environment, vol. 130, 2013, pp. 294-304.
Yang, Jiachuan, et al. "Enhancing Hydrologic Modelling in the Coupled Weather Research and
Forecasting–Urban Modelling System." Boundary-Layer Meteorology, vol. 155,
no. 1, 2014, pp. 87-109.
Yu, Chenjie, et al. "Characterizing the Particle Composition and Cloud Condensation Nuclei
from Shipping Emission in Western Europe." Environmental Science & Technology,
vol. 54, no. 24, 2020, pp. 15604-15612.
Yu, X.; Venecek, M.; Kumar, A.; Hu, J.; Tanrikulu, S.; Soon, S.- T.; Tran, C.; Fairley, D.;
Kleeman, M. J. “Regional Sources of Airborne Ultrafine Particle Number and Mass
Concentrations in California.” Atmospheric Chemistry and Physics, 2019, 19,
14677−14702
Zanobetti, Antonella, and Joel Schwartz. "The Effect of Particulate Air Pollution on Emergency
Admissions for Myocardial Infarction: A Multicity Case-Crossover
Analysis." Environmental Health Perspectives, vol. 113, no. 8, 2005, pp. 978-982.
Zhang, Jiachen, et al. "Systematic Comparison of the Influence of Cool Wall versus Cool Roof
Adoption on Urban Climate in the Los Angeles Basin." Environmental Science &
Technology, vol. 52, no. 19, 2018, pp. 11188-11197.
107
Zhang, Jiachen, et al. "Investigating the urban air quality effects of cool walls and cool roofs in
Southern California." Environmental Science & Technology, 53.13 (2019): 7532-7542.

108
Appendices

Appendix for Chapter 2

1 Activity Analysis
Data sources for the activity analysis are summarized in Table A.1. Automatic
Identification System (AIS) location reports use the WGS-84 coordinate reference system.
Vessel type codes from the US Coast Guard were used to extract data for relevant vessel types
(AIS Vessel Type Codes). Relevant harbor craft types included tugboats, passenger boats, and
fishing boats. It should be noted that the US Coast Guard does not require all self-propelled
vessels less than 1600 gross tons to be equipped with AIS; hence, small passenger boats and pilot
boats are not included in this analysis (Navigation Safety Regulations). However, all fishing
boats and passenger boats measured in the study did have AIS since they were observable on the
Marine Traffic smart phone application. The passenger ship AIS vessel type codes include
ferries, excursion vessels, and cruise ships. Cruise ships were removed from the harbor craft
passenger vessel category using a hobbyist database of all port calls made by cruise ships in
Southern California (2017 Cruise Ship Schedules for all US Ports).

Although emissions from
large ocean-going vessels were not measured in this study, vessel activity of cruise ships, cargo
ships, and tankers was analyzed to compare against harbor craft activity. These vessel types are
hereafter referred to as large ocean-going vessels.

Table A.1 Summary of data used in the activity analysis
Dataset Description Use Source
109
AIS Data
CSVs containing AIS
reports for UTM Zone
11 from Jan. 1 to Dec.
31, 2017
Used to characterize
vessel activity in
Southern California
waters
Marine Cadastre Data Registry
https://marinecadastre.gov/ais/
Vessel Type
Codes
Table of AIS vessel
type codes and
definitions
Used to filter relevant
vessel types from the
bulk dataset
AIS Vessel Type Codes
https://marinecadastre.gov/ais/faq/
NOAA
Maritime
Boundaries
Shapefile with layers
of submerged state
lands, territorial sea,
and exclusive
economic zone for the
entire United States
Used as spatial buffers
to identify vessel
activity in territorial sea
and contiguous zone of
CA waters
U.S. Maritime Limits and Boundaries
https://nauticalcharts.noaa.gov/data/us-
maritime-limits-and-
boundaries.html#general-information
California
State
Boundary
Shapefile with the
official state lines of
California
Used as a spatial buffer
to identify vessel
activity in submerged
state lands
California Open Data
https://data.ca.gov/dataset/ca-
geographic-
boundaries/resource/3db1e426-fb51-
44f5-82d5-a54d7c6e188b

