kong. - theses.lib.polyu.edu.hk · environment, c. 2-c. 4. rono. 2. decreased considerably as a...
TRANSCRIPT
Copyright Undertaking
This thesis is protected by copyright, with all rights reserved.
By reading and using the thesis, the reader understands and agrees to the following terms:
1. The reader will abide by the rules and legal ordinances governing copyright regarding the use of the thesis.
2. The reader will use the thesis for the purpose of research or private study only and not for distribution or further reproduction or any other purpose.
3. The reader agrees to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
IMPORTANT
If you have reasons to believe that any materials in this thesis are deemed not suitable to be distributed in this form, or a copyright owner having difficulty with the material being included in our database, please contact [email protected] providing details. The Library will look into your claim and consider taking remedial action upon receipt of the written requests.
Pao Yue-kong Library, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
http://www.lib.polyu.edu.hk
PHOTOCHEMICAL FORMATION OF ALKYL NITRATES AND
THEIR IMPACTS ON OZONE PRODUCTION IN HONG KONG
LYU XIAOPU
PhD
The Hong Kong Polytechnic University 2018
The Hong Kong Polytechnic University
Department of Civil and Environmental Engineering
PHOTOCHEMICAL FORMATION OF ALKYL NITRATES AND THEIR IMPACTS ON OZONE PRODUCTION IN HONG
KONG
LYU XIAOPU
A thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy
September 2017
I
CERTIFICATE OF ORGINALITY
I hereby declare that this thesis is my own work and that, to the best of my knowledge and
belief, it reproduces no material previously published or written, nor material that has been
accepted for the award of any other degree or diploma, except where due acknowledgement
has been made in the text.
(Signed)
LYU XIAOPU (Name of student)
II
ABSTRACT
Alkyl nitrates (RONO2) are important constituents of organic nitrates and atmospheric odd
nitrogen. The widely recognized sources of RONO2 include emissions from equatorial oceans,
biomass burning and photochemical formation. Since RONO2 stabilize the alkyl and nitrate
groups in the molecules, they generally pose non-negligible impact on atmospheric
photochemistry in the processes of their formation and degradation. To understand the
spatiotemporal patterns of RONO2 and their parent hydrocarbons in Hong Kong, the formation
routes of RONO2 and the impacts of RONO2 photochemistry on ozone (O3) production, whole
air samples were collected over a wide range of environments in Hong Kong through the first
sampling in 2001 to the latest sampling in 2014. Based on the concentrations of speciated C1-
C5 RONO2, the photochemical formation, evolution and degradation of RONO2, as well as
their impacts on O3 production were studied in this thesis.
Results indicated that the C1-C5 RONO2 increased significantly in Hong Kong during the past
15 years (2001-2014), which was mainly attributable to the increased abundances of parent
hydrocarbons and enhanced oxidative capacity of the atmosphere (e.g. increased O3). The
spatially disproportional distributions of the parent hydrocarbons and regional transport from
inland Pearl River Delta led to higher levels of RONO2 in northwestern Hong Kong than in
east of Hong Kong.
With regard to observed changes in parent hydrocarbons of RONO2, this work mainly focused
on the evaluation of a program implemented by Hong Kong Environmental Protection
Department for the purpose of reducing the emissions of volatile organic compounds (VOCs)
and nitrogen oxides (NOx) from liquefied petroleum gas (LPG)-fueled vehicles. Overall, this
program effectively decreased the emissions of VOCs from LPG-fueled vehicles, including the
parent hydrocarbons of RONO2 (such as propane and n-butane). Due to the concurrent
III
emission reduction of NOx, the atmospheric oxidative capacity (represented by O3, OH and
HO2) slightly increased after the program. In the highly vehicle-populated roadside
environment, C2-C4 RONO2 decreased considerably as a result of the substantial reductions of
parent hydrocarbons. However, RONO2 at the roadside sites were still not obviously lower than
at the other locations in Hong Kong.
This work for the first time updated an advanced protocol in a photochemical box model
describing atmospheric processes of speciated RONO2, based on master chemical mechanism
(a near-explicit chemical mechanism). With the appropriate settings of branching ratios and
dry deposition velocities, the model well reproduced the observed RONO2. A branching ratio
of 0.0003 was suggested as the most appropriate value for CH3O2 reacting with NO to form
CH3ONO2, with an estimated uncertainty of less than 50%. The first application of the model
at a coastal site in Hong Kong confirmed that the gas-phase formations of RONO2 were
dominated by the pathway of alkylperoxy radicals (RO2) reacting with nitric oxide (NO) for
C2-C5 RONO2. However, the association between methoxy radicals (CH3O) and nitrogen
dioxide (NO2) made considerable contribution to CH3ONO2. RONO2 formation led to the
reductions of O3 production at this coastal site, due to the extraction of oxidative radicals from
the atmosphere. As a temporary nitrogen reservoir, C1-C5 RONO2 constituted approximately
4% of the total nitrogen as estimated by the model.
Regional transport and meso-scale circulation were important factors elevating RONO2 levels
at the mountainous site where the parent hydrocarbons were much less abundant than at a
coupled urban site. In addition, the oxidative capacity of the atmosphere at the mountainous
site was stronger than that at the urban site, as identified by the photochemical box model
incorporating master chemical mechanism (PBM-MCM). As a result, the oxidation efficiencies
of parent hydrocarbons at the mountainous site were higher than at the urban site, thus
increasing the production of RO2 radicals and the production of RONO2. This led to
IV
comparable or even higher concentrations of some RONO2 at the mountainous site, but not for
all the RONO2 species.
Furthermore, the relationships between RONO2 production and their precursors were
comprehensively studied with the aid of the PBM-MCM model. The isopleths of RONO2
production clearly demonstrated the VOC-limited and NOx-limited regime in controlling
RONO2 formation. The dividing ratio of TVOC/NOx between NOx-limited and VOC-limited
regime for C2-C4 RONO2 shifted from 8.7/1 ppbv/ppbv to 10.0/1 ppbv/ppbv from the urban
environment to the mountainous environment. However, the ratio was decreased to 3.1/1 and
2.4/1 ppbv/ppbv for the formation of CH3ONO2 at the urban and mountainous site, respectively.
This discrepancy was mainly caused by the pathway of CH3O reacting with NO2 that
contributed significantly to the production of CH3ONO2 under high NOx. RONO2 formation
led to the reduction of O3 at both the urban and the mountainous sites. Moreover, different
mechanisms were found in terms of the impacts of RONO2 degradation on O3 production,
including NO2 stimulating, NO2 suppressing and radicals stimulating. In addition, model
simulations revealed the greater impacts of organic nitrates formed from the precursors of
higher O3 formation potentials (e.g. alkenes, isoprene and aromatics) on O3 production, which
needs further study in the future.
In summary, this work addressed knowledge gaps in RONO2 research and made contributions
to the scientific community, including the development of the RONO2 module in the PBM-
MCM model, the verification of RONO2 formation pathways, the first isopleths showing the
RONO2-VOC-NOx relationships, and the unprecedented finding of the mechanisms of RONO2
degradation influencing O3 formation, including NO2 stimulation, NO2 suppression and RO
stimulation.
V
THE NOVELTY OF THIS STUDY
In the scientific community, RONO2 has been recognized as a non-negligible constituent of
odd nitrogen, and a temporary nitrogen reservoir. Previous studies on RONO2 mainly focus on
the kinetic studies of their reaction rates (including reactions involved in their formation and
degradation) and their sources on the basis of observation data. Although RONO2 has a close
relationship with O3, a quantitative evaluation of the interactions between them has been
seldom conducted in model studies, possibly due to the uncertainties in chemical mechanisms
of RONO2 and lack of sufficient validation of the related mechanisms in many existing
chemical transport models and chemical box models. Hence, the overall objective of this study
is to address significant knowledge gaps and advance understanding of atmospheric RONO2.
The novelty of this study includes but is not limited to the following aspects:
(i) The long-term observations of RONO2 over the entire territory of Hong Kong indicated that
C1-C4 RONO2 increased in the past 15 years, which might be caused by the increases of their
parent hydrocarbons in the atmosphere and the enhanced oxidative capacity of the atmosphere.
In addition, the spatial patterns clearly implied the regional transport of RONO2 to Hong Kong.
(ii) The module describing the atmospheric processes of RONO2 was developed and integrated
into the PBM-MCM model, which experienced a decade’s development and upgrade by our
group. The RONO2 module considers the impacts of the atmospheric processes, including
background concentration, photochemical formation, degradation and dry deposition, on the
fate of RONO2. In particular, a most appropriate branching ratio of 0.0003 was determined for
the formation of CH3ONO2 from CH3O2 reacting with NO, with the uncertainty less than 50%.
For the larger (≥C2) RONO2, the model parameters were based on the values reported in the
VI
literature. With the most appropriate settings, the model well captured observed variations of
RONO2 in ambient air.
(iii) The relative contributions of RO2 reacting with NO (RO2+NO) and RO reacting with NO2
(RO+NO2) to RONO2 production have been seldom quantified in previous studies. With the
aid of the PBM-MCM model, we found that RO2+NO was the dominant route for C2 – C5
RONO2. However, the pathway of RO+NO2 made considerable contribution to CH3ONO2,
particularly in polluted environments. Furthermore, this study advanced our understanding on
the production of RONO2. The critical role of the oxidative capacity of atmosphere in RONO2
formation was emphasized, in addition to the abundances of precursors and the branching ratios.
We showed that the production of RONO2 was significantly influenced by the oxidative
capacity of the atmosphere, which determined the oxidation efficiencies of parent
hydrocarbons. For example, higher atmospheric oxidative capacity at a mountainous site led to
comparable or even higher productions of RONO2, despite less abundant RONO2 precursors.
(iv) This study established the non-linear relationships between RONO2 and their precursors,
expressed as isopleths of RONO2 production. The isopleths clearly described the RONO2-
VOC-NOx relationships over a wide range of environments. Specifically, the NOx-limited and
VOC-limited regimes in RONO2 formation and the corresponding dividing ratios of
TVOCs/NOx were determined. Moreover, the impacts of RONO2 degradation on O3 production
were studied in different environments. In the NOx-limited regime, the impact was manifested
as NO2 stimulating, resulting in O3 increment as a consequence. However, it shifted to NO2
suppressing (O3 decrease) and RO stimulating (O3 increase) under relatively higher and lower
ratios of TVOCs/NOx in the VOC-limited regime, respectively. To our best knowledge, this is
the first study that explored the diverse mechanisms of RONO2 photochemistry which
influenced O3 formation.
VII
PUBLICATIONS 1. Lyu, X.P., Chen, N., Guo, H. *, Zhang, W.H., Wang, N., Wang, Y., and Liu, M., 2016.
Ambient volatile organic compounds and their effect on ozone production in Wuhan, central China. Science of The Total Environment 541, 200-209. (IF: 3.976)
2. Lyu, X.P., Ling, Z.H., Guo, H. *, Saunders, S.M., Lam, S.H.M., Wang, N., Wang, Y., Liu, M., and Wang, T., 2015. Re-examination of C 1–C 5 alkyl nitrates in Hong Kong using an observation-based model. Atmospheric Environment 120, 28-37. (IF: 3.459)
3. Lyu, X.P., Liu, M., Guo, H. *, Ling, Z.H., Wang, Y., Louie, P.K.K., and Luk, C.W.Y., 2016. Spatiotemporal variation of ozone precursors and ozone formation in Hong Kong: Grid field measurement and modelling study. Science of The Total Environment 569, 1341-1349. (IF: 3.976)
4. Lyu, X.P., Guo, H. *, Simpson, I.J., Meinardi, S., Louie, P. K., Ling, Z., Wang, Y., Liu, M., Luk, C.W.Y., Wang, N., and Blake, D.R., 2016. Effectiveness of replacing catalytic converters in LPG-fueled vehicles in Hong Kong. Atmospheric Chemistry and Physics 16(10), 6609-6626. (IF: 5.114)
5. Lyu, X.P., Chen, N., Guo, H. *, Zeng, L.W., Zhang, W.H., Shen, F., Quan, J.H., and Wang, N., 2016. Chemical characteristics and causes of airborne particulate pollution in warm seasons in Wuhan, central China. Atmospheric Chemistry and Physics 16(16), 10671-10687. (IF: 5.114)
6. Lyu, X.P., Zeng, L.W., Guo, H. *, Simpson, I.J., Ling, Z.H., Wang, Y., Murray, F., Louie, P.K.K., Saunders, S.M., Lam, S.H.M., and Blake, D.R. 2016. Evaluation of the effectiveness of air pollution control measures in Hong Kong. Environmental Pollution 220, 87-94. (IF: 4.839)
7. Lyu, X.P., Guo, H. *, Wang, N., Simpson, I.J., Cheng, H.R., Zeng, L.W., and Blake, D.R., 2017. Modelling of C1-C4 alkyl nitrate photochemistry and their impacts on O3 production in urban and suburban environments of Hong Kong. Journal of Geophysical Research doi: 10.1002/2017JD027315. (IF: 3.318)
8. Lyu, X.P., Guo, H. *, Cheng H.R., Wang, X.M., Ding, X., Lu, H.X., Yao, D.W., and Xu, C., 2017. Observation of SOA tracers at a mountainous site in Hong Kong: Chemical characteristics, origins and implication on particle growth. Science of The Total Environment 605, 180-189. (IF: 3.976)
9. Lyu, X.P., Guo, H. *, Cheng, H.R., and Wang, D.W., 2017. New particle formation and growth at a suburban site and a background site in Hong Kong. Chemosphere 193, 664-674. (IF: 4.205)
10. Wang, H. #, Lyu, X.P. # (equal contribution), Guo, H. *, Wang Y., Zou, S., Ling, Z., Wang, X., Jiang, F., Zeren, Y., Pan, W., Huang, X. and Shen, J., 2018. Ozone pollution around a coastal region of South China Sea: Interaction between marine and continental air. Atmospheric Chemistry and Physics 18, 4277-4295.
11. Lyu, X.P., Wang, N., Guo, H., Xue, L., Pan, W., Zeren, Y., Cheng, H., Han, L., and Zhou, Y., 2018. Causes of a continuous summertime O3 pollution event in Ji’nan, a central city in North China Plain. Journal of Geophysical Research, under revision.