Three buffers were constructed that represent state-owned tidelands, the territorial sea,
and the contiguous zone of California waters. These zones are established maritime boundaries
with specific legislative implications and were selected to increase the policy relevance of the
activity analysis. State-owned tidelands extend from the coastline to an area offshore known as
110
the baseline, a low water line along the coast that is approximately 3 nautical miles from shore
(US Maritime Limits and Boundaries). The state-owned tideland buffer was extracted from a
shapefile of California’s state boundaries by masking out land and clipping (i.e., finding the
intersection of two overlying polygons) the remaining polygon to UTM Zone 11 (see red dashed
boundary in Figure 1A of the main text). The territorial sea extends 12 nautical miles beyond the
baseline and the contiguous zone is the area extending 12 nautical miles beyond the territorial
sea (see orange and yellow zones, respectively, in Figure 1A of the main text). Buffers for these
zones were constructed by clipping the NOAA national maritime boundaries to UTM Zone 11
(US Maritime Limits and Boundaries). Buffers were exported as shapefiles for use in the activity
analysis.
The methodology used to compute total operating time in each buffer zone is described
here. First, AIS reports were grouped by vessel type code for the relevant harbor craft and large
ocean-going vessels listed above. Then, each vessel type was sorted by vessel identifier (MMSI)
and time stamp. For each unique vessel, the change in time (dt) between AIS reports was found
by subtracting consecutive timestamps. When dt was greater than the AIS reporting frequency of
anchored vessels (~180 seconds), it was assumed that vessels were turned off between AIS
reports and dt was set to 0 seconds (U.S. Coast Guard). The dynamic data in AIS reports were
also used to compute calculated speed over ground (SOG) since previous studies have cited that
SOG values reported in the AIS data may be erroneous (Robards et al., 2016, Chen et al., 2016).
When dt was greater than 0 seconds, calculated SOG was computed by finding the vincenty
ellipsoid distance between consecutive location reports and dividing by dt; calculated SOG was
set to 0 m/s when dt = 0. Likewise, the “rate of turn” was calculated by finding the change in
111
course over ground and dividing by dt. Total operating time spent in each buffer zone was
computed for each vessel type by summing dt values per zone.
The international Maritime Organization (IMO) defines anchored vessels as those
travelling less than 3 knots (1.54 m/s) and at-berth vessels as those traveling slower than 1 knot
(0.51 m/s; Third IMO GHG Study). However, analyzing the distribution of calculated SOG by
vessel type, shown in Figure A.1, suggests that different vessel types operate at different
characteristic speeds, reflecting the different operating modes typically engaged by each vessel
type. Therefore, we employed custom definitions of stationary time for each vessel category,
descriptions of which are provided in Table A.2. Furthermore, a previous vessel traffic study in
California found evidence that environmental regulations, economic events, and fuel prices
impact ship speed on regional scales (Moore et al., 2018). This suggests that vessel speeds are
not only vessel-type dependent, as observed in the current study, but likely also location
dependent, and further validates our decision to employ custom speed thresholds for operating
mode definitions.
112

Figure A.1: Distribution of the calculated SOG for one month of AIS data for vessels in UTM
Zone 11.

113
Table A.2 Definitions of the stationary operating mode used for each vessel type
Vessel Type
Stationary
Definition
Justification
Tug/Tow Boats
Calculated SOG <
0.1 m/s &
Rate of turn <
1°/sec
When tugboats maneuver large ships into harbor, they
move at extremely slow speeds while significantly
changing their course. The calculated SOG threshold is
similar to that used by Chen et al. (2020) and the rate of
turn threshold was selected from trial and error in qGIS.
Passenger Boats
Calculated SOG <
1.54 m/s

SOG less than 3 knots was used in previous studies to
define at-berth activity. Using this speed threshold results in
clusters of passenger boats in near shore areas without
flagging linear ship tracks that would be characteristic of
cruising.
Fishing Vessels
Calculated SOG <
0.05 m/s

Fishing vessels maneuver very slowly while trawling, line
fishing, and seine fishing. When calculated SOG was less
than 0.05 m/s, stationary points did not form linear patterns
associated with fishing activity.
Cruise Ships
Calculated SOG <
0.1 m/s
When navigating shallow water and busy ports, these
vessels maneuver extremely slowly with the aid of
tugboats. This speed threshold flags vessels as stationary Cargo Ships
Calculated SOG <
0.1 m/s
114
Tankers
Calculated SOG <
0.1 m/s
without flagging linear paths from vessels moving slowly
into ports.

A brief explanation of the operating mode definitions used for each vessel type is
described here and sample activity maps are shown in the Figures S2 through S7. Stationary
operating mode definitions were determined from a trial and error approach using rule-based
symbology in qGIS version 3.10. When data points were clustered instead of forming linear
tracks, the stationary operating mode definition employed was considered reasonable. Passenger
boats clustered using the IMO 1.54 m/s threshold (see Figure A.2) for at-berth vessels unlike all
other vessel types, which formed linear tracks when using this same speed threshold. The linear
features observed in fishing vessels formed patterns similar to those identified as characteristic of
trawling, longline fishing, and purse seine fishing (de Souza et al., 2016). A relatively slow speed
threshold of 0.05 m/s was needed to decrease the number of linear features observed in stationary
fishing vessels (Figure A.3). Characteristic of tugboat activity is maneuvering very slowly while
significantly changing course (Chen et al., 2020). Therefore, a calculated rate of turn threshold
of 1°/sec was used in conjunction with a calculated SOG threshold of 0.1 m/s to further refine
tugboat idling definitions (Figure A.4). Chen et al. (2020) employed a similar speed threshold to
identify stationary tugboats and likewise used a change in course over ground to identify
maneuvering tugboats in the Tiajan Port of China. Lastly, large ocean-going vessels travel very
slowly when being maneuvered into port by tugboats, so a speed threshold of 0.1 m/s was used
to identify stationary cruise ships, cargo ships, and tankers (Figures S5-S7). A summary of these
custom operating mode definitions is included in Table A.2.
115
Table A.3 summarizes the difference in estimated mobile time for each vessel type found
using the custom speed thresholds used in the current study versus the universal at-berth
threshold used in the IMO Third Green House Gas Study (2014). These findings suggest that
previous studies may be overestimating idling emissions and underestimating maneuvering and
cruising emissions, especially for slow-moving vessels like tugboats, cruise ships, cargo vessels,
and tankers.

Figure A.2: Sample passenger boat activity.

116

Figure A.3: Sample fishing boat activity.
117

Figure A.4: Sample tugboat activity.

118

Figure A.5: Sample cruise ship activity.
119

Figure A.6: Sample cargo ship activity.
120

Figure A.7: Sample tanker activity.

121
Table A.3: Comparison of estimated stationary hours using 3-knot threshold versus the custom
stationary definitions employed in the current study
Vessel Type
Yearly stationary hours
in study area using a 3-
knot threshold
Yearly stationary hours
in the study area
estimated using custom
thresholds
Percent change in
stationary hours
estimate
Tug/Tow Boats 218,509 145,398 33.5%
Fishing Vessels 48,376 43,919 9.2%
Passenger Boats 327,084 327,084 0%
Cruise Ships 2,898 2,476 14.6%
Cargo Ships 176,744 159,502 9.8%
Tankers 89,751 79,549 11.4%
Note that passenger boats had no change in yearly stationary hours because the 3-knot speed
threshold was determined sufficient for identifying stationary time for this vessel type.

Appendix B: Onshore Measurement Location and Dates
Pier F in the Port of Long Beach was selected for onshore measurements because of its
frequent close-to-shore harbor craft traffic, as is shown in the sample AIS data in Figure B.1.
Some additional measurements of stationary vessels were made at the fire pier in the Port of Los
Angeles and at Pier G in the Port of Long Beach. Measurement dates and descriptions are
included in Table B.1.
122

Figure B.1: The San Pedro Bay Port Complex (left) and twenty-four hours of harbor craft
activity near the Pier F sampling site (right).
Table B.1: Onshore Measurement Dates
Sampling Date Location
19 June 2019 Pier F
26 June 2019 Pier F
27 June 2019 Pier F
22 July 2019 Pier F
16 August 2019 Pier F
21 August 2019 Pier F
22 November 2019
Dock measurements at the Fire Pier in the Port of Los
Angeles
17 January 2020 Dock measurements at Pier G in the Port of Long Beach
123
27 January 2020 Pier F

Appendix C: Onboard Measurements
Additional sampling was conducted onboard a passenger boat and charter fishing boat
throughout typical operating conditions on July 2 and 15, 2019. The passenger boat had a “wet”
exhaust system, wherein exhaust gas is cooled with sea water before exiting the tail pipe at the
back of the boat. The fishing vessel had a vertical, “dry” exhaust stack, meaning no sea water
was injected into the exhaust. For both vessels, exhaust pipes from main engines (used for
propulsion) and auxiliary engines (used for onboard electricity generation) were spaced close
together. Consequently, emission factors from onboard measurements are interpreted to be a
composite of main and auxiliary engine emissions unless otherwise noted. While each vessel
operated, the plume capture method was simulated by holding a sample line within three meters
of the boats’ exhaust for approximately 1 to 3 seconds and then pulling the sample line out of the
exhaust to capture a baseline concentration. Plumes were measured immediately upon exiting the
tail pipe. Results from onboard measurements are shown in Figure C.1 and are discussed in the
main text.
124

Figure C.1: Distribution of BC emission factors measured on-board a passenger boat with wet
exhaust and a fishing boat with dry exhaust.