VIII
12. Zeng, L., Lyu, X., Guo, H., Zou, S., and Ling, Z., 2018. Photochemical formation of C1-C5 alkyl nitrates in suburban Hong Kong and over South China Sea. Environmental Science and Technology doi:10.1021/acs.est.8b00256. (IF: 6.198)
13. Ling, Z., Guo, H. *, Simpson, I. J., Saunders, S. M., Lam, S. H. M., Lyu, X.P., and Blake, D. R., 2016. New insight into the spatiotemporal variability and source apportionments of C1–C4 alkyl nitrates in Hong Kong. Atmospheric Chemistry and Physics 16, 8141-8156, doi:10.5194/acp-16-8141-2016. (IF: 5.114)
14. Wang, N., Lyu, X.P., Deng, X.J. *, Guo, H., Deng, T., Li, Y., Yin, C. Q., Li, F., and Wang, S.Q., 2016. Assessment of regional air quality resulting from emission control in the Pearl River Delta region, southern China. Science of the Total Environment 573, 1554-1565. (IF: 3.976)
15. Guo, H. *, Ling, Z.H., Cheng, H.R., Simpson, I.J., Lyu, X.P., Wang, X.M., Shao, M., Lu, H.X., Ayoko, G., Zhang, Y.L., Saunders, S.M., Lam, S.H.M., Wang, J.L., and Blake, D.R., 2017. Tropospheric volatile organic compounds in China. Science of The Total Environment 574, 1021-1043. (IF: 3.976)
16. Zhan, L., Lin, T., Wang, Z., Cheng, Z., Zhang, G., Lyu, X., and Cheng, H., 2017. Occurrence and air–soil exchange of organochlorine pesticides and polychlorinated biphenyls at a CAWNET background site in central China: Implications for influencing factors and fate. Chemosphere 186, 475-487. (IF: 4.205)
17. Wang, Y., Wang, H., Guo, H.*, Lyu, X.P., Cheng, H.R., Ling, Z.H., Louie, P.K.K., Simpson I.J., Meinardi, S., and Blake, D.R., 2017. Long-term O3-precursor relationships in Hong Kong: Field observation and model simulation. Atmospheric Chemistry and Physics 17, 10919-10935. (IF: 5.114)
18. Wang, Y., Guo, H., Zou, S., Lyu, X., Ling, Z., Cheng, H., Zeren, Y., 2018. Surface O3 photochemistry over the South China Sea: Application of a near-explicit chemical mechanism box model. Environmental Pollution 234,155-166. (IF: 4.839)
IX
ACKNOWLEDGEMENTS
I have been truly blessed to spend the three years of good time in The Hong Kong Polytechnic
University (PolyU), which will be an unforgettable memory for my lifetime. Firstly, I would
like to thank my supervisor, Professor Hai Guo, for providing me all necessary resources for
my PhD study. More importantly, I must show my great appreciation to him for his rich
knowledge, patience and kindness in guiding my PhD study, and his cares in my livelihood.
What deeply impressed me in the past three years was that the light was always on in his office
till midnight, and he would like to spend a whole night discussing scientific issues with me
without any complaint. In many cases, I prefer to say we are the comrades in a battle.
My sincere thanks also goes to our collaborators, including Dr. Peter Louie in Hong Kong
Environmental Protection Department, Dr. Shengwen Liang in Wuhan Environment
Monitoring Center, Dr. Nan Chen in Hubei Provincial Environment Monitoring Center for their
support and recognition; to the other experts who gave valuable suggestions to my PhD study,
including Prof. Allen Goldstein in University of California, Berkeley, Profs. Tao Wang, Shun-
Cheng Lee and Xiangdong Li in CEE of PolyU, Prof. Weihao Zhang and Dr. Hairong Cheng
in Wuhan University; to the staff in our department and university, particularly Mr. Tam in air
lab, Emmy in general office and Elsie in research office who also provided excellent support
to my study.
I am also grateful to my team members and friends, Yu Wang, Nan Wang, Ming Liu, Zhenghao
Ling, Dawei Wang, Dawen Yao, Lewei Zeng and others. They helped me a lot in my study and
daily life. In particular, Dr. Zhenghao Ling, a lecturer now in Sun Yat-sen University, taught
me how to construct the models, use the data analysis tools and conduct in-depth data analysis
at the early stage of my PhD study. Specifically, I would like to acknowledge his permission
to having our joint research as an important part of this thesis (Chapter 7).
X
Last but not the least, I would like to thank my family for their support and trust. No matter
where I am and what I do, they are always my strongest backup force.
XI
CONTENTS CERTIFICATE OF ORGINALITY .................................................................................................... I ABSTRACT .......................................................................................................................................... II THE NOVELTY OF THIS STUDY ................................................................................................... V PUBLICATIONS .............................................................................................................................. VII ACKNOWLEDGEMENTS ...............................................................................................................IX CONTENTS.........................................................................................................................................XI Chapter 1 Overview .............................................................................................................................. 1
1.1 Introduction ............................................................................................................................. 1 1.2 Scope of this study ................................................................................................................... 3 1.3 Outline of the thesis................................................................................................................. 5
Chapter 2 Literature Review ............................................................................................................... 7 2.1 Organic nitrates ...................................................................................................................... 7 2.2 Abundance of RONO2 ............................................................................................................. 9 2.3 Sources of RONO2 ................................................................................................................. 11
2.3.1 Oceanic emission ........................................................................................................ 11 2.3.2 Biomass burning ......................................................................................................... 12 2.3.3 Photochemical formation of RONO2 ........................................................................ 13
2.4 Sinks of RONO2 ..................................................................................................................... 18 2.5 Relationships between RONO2 and O3................................................................................ 20 2.6 Air quality studies in Hong Kong ........................................................................................ 22
Chapter 3 Methodology ...................................................................................................................... 28 3.1 Description of sampling sites ................................................................................................ 28 3.2 Sample collection and chemical analysis ............................................................................. 31
3.2.1 VOCs collection .......................................................................................................... 31 3.2.2 Offline and online analysis of VOCs ......................................................................... 32 3.2.3 Continuous measurement of inorganic trace gases ................................................. 35
3.3 PBM-MCM model................................................................................................................. 36 3.3.1 Basic structure of PBM-MCM .................................................................................. 36 3.3.2 Atmospheric processes of RONO2 in PBM-MCM .................................................. 38 3.3.3 Application of PBM-MCM ........................................................................................ 40
3.4 Positive matrix factorization model ..................................................................................... 41 3.5 Other tools and calculation techniques ............................................................................... 43
3.5.1 Calculation of VOCs diurnal patterns ..................................................................... 43 3.5.2 Removal of background concentrations ................................................................... 47 3.5.3 Eliminating interferences of non-local air masses ................................................... 48 3.5.4 Calculation of relative incremental reactivity ......................................................... 50
Chapter 4 Spatiotemporal variations of RONO2 and their parent hydrocarbons in Hong Kong 52 4.1 Long term variations of RONO2 in Hong Kong ................................................................. 53 4.2 Spatial distributions of RONO2 in Hong Kong .................................................................. 59 4.3 Impact of LPG program on parent hydrocarbons ............................................................. 63
4.3.1 Variations of the observed parent hydrocarbons .................................................... 64 4.3.2 Source apportionments of VOCs .............................................................................. 67
4.4 Impact of LPG program on oxidative capacity .................................................................. 72 4.4.1 Model validation and O3 simulation ......................................................................... 72 4.4.2 Impact of the program on O3 formation .................................................................. 76
4.5 Spatial characteristics of photochemical reactivity ............................................................ 77 4.6 Implication on RONO2 abundances .................................................................................... 79 4.7 Sub-conclusions ..................................................................................................................... 83
Chapter 5 Impacts of replacing catalytic converters in LPG-fueled vehicles on the abundances of RONO2 and their parent hydrocarbons in Hong Kong ................................................................... 84
5.1 Variations of LPG-related VOCs and NOx during the intervention program ................. 86 5.1.1 Concentrations of primary LPG-related VOCs and NOx ....................................... 86 5.1.2 Temporal variations of primary LPG-related VOCs and NOx .............................. 89
XII
5.2 Variations of LPG contributions to VOCs and NOx .......................................................... 95 5.2.1 Source identification .................................................................................................. 95 5.2.2 Source contribution .................................................................................................... 97
5.3 Impact of the intervention program on O3 production .................................................... 101 5.3.1 O3 Simulation............................................................................................................ 101 5.3.2 Net O3 production .................................................................................................... 102
5.4 Photochemical reactivity .................................................................................................... 103 5.4.1 OH, HO2 and their formation/loss rates in whole air ........................................... 103 5.4.2 Influence of the program on photochemical reactivity ......................................... 107
5.5 Improvement of the intervention program ....................................................................... 109 5.5.1 O3-VOCs-NOx sensitivity in the whole air ............................................................. 109 5.5.2 O3-VOCs-NOx sensitivity in LPG ........................................................................... 112
5.6 Implication on RONO2 formation ..................................................................................... 114 5.7 Sub-conclusions ................................................................................................................... 117
Chapter 6 Re-examination of C1-C5 alkyl nitrates in Hong Kong using an observation-based model .................................................................................................................................................. 119
6.1 Mixing ratios and seasonal patterns of RONO2 ............................................................... 120 6.2 Construction of the PBM-MCM model............................................................................. 122
6.2.1 Examination of branching ratios ............................................................................ 122 6.2.2 Other settings ............................................................................................................ 123
6.3 Pathways to RONO2 ............................................................................................................ 125 6.3.1 Source apportionment of RONO2 ........................................................................... 125 6.3.2 Photochemical pathways of RONO2 ....................................................................... 134
6.4 Impact on O3 formation ...................................................................................................... 140 6.4.1 Net O3 production .................................................................................................... 140 6.4.2 Nitrogen partitioning ............................................................................................... 142
6.5 Sub-conclusions ................................................................................................................... 143 Chapter 7 New insight into the spatiotemporal variability and source apportionments of C1-C4 alkyl nitrates in Hong Kong ............................................................................................................. 145
7.1 Descriptive statistics of RONO2 and their parent hydrocarbons .................................... 145 7.2. Sources of RONO2 .............................................................................................................. 155
7.2.1. Photochemical evoluation of RONO2 .................................................................... 155 7.2.2. Source apportionment of RONO2 .......................................................................... 162 7.2.3. Contributions of mesoscale circulation, in-situ formation and regional transport to RONO2 at TMS ............................................................................................................. 168
7.3. Relationship between RONO2 and O3 .............................................................................. 171 7.4 Sub-conclusion ..................................................................................................................... 172
Chapter 8 Modelling of C1-C4 alkyl nitrate photochemistry and their impacts on O3 production in urban and suburban environments of Hong Kong .................................................................... 174
8.1 Modelling of C1-C4 RONO2 ................................................................................................ 175 8.1.1 Overview of RONO2 sources ................................................................................... 175 8.1.2 Model construction .................................................................................................. 177 8.1.3 Modelling of CH3ONO2 ........................................................................................... 183 8.1.4 Model validation ....................................................................................................... 188
8.2 Secondary RONO2 formation ............................................................................................ 193 8.2.1 RONO2 formation at TMS and TW ........................................................................ 193 8.2.2 Isopleths of RONO2 formation ................................................................................ 196 8.2.3 Lower thresholds of TVOCs/NOx for CH3ONO2 ................................................... 201
8.3 Impacts on O3 production .................................................................................................. 205 8.3.1 During RONO2 formation ....................................................................................... 205 8.3.2 During RONO2 degradation.................................................................................... 207 8.3.3 Impacts on O3 in different environments ............................................................... 209
8.4 Simulation of ≥C5 RONO2 ................................................................................................ 215 8.5 Sub-conclusions ................................................................................................................... 219
Chapter 9 Conclusions and suggestions for future studies ............................................................ 221
XIII
9.1 Conclusions .......................................................................................................................... 221 9.2 Suggestions for future studies ............................................................................................ 224
References .......................................................................................................................................... 228
1
Chapter 1 Overview 1.1 Introduction Photochemical pollution has become one of the most significant environmental issues in
developing and even some developed countries over the world [Guo et al., 2013; Sharma et al.,
2013; Lin et al., 2014]. As typical secondary air pollution, photochemical smog is generally
recognized with high concentrations of ozone (O3). In China, O3 is often found to exceed the
national standard (hourly upper limit of 160 μg/m3 and 200 μg/m3 for level I and level II of the
national standard, respectively) [Zhang et al., 2008; Wang et al., 2009; NAAQS, 2012]. Due to
adverse effects of O3 on human health, yield of crops, climate change and atmospheric
environment [Fiscus, et al., 2005; Bell, 2006; Ainsworth et al., 2012], it is urgent for the
Chinese government to mitigate the nation-wide photochemical pollution, particularly in South
China where the emissions of O3 precursors are intensive and meteorological conditions favor
O3 formation [Chan and Chan, 2000; Tang et al., 2007]. Despite many air pollution control
measures, O3 attainment is still unreachable in this region, in contrast to the obvious
improvement of airborne fine particle pollution [Xue et al., 2014; Ma et al., 2016]. Therefore,
full understanding of O3 formation mechanisms and the potential factors influencing O3
production are required to effectively control O3 pollution. In this context, alkyl nitrates attract
the attention of atmospheric scientists given their close relationships with O3 in the
photochemical processes [Flocke et al., 1991; Horowitz et al., 2007].
Alkyl nitrates are a group of organic nitrates which are formed simultaneously with O3 through
branched chain reactions. Since RONO2 and O3 share the same precursors, they generally have
moderate to good correlations with each other. According to the incomplete survey on
literatures, the correlation coefficients ranged from 0.46 to 0.85 [Simpson et al. 2006; Perring
et al., 2013; Ling et al., 2016]. During the formation of RONO2, the oxidative radicals and
nitrogen oxides (NOx) are stabilized in RONO2. The budget changes of these O3 precursor
2
radicals are expected to influence O3 production in the source region. For example, studies
suspected that RONO2 formation in urban areas generally lead to O3 reduction, because the
production rate of O3 decreases with the stabilization of alkyl peroxy radicals in RONO2
[Perring et al., 2010; Aruffo et al., 2014]. Conversely, a case study on the impact of fuel
substitution on O3 production indicated that O3 increased after the replacement of liquefied
petroleum gas (LPG) as the vehicle fuel, which lowered the branching ratios leading to RONO2
formation [Farmer et al., 2011]. On the other hand, the oxidative radicals and NOx combined
in RONO2 can be released through the photolysis and OH oxidation of RONO2. Subsequently,
the released radicals and NOx are capable of fueling O3 formation. This effect was hypothesized
to be an important mechanism influencing O3 formation in the remote regions where local
sources are sparse [Neu et al., 2008].
Contradictory to the critical roles of RONO2 in the photochemical system, the detailed
photochemistry of RONO2 is not well understood by the scientific community. Although it is
widely recognized that RO2 reacting with NO and RO reacting with NO2 are the main formation
pathways of gas phase RONO2, the relative importance of these two pathways is uncertain, due
mainly to the uncertainties in calculating/modelling RO2 and RO, the reaction rates between
ROx and NOx, and the wide range of the branching ratios leading to RONO2 formation in
literature [Atkinson et al., 1982a; Simpson et al., 2006]. In addition, RO2 reacting with NO
leads to the formation of O3 and RONO2. The yields of RONO2 depend upon the branching
ratios between the two routes. Laboratory experiments indicated that the branching ratios were
temperature and pressure dependent [Atkinson et al., 1987]. Moreover, it was found that with
the increase of carbon number in RO2 radicals, the branching ratios increased [Atkinson et al.,
1982b]. Based on the experiments, formulas were proposed to calculate the branching ratios
for individual RONO2 [Carter and Atkinson, 1985]. Despite the previous work, the branching
ratios for specific RONO2 determined or used in previous studies were generally different, and
3
varied in a large range. In other words, the real branching ratios for RONO2 formation in a
certain photochemical system need to be re-examined, according to the field observation and
model simulation. Another common phenomenon observed in the field measurements of
RONO2 was that the observed RONO2 (particularly for methyl nitrate) often exceeded what
can be explained by the kinetic calculation [Simpson et al., 2006; Wang et al., 2013]. Therefore,
it is of high necessity to build up the exact relationships between RONO2 and their precursors.
Furthermore, the alternative formation pathways may be proposed if any.
As introduced earlier, the interactions between RONO2 and O3 occur during the formation and
degradation of RONO2. Since RONO2 are generally formed in the urban areas where O3
formation is limited by VOCs, the impacts of RONO2 formation on O3 production are more
likely associated with the stabilization of oxidative radicals in RONO2 molecules. If so, the
impacts will also relate to the reactivity of the parent hydrocarbons of RONO2. In contrast to
RONO2 formation, the impacts of RONO2 degradation on O3 production are more relevant in
the receptor regions. However, the released radicals and NOx from RONO2 degradation may
have distinct effects on O3 formation in different environments. Therefore, it is necessary to
study the impacts of RONO2 degradation on O3 production over a wide range of environments.
So far, the issues on RONO2 photochemistry influencing O3 formation are generally studied
based on theoretical calculation or observation in scientific community, lacking consideration
of the complicated photochemical reactions and/or quantification of the impacts of RONO2
photochemistry on O3 production.
1.2 Scope of this study This study includes description of the spatiotemporal patterns of RONO2 and their parent
hydrocarbons in Hong Kong, which provides us a basic understanding of the abundances of
RONO2 in the most highly developed region in South China. On one hand, with the increase
4
of vehicle population and implementation of air pollution control measures (e.g., promotion of
LPG as the vehicle fuel), the profiles of ambient air pollutants in Hong Kong experienced
substantial changes during the past decades. Correspondingly, the abundances and composition
of RONO2 might also change. On the other hand, the spatially variable emissions of VOCs in
Hong Kong and the potential influences of regional transport might lead to the disproportionate
distributions of RONO2 over the territory of Hong Kong. Therefore, it is essential in
understanding the relationships of RONO2 to the parent hydrocarbons to determine the
spatiotemporal patterns of RONO2 in Hong Kong. A second part of this study examines the
effects of a recent air pollution control measure on the abundances of RONO2 and their
precursors. These impacts were evaluated from the perspective of impacts of the program on
parent hydrocarbons, atmospheric oxidative capacity and RONO2 formation. The third subject
of this study deals with the sources of RONO2 in Hong Kong. Although the different types of
sources of RONO2 have been well known, the specific and quantitative source contributions to
RONO2 in this region were determined for the first time. Another major area of study in this
thesis is the formation pathways of RONO2 and the factors influencing RONO2 production.
Specifically, the contributions of the pathways of “RO2+NO” and “RO+NO2” to individual
RONO2 through C1 to C5 were determined. Then, the effects of branching ratio, precursors and
atmospheric oxidative capacity on RONO2 production were studied; The current study
examines the relationships between RONO2 and their precursors. The production of RONO2
depends strongly on the abundances of precursors. However, it needs investigation if the
response of RONO2 to the changes of their precursors is linear. Therefore, RONO2 production
was simulated with a matrix of parent hydrocarbons and NOx over a wide range of
concentrations to examine this response relationship. Finally, the impacts of RONO2
photochemistry on O3 production is studied in this thesis. The impacts were examined in the
processes of RONO2 formation and degradation (photolysis and OH oxidation). In addition,
5
this study also demonstrated the importance of isoprene nitrates and aromatics nitrates in
influencing O3 formation.
1.3 Outline of the thesis To better present the work in my PhD study, the thesis is divided into 9 chapters, including
“Overview”, “Literature review”, “Methodology”, 5 chapters presenting the results and
discussions, and one chapter to conclude the whole study (“Conclusions and Suggestions”).
Chapter 1 “Overview” included three subsections. The “Introduction” subsection briefly
summarized the previous studies of RONO2 and the knowledge gaps, and the subsection
“Scope of this study” illustrated the research contents of this work. Then the structure and
organization of the thesis were given in subsection “Outline of the thesis” for readers’
convenience.
Chapter 2 “Literature review” introduced the previous studies on RONO2 from the perspectives
of organic nitrates, abundances, sources and sinks of RONO2, relationship between RONO2
and O3, and air quality studies in Hong Kong.
Chapter 3 “Methodology” described the details of sample collection, chemical analysis and
quality control of the data. Besides, the model construction and application, as well as other
tools and calculations, were also introduced in this section.
Chapter 4 “Spatiotemporal variations of RONO2 and their parent hydrocarbons in Hong Kong”
studied the long term trends of C1-C5 RONO2 in Hong Kong, the spatial distributions of C1-C5
RONO2 over the whole territory of Hong Kong and the impacts of an air pollution control
measure on the abundances of RONO2 and their precursors.
Chapter 5 “Impacts of replacing catalytic converters in LPG-fueled vehicles on the abundances
6
of RONO2 and their parent hydrocarbons in Hong Kong” further examined the effects of this
air pollution control measure on RONO2 and their parent hydrocarbons. For this study, more
comprehensive data were collected at a roadside site heavily influenced by vehicle exhaust,
which better reflected the changes of air composition from vehicle emissions.
Chapter 6 “Re-examination of C1-C5 alkyl nitrates in Hong Kong using an observation-based
model” described the process of construction of the RONO2 module in the PBM-MCM model.
Furthermore, the model was employed to examine the contributions of different reaction
pathways to each C1-C5 RONO2. It was found that the reaction pathway “RO2+NO” dominated
the formation of C2-C5 RONO2.
Chapter 7 “New insight into the spatiotemporal variability and source apportionments of C1-
C4 alkyl nitrates in Hong Kong” discussed the sources of C1-C4 RONO2 at two sites in Hong
Kong.
Chapter 8 “Modelling of C1-C4 alkyl nitrate photochemistry and their impacts on O3 production
in urban and suburban environments of Hong Kong” focused on the mechanisms of RONO2
formation and the impacts of RONO2 photochemistry on O3 production. The study was based
on the data collected at two sites in Hong Kong, however the results were extended to a wide
range of environments through model simulations.
Chapter 9 “Conclusions and suggestions” concluded the main findings and the original and
novel contributions of this work to the scientific community. In addition, suggestions for future
studies on RONO2 were also proposed.
7
Chapter 2 Literature Review 2.1 Organic nitrates Organic nitrates are a cluster of organic compounds that contain the covalently-bound nitrate
group. They are important constituents of reactive odd nitrogen (NOy) in the troposphere, and
participate in nitrogen cycling. The instrumental measurement of NOy includes nitric oxide
(NO), nitrogen dioxide (NO2), nitric acid (HNO3), nitrous acid (HONO), dinitrogen pentoxide
(N2O5), nitrate (NO3), peroxyacetyl nitrate (PAN) and other peroxyacyl nitrates, peroxynitric
acid (HO2NO2), alkyl and multifunctional nitrates (RONO2) and particulate nitrates. Among
them, PAN and RONO2 are the most commonly reported organic nitrates in the atmosphere.
Kastler et al. [2000] quantified the mixing ratios of 43 alkyl mononitrates, 24 alkyl dinitrates
and 19 hydroxy alkyl nitrates in 5 US cities, and found that besides alkyl mononitrates, alkyl
dinitrates and hydroxyl alkyl nitrates were also important constituents of NOy in urban air.
Studies [Buhr et al., 1990; Atlas et al., 1992; Singh et al., 1992] indicated that alkyl nitrates
generally accounted for a small part (<20%) of NOy. Day et al. [2003] reported that 10-20% of
NOy was contributed by the total RONO2 in winter months when the total NOy was very low.
However, the fraction can be elevated to a large extent in specific regions, particularly for those
remote from the anthropogenic sources, such as over the equatorial oceans [Ridley et al., 1990;
Day et al., 2003]. Alkyl nitrates generally include alkyl mononitrates, hydroxynitrates and
dinitrates, which contain one nitro group, nitro and hydroxyl groups and two nitro groups,
respectively. Kastler et al. [2000] quantified the mixing ratios of 43 alkyl mononitrates, 24
alkyl dinitrates and 19 hydroxy alkyl nitrates in 5 US cities, and found that besides alkyl
mononitrates, alkyl dinitrates and hydroxyl alkyl nitrates were also important constituents of
NOy in urban air. O'Brien et al. [1995] reported that alkyl and multifunctional organic nitrates
accounted for 0.5-3.0% of NOy at a rural site in Ontario, with alkyl mononitrates,
hydroxynitrates and dinitrates representing 82%, 16% and 2% of the total measured alkyl
nitrates. Since organic nitrates are initially formed through the reactions between VOCs and
8
NOx, they are also temporary reservoirs of reactive radicals and NOx, both of which are
essential components in the photochemical system. As a result, organic nitrates are capable of
influencing the formation of secondary products originally generated from the reactions
between VOCs and NOx, such as ozone (O3) and secondary organic aerosols (SOA). Besides,
the gas-particle partitioning of organic nitrates is also of significance in air and aerosol studies.
Overall, organic nitrates play critical roles in atmospheric environment, climate change and
human health.
As one of the representative organic nitrates, PAN was firstly detected and identified in ambient
air in the study of photochemical smog in 1950s. Similar to O3, PAN is formed from the
reactions between VOCs and NOx in presence of solar radiation [Singh et al., 1981]. Therefore,
it is also an excellent indicator of photochemical pollution in city plumes. PAN is quite
sensitive to temperature and prone to thermal decomposition under heat [Cox and Roffey, 1977],
even though the molecule is to some extent stabilized by the carbon-oxygen double bond.
Another typical member in the organic nitrates pool is RONO2. RONO2 are a group of organic
nitrates in which the nitrogen is stabilized in the molecular structure of R-O-NO2. As an
important constituent of reactive odd nitrogen (NOy), particularly in areas far from urban
sources [Buhr et al., 1990; Murphy et al., 2006], RONO2 participate in nitrogen cycling as
temporary nitrogen reservoirs due to their long atmospheric lifetimes [Clemitshaw et al., 1997].
In contrast to the poor thermal stability of PAN, RONO2 are relatively stable in the atmosphere,
not prone to transform into particle phase and insoluble in water. Therefore, the lifetimes of
RONO2 range from several days to months. Despite this, PANs are typically more abundant
than RONO2 in the atmosphere, particularly in the free troposphere where the lower
temperature favors the survival of PANs [Beine et al., 1996; Singh et al., 1998]. The
contribution of PANs to NOx in the remote areas has also been well recognized [Moxim et al.,
1996]. However, the long lifetimes of RONO2 enable them to live and transport in a long
9
distance, further regulating the photochemistry in downwind areas through their degradation
and release of oxidative radicals and NOx.
2.2 Abundance of RONO2 The mixing ratios of RONO2 generally range from several to hundreds pptv. The short chain
(C1-C3) RONO2 are often found to be the most abundant RONO2 in the remote marine
atmosphere, due to their long lifetimes and high marine emissions. During a cruise on the
equatorial Pacific Ocean, Atlas et al. [1993] reported that isopropyl nitrate was the most
abundant RONO2 above the sea (~5 pptv), followed by butyl nitrates, ethyl nitrate and 2-
methyl-3-butyl nitrate, while methyl nitrate was not measured in that study. For methyl nitrate,
numerous studies identified that its mixing ratio was in the range of 5-50 pptv over the oceans.
Walega et al. [1992] reported the average mixing ratio of 4 pptv for methyl nitrate on the island
of Hawaii. Then, within the marine boundary layer (MBL) near the equator, the mixing ratio
of methyl nitrate was found to be up to 50 pptv or greater [Blake et al., 1999]. High levels of
methyl nitrate (25 pptv) were also reported at a coastal site in Hong Kong [Simpson et al.,
2006]. Furthermore, owing to its lifetime of approximately one month, methyl nitrate can
penetrate the troposphere and enter the lower stratosphere, where the larger RONO2 molecules
with more carbon numbers are below analytical detection limits [Flocke et al., 1998b]. In
continental areas, C3-C4 alkyl nitrates are often observed to be the highest due to the combined
effect of the branching ratio and the abundance of parent hydrocarbons [Darnall et al., 1976;
Atkinson et al., 1987; Wang et al., 2013]. For example, Katzenstein et al. [2003] indicated that
the mixing ratio of 2-butyl nitrate (68 pptv) was the highest among all the RONO2 in south-
central Kansas. At a rural site in Ontario, Shepson et al. [1993] also found the dominance of 2-
propyl and 2-butyl nitrates in the total RONO2.
10
As the nitrogen reservoir, RONO2 is an important constituent of NOy. Field measurements and
model simulations identified that nitrogen bound in RONO2 cannot be neglected near the
Earth’s surface. Although many studies [O’Brien et al., 1995; Flocke et al., 1998b; Thornberry
et al., 2001] indicated that RONO2 only accounted for <3% of NOy in the continental boundary
layer, Atherton and Penner [1988] calculated that 5% of NOx could be converted to RONO2.
Flocke et al. [1991] indicated that in some cases the fraction of RONO2 in NOy could be as
high as 7%. However, the individual methyl nitrate can even account for 10% of NOy over the
remote equatorial ocean characteristic of clean air, according to Walega et al. [1992]. The
aircraft measurement over the Southern Ocean indicated that methyl nitrate was a major
component of NOy at the altitude of lower than 1 km, particularly near the equator [Blake, et
al., 1999]. This proportion of RONO2 in NOy even increased to 20-80% in the MBL over
equatorial Pacific [Blake et al., 2003; Talbot et al., 2003].
C1-C5 RONO2 are the main constituents of the total RONO2 over the world, no matter in the
polluted or clean atmospheres. The measurements at Mauna Loa and in the marine boundary
layer over the Pacific indicated that methyl nitrate was the dominant species in total RONO2
[Atlas et al., 1992; Walega et al., 1992]. The dominance of methyl nitrate was also observed
over the equatorial, where the highest mixing ratio of methyl nitrate reached ~50 pptv, several
folds higher than those of C2-C4 RONO2 [Blake et al., 1999, 2003]. In addition to methyl nitrate,
2-propyl and 2-butyl nitrates also had the highest mixing ratios in the island of Hawaii among
all the ≥C3 RONO2 [Atlas et al., 1992]. Over the equatorial Pacific Ocean from 10°S to 15N
latitude and 144 to 165 W longitude, Atlas et al. [1993] reported that C3 RONO2 were the most
abundant species among the ≥C2 RONO2. In the areas that were more influenced by
anthropogenic activities, it was more common that C3-C4 RONO2 accounted for the largest
fractions in total RONO2. Flocke et al. [1998b] measured the mixing ratios of RONO2 at
Schauinsland station in the Black Forest during 1990-1991, and found that 2-propyl, 2-butyl
11
and methyl nitrates were the most abundant RONO2 species, while the levels of ≥C7 nitrates
were very low. The dominance of 2-propyl and 2-butyl nitrates in total RONO2 in continental
air masses has been extensively reported in previous studies, including the measurements of
C1-C5 RONO2 at Chebogue Point, NovaScotia [Roberts et al., 1998], the year-round
observations of C1-C5 RONO2 at a coastal site in Hong Kong [Simpson et al., 2002]. Overall,
methyl nitrate is the most predominant RONO2 species over the equatorial oceans, while C3-
C4 RONO2 occupy the first place among RONO2 in continental areas, which closely relate to
the sources and lifetimes of RONO2. Oceanic emission has been known as an important source
of C1-C3 RONO2, particularly near the equator. Due to the long atmospheric lifetime of methyl
nitrate (~ 1 month), it is generally observed with the highest mixing ratio among all the RONO2
species. In contrast, photochemical formation leads to the most abundances of C3-C4 RONO2,
because of the high levels of their parent hydrocarbons and relatively large branching ratios for
RONO2 formation through RO2 reacting with NO. Since they share the same formation
pathways with O3 and serve as the nitrogen reservoir, C1-C5 RONO2 may play important roles
in atmospheric chemistry in both urban and remote atmospheres.