Appendix D: Instrumentation
A LI-COR LI-840A CO2/H2O Gas Analyzer was used to measure CO2 emissions at 1 Hz.
An external pump with a flow rate of 110 cc/min drew air through a ~6 m long sample line made
of Teflon tubing and through a 1-micron filter before reaching the sample chamber of the LI-
840A. This instrument works by shining infrared light through a channel filled with the sampled
air to a detector at the end of the optical path. The detector then measures the intensity of the
remaining light and derives the mole fraction of CO2 in air (LI-840A Instruction Manual).
Black carbon (BC) mass concentrations were measured in parallel to CO2 measurements
at 1 Hz using a custom-built Aerosol Black Carbon Detector (ABCD) developed by University
125
of California, Berkeley (Caubel et al., 2018, Caubel et al., 2019). In this instrument, two LEDs
shine light centered at 880 nm, where BC is the predominant absorbing aerosol species, through
the sample filter and a blank reference filter, to two detectors that determine attenuation. Light
attenuation is then related to BC concentration using a mass attenuation coefficient (MAC) for
BC of 12.5 m
2
/g. This MAC value has been widely used in previous studies that use similar
filters to those described above (Caubel et al., 2018). An example plume intercept is show in
Figure D.1.

Figure D.1: An example of a plume intercept from onshore measurements. The yellow regions
denote the area calculated to determine the emission factor.

126
During three of the ten days of onshore measurements, a single particle soot photometer
(SP2) was used in parallel with the ABCD and LI-840A to measure size-resolved refractory
black carbon (rBC) mass and number concentrations. The SP2 detects rBC particles with mass
equivalent-diameters between ~70 and 500 nm. This instrument uses a laser with a wavelength
of 1064 nm to induce incandescence of individual rBC particles, which is then related to rBC
mass (Single Particle Soot Photometer User Manual). Light scattering is simultaneously
measured for each particle to determine particle size.

Appendix E: Moving Average Data Smoothing Technique
BC concentrations from onshore measurements were often low relative to baseline
concentrations and instrument noise, so data were smoothed using a centered, moving average.
In this smoothing approach, each individual data point was replaced by the arithmetic mean of
the neighboring data points within a given window. Smoothed BC plumes matched well with
corresponding CO2 plumes when superimposed, even in instances where peak BC concentrations
were very low (Figure E.1). To determine an optimal window size (i.e., the number of data
points used for the moving average), we performed statistical and graphical analyses.
127

Figure E.1: Example of a low concentration BC peak (orange) smoothed with a 40-second
moving average and the corresponding CO2 signal (blue).

BC emission factors were calculated for five plumes sampled during onshore
measurements using smoothed data with window widths ranging from 5 to 100 seconds (Figure
E.2). Note that the measurements were made at 1 Hz, so each second of sampling corresponds to
one data point. BC emission factors for each plume showed the most variability by window
width for widths less than 30 seconds and greater than 50 seconds. Explanations for the observed
variability are discussed below.
128

Figure E.2: BC emission factors for five different plumes that were computed using smoothed
data with varying window widths. The moving average window size corresponds to the number
of data points used to smooth the BC data.

Noise in the measurements was analyzed by smoothing 200 seconds of ambient ABCD
measurements (i.e., without a plume present) using window widths of 2 to 100 seconds and
calculating the resulting standard deviation (Figure E.3). The standard deviation decreased
exponentially until window size reached 45 seconds. When window widths were increased
beyond 45 seconds, the standard deviation of the ambient ABCD measurements responded
approximately linearly to increased window size. Therefore, the greatest reduction in noise per
increase in window width was achieved at window widths less than 45 seconds long.
129

Figure E.3: Noise observed in 200 seconds of ABCD measurements of ambient BC after
smoothing data with different window widths.

Recall that the moving baseline BC concentrations used to compute emission factors (see
Equation 1 in the main text) were determined by linearly interpolating between the 10 second
averages preceding and following each plume capture. By definition, the baseline concentration
should be lower than the plume capture concentration, and hence the baseline subtraction should
result in only positive BC concentrations. The effect of instrument noise on this baseline
subtraction was analyzed by smoothing individual plume captures using window widths that
ranged from 10 to 100 seconds. Smoothed plumes and their corresponding baseline BC
concentrations were plotted for each window size, as is illustrated by the example comparison
130
shown in Figure E.4. When plumes were smoothed using window widths less than 30 seconds,
BC concentrations (black line) were occasionally lower than the computed baseline (red dashed
line) . As window width increased, the noise observed in BC data decreased and fewer BC
concentrations fell below the computed baseline. Increasing the window width did not
significantly affect the baseline computed for each plume, further corroborating the analysis on
smoothing ambient measurements that was discussed above and is shown in Figure E.3. Taken
together, these results suggest that the variability observed in emission factors for plumes
smoothed with narrow window widths is likely due to instrument noise influencing the baseline
subtraction in the emission factor calculations.
131

Figure E.4: Comparison of the moving BC baseline (red dashed line) calculated for Plume 4
(black) when data were smoothed using a 20-second moving average (A) versus a 40-second
moving average (B).
BC plumes from onshore measurements were analyzed graphically after data were
smoothed using window widths of 15 to 100 seconds (Figure E.5). As window widths became
wider, BC plumes became broader. At wide window widths greater than 60 seconds, plumes
became so broad that they intersected with nearby plumes such that individual emission factors
could not be calculated. At mid-range window widths, plumes did not “spread” enough for
nearby plumes to affect the emission factor calculation. This at least partly explains the
132
variability observed in the emission factors for plumes smoothed with broad window widths in
Figure E.2Error! Reference source not found..
From these analyses, a 40-second moving average was selected for smoothing onshore
data. At this window width, a significant amount of instrument noise is reduced (Figure E.3)
without merging adjacent plumes (Figure E.5). BC data smoothed at this window width are
graphically similar to corresponding CO2 plumes, even in cases where peak BC concentrations
are very low (Figure E.1). These observations suggest that a 40-second moving average is an
effective window size to differentiate the measured BC signal from instrument noise.

Figure E.5: Comparison of Plume 2 smoothed with different window widths.

Appendix F: CO2 Analyzer Correction Factor
The Licor LI-840A CO2/H20 gas analyzer has a known sampling artifact where CO2 is
overestimated when measuring plumes with high peak CO2 concentrations. Lab experiments
133
were conducted in a study currently in preparation to investigate the response of various CO2
analyzers to CO2 concentrations up to 15,000 ppm using certified gas cylinders (Sugrue et al., in
preparation). At concentrations greater than 15,000 ppm, measurements could no longer be
validated within 5% of the calibration concentration and were therefore excluded from analysis.
Plume area measured using an LI-820, a CO2 analyzer with the same specifications as those of
the LI-840A, was compared to that measured by a higher-grade instrument, the LI-7000. The LI-
820 overestimated plume area relative to the LI-7000 and hence resulted in underestimated BC
emission factors. BC emission factors from onboard measurements were adjusted using the
linear relationship between the LI-7000 and LI-820 peak areas, resulting in a correction factor of
1.24.
Appendix G: Filter Loading Correction Factor
Filter-based optical measurements of BC are subject to a sampling artifact known as the
filter loading effect where BC concentrations are underestimated as the filter becomes
progressively more loaded with absorptive material. To minimize this effect, ABCD filters were
changed whenever attenuation reached 70%, and a correction was applied to all measurements
using Equation G.1 from Caubel et al. (2019).
In Equation G.1, BC is the corrected BC concentration (μg/m
3
), BC0 is the original concentration
(μg/m
3
), 𝑎 is an empirical correction factor (0.64), and ATN is attenuation (unitless).
𝐵𝐶 =
𝐵𝐶
3
𝑎 ×expN−
𝐴𝑇𝑁
100
Q+(1−𝑎)

Equation G.1

Appendix H: rBC Size Distributions
134
Mass and number based rBC size distributions were measured for fourteen plume
intercepts from onshore measurements. The average baseline-subtracted mass and number size
distributions are shown in relative units in Figure 3 of the main text. The baseline-subtraction did
not considerably affect the rBC mode diameters (Dp) observed in the number or mass size
distributions, as can be observed by comparing the blue and black lines in Figure 3 in the main
text. The arithmetic mean rBC number emission factor was (6.9 ± 6.3) x 10
13
#/kg (for rBC
diameter ~70 to ~500 nm).
The rBC mode diameter of the baseline-subtracted number distributions ranged from 72.2
to 117.9 nm, with an average of 86.8 nm. This is slightly smaller than that reported by Buffaloe
et al. (2014) of 98 nm. The rBC mode diameter of the baseline-subtracted mass distributions of
the fourteen plumes ranged from 110.9 to 314.5 nm, with an average of 125.4 nm. This is
considerably smaller than the average volume-weighted mode observed in CalNex (Dp= 175.3
nm; Buffaloe et al., 2014). The discrepancies observed are likely due to two reasons. First, the
CalNex study applied a size-dependent correction factor to account for the fall-off in SP2
detection efficiency for particles smaller than 100 nm, which was not done in the current study
(Buffaloe et al., 2014). This correction factor resulted in bimodal mass and number size
distributions with one mode observed in the <60 nm size range and a second mode observed
around the mode diameters mentioned above. Before the correction factor was applied, the <60
nm peak was not present in the number or mass size distributions reported by CalNex (Buffaloe
et al., 2014). Secondly, the CalNex study measured rBC from both harbor craft and large ocean-
going vessels, which may have influenced the size distributions observed. To the author’s
knowledge, the current study presents the first rBC size distributions measured specifically from
harbor craft emissions.
135
The arithmetic mean rBC number emission factor was (6.9 ± 6.3)×10
13
#/kg (for rBC
diameter 70–500 nm). This is approximately two orders of magnitude less than the PN emission
factor reported for the 2013 heavy-duty diesel drayage truck fleet (2.47 ± 0.48)×10
15
g/kg and
three orders of magnitude less than the PN emission factor reported for locomotives of (2.1 ±
1.5)×10
16
#/kg (Preble et al., 2015, Krasowsky et al., 2015). These large differences may be
driven by a difference in measurement technique, though, given that the SP2 reports BC rather
than total PN and has a different cut-off diameter than the instruments used in the other studies.