2.3 Sources of RONO2
2.3.1 Oceanic emission
So far, the identified sources of RONO2 include oceanic emission, biomass burning (although
RONO2 emitted from biomass burning are also photochemically formed, the process is much
faster) and photochemical formation. Atlas et al. [1988] for the first time discovered the
presence of ≥C3 RONO2 in the atmosphere. Subsequently, Atlas et al. [1993] proposed that
oceanic emission might be a source of RONO2 in light of the unexpectedly high levels of
RONO2 and good agreements between RONO2 and marine tracers over the equatorial Pacific
Ocean. Through the measurements of methyl and ethyl nitrates in both seawater and the
12
atmosphere over the Atlantic Ocean, Chuck et al. [2002] firstly found the evidence that
equatorial oceans are the emitters of alkyl nitrates where the super-saturation of both species
reached 800%. The marine source of RONO2 was repeatedly confirmed by the following
studies. For example, very high level of methyl nitrate (up to 50 pptv) was detected over the
tropical Pacific Ocean, which was thought to be originated from the Pacific waters [Blake et
al., 2003]. Also in the tropical Pacific Ocean, Dahl et al. [2005] found the extremely high
saturation (up to 2000%) of light (C1-C3) RONO2, and indicated that the air-sea exchange was
sufficient to explain the observed RONO2 in this region, with the consideration of net output
to the surrounding areas.
For the formation mechanisms of oceanic RONO2, Chuck et al. [2002] found that the
concentration of methyl nitrate decreased with the decreasing of chlorophyll, suggesting the
algal production of RONO2. In addition, many evidences pointed to the radical formation of
oceanic RONO2. For example, methyl and ethyl nitrates in Atlantic Ocean showed maximum
concentrations when the water was heavily polluted by dissolved organic matter [Chuck et al.,
2002]. Furthermore, laboratory experiments identified the aqueous production of RONO2 via
the reactions between peroxy radicals (RO2) and NO [Dahl et al., 2003]. With the analysis of
depth profiles of RONO2 in the tropical Pacific Ocean, Dahl et al. [2007] demonstrated that
radical production (RO2+NO) was most likely the route to form deep water RONO2, but
biological activity cannot be eliminated. As of this writing, a quantitative description of the
various formation mechanisms of oceanic RONO2 remain uncertain.
2.3.2 Biomass burning
Biomass burning as the source of RONO2 was firstly discovered by Lobert et al. [1991] in
laboratory fire experiments. The direct evidences of RONO2 emission from biomass burning
through field measurement had not been obtained until Simpson et al. [2002] observed the
emissions of C1-C4 RONO2 from savanna burning. Within 3 meters of the fires in Northern
13
Australia, the methyl nitrate was found to be as high as 3297 pptv, while the local background
methyl nitrate was only 5.4±0.2 pptv [Simpson et al., 2002]. Consistently, Blake et al. [2003]
observed enhancements of C3 and C4 RONO2 in the southern-tropics by biomass burning,
although the influence was minimal. However, Flocke et al. [1991] and Blake et al. [1999]
insisted that biomass burning was not a major source of methyl nitrate.
Based on the good correlations between the emission factors of parent hydrocarbons, alkyl
nitrates and the carbonyl compounds generated from the oxidation of RO, Friedli et al. [2001]
inferred that the formation mechanism of RO2 associating with NO was applicable to RONO2
formation in biomass burning. However, Simpson et al. [2002] indicated that this route cannot
explain the high production of fresh RONO2, even in full consideration of the strong emissions
of parent hydrocarbon, NOx and hydroxyl radical (OH) from biomass burning. Instead, the
combination between alkoxy radical (RO) and NO2 was proposed to be the additional pathway
of RONO2 formation in biomass burning. The decomposition of large RO radicals and the self-
consumption of RO2 (RO2+RO2→2RO+O2) provide sufficient RO to associate with NO2 and
generate RONO2. However, there have been no experimental or modelled studies on validation
of this formation mechanism until the present study. It is noteworthy that the emission ratios
of RONO2 relative to CO (smoldering mode) and CO2 (flaming mode) ranged from
(4.0±0.5)×10-8 for n-propyl nitrate to (6.8±2.3)×10-7 for methyl nitrate in smoldering mode and
from (4.9±1.2)×10-9 for n-propyl nitrate to (4.7±0.5)×10-7 for methyl nitrate in flaming mode,
according to the measurement near the savanna burning in Australia by Simpson et al. [2002].
It therefore can be expected that biomass burning may be an important source of RONO2 in the
regions where open fires are common, such as in south China, India and Pakistan.
2.3.3 Photochemical formation of RONO2
In addition to marine sources and biomass burning, it is widely recognized that photochemical
formation is one of the main sources of RONO2 [Atkinson et al., 1987; Carter and Atkinson,
14
1989a; Arey et al., 2001]. It makes considerable contribution to RONO2 in the continental
regions where oceanic emission is weak and biomass burning is rare. The following section
introduces the photochemical formation of RONO2 in detail.
Two main photochemical formation mechanisms have been proposed to explain observed
RONO2 levels in urban settings, including RO2 reacting with NO [Darnall et al., 1976; Carter
and Atkinson, 1989a] and RO reacting with NO2 [Atkinson et al., 1982a], which are well
documented and play dominant roles in RONO2 formation, particularly in daytime hours. A
sequence of reactions is involved in the formation pathway of RO2+NO, which are expressed
as reactions (1)-(4). OH initiated oxidation of parent hydrocarbons (reaction (1)) is the rate
limiting step of these sequential reactions. The generated R radical quickly combines with
oxygen (O2) and produces RO2 radicals (reaction (2)). Subsequently, two complementary
reactions occur between RO2 and NO, leading to the formation of RONO2 (reaction (3)) and
O3 (reaction (4)), respectively [Atkinson, 1990]. In reaction (3), RO2 initially associates with
NO and forms ROONO* in vibrational excitation mode. ROONO* then transforms to the
vibrationally excited RONO2*, which decomposes and regenerates RO and NO2 or stabilize to
RONO2 in molecular media [Darnall et al., 1976]. To quantitatively evaluate the relative
importance of reaction (3) and reaction (4), a branching ratio (α) is introduced, which represents
the fraction of RO2 forming RONO2 through reaction (3) in total RO2. In this case, the
branching ratio is a measure of the yield of RONO2. It is defined as the ratio of k3/(k3+k4)
[Atkinson, 1990], where k is the reaction rate constant. Generally, the branching ratio increases
with increasing pressure and decreasing temperature [Atkinson et al., 1987]. In addition, more
complicated molecular structures of RO2 and/or RO2 with higher carbon number tend to have
higher branching ratios. For example, the branching ratios for n-alkanes increase from ≤1% for
ethane to ~33% for n-octane, with an upper limit of ~35% for larger n-alkanes [Atkinson et al.,
1982a]. This theory was repeatedly confirmed by kinetic calculations and model simulations
15
[Bertman et al., 1995; Simpson et al., 2006]. In combination with laboratory studies, Carter
and Atkinson [1985] put forward Formulas 1-3 for calculating branching ratios under the
atmospheric conditions in the stratosphere, and suggested a correction factor of 3 to methyl
nitrate. However, in urban settings the smaller-chain RONO2, particularly methyl and ethyl
nitrate, often have mixing ratios higher than what can be explained by the reaction of RO2 with
NO [Flocke et al., 1998a]. Depending on the location, direct marine emissions could be one
reason [Atlas et al., 1993]. In addition, the exact branching ratios for these RONO2 remain
uncertain. For example, Lightfoot et al. [1992] proposed an upper limit of 0.005 for methyl
nitrate in the lower troposphere. Through RONO2 observations in the lower stratosphere,
Flocke et al. [1998b] found a much lower methyl nitrate branching ratio of 5-10×10-5, meaning
it could only reach a maximum of 0.0003 even under surface conditions, when applying the
adjusted factor of 3 to the branching ratio in lower stratosphere. Simpson et al. [2006] accepted
the upper limit of 0.0003, and indicated that RO reacting with NO2 was the main pathway of
methyl nitrate in highly polluted environments. However, according to the formulas proposed
by Carter and Atkinson [1985], the branching ratio for methyl nitrate was approximately 0.001,
which was also adopted by the master chemical mechanism (MCM, accessible at
http://mcm.leeds.ac.uk/MCM/). Overall, the branching ratios for RONO2 formation remain to
be further examined.
RH + OH → R + H2O (1), k1
R + O2 → RO2 (2), k2
RO2 + NO → RONO2 (3), k3, α
RO2 + NO → RO + NO2 (4), k4, 1-α
RO + NO2 → RONO2 (5), k5
16
α = [𝑌0
300[𝑀](𝑇
300)
−𝑚0
1+𝑌0
300[𝑀](𝑇
300)
−𝑚0
𝑌∞300(
𝑇300
)−𝑚∞
]𝐹𝑧 (Formula 1)
z = {1 + [log𝑌0
300[𝑀](𝑇
300)
−𝑚0
𝑌∞300(
𝑇
300)
−𝑚∞ }]
2
}−1 (Formula 2)
𝑌0300 = 𝛽𝑒𝛾𝑛 (Formula 3)
where T is the temperature (K); M represents the number of molecules (molecules/cm3) and n
is the carbon number in RO2. The values of the constants β (1.95 × 10-22 cm3/molecule), 𝛾
(0.947), 𝑌∞300 (0.435), 𝑚0 (2.99), 𝑚∞ (4.69) and 𝐹 (0.556) are all from Carter and Atkinson
[1985]. On the basis of the calculated results, the branching ratios for the primary and tertiary
RO2 radicals are calibrated by a factor of 0.4 and 0.25, respectively.
Although the pathway of RO2 reacting with NO is well recognized as the major contributor to
continental RONO2, it cannot explain the high observed RONO2 in many cases, even in
consideration of RONO2 emissions from ocean and biomass burning and the variations of
branching ratio in reasonable range. Namely, alternative formation pathways exist, among
which RO reacting with NO2 (reaction (5)) is generally accepted. However, the contribution of
this pathway to large (≥4) RONO2 is minor, due to the high decomposition and isomerization
tendency of large RO radicals. In contrast, the light (≤3) RONO2 generated from this route are
generally non-negligible or even significant. For example, Simpson et al. [2006] indicated that
the excessive methyl nitrate that cannot be explained by CH3O2+NO and primary emissions at
a coastal site in Hong Kong was attributable to CH3O reacting with NO2. Coincidently,
Archibald et al. [2007] confirmed that this pathway becomes important at about 10 ppb of NO2,
and dominant at about 35 ppb, based on MCM simulations for European conditions. However,
whether this conclusion is universally applicable still needs further examination.
17
Upon these two formation pathways, the routes to form RONO2 are still complicated, due to
the diverse origins of RO2 and RO radicals. For example, Bertman et al. [1995] established the
relationship between RONO2 and parent hydrocarbons. However using this method, the
observed RONO2 are generally higher than the predicted RONO2 calculated based on the
concentration of parent hydrocarbon, OH and a reasonable branching ratio [Russo et al., 2010;
Worton et al., 2010], particularly for the light RONO2. In addition to the primary emission and
background concentration, a likely cause of this discrepancy is that the radicals (RO2 and RO)
were underestimated with the consideration of only parent hydrocarbons as the source of them.
In this case, many sources of radicals were missed, such as the photolysis of carbonyls, the
oxidation of peroxides and carboxylic acids or the decomposition of larger radicals [Sommariva
et al., 2008]. Therefore, to better describe the formation pathways of RONO2, an explicit
chemical mechanism containing the degradation of parent hydrocarbons as well as more
chemical compounds is needed.
As can be expected, the production of RONO2 depends strongly upon the abundances of RO2
and RO radicals, which are representative of the oxidative capacity of the atmosphere and
correlate well with OH and HO2. As such, the factors influencing the strength of oxidative
capacity (e.g., relative abundances of VOCs and NOx) may also impact the production of
RONO2.
In addition to the pathways introduced above, the other formation mechanisms of RONO2
include but are not limited to reactions between organic aerosols and particle-phase nitrates
and NO3 initiated oxidation of RO2 at nighttime [Carslaw et al., 1997; Worton et al., 2010].
However, these pathways are generally not included in the chemical transport models and
chemical box models, hampering their validation and application in modelling study of RONO2.
18
To conclude, numerous questions about the photochemical formation of RONO2 remain to be
answered, including the pathway contributions to RONO2 production in different environments,
the influencing factors of RONO2 formation and the RONO2-precursors relationships.
2.4 Sinks of RONO2
The sinks of gas-phase RONO2 consist of photolysis, oxidation by OH (reaction (7a) and (7b)),
dry deposition and transformation to particle phase.
In the presence of sunlight, RONO2 photolyzes and generates RO and NO2 (reaction (6)). The
photolysis rates of primary and secondary RONO2 are generally lower than tertiary RONO2,
and the photolysis frequencies are highly temperature dependent [Luke et al., 1989]. The
relative importance of photolysis as a sink of RONO2 decreases with increasing carbon number
in RONO2, [Atkinson et al., 1982a; Bertman et al., 1995; Clemitshaw et al., 1997].
The oxidation of RONO2 by OH involves two mechanisms, i.e., hydrogen abstraction (reaction
7a) and hydroxyl addition (reaction 7b) [Nielsen et al., 1991]. In the case of hydrogen
abstraction, a hydrogen in group R is initially abstracted by OH, generating ·RONO2 and H2O.
The unsteady state ·RONO2 molecule will further decompose and produce carbonyls and NO2.
In contrast, OH can be also added on the nitrogen atom of strong electronegativity, forming the
adduct of RON(OH)O2, which further degrades and generates RO and HONO2. In contrast to
photolysis, the importance of OH oxidation generally increases with carbon number increasing
in RONO2.
RONO2 + hv → RO + NO2 (6)
RONO2 + OH → ·RONO2 + H2O → carbonyls + NO2 + H2O (7a)
RONO2 + OH → RON(OH)O2 → RO + HONO2 (7b)
19
Dry deposition is an important pathway that brings atmospheric nitrogen into ecosystems. For
example, studies at Harvard forest found that the deposition of unmeasured organic nitrogen
was responsible for the discrepancy between total NOy and the sum of NOy species [Lefer et
al., 1999; Horii et al., 2004]. Through analysis of diurnal patterns of RONO2, other trace gases
and wind speed, Russo et al. [2010] inferred that dry deposition contributed to the lowest
methyl nitrate in early morning in New England, but no similar dry deposition was noted
for >C2 RONO2. However, the significance of dry deposition closely relates to the local surface
type and mixing ratios of RONO2. With a global modelling of methyl nitrate, Williams et al.
[2014] indicated that the tropical oceans featuring high emissions of methyl nitrate also had the
highest deposition loss (~50%), while the lowest deposition occurred in the regions of low
methyl nitrate emissions, such as Sahara and Antarctica. This highlights the necessity of
measuring the deposition rate of RONO2 over a wide range of surface types and the ambient
mixing ratios of RONO2, to guide the accurate simulation of RONO2. Importantly, the
deposition rate of individual RONO2 species, particularly for the light RONO2, is far from clear.
As an alternative to actual dry deposition rates for RONO2, many modelling studies [Neff et al.,
2002; Williams et al., 2014] used the deposition rates of other organic nitrates, such as PANs,
carbonyl nitrates and hydroxyl alkyl nitrates, to describe the deposition of RONO2, which
might introduce a great uncertainty to the simulations. For example, O’Brien et al. [1995] stated
that the dry deposition rates of hydroxyl alkyl nitrates might be much higher than those of alkyl
nitrates, providing an efficient removal route of atmospheric nitrogen despite the low mixing
ratios of these polar multifunctional nitrates.
The partitioning between gas and particle phase RONO2 is also a loss of RONO2. For example,
Fry et al. [2009] found that the gas phase organic nitrates produced by NO3 oxidation of
monoterpenes accounted for 0.5-8% of SOA in global scale. Similarly, Rollins et al. [2013]
estimated that ~21% of total RONO2 was in particle phase according to the measurements of
20
RONO2 in both gas and particle phases in California. However, this pathway only becomes
significant for RONO2 with complex molecular structure and high carbon number (e.g.,
biogenic VOCs and long chain alkanes), which features relatively low volatility.
2.5 Relationships between RONO2 and O3 As introduced above and shown in reactions (3) and (4), the formations of RONO2 and O3 share
the same precursors. Therefore, RONO2 always correlate well with O3 under the condition that
photochemical formation dominates the source of RONO2 [Flocke et al., 1991; Horowitz et al.,
2007]. Theoretically, the relationship between total RONO2 and O3 can be expressed as
𝑂3
𝑡𝑜𝑡𝑎𝑙 𝑅𝑂𝑁𝑂2=
2(1−𝛼)
𝛼 , where α represents the average branching ratio of total RONO2 [Day et
al., 2003]. This is because that the majority of RO produced in reaction (4) will be further
oxidized by O2 to generate HO2. As one of the strongest oxidants in the atmosphere, HO2 is
capable of transferring an O atom to NO, which is oxidized to NO2. The NO2 molecules
generated in reactions (4) and (9) both have the potential to react with O2 to form O3. In this
case, one molecule of RO2 in reaction (4) produces 2 molecules of O3. In combination with the
possibilities of RO2 participating in reactions (3) and (4), the theoretical ratio between O3 and
total RONO2 should be 𝑂3
𝑡𝑜𝑡𝑎𝑙 𝑅𝑂𝑁𝑂2=
2(1−𝛼)
𝛼.
RO + O2 → HO2 + carbonyls (8)
HO2 + NO → NO2 + OH (9)
NO2 + O2 → NO + O3 (10)
In reality, this quantitative relationship might be biased due to the fact that the species between
RO2 and O3 all involve many reactions. For example, besides reacting with O2, a part of RO
decomposes, isomerizes or reacts with other radicals. As the major sink of atmospheric nitrogen,
21
the formation of HNO3 also consumes NO2. Despite these interferences, good correlations
between RONO2 and O3 were repeatedly reported [Roberts et al., 1998; Day et al., 2003;
Simpson et al., 2006].
Furthermore, interactions must exist between the formation of RONO2 and O3, as they compete
for the same precursors in the source region. In the receptor region, RONO2 may degrade and
release the oxidative radicals and NOx, which fuel or suppress O3 formation dependent upon
the local photochemistry. Based on correlations between Ox and ∑ANs, Aruffo et al. [2014]
stated that RONO2 played important roles in O3 formation in both urban and suburban London.
O3 production was expected to be reduced in the processes of RONO2 formation under high
NOx and intensive VOC oxidation. Perring et al. [2010] indicated that the peak O3 production
in the near-field of Mexico City (source region of RONO2) was reduced by as much as 40%
due to the formation of total RONO2. Farmer et al. [2011] even claimed that VOC reductions
might cause an O3 increment because the branching ratios of RONO2 formation decreased
when fuels containing low boiling point VOCs products were used. This view was also
confirmed by Perring et al. [2013], in which a 20% reduction of VOCs led to an 8% O3
increment due to the unexpected reduction of the average branching ratio for total RONO2 from
8% to 4%. Despite the previous studies, they are quite insufficient and the drawbacks are
obvious. For instance, most of the studies did not directly link the formation of RONO2 to O3
concentration. Instead, the production of O3 was calculated using the simplified formulas (4)-
(5) [Aruffo et al., 2014]. Briefly, the reactions of HO2+NO and RO2+NO are treated as the
production pathways of O3 (PO3), which however is removed through OH reacting with NO2
and HO2 reacting with O3 (LO3). The formation of RONO2 was added as a negative term to PO3.
This approach actually simplifies the O3 formation mechanism, which may be much more
complicated in real atmospheric environments. Specifically, the reactions between O3 and OH,
O3 and alkenes and the photolysis of O3 also serve as important losses of the ambient O3 [Bates
22
and Nicolet, 1950; Michelsen et al., 1994]. Furthermore, the diversity and abundances of the
radicals and intermediates applied in formulas 4-5 are difficult to be accurately estimated in
these theoretical calculations due to their complicated interactions with other compounds in the
atmosphere. Therefore, uncertainties are inevitably inherent in the quantification of the impacts
of RONO2 photochemistry on O3 production through the theoretical calculations. The
application of the PBM-MCM model in this study will help to better understand the
relationships between RONO2 and O3. On the other hand, the impacts of RONO2 degradation
on O3 production were seldom studied, which will be comprehensively discussed over a wide
range of environments with the aid of the PBM-MCM model.
𝑃𝑂3 = 𝑘𝐻𝑂2+𝑁𝑂[𝐻𝑂2][𝑁𝑂] + ∑ 𝑘𝑅𝑂2𝑖+𝑁𝑂[𝑅𝑂2𝑖][𝑁𝑂] − 𝑃𝑅𝑂𝑁𝑂2 (Formula 4)
𝐿𝑂3 = 𝑘𝑂𝐻+𝑁𝑂2+𝑀[𝑂𝐻][𝑁𝑂2][𝑀] + 𝑘𝐻𝑂2+𝑂3[𝐻𝑂2][𝑂3] (Formula 5)
2.6 Air quality studies in Hong Kong Hong Kong, a special administrative region (SAR) of China, is a highly developed coastal
metropolis in southern China. It is a part of the greater Pearl River Delta (PRD) region with
other nine mainland cities and Macau (https://en.wikipedia.org/wiki/Pearl_River_Delta).
Geographically, to its north Hong Kong adjoins Shenzhen, a fast developing city in China,
while to its east, south and west Hong Kong is surrounded by the Pearl River Estuary (PRE)
and the South China Sea (SCS). The geographic coordinates of Hong Kong are in the range of
22.13-22.58° N and 113.82-114.52° E, equivalent to the territory of 1.1×103 km2. Despite
limited area, the vegetation coverage rate in Hong Kong is as high as 70%. On the other hand,
the population is 7.2 million and the number of vehicles is around 0.7 million in Hong Kong,
making it one of the most crowded cities in the world. The weather in Hong Kong is mainly
influenced by subtropical monsoon climate. Namely, in warm and humid season (April-
23
September), the southerly winds from SCS prevail in Hong Kong, which bring in clean marine
air diluting local air pollutants, except for the occasional photochemical pollution induced by
tropical cyclones. In contrast, in cool and dry season (October – March), the dominant northerly
winds deliver polluted air from inland PRD to Hong Kong, burdening local air pollution.
Street level emissions and photochemical smog are two typical air pollution problems in Hong
Kong. On one hand, extensive and intensive studies have been conducted on vehicle emissions
[Li et al., 2001; Chan et al., 2004; Cheng et al., 2010a; Guo et al., 2011a], source identifications
of primary air pollutants [Lee et al., 1999; Guo et al., 2004; Guo et al., 2007; Lau et al., 2010],
local formation mechanisms of O3 and SOA [So and Wang et al., 2003; Cheng et al., 2010b;
Ling et al., 2013], effects of synoptic conditions on air quality [Chan and Chan, 2000; Huang
et al., 2005], and regional transport [Louie et al., 2005; Jiang et al., 2008; Guo et al., 2009,
2013]. VOCs is a hot topic in air quality study in Hong Kong, due to their critical roles in
human health and the formation of O3 and SOA. Numerous studies in the early 2000s put their
emphases on the abundances of speciated VOCs in Hong Kong. As early as 2000, Sin et al.
[2000] reported the mixing ratios of 42 VOCs at two urban sites in Hong Kong. While the
individual VOCs varied in the range of 0.2-5 ppbv in annual average mixing ratio, toluene was
found to occasionally exceed 20 ppbv in some samples. Subsequently, Lee et al. [2002] pointed
out that benzene, toluene, ethylbenzene and xylenes were the main constituents of VOCs in
urban Hong Kong, with toluene as the most abundant VOC species mainly attributable to
vehicle emissions. The high concentrations of aromatics in the ambient air of Hong Kong were
also confirmed by Chan et al. [2002] who found that the highest level of benzene and toluene
even reached 36.9 and 77.9 ppbv in the roadside microenvironments of Hong Kong,
respectively. In fact, it was found that aromatics were always the most abundant VOCs in the
air of Hong Kong in the early 2000s, with toluene ranking the first [Guo et al., 2004 and 2007;
Ho et al., 2004]. To improve air quality, diesel used in taxis and public light buses was replaced
24
by LPG in Hong Kong from August 2000 and August 2002, respectively
(http://www.epd.gov.hk/epd/english/environmentinhk/air/air_maincontent.html). As a result,
the LPG-related VOCs including propane and n/i-butanes increased remarkably since then, in
contrast to the decrease of aromatics. Ou et al. [2015] summarized the variations of VOCs at
an suburban site in Hong Kong from 2005 to 2013, and indicated that C3-C5 alkanes increased
with the rate of 0.0015 μg·m-3/year, while aromatics decreased by 0.006 μg·m-3/year.
Specifically, toluene experienced a statistically significant (p<0.05) reduction (rate = -0.001
μg·m-3/year) in this period. A grid study [Lyu et al., 2016a] on the abundances and
compositions of O3 precursors in Hong Kong during 2013-2014 revealed that propane (7.7±1.9
ppbv), n-butane (11.2±3.1 ppbv) and i-butanes (6.4±1.7 ppbv) dominated over toluene (3.4±0.5
ppbv) in both roadside and urban atmospheres of Hong Kong.
For the sources of VOCs in Hong Kong, a lot of studies spanning from the suburban site to the
roadside site have been carried out by Prof’ Guo Hai’s group during the past 15 years. Overall,
vehicle emissions and solvent usage were identified as the main sources of VOCs in Hong
Kong, despite the varying source contributions among different sites and different studies. For
example, Guo et al. [2004] indicated that 39-48% of VOCs in urban Hong Kong could be
attributable to vehicle emissions. In addition, Solvent usage, LPG usage and industrial
emissions accounted for 32-36%, 11-19% and 5-9%, respectively. It should be noted that the
industries with high emission of air pollutants (including VOCs) are very rare in Hong Kong,
which explained the limited contribution of industrial emissions to VOCs. Similarly, Guo et al.
[2007] examined the sources of VOCs at an urban site, two suburban sites and a rural site in
Hong Kong. The results showed that vehicle emissions were responsible for a great proportion
of VOCs in both urban (65±36%) and suburban environments (50±28% and 53±41%).
Consistent results were acquired in the following studies [Guo et al., 2011b; Ling and Guo,
2014; Ou et al., 2015; Lyu et al., 2016 a, b]. In fact, the other research groups in Hong Kong
25
also echoed our finding that vehicle exhausts and solvent usage accounted for most of VOC
emissions in Hong Kong [So and Wang, 2004; Zhang et al., 2008; Lau et al., 2010].
Furthermore, the roles of VOCs, alongside with NOx and CO, in O3 formation have also been
intensively studied in Hong Kong. Zhang et al. [2007] indicated that O3 formation was
generally limited by VOCs in most areas of Hong Kong where high concentration of NO
suppressed O3 formation. Aromatics, particularly xylenes and toluene, dominated local O3
production. At a suburban site in Hong Kong, Cheng et al. [2010b] indicated that O3 formation
was most sensitive to formaldehyde, acetaldehyde and isoprene, although aromatics also played
important roles in O3 formation. Similar results were reported by Ling et al. [2013] who found
that isoprene regulated O3 production significantly at this site. Further, Ling and Guo [2014]
calculated the relative incremental reactivity of VOCs in specific sources, and revealed that
xylenes emitted from solvent usage, isoprene from biogenic emission and toluene from
gasoline vehicles were the main contributors to O3 production at an urban site in Hong Kong.
Lyu et al. [2016a] investigated the spatial characteristics of the relationships between O3 and
its precursors, and confirmed that O3 formation in urban and suburban Hong Kong was in VOC-
limited regime, while it was co-limited by VOCs and NOx in rural area. In contrast to urban
and suburban areas where O3 formation was dominated by alkenes and aromatics, the rural area
featured high O3 formation potential of isoprene. Further studies [Xue et al., 2014; Wang et al.,
2017] on the long-term variation trend of O3 in Hong Kong have also been carried out. Xue et
al. [2014] indicated that the production of local Ox (Ox = O3 + NO2) decreased with the rate of
0.26 ± 0.25 ppbv/h/year from 2002 to 2012 in Hong Kong. However, the decreasing trend
was overridden by the regional transport which increased continuously during the study period,
causing the upward trend of the observed O3 in Hong Kong. Consistently, the increasing trend
of regionally transported O3 was confirmed by Wang et al. [2017], who quantified the growth
rate of autumn O3 attributable to regional transport in Hong Kong as 1.09 ppbv/year over the
26
period of 2005 – 2014, while the locally produced O3 was identified to be decreased by 0.39
ppbv/year.
In addition to the locally produced O3, many other air pollutants (such as particulate matters,
SO2, CO and NO) have also shown decreasing trends in Hong Kong [HKEPD, 2016; Lyu et al.,
2017]. We found that the improvement of air quality in Hong Kong was mainly benefited from
the air pollution control measures taken by the Hong Kong government [Lyu et al., 2016b,
2017]. The implementation of air pollution control programs in Hong Kong can be traced to
the early 2000s, such as the substitution of diesel by liquefied petroleum gas (LPG) as the fuel
of taxis and light public buses enforced in 2000 and 2002, respectively, the solvent program
constraining the upper limits of VOC contents in solvent products which came into effect in
2007 and 2010 in phases, the diesel commercial vehicle (DCV) program eliminating high
emission DCVs or upgrading the emission standards from 2007 till the present, and the LPG
program replacing catalytic converters on LPG fueled vehicles implemented between
September 2013 and May 2014. Details about these programs can be found in
http://www.epd.gov.hk/epd/english /environmentinhk/air/air_maincontent.html. Due to the
implementation of the solvent and DCV program, VOCs emitted by solvent usage and diesel
exhaust experienced significant (p<0.05) reductions through 2005 to 2013 in Hong Kong, with
the rate of 204.7±39.7 ppbv/year and 304.5±17.7 ppbv/year, respectively [Lyu et al., 2017]. In
fact, the emission reductions of light alkenes and aromatics from solvent usage and DCVs were
the key factors leading to the decrease of locally produced O3 [Xue et al., 2014, Wang et al.,
2017]. More obvious was the effectiveness of the LPG program on emission reduction of VOCs.
Lyu et al. [2016b] confirmed that the contribution of LPG fueled vehicles to VOCs at a roadside
site in Hong Kong was reduced from 51.5±0.1% before to LPG program to 39.1±0.1% during
the program. Furthermore, some of these programs were closely related to the photochemistry
of RONO2. For example, the ambient concentrations of ethane, propane and n/i-butanes might
27
be significantly influenced by the DCV and LPG programs, while they are all parent
hydrocarbons of RONO2. As a specific example, the impacts of the LPG program on RONO2
production will be discussed in this thesis.
Studies on RONO2 are quite limited in southern China. To my knowledge, “long-term
atmospheric measurements of C1-C5 alkyl nitrates in the Pearl River Delta region of southern
China” by Simpson et al. [2006] was the only RONO2 study in this region, which focused on
the abundances of RONO2 and their relationships with O3. This study aims to fill up the
knowledge gap of RONO2 in Hong Kong and address some key problems related to RONO2.