Appendix I: Skewness of Tugboat Emissions
Figure I.1 illustrates the skewness of tugboat emissions by operating mode. BC emissions
were most skewed when tugboats were operating without load. This might reflect greater
operational variability in the without load operating mode since tugboats are able to cruise and
accelerate freely when not attached to cargo barges and tankers with tow lines.

Figure I.1: Cumulative BC emission factor distributions of tugboat measurements by operating
mode. If normally distributed, these data would plot in a straight line. Note that the distribution
136
shown here changes minimally when repeat measurements per vessel are averaged to show the
fraction of vessels (as opposed to plumes) from dirtiest to cleanest.

137
Appendix for Chapter 3

Text S1. Chemistry Schemes
The chemistry schemes selected included the Madronich TUV photolysis scheme
(Madronich, 1987), the RACM-ESRL scheme for gas phase chemistry (Kim et al., 2009), and
the MADE/VBS aerosol scheme (Ackermann et al., 1998; Ahmadov et al., 2012). Lastly, a
modified version of the Wesley dry deposition scheme was used for compatibility with the 33-
class land cover scheme described in the Methodology (Fallmann et al., 2016).

Text S2. Boundary conditions
WRF-chem as applied here requires initial and boundary meteorological and chemical
conditions. The North American Regional Reanalysis (NARR) dataset was used for initial and
boundary meteorology conditions for all three domains (Mesinger et al., (2006). Chemistry
boundary conditions were generated for the outermost domain and chemistry initial conditions
were generated for all three domains using the Model for Ozone and Related chemical Tracers
Version 4 (Emmons et al., 2010).

Text S3. Anthropogenic emissions
Gridded anthropogenic emissions datasets are required for running WRF-Chem. For the
outer two domains, an emissions inventory from the California Air Resource Board (CARB) was
used for areas within California and the National Emissions Inventory (NEI) from the U.S.
Environmental Protection Agency (EPA) was used for areas outside of California that were still
within the simulation domains (CARB; US EPA, 2014). The CARB emissions dataset is
composed of hourly emissions for the year 2012 at 4-km resolution and represents the most up-
138
to-date emissions data set available. These emissions inventories were regridded to match the
model domains and chemical speciation was converted to align with the RACM-ESRL and
MADE/VBS mechanisms following Li et al. (2019).

Text S4. Biogenic emissions
Biogenic emissions were calculated online with meteorology using the Model of
Emissions of Gases and Aerosols from Nature (MEGAN) version 2.06 (Guenther et al., 2006).
MEGAN uses a gridded emissions dataset for isoprene emissions and a look up table for non-
isoprene species of biogenic volatile organic compounds (BVOCs). Then, BVOC emissions are
calculated by adjusting base emission factors to account for deviations from standard
temperature, pressure, and relative humidity. Additional adjustments are made to account for
canopy loss, canopy production, and activity (Guenther et al., 2006).

The default MEGAN base
emission factors, plant functional type distributions, and leaf area index data were used for all
simulations.

139

Figure S1. The two-way nested domains used in the model simulations (left) and the spatial
distribution of the urban fraction in the innermost domain (right).

140

Figure S2. Panel (a) shows the MODIS-derived albedo used in the Baseline and GVF50
simulations. Panel (b) shows the albedo used in the Baseline_Albedo simulation that was
calculated by the Noah Land Surface Model using the green vegetation fraction of current day
land cover. Panel (c) and Panel (d) show the change in albedo from increasing urban vegetation
by 50% in GVF50_Low_Albedo and GVF50_High_Albedo compared to Baseline_Albedo.

141

Figure S3. The diurnal cycle of near surface wind speed in urban grid cells for the baseline
simulations (a) and the changes to windspeed modeled in the urban greening simulations
compared to their respective baselines (b).