Specifically, the objectives of this work are to analyze the spatiotemporal variations of RONO2
and their parent hydrocarbons in Hong Kong (Chapter 4); to study the impacts of LPG program
on the parent hydrocarbons of RONO2 (Chapter 5); to explore the sources and photochemical
evolutions of RONO2 in Hong Kong (Chapter 6); to examine the formation pathways of
RONO2 (Chapter 7); to investigate the productions of RONO2 as a function of VOCs and NOx
(Chapter 8); and to figure out the impacts of RONO2 photochemistry on O3 formation (Chapter
8).
28
Chapter 3 Methodology 3.1 Description of sampling sites In this study, whole air samples (WAS) were collected at different sites in Hong Kong from
2001 to 2010. These sites distributed in different areas of Hong Kong, consisting of a
regional background/coastal site in southeastern Hong Kong (Hok Tsui, referred to as HT
hereafter), a rural/coastal site in southwestern Hong Kong (Tai O Island, referred to as Tai O
hereafter), a regional background/rural/coastal site in northeastern Hong Kong (Tap Mun,
referred to as TM hereafter), a suburban site in southwestern Hong Kong (Tung Chung,
referred to as TC hereafter), a suburban site in northwestern Hong Kong (Yuen Long, referred
to as YL hereafter), an urban site in southern Hong Kong (Central Western, referred to as CW
hereafter), an urban site in the centre of Hong Kong (Tsuen Wan, referred to as TW hereafter),
a roadside site in urban centre of Hong Kong (Mong Kok, referred to as MK hereafter), and a
mountainous site in the centre of Hong Kong (Mt. Tai Mo Shan, referred to as TMS
hereafter). Figure 3.1 shows the geographical locations of the sites, of which the detailed
introductions of the coordinates, sampling periods, surrounding environments, and pollution
sources are given in Table 3.1.
Whole air samples were also collected on 27 September 2013 and 24 September 2014,
respectively, at 24 roadside sites, 4 general sites and 2 background sites over the whole
territory of Hong Kong. These samples are of help to study the spatial distributions of
RONO2 in Hong Kong (shown in Chapter 4). Brief description of these sites will be given in
Chapter 4.
Data obtained at all these sites were used to study the spatiotemporal variations of RONO2 in
Hong Kong in Chapter 4. More specifically, the samples were grouped according to the
sampling year. In each group, the observed RONO2 were averaged, which were further
plotted against the sampling time to study the long-term trends of RONO2 in Hong Kong.
29
However, only the samples collected on 27 September 2013 and 24 September 2014 at the 30
sites in Hong Kong were used to get the spatial distributions of RONO2 in Hong Kong. In
Chapter 5, we focused on the variations of RONO2 and their precursors from before to during
an intervention program (replacing catalytic converters on LPG-fueled vehicles), the VOCs
continuously measured at MK from September 2013 to June 2014 and RONO2 data analyzed
from the canister samples collected at the same site during the sample period were used.
Chapter 6 investigated the pathways contributing to RONO2 formation during O3 pollution
events at a coastal site in Hong Kong. Data collected at Tai O between August 2001 and
December 2002 were used in this Chapter. Chapters 7 and 8 put emphases on the sources and
formation mechanisms of RONO2 at a mountainous site (TMS) and an urban site (TW) in
Hong Kong, with the application of VOCs and RONO2 data acquired at TMS and TW during
September-November, 2010.
Figure 3.1 Geographical locations of the sampling sites. The border line between Hong Kong
and mainland China is highlighted in yellow.
Table 3.1 Description of the sampling sites
30
No. Site Coordinates Sampling
period
Surrounding environments/pollution
sources
1 HT 22.22°N,
114.25°E
March 2001-
April 2002
A background/coastal site; ~25 km distance
from urban centre to the northwest; free of
local air pollutant emissions, but affected
by transport from PRD and Hong Kong.
2 Tai O 22.25°N,
113.85°E
August 2001-
December 2002
A suburban/coastal site; ~30 km away from
the urban center of Hong Kong to the
north; ~30 km from Macau to the west, and
at the mouth of PRE; low local air pollutant
emissions, but affected by air pollution at
PRE.
3 TM 22.47°N,
114.36°E
August 2002-
August 2003,
September
2006-July 2007
A background/rural site, surrounded by
country parks, and upwind of Hong Kong
in autumn/winter seasons.
4 TC 22.29°N,
113.94°E
August 2002-
August 2003,
September
2006-July 2007
A mixed residential and commercial area;
to the northwest ~3 km from Hong Kong
international airport, and to the northeast
~20 km away from urban Hong Kong. The
main local sources of air pollutants are the
airport, nearby highways to the airport,
local residential activities, as well as air
transport from inland PRD and urban Hong
Kong.
31
5 YL 22.44°N,
114.02°E
August 2002-
August 2003,
September
2006-July 2007
A mixed residential and commercial area;
to the southeast, ~10 km from urban areas
of Hong Kong, and to the north ~10 km
from urban Shenzhen.
6 CW 22.29°N,
114.14°E
August 2002-
August 2003,
September
2006-July 2007
An urban site in Hong Kong Island, and
local traffic is the main source of air
pollutants.
7 TW 22.37°N,
114.11°E
September-
November 2010
An urban site in New Territories of Hong
Kong, and local traffic is the main source
of air pollutants.
8 MK 22.32°N,
114.17°E
September
2013-June 2014
A roadside site with heavy traffic and
surrounded by dense tall buildings; local
traffic is the main source of air pollutants.
9 TMS 22.41°N,
114.12°E
September-
November 2010
At the mountainside (height: 640 m) of the
highest mountain (height: 957 m) in Hong
Kong; a receptor site of local and regional
air pollution under mesoscale circulations
and trans-boundary transport.
3.2 Sample collection and chemical analysis
3.2.1 VOCs collection
The sample collection and chemical analysis of VOCs were carried out offline or online. For
the offline approach, the WASs were collected at the sites introduced above during the
corresponding periods (see Table 3.1) using the evacuated 2 liter stainless steel canisters. The
32
inner walls of the canisters were electro-polished and preprocessed with 10 Torr of degassed,
distilled water to quench the active surface sites. These treatments were to alleviate the loss of
chemical molecules on the inner walls. Prior to each sampling campaign, the canisters were
pre-cleaned with humidified pure nitrogen gas (N2) for at least five times. Specifically, the
canister was filled with N2 to 206.8 kPa and evacuated to 1.1×10-4 kPa in a round of cleaning,
which was repeated for five times and the ultimate inner pressure was controlled to be less than
0.05 mm Hg. To test for any contamination in the canister, the randomly selected well prepared
canister was filled with pure N2 and stored for at least 24 h, followed by analysis with the VOC
analytical protocol identical to that used in analysis of WASs. The standard of acceptance was
that all the target compounds were not found or were under the corresponding method detection
limits (MDLs). During sampling, a valve was interfaced with the inlet of the canister to
maintain certain duration of sampling, which was ~1 min, ~3 min and ~1 hour in different
campaigns. After sampling, the canisters were delivered to University of California, Irvine
(UCI) for chemical analysis within two weeks.
3.2.2 Offline and online analysis of VOCs
The Rowland/Blake group (Nobel laureate laboratory) at UCI has rich experience in analyzing
ambient VOCs including hydrocarbons, halocarbons, carbonyls and organic nitrates, and has
made remarkable contributions to the scientific community in the field of VOC analysis
techniques. The whole analysis system is composed of a pre-concentration unit, three gas
chromatographs (GCs) and six detectors. During each analysis, ~700 Torr of air sample was
introduced into a manifold immersed in a liquid nitrogen bath to remove the highly volatile
compounds, such as N2, O2 and Ar, but trap the less volatile species including hydrocarbons,
halocarbons, carbonyls and organic nitrates. Subsequently, the sample loop was immersed in
the hot water of ~80 °C to volatilize the trapped compounds, which were then flushed into a
helium carrier flow and split into five streams [Colman et al., 2001]. Sample in each stream
33
was separated with a GC column and analyzed an individual detector (methane and CO were
separately analyzed) [Simpson et al., 2003]. Table 3.2 lists the configurations of GCs,
chromatographic columns and detectors for the separation and detection of different
compounds.
Table 3.2 Configurations of GCs, chromatographic columns and detectors for the separation
and detection of different compounds.
Compounds GC Chromatographic column Detector
Halocarbons &
organic nitrates
HP-6890 (GC1) J&W DB-5 + RESTEK
1701
ECD 1
Sulphur
compounds
HP-6890 (GC1) DB-5ms MSD 2
Hydrocarbons HP-6890 (GC2) J&W DB-1 FID 3
Hydrocarbons HP-6890 (GC3) J&W GS-Alumina PLOT
+ DB-1
FID
Halocarbons and
organic nitrates
HP-6890 (GC3) RESTEK 1701 ECD
Methane & CO * HP 5890 Porapak Q FID
1 Electron capture detector, 2 Mass spectrometer detector, 3 flame ionization detector. * Methane
and CO are separately analyzed.
To control data quality, the working standard and calibration standard of VOCs were employed,
which were run every 2 hours and at least twice daily, respectively. The precision and accuracy
for the measurements were on average 3% and 5%, respectively, with MDL of 3 pptv. Details
about the MDL, precision and accuracy for the measurements of each species can be found in
Simpson et al. [2010]. Specifically for RONO2, the precision was not lower than 2% and 10%
34
for mixing ratios above and below 5 pptv, respectively. The accuracy of RONO2 measurement
was 10-20%, and the MDL was 0.01 pptv. It is worth noting that the RONO2 calibration scale
changed in 2008. According to the new scale, a factor of 2.13, 1.81, 1.24, 1.17 and 1.13 was
applied to C1, C2, C3, C4 and C5 RONO2 measured before 2008, respectively [Simpson et al.,
2011].
For the online measurements of VOCs including alkanes, alkenes and aromatics (RONO2
cannot be measured by this system), a real-time gas chromatograph (GC) unit (Syntech Spectra
GC 955 611/811, the Netherlands) was employed to quantify the concentrations of speciated
VOCs every 30 minutes. A built-in computerized program, including auto-linearization, auto-
calibration and calibration with span gas, was adopted to control the quality of the data. The
accuracy and precision of VOC measurements were 5-20% and 1.2-10.1%, respectively. The
accuracy was based on weekly span checks, monthly calibration and annual auto-linearization
using the National Physical Laboratory (NPL) span gas. The precision was based on quarterly
precision check results (the 95% probability limits for the integrated precision based on weekly
precision check results of the latest 3 months). Moreover, the online-measurements of VOCs
were regularly compared with whole-air canister samples collected and analyzed by UCI. Good
agreements were identified for the alkanes (e.g., R2 = 0.95 and 0.85, slope = 1.14 and 0.97 for
propane and butanes, respectively), while the agreements for the more reactive alkenes and
aromatics were also reasonable (e.g., R2 = 0.64 and 0.94, slope = 1.34 and 0.86 for propene
and toluene, respectively). The MDLs of VOCs varied between 15 and 1186 pptv. In this thesis,
except that VOCs at MK during September 2013 – May 2014 were detected by on-line analysis
system, all the other samples were collected by canisters and analyzed by the off-line GC-
MSD/FID/ECD system in UCI.
The collection of RONO2 samples spanned a long time from as early as 2001 to the latest 2014.
I only participated in sample collections in the grid study in 2013 – 2014. However, a set of
35
consistent quality control and assurance (QA/QC) protocol has been strictly abided by in all
the sampling campaigns, as introduced above. More importantly, the log sheets for all the
sampling campaigns are preserved and available for me to check when the data quality is
questioned. In fact, I participated in one and led another three sampling campaigns during my
PhD study, including a large sampling campaign to collect hazardous air pollutants in 18 cities
over the whole territory of China. During these four sampling campaigns, the QA/QC protocols
obeyed were the same with those applied in previous samplings. Thus, I believe that I was well
trained the skills and standard operations in sample collection. The data obtained from the
sampling campaigns I participated in will be used for future studies on RONO2, O3 and SOA
in our group.
For readers’ awareness, methyl, ethyl, i-propyl, n-propyl and 2-butyl nitrate in this study are
expressed as C1, C2, 2-C3, 1-C3 and 2-C4 RONO2, or CH3ONO2, C2H5ONO2, 2-C3H7ONO2, 1-
C3H7ONO2 and 2-C4H9ONO2, respectively.
3.2.3 Continuous measurement of inorganic trace gases
The inorganic trace gases (SO2, CO, NO, NO2 and O3) were continuously monitored by a set
of online instruments deployed either at the air quality monitoring stations (AQMSs) by Hong
Kong Environmental Protection Department (HKEPD) or at the other sites by our group. The
main analysis techniques rather than instrumentations which might vary among the sites are
introduced here. The CO was analyzed using the method of non-dispersive infra-red absorption
with gas filter correlation. The SO2 was detected with the technique of UV fluorescence. The
NO-NO2-NOx were measured by chemiluminescence technique; and O3 was monitored with
the UV absorption method. Data monitored at AQMSs were accessible at
http://epic.epd.gov.hk/EPICDI/air/station/. The detection limits of CO, SO2, NO, NO2 and O3
were 50.0, 1.0, 0.5, 0.5 and 2.0 ppbv, respectively, which were 30.0, 0.4, 0.4, 0.4 and 0.6 ppbv,
respectively, with our own measurements. HKEPD implemented standard quality controls on
36
the measurements, including checking the precision in accordance with EPD’s quality manuals
and assessing the accuracy by performance audits. In the most recently published air quality
assessment report in 2015, the precision of CO, NO2 and O3 in the HKEPD network stations
ranged between -4.5% and 5.4%, or well within the control limits of ±15% for gaseous
pollutants [HKEPD, 2015a]. The accuracy of these pollutants ranged from -11.5 to 8.5%, also
within the control limits of ±20%. Measurements at the sites other than AQMSs were also
subject to regular calibrations with the NIST (National Institute of Standards and Tech)
traceable standard (Scott-Marrin, Inc.). The standard contains 156.5 ppmv CO, 15.64 ppmv
SO2 and 15.55 ppmv NO, which were diluted with a dynamic calibrator (Environics, Inc.,
Model 6100) for calibration. O3 calibration was carried out with the combined application of
the calibrator and an O3 generator (API Model 701H). The precision of the inorganic trace gas
measurements was generally less than 0.5%.
3.3 PBM-MCM model
3.3.1 Basic structure of PBM-MCM
In this study, a photochemical box model incorporating master chemical mechanism (PBM-
MCM) was developed to simulate the photochemistry of RONO2 and study their impacts on
O3 production. The PBM is composed of the following modules: (1) parameter/variable
definition, (2) data input, (3) initial condition setting, (4) photolysis rate calculation, (5)
chemical mechanism, (6) dry deposition setting, (7) aloft exchange setting and (8) output. The
parameters and variables appeared in the model protocol must be defined in the
parameter/variable definition module, or they will be treated as unknown characters and lead
to the termination of modelling. It should be noted that the parameters are constrained by the
input data, meaning that the values of the parameters will be unchanged and consistent with
those input into the model if any over the course of modelling. Differently, the value of
37
variables depends upon the initial conditions, input data and the iterative calculation on basis
of the differential equations compiled in the model. Data input module includes the observed
data with any time resolution. Generally, both the environmental and meteorological data are
needed to drive the modelling. In the module of initial condition setting, the initial
concentrations and conditions to initiate the model (such as start time, duration of sunlight and
height of mixing layer) can be assigned. Photolysis rates of air pollutants are important factors
influencing the oxidative capacity of the atmosphere. In the PBM-MCM model, the photolysis
rates are expressed as functions of latitude, solar declination angle and the time of the day.