Figure S4. The average daily change in surface temperature (a) and 2m air temperature (b) for
the GVF50_High_Albedo simulation.
142

Figure S5. Hourly 2m air temperature data from the Baseline simulation compared to
observational data from the US EPA Air Quality System network of sensors.

143
Appendix for Chapter 4

S1 Biogenic Emissions Scenarios
As described in Section 2.3.2 of the main text, we adjusted the gridded input data used by
MEGAN to simulate increased BVOC emissions from urban greening. Figure S1 shows the
gridded isoprene emission factors used in the different BVOC emissions scenarios. For
monoterpene emissions, the plant functional type distributions were adjusted, which are shown in
Figure S2.

Figure S1: The gridded isoprene emission factors for (a) the baseline biogenic emissions
scenario, (b) the low isoprene emission scenario and (c) the high isoprene emission scenario.
Panel (a) was used as an input for the following simulations: Baseline, Baseline_LA100, GVF50,
and GVF50_LA100. Panel (b) was used as an input for BVOC_Low and BVOC_Low_LA100.
Panel (c) was used as an input for BVOC_Med, BVOC_Med_LA100, BVOC_High, and
BVOC_High_LA100.

144

Figure S2: The left-hand column shows the change in the plant functional type distribution for
broadleaf trees compared to baseline simulations and the right-hand column shows changes in
the needleleaf tree distribution. Note that Baseline, Baseline_LA100, GVF50, and
GVF50_LA100 used MEGAN defaults for the plant functional type distributions.

145
S2 Model validation results
We validated our model results for the Baseline simulation against observational data
from the US EPA Air Quality System network (US EPA Air Quality Systems Data Mart). Figure
S3 and S4 show timeseries comparison for 2m air temperature and hourly O3 concentrations,
respectively. Table S1 shows the summary statistics for compared against recommended model
performance benchmarks from Emery et al. (2017).

Figure S3: The mean hourly 2m air temperature observations from EPA Air Quality Systems
compared to the corresponding hourly outputs from the Baseline simulation.
146

Figure S4: The mean hourly O3 concentrations from EPA Air Quality Systems compared to the
corresponding hourly outputs from the Baseline simulation.

Table S1: Model performance compared to recommended benchmarks (Emery et al., 2017)
Variable Metric Recommended Benchmark Results
Daily 8-hour O3
NMB <±15% -0.22
NME <25% 0.26
r >0.5 0.65
Hourly O3
NMB <±15% -0.19
NME <25% 0.31
r >0.5 0.76
147
Hourly NO2
NMB <±65% -0.07
NME <115% 0.53
PM2.5
NMB <±30% -0.14
NME <50% 0.36
r >0.4 0.14
Hourly Temp
MB <0.5 -0.44
ME <2 K 2.09

S3 Changes to meteorology from urban greening
Increasing urban vegetation by 50% led to modest 2m air temperature changes in the
urban greening simulations. Since aerosol effects on temperature were not included in the model
configuration, temperature changes were approximately equal in all of the urban greening
simulations, as illustrated by Figure S5. Increasing urban vegetation also increased surface
roughness. This resulted in slightly decreased near surface wind speeds in urban grid cell, as
shown in Figure S6.
148

149

Figure S5: The mean change in 2m air temperature for each simulation compared to their
respective baselines.

Figure S6: The mean diurnal cycle of wind speed for urban grid cells in the baseline simulations
and in the urban greening simulations.

S4 Changes in nighttime O3
As discussed in Section 4.3.4 of the main text, urban greening led to reduced nighttime
urban O3 concentrations and increased O3 concentrations in nonurban areas. Mean nighttime
changes in O3 are shown in Figure S7. Changes in nighttime NOx concentrations are consistent
with enhanced O3 titration in urban areas. Fig S8 shows that nighttime changes in O3 were highly
correlated to nighttime changes in NOx. Spatially, the areas with decreased nighttime O3 were
visually consistent with areas with increased NOx concentrations, shown in Figure S9. Finally,
150
nighttime changes in O3 were somewhat correlated with changes in CO (Figure S10), which can
be used as a first approximation for changes in ventilation.
151

152

Figure S7: The average change in nighttime O3 concentrations for each simulation.

Figure S8: Linear regression between nighttime changes in NOx concentrations and nighttime
changes in O3 concentrations. The top panel shows results for the current day anthropogenic
emissions scenario and the bottom panel shows results for the aggressive electrification
emissions scenario.

153

154

Figure S9: Spatial distributions of the average change in nighttime (20:00 – 06:00 LST) NOx
concentrations.

155
Figure S10: Linear regression between nighttime changes in CO concentrations, which can be
used as a first approximation for ventilation, and nighttime changes in O3 concentrations. The top
panel shows results for the current day anthropogenic emissions scenario and the bottom panel
shows results for the future anthropogenic emissions scenario.