However, this is an ideal condition without considering the effects of overhead ozone column,
surface albedo, cloud optical depth (COD) and aerosol optical depth (AOD) on the strength of
solar radiation. Therefore, the photolysis rates calculated by the PBM-MCM model are
calibrated according to those estimated by Tropospheric Ultraviolet and Visible Radiation
(TUV) model [Madronich and Flocke, 1997]. The parameters in the TUV model are set on the
basis of real measurements (such as latitude, longitude, date and time) or assumptions (such as
COD and AOD), with the goal of reproducing the observed solar radiation. The photolysis rates
were calibrated daily according to the monitored solar radiation. With the input of location,
time period and the initial default COD and AOD, the TUV model calculates and outputs the
photolysis rates of the important air pollutants (such as O3, NO2, HONO and HCHO) and the
simulated solar radiation. Then, the COD and AOD are adjusted to make the simulated solar
radiation progressively approach the observed one. The photolysis rates are adopted in the case
that the simulated solar radiation is most close to the observation. The chemical mechanism
module is constructed with the near-explicit and species-based master chemical mechanism,
which includes 17,242 reactions of 5,836 species in the latest released version (MCM v3.2). It
is noteworthy that our group has participated in developing the photochemical mechanisms of
many important VOCs in MCM, including biogenic VOCs and RONO2. The atmospheric
38
processes of RONO2 compiled in MCM are introduced in details below. The dry deposition
and substance exchange between lower troposphere and free troposphere (aloft exchange) are
set with parameterized velocities. The output module enables to extract the values of
parameters and variables in any resolution. Though the dry deposition rates and concentrations
of air pollutants in the free troposphere were set according to the literatures, they would
inevitably bring some uncertainties to the modeling results. For example, the dry deposition
rate of O3 is generally lower over the waters than above the terrestrial regions. However, we
believe that the influences of these settings were minor in the urban areas where the abundances
of air pollutants were dominated by local emissions and photochemistry.
3.3.2 Atmospheric processes of RONO2 in PBM-MCM
The atmospheric processes of RONO2 considered in PBM-MCM include the formation,
degradation, dry deposition and exchange between lower and free troposphere. The formation
of RONO2 is described as RO2 reacting with NO and RO reacting with NO2. The reaction
coefficient of RO2 reacting with NO is based on the recommendations by Eberhard and
Howard [1996, 1997], Eberhard et al. [1996] and Atkinson et al. [1999]. Specifically, it is
presented as a function of temperature (T) as 2.3×10-12×e(360/T), 2.55×10-12×e(380/T), 2.7×10-
12×e(360/T), 2.9×10-12×e(350/T) and 2.7×10-12×e(360/T), 2.7×10-12×e(360/T) and 2.7×10-12×e(360/T) in
unit of cm3 molecule-1 s-1 for the formation of C1, C2, 2-C3, 1-C3,2-C4, 3-C5 and 2-C5 RONO2,
respectively. The branching ratio (α) was an important factor influencing the production of
RONO2, which will be discussed in detail in Chapter 8. The reaction coefficient of RO reacting
with NO2 was set according to the International Union of Pure and Applied Chemistry (IUPAC)
recommendations in 2011 (https://www.iupac.org/publications/pac/ reports/year/2011/),
namely 1.5×10-11, 2.8×10-11, 3.4×10-11, 3.6×10-11 and 8.6×10-12×e(400/T) in units of cm3
molecule-1 s-1 for the formation of C1, C2, 2-C3, 1-C3 and 2-C4 RONO2, respectively. Due to
39
the decomposition and isomerization of C5 RO radicals, the reactions between C5 RO and NO2
were not considered as the formation pathways of C5 RONO2 in PBM-MCM model.
The degradation (chemical losses) of RONO2 includes photolysis and OH initiated oxidation.
The photolysis rates in the MCM were treated as a function of solar zenith angle, which is fully
described in Jenkin et al. [1997]. The parameterizations of the functions were based on the
recommendations or averages of previous studies, such as Roberts and Fajer [1989], Zhu and
Ding [1997], and the IUPAC recommendations of 2011. The reaction coefficient of OH
initiated oxidation was adopted from the IUPAC recommendations of 2011, and the preferred
values were set as 4.0×10-13×e(-845/ T), 6.7×10-13×e(-395/T), 6.2×10-13×e(-230/T), 5.8×10-13 and
8.6×10-13 in units of cm3 molecule-1 s-1 for C1, C2, 2-C3, 1-C3 and 2-C4 RONO2, respectively.
The reasonableness of these settings can be found at the MCM website
(http://mcm.leeds.ac.uk/MCM/).
The dry deposition and aloft exchange of RONO2 are parameterized with the settings
appropriate for Hong Kong, such as the dry deposition velocities and the concentrations of
RONO2 in the free troposphere of Hong Kong. In chapter 8, different dry deposition velocities
within the range of reference values reported in literatures were tested in simulations of
different RONO2, through which the most appropriate velocities were obtained by comparison
to observations. The concentrations of RONO2 in the free troposphere of Hong Kong were
identical to the settings in Lam et al. [2013]. Furthermore, sensitivity analysis indicated that
aloft exchange did not significantly influence RONO2 modelling. For example, 10% change of
RONO2 concentrations in the aloft layer only caused less than 2% variation of the simulated
RONO2 at a mountainous site (TMS) in Hong Kong. Therefore, the aloft settings in Lam et al.
[2013] at a mountainous site (TMS) in Hong Kong were applied in RONO2 modeling at the
other sites in this thesis.
40
3.3.3 Application of PBM-MCM
In this study, the observed VOCs, inorganic trace gases and meteorological parameters were
used to construct PBM-MCM, an observation based photochemical box model. Five inorganic
trace gases including SO2, CO, NO, NO2 and O3, as well as 2 meteorological parameters
(temperature and relative humidity), were the input of the model throughout the study.
However, the number and species of VOCs might differ among the simulations at different
sites, dependent upon the availability of VOC data in different measurements. In chapters 4-5,
the model was constructed with 27 VOCs obtained from online measurements, while 63 VOCs
obtained from canister samples were used in the RONO2 simulations in chapters 7-8. The
resolutions of the model input and output were set as hourly.
Prior to this study, PBM-MCM model has been successfully used in the simulations of O3
production and photochemical reactivity in Hong Kong by our group [Lam et al., 2013; Ling
et al., 2014]. The performance of this model was evaluated using a statistical parameter, i.e.,
index of agreement (IOA), which was calculated using formula (3-1). Within the range of 0-1,
higher values of IOA are generally indicative of better agreements between two sets of data
(e.g., the observed and simulated data) [Willmott et al., 1985; Legates et al., 1999]. To
guarantee the model performance, the simulations were regarded as acceptable only when IOAs
between the observed and simulated values exceeded 0.65.
IOA = 1 −∑ (𝑂𝑖−𝑆𝑖)2𝑛
𝑖=1
∑ (│𝑂𝑖−�̅�│+│𝑆𝑖−�̅�│)2𝑛1𝑖=1
Formula (3-1)
where 𝑂𝑖 and 𝑆𝑖 are the hourly observed and simulated values, and �̅� represents the average of
observed values in n samples.
The output uncertainty of the PBM-MCM model derived from two parts, i.e., uncertainties of
the input species and uncertainty inherent to the chemical mechanism. In this study, 27 - 63
VOCs, 5 trace gases and 2 meteorological conditions (temperature and relative humidity) were
41
used to construct the model. The measurement accuracy for VOCs ranged from 2% for methane
to 20% for dimethyl sulfide and some other compounds. Details about the accuracy for each
species can be found in Simpson et al. [2010]. The uncertainty of temperature, relative humidity
and the chemical mechanism (MCM) was roughly estimated as 5%, 5% and 10%, respectively.
Mean root square error (MRSE) was calculated for the accuracies of the input species and
uncertainty of MCM, using formula 3-2 [Willmott, 1982], which was regarded as the
uncertainty of the PBM-MCM model.
MRSE = √∑ 𝑋2𝑛
1
𝑛
2 (Formula 3-2)
where X represents the individual uncertainty of each component (total of n).
3.4 Positive matrix factorization model Positive matrix factorization (PMF) is one of the most commonly used receptor models (top-
down approach) in source apportionment of VOCs and particulate matter [Lee et al., 1999;
Brown et al., 2007]. Consistent with other receptor models such as CMB and Unmix, the
operation principle of PMF is to solve the chemical mass balance equation between the
observed ambient concentrations and source emissions. Briefly, as a multivariate factor
analysis tool, PMF decomposes the input matrix (X) into matrices of factor contribution (G)
and factor profile (F) in p sources (Equation 3-1). Differing from CMB that the source types
are known in advance, the sources extracted from PMF need to be identified by the users
according to the factor profile and emission characteristics of sources [Paatero and Tapper,
1994; Paatero, 1997]. In this study, PMF was utilized to resolve the sources of VOCs in Hong
Kong, with or without the inclusion of RONO2 depending upon the purposes and data
availability. The hourly concentrations of VOCs were included in the input matrix. Values
below the MDLs were replaced with the half of the MDLs. The uncertainties were set
42
√(10% × 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛)2 + 𝑀𝐷𝐿2 and 5
6× 𝑀𝐷𝐿 as for the samples with the
concentrations higher and lower than MDLs, respectively. The signal-to-noise (S/N) ratios
were all greater than 1, indicating “good” signal for all the species involved in source
apportionment, according to the PMF 5.0 User Guide [Norris et al., 2014]. Samples with any
missing values were excluded.
𝑥𝑖𝑗 = ∑ 𝑔𝑖𝑘𝑓𝑘𝑗 + 𝑒𝑖𝑗𝑝𝑘=1 (Equation 3-1)
where 𝑥𝑖𝑗 is the concentration of jth species in ith sample, 𝑔𝑖𝑘 represents factor contribution of
kth source to ith sample (by mass), 𝑓𝑘𝑗 denotes the percentage contribution of kth source to jth
species (by percentage), and 𝑒𝑖𝑗 is the residual of 𝑥𝑖𝑗.
To obtain the most reasonable apportionment results, including the number of factors and the
best solutions, the following criteria were executed: (1) The objective function Q got the lowest
value (Equation 3-2); (2) The ratio between Qtrue and Qrobust was lower than 1.5 under the
condition of data uncertainties introduced above (Qtrue and Qrobust are the goodness-of-fit
parameters calculated including all points and excluding the samples with uncertainty-scaled
residual>4, respectively); (3) Good agreements were shown between the predicted and
observed concentrations for total and individual VOCs; (4) The residuals were normally
distributed between −3 and 3; and (5), no correlation was found between the factors, which
was achieved by examining the G-space plots and controlled by the FPEAK model runs.
Q = ∑ ∑ [𝑥𝑖𝑗−∑ 𝑔𝑖𝑘𝑓𝑘𝑗
𝑝𝑘=1
𝑢𝑖𝑗]2𝑚
𝑗=1𝑛𝑖=1 (Equation 3-2)
where 𝑢𝑖𝑗 is the uncertainty of jth species (total of m) in ith sample (total of n). The others are
the same with those in Equation 3-1.
43
The uncertainties of model outputs were estimated using the bootstrap method integrated in
PMF model. It should be noted that all the error bars given in this study represent 95%
confidence intervals (C.I.s) unless otherwise specified.
3.5 Other tools and calculation techniques
3.5.1 Calculation of VOCs diurnal patterns
As introduced above, two VOC canister samples per day were collected at the 24 roadside sites
and 6 general sites on September 27, 2013 and September 24, 2014, respectively. The samples
were collected at ~10:00 and ~15:00, respectively. To investigate the photochemical behaviors
of VOCs with the PBM-MCM model, data with higher time resolutions were expected to
support more accurate simulations, although PBM-MCM did not require the time resolution of
data. However, the online measurement of VOCs with hourly resolution was only available at
3 out of 24 sites (MK, TC and YL) Hence, the hourly concentrations of VOCs in a diurnal cycle
were estimated based on the instant concentrations in only two canister samples. The
procedures to obtain hourly VOC profiles from two samples were similar to the method
described in Zhang et al. [2007]. Briefly, the method was based on the mass conservation of a
species inside a fixed Eulerian box, namely, the Eulerian box model [Seinfeld and Pandis,
1997]. The entraining equations are as follows:
iii
idi
ii CCC
tHR
tHq
dtdC
0
,
)()( for 0
dtdH (Equation 3-3)
dtdH
tHCCCC
CtH
RtH
qdt
dC iaiii
iid
iii
)()()(
0,
for 0
dtdH (Equation 3-4)
where, Ci: Concentration of species i (µg m-3),
qi : Emission rate of species i (µg m-2 s-1),
44
H(t): Mixing height as a function of time t (m),
Ri: Chemical destruction rate of species i (µg m-3 s-1),
υd,i: Dry/wet deposition rate of species i (m s-1),
C0 i : Background concentration of species i (µg m-3),
τγ: Residence time of air over the area (s),
Ca i : Concentration of species i above the boundary layer (µg m-3).
Equations (3-3) and (3-4) mathematically described the concentration of species above a given
area, assuming that the corresponding airshed was well mixed, accounting for emissions,
chemical reactions, removal, advection of material in and out of the airshed, and entrainment
of material during growth of the mixing layer. Before the numerical solutions of equations (3-
3) and (3-4) are solved with Gear's backward differentiation formula [Jacobson, 2005], the
parameters in equations (3-3) and (3-4) need to be determined. The first term in both equations
( 𝑞𝑖
𝐻(𝑡)) describes the concentrations of air pollutants elevated by the emission and mixing within
the boundary layer. The emission rates of air pollutants (qi) calculated from the emission
inventory may introduce uncertainties into the calculation of the diurnal profiles. The second
term (Ri) stands for the loss rates of air pollutants through chemical reactions, which closely
relates to the full and explicit consideration of the chemical reactions. The third term ( 𝑉𝑑,𝑖
𝐻(𝑡)𝐶𝑖)
represents the dry deposition losses in the mixing layer, depending upon the accurate estimates
of the dry deposition velocities and the heights of mixing layer. 𝐶𝑖0−𝐶𝑖
𝜏𝑦 is the advection term
which represents the change of air pollutants concentrations due to horizontal advection. The
last term in equation 3-4 (𝐶𝑖𝑎−𝐶𝑖
𝐻(𝑡)
𝑑𝐻
𝑑𝑡) stands for the change of air pollutants concentrations due
to exchange with free troposphere.
45
Emission Rates: the Hong Kong emission inventory of total anthropogenic VOCs from
different sources in 2013 was used to estimate the annual emission amount [HKEPD, 2015b].
This annual emission amount was then equally allocated to 52 weeks and the area of 1104
square kilometers covering Hong Kong territory. The area of Hong Kong was obtained from
Censtatd [2016], while the days were classified as weekdays and weekends, and the emission
factors of each day of a week were determined by Cardelino [1998]. Hence, the daily initial
emission amount of total VOCs per unit area was calculated. This value was multiplied by the
typical profile of ambient VOCs, which was obtained by averaging all canister data at different
sites, to derive the daily initial emission rates of speciated VOCs at different sites. The diurnal
variations of the anthropogenic VOC emissions were estimated according to source types.
Industry and power generation were assumed to have no diurnal variations, while mobile
emission had the same pattern as traffic flow in Hong Kong [Xia and Shao, 2005; Lam et al.,
2006]. The diurnal variation of biogenic VOCs (i.e., isoprene) was estimated by considering
the temperature variations and the best fit value coefficient TM, which affected the predicted
emission behavior at high temperatures [Guenther, 1993, 1999]. In this way, the speciated
VOCs emission rates from different sources were determined.