S5: Population weighted changes in DM8HO3
Here, we describe the geospatial analysis used to find population weighted changes daily
maximum 8-hour O3 (DM8HO3) concentrations. First, population counts for each census tract
were joined to a shapefile containing census tracts for the state of California for the year 2021 by
matching the GEOIDs . Next, the center points of each WRF grid cell in the innermost domain
were spatially joined to the census tracts that contained them using the geoprocessing “clip” tool.
Finally, for each census tract, the mean change in DM8HO3 was calculated for the WRF grid
cells that they contained. The resulting population weighted distributions in DM8HO3 are shown
in Figure 4-8 of the main text. While the census tracts contained varying numbers of WRF grid
cells, no correlation was observed between sample size and the average change in DM8HO3, as
shown in Figure S11.

156

Figure S11: Sample size of WRF grid cells versus the mean change in DM8HO3 calculated for
each census tract.

Abstract (if available)

Abstract Urban areas are warmer than their rural surroundings due to the Urban Heat Island (UHI) effect and experience elevated air pollutant concentrations. As anthropogenic climate change progresses and urban populations continue to grow, the UHI effect will be exacerbated and human exposure to both extreme heat and air pollution will be affected. Thus, it is more important now than ever before to find synergistic solutions that mitigate both urban air pollution and urban heat. Through field measurements and atmospheric modeling, this dissertation aims to inform effective, optimized policy that reduces heat stress and improves urban air quality. First, I describe an extensive measurement campaign that characterizes black carbon emissions from harbor craft, a category of vessels that operate close to shore, contributing to particulate matter pollution in coastal communities. Black carbon is not only harmful to human health, but also contributes to global and regional warming through direct and indirect effects. In Chapter 2, I shift focus to UHI mitigation and use atmospheric modeling to quantify the competing warming effects of urban greening. Here, I reveal the critical role that albedo plays in determining the temperature effects of urban greening. Finally, in Chapter 3, I investigate O3 sensitivity to the chemical and physical effects of urban greening under a current day and reduced anthropogenic emissions scenario. This dissertation underscores the intricate interplay between urban meteorology and atmospheric chemistry, revealing mitigation pathways to equitably reduce urban heat and air pollution exposure while avoiding unintended trade-offs of individual mitigation strategies.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

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

PDF

Impacts of heat mitigation strategies and pollutant transport on climate and air quality from urban to global scales

PDF

Identifying and mitigating the effects of urban heat islands in California

PDF

Evaluating the role of energy system decarbonization and land cover properties on urban air quality in southern California

PDF

Using demand-side management for decarbonization: developing methods to quantify the impact of altering electricity consumption patterns

PDF

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

PDF

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

PDF

Examining exposure to extreme heat and air pollution and its effects on all-cause, cardiovascular, and respiratory mortality in California: effect modification by the social deprivation index

PDF

Residential electricity demand in the context of urban warming: leveraging high resolution smart meter data to quantify spatial and temporal patterns…

PDF

Developing a replicable approach for the creation of urban climatic maps for urban heat island analysis: a case study for the city of Los Angeles, California

PDF

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

PDF

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

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

PDF

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

PDF

Investigating the role of climate in affecting residential electricity consumption through high spatiotemporal resolution observations

PDF

Evaluating energy consuming behaviors and the sufficiency of urban systems in the context of extreme heat hazards

PDF

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

PDF

Urban air pollution and environmental justice: three essays

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

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

PDF

Mitigating the urban heat island effect: thermal performance of shade-tree planting in downtown Los Angeles

Asset Metadata

Creator Schlaerth, Hannah Lynn (author)

Core Title From tugboats to trees: investigating the coupled systems of urban air pollution and meteorology

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Environmental Engineering

Degree Conferral Date 2023-08

Publication Date 01/10/2024

Defense Date 05/08/2023

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

Tag Air pollution,albedo,atmospheric science,black carbon,climate change,OAI-PMH Harvest,urban greening,urban heat island,WRF

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Silva, Sam J. (committee chair), Berelson, William (committee member), Sanders, Kelly (committee member)

Creator Email hannahschlaerth@gmail.com,hschlaer@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113263168

Unique identifier UC113263168

Identifier etd-SchlaerthH-12054.pdf (filename)

Legacy Identifier etd-SchlaerthH-12054

Document Type Dissertation

Format theses (aat)

Rights Schlaerth, Hannah Lynn

Internet Media Type application/pdf

Type texts

Source 20230710-usctheses-batch-1065 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

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

Repository Email cisadmin@lib.usc.edu