Mixing Height Profile: The mixing height was estimated using the Holzworth method
[Holzworth, 1967]. The Holzworth method provides twice-per-day (morning and afternoon)
mixing heights based on calculations using routine upper-air data and minimum and maximum
temperature of the day. The upper air sounding data were obtained from the University of
Wyoming (http://weather.uwyo.edu/upperair/sounding.html). The minimum temperature was
determined from the data of King's Park station operated by Hong Kong Observatory (HKO)
(http://www.hko.gov.hk/) for the time period of 0200-0600 local standard time (LST). Here,
we followed the method of Zhang et al. [2007], which also calculated the mixing height in
Hong Kong by using “plus 2 °C” to the morning minimum surface temperature to calculate the
46
morning mixing height. The afternoon mixing height was calculated using the maximum
surface temperature observed at 1200-1600. The hourly mixing heights, often used in
regulatory dispersion modeling, were interpolated from these twice-per-day estimates. The
recommended interpolation procedure is provided in the user’s guide for the Industrial Source
Complex (ISC) dispersion model [USEPA, 1985].
The dry/wet deposition rate and the concentration above the boundary layer were assumed to
be zero for all VOCs, because almost all the sites in this sampling campaign were on ground
level. Background concentrations of VOCs were expressed as geometric mean concentrations
at general sites. Residence time of air over the area was the ratio of length of the box to wind
speed, and the time-dependent wind speed was determined by curve fitting using the in-situ
hourly wind speed obtained from the HKO. Since length of the box and emission rates of VOCs
at different sites were different and the VOC chemical destruction rates were unknown, we
adopted an iterative approach to determine these parameters based on canister data at 10:00
and 15:00. We first used a typical OH profile in clean marine atmosphere [Creasey et al., 2003]
and the initial emission rate of propane to calculate the temporal variations of propane from
10:00 to 15:00 with the box length ranging from 0 to 60 km (i.e., beyond the longest range of
Hong Kong territory). Propane was selected due to its high concentration and lower reactivity
with OH compared to alkenes. The optimal emission rate and the length of box were adjusted
by matching the calculated propane level at 15:00 with the observed value, using a 5%
agreement for consistency. The ratio of optimal emission rate to initial emission rate of propane
was defined as emission rate factor. The temporal variation of more reactive propene from
10:00 to 15:00 was calculated in the same way, but the length of box was fixed and the emission
rate of propene was modified by multiplying the initial emission rate of propene by the
emission rate factor determined by propane above. As such, the original OH profile used above
was refined to fit the real situation in Hong Kong. The refined OH profile was then used to
47
recalculate the temporal variation of propane. The entire procedure was called iteration.
Iterations were repeated until convergence was obtained. Thus the length of box, emission rates
of VOCs and OH profiles at different sites were obtained via this iterative approach. This
approach was employed to determine the diurnal cycle of VOCs at the sites where continuously
measured VOCs were not available, and the data were only used in Chapter 4 to simulate the
O3 productions in different areas of Hong Kong before and after the replacement of the catalytic
converters on LPG-fueled vehicles.
3.5.2 Removal of background concentrations
In Chapter 5, we examined the effectiveness of replacing catalytic converts on emission
reduction of VOCs and trace gases from LPG-fueled vehicles at a roadside site (MK) in Hong
Kong. The measured concentrations of VOCs and trace gases are composed of the backgrounds,
primary emissions and secondary formation (applicable to O3 and NO2). To obtain the
concentrations of VOCs and NOx emitted from primary sources at the roadside sites, the
background concentrations and the concentrations elevated by the secondary formation (if
applicable) in this study were excluded using equations (3-5) - (3-9) [Takekawa et al., 2013]:
[𝑉𝑂𝐶] 𝑝𝑟𝑖𝑚. = [𝑉𝑂𝐶] 𝑜𝑏𝑠. − [𝑉𝑂𝐶] 𝑏𝑔. (Equation 3-5)
[𝑁𝑂2]𝑠𝑒𝑐. = [𝑂3]𝑏𝑔. − [𝑂3]𝑜𝑏𝑠. (Equation 3-6)
[𝑁𝑂2]𝑝𝑟𝑖𝑚. = [𝑁𝑂2]𝑜𝑏𝑠. − [𝑁𝑂2]𝑠𝑒𝑐. − [𝑁𝑂2]𝑏𝑔. (Equation 3-7)
[𝑁𝑂𝑥]𝑝𝑟𝑖𝑚. = [𝑁𝑂𝑥]𝑜𝑏𝑠. − [𝑁𝑂𝑥]𝑏𝑔. (Equation 3-8)
[𝑁𝑂]𝑝𝑟𝑖𝑚. = [𝑁𝑂𝑥]𝑝𝑟𝑖𝑚. − [𝑁𝑂2]𝑝𝑟𝑖𝑚. (Equation 3-9)
where [xx]obs., [xx]prim., [xx]sec. and [xx]bg. represent the observed concentrations, the
concentrations emitted from primary sources, secondary formation and the backgrounds,
respectively. In this study, the hourly measured VOCs at HT were considered as the
48
background levels for VOCs. For O3 and NOx, the 8-hr averages at TM were treated as
background values because (1) O3 and NOx are highly reactive, and the use of 8-hr averages
would minimize the impact of abrupt changes; (2) TM is located at the upwind location of Hong
Kong, and is a rural coastal site; (3) 8-hr averages of O3 and NOx showed moderate to good
correlations (R2 = 0.75 and 0.57 for O3 and NOx, respectively) between TM and HT, and not
all the data were available at HT during the entire study period. It should be noted that the
removal of background concentrations from the observed VOCs and trace gases was only
applied to obtain the actual emissions of LPG-fueled vehicles in Chapter 5, but not in
simulation of O3 production and the whole air photochemistry before and during the LPG
program.
3.5.3 Eliminating interferences of non-local air masses
To investigate the impacts of a LPG program on the concentrations of RONO2 parent
hydrocarbons, MK was regarded as an appropriate site, because (i) it was a roadside site,
sensitive to the changes of vehicle emissions, and (ii) data of VOCs and trace gases were
complete at this site. As a roadside site, MK was expected to be strongly influenced by fresh
vehicular emissions. VOC ratios can indicate the relative ages of air masses and regional
transport of air pollutants. For example, the ratios of VOCs with higher reactivity to those with
lower reactivity (e.g. toluene/benzene and xylenes/ethylbenzene) imply more recent emissions
when values are higher, and these ratios have been extensively used in previous studies [Ho et
al., 2004; Guo et al., 2007] to segregate fresh and aged emissions. Therefore, ratios of
toluene/benzene and xylenes/ethylbenzene were calculated here to roughly assess the relative
age of air masses at MK, compared to that in the other environments (such as in the suburban
and background areas). Table 3.3 summarizes the ratios at MK (roadside site), TC and YL (both
are general ambient sites), and HT (background site) in Hong Kong from October 2012 to May
2014. The ambient VOCs at all the four sites were simultaneously measured. The ratios of
49
toluene/benzene and xylenes/ethylbenzene at MK were significantly higher than those at the
other three sites (p<0.05). Furthermore, these ratios at MK were even higher after the
background values were deducted. Therefore, the concentrations of air pollutants observed at
MK with the subtraction of background levels could well represent the local emissions.
Table 3.3 Ratios of toluene/benzene and xylenes/ethylbenzene with 95% confidence intervals
at MK, TC, YL and HT sites in Hong Kong from October 2012 to May 2014 (unit: ppbv/ppbv)
Toluene/Benzene Xylenes/Ethylbenzene
MK (roadside site) 4.1 ± 0.04 3.0 ± 0.02
MK* (roadside site) 11.3 ± 0.9 5.7 ± 0.2
TC (general site) 0.7 ± 0.1 -
YL (general site) 3.7 ± 0.04 2.4 ± 0.02
HT (background site) 2.5 ± 0.1 1.8 ± 0.03
MK* refers to VOCs at MK with the background values subtracted.
However, regional and super-regional transport cannot be completely eliminated only with the
VOC ratios, and the influence of regional/super-regional air should be further examined when
studying the emissions of local LPG-fueled vehicles. Using the propane/CO ratio method
proposed by Guo et al. [2006], i.e., the ratio range of 5.0-300 pptv/ppbv for air masses in Hong
Kong, the influence of regional/super-regional air masses on local air was evaluated. Briefly,
the air masses with propane/CO ratios lower than 5.0 pptv/pptv were treated to be from inland
PRD, where CO was higher and propane was lower than those in Hong Kong due to the more
intensive emissions of CO in inland PRD (e.g. bad situation of vehicle engines, biomass/biofuel
burning, coal combustion and etc.) and the use of LPG as vehicle fuels in Hong Kong. Figure
3.2 shows the concentrations of propane and CO at MK from 2011-2014. The propane/CO
ratios of ~96% air masses were between 5.0 and 300 pptv/ppbv. Further inspection of the values
50
of LPG-related VOCs and NOx with ratios of 5.0-300 showed insignificant differences (p>0.1)
from those with ratios beyond 5-300, suggesting that the air masses at MK were seldom
influenced by regional/super-regional air. Moreover, wind speeds were lower than 2.0 m/s for
more than 95% of the time (i.e., calm or light air according to the Beaufort Wind Scale)
(http://www.spc.noaa.gov/faq/tornado/beaufort.html), and the values of LPG-related VOCs and
NOx in all air masses were not different from those in the air masses with wind speeds lower
than 2.0 m/s (p>0.1), confirming a minor impact of regional/super-regional pollution on air at
MK. Indeed, the prevailing wind direction was from the east, where it is the local urban core
of Hong Kong.
Figure 3.2 Hourly average mixing ratios of propane versus CO at MK during 2011-2014.
3.5.4 Calculation of relative incremental reactivity
Relative incremental reactivity (RIR), initially proposed by Carter and Atkinson [1989b], has
been extensively used to describe the relationship between O3 and its precursors, i.e., VOCs,
NOx and CO. Although the observation-based models incorporating carbon bond mechanisms
were often utilized to simulate the O3 production rate [Ling et al., 2011; Zhang, et al., 2007;
Martien et al., 2003], a more explicit PBM-MCM model was applied in this study. The RIR
51
and RIR (weighted concentration) were calculated using equations (3-10) and (3-11) [Ling et
al., 2011]:
RIRS(X) =[PS
O3−NO(X)−PSO3−NO(X−∆X)]/PS
O3−NO(X)
∆S(X)/S(X) (Equation 3-10)
RIR (weighted concentration) = RIR(X) × concentration (Equation 3-11)
where, PSO3−NO(X) and PS
O3−NO(X − ∆X) represent the original O3 production rate, and that
in the scenario, with the a hypothetical change (∆X) (10% in this study) in source/species X,
respectively, both of which considered O3 titration by NO. ∆S(X) is the change in the
concentration of X ( S(X) ). The “concentration” refers to the observed or PMF-extracted
concentration of source/species X.
52
Chapter 4 Spatiotemporal variations of RONO2 and their parent hydrocarbons in Hong Kong
In this chapter, the long term trends and spatial distribution of RONO2 in Hong Kong were
studied. In addition, the impacts of an intervention program aiming to improve air quality in
Hong Kong on the abundances of RONO2 and their parent hydrocarbons were evaluated.
Overall, RONO2 increased significantly in the past 15 years in Hong Kong, due mainly to the
increased abundances of parent hydrocarbons. The concentrations of RONO2 were generally
higher in northwestern Hong Kong and lower in eastern Hong Kong, which were likely related
to regional transport and lower concentrations of parent hydrocarbons, respectively.
Furthermore, the parent hydrocarbons emitted from LPG-fueled vehicles decreased due to the
program, while RONO2 increased. An important factor leading to the increases of RONO2 was
the formation of RONO2 from emissions of parent hydrocarbons from sources other than LPG
fueled vehicles. Also, factors that might also accelerate the formation of RONO2 were the
meteorological conditions at the site and the atmospheric oxidative capacity, which was
enhanced in the roadside environments due to the LPG program. However, it is noteworthy
that great uncertainties were inherent to the study on the impacts of LPG program. On one hand,
samples were only collected for one day before the program and one day after the program. On
the other hand, although 24 out of the 30 sites were at the roadsides, the vehicle densities were
not very high at many of these sites, particularly for the LPG-fueled vehicles (taxis and public
light buses). To more comprehensively understand the impacts of LPG program on RONO2
and their parent hydrocarbons, further analyses will be given in Chapter 5 based on continuous
measurements of RONO2 and their parent hydrocarbons at a crowded roadside site through the
entire duration of the program.
53
4.1 Long term variations of RONO2 in Hong Kong Figure 4.1 shows the average mixing ratios of RONO2 and their parent hydrocarbons in Hong
Kong during different sampling campaigns. Table 4.1 summarizes the details of the sampling
campaigns including the sites and sampling durations. It was found that both RONO2 and the
parent hydrocarbons varied largely among the sampling campaigns. To understand the
historical variations of RONO2 in Hong Kong, the sampling campaigns have been grouped
according to the time. For example, RONO2 observed at HT and Tai O between 2001 and 2002
are averaged to represent the atmospheric abundances of RONO2 in that time (2001-2002).
Figure 4.2 plots the time dependent variations of RONO2 and the parent hydrocarbons in Hong
Kong. 2-/3-C5H11ONO2 are not presented, because they were only analyzed in the samples
collected during the periods of 2001-2002 and 2013-2014. In fact, 2-C5H11ONO2 and 3-
C5H11ONO2 experienced significant (p<0.05) increases from 6.9 ± 0.5 and 6.2 ± 0.5 pptv in
2001-2002 to 12.1 ± 0.6 and 8.5 ± 0.4 pptv in 2013-2014. The mixing ratios of C1-C4 RONO2
in the atmosphere of Hong Kong all increased remarkably (p<0.05) in the past 15 years.
Furthermore, the increase rate of 2-C4H9NO3 was the highest (1.53 pptv/year), followed by 2-
C3H7NO3, CH3NO3, C2H5NO3 and 1-C3H7ONO2. This coincided with the fact that 2-C4H9NO3
was generally the highest in the atmosphere, due to the combined influence of increasing
branching ratio and decreasing abundance of parent hydrocarbon with the increase of carbon
number in RO2 radicals. In addition, the observed mixing ratios of the parent hydrocarbons
also increased substantially from 2001 to 2014 in Hong Kong (see Figure 4.2). As such, we
inferred that the increase of RONO2 (at least 2-C4H9NO3) was closely associated with
secondary formation of RONO2. Certainly, it should be noted that the sites involved in the
sampling campaigns varied from the background site (HT) to the roadside sites (e.g., MK),
which might reduce the interpretability of the long-term trends due to the differences in sources
of RONO2, emission strength of RONO2 and their parent hydrocarbons and the age of air
54
masses. However, although the concentrations of RONO2 had inter-sites differences, the
sampling campaigns almost covered all territories with different land uses in Hong Kong. In
addition, according to the spatial distributions of RONO2 (see section 4.2), samples collected
at non-roadside sites in earlier periods (e.g., samples collected at Tai O and TC between 2001
and 2003) were more likely laden with comparable or even higher concentrations of RONO2
than those collected at the roadside sites in later periods, as vehicle emission is not a direct
source of RONO2 and photochemical formation of RONO2 can be inhibited in roadside
environment where the atmospheric oxidative capacity is weakened by high NOx. Namely, the
site differences were not the predominant factor leading to the increases of RONO2.
Furthermore, insight was given to RONO2 measured at TC, CW, YL and TM during two
comparative periods, i.e., 2002-2003 and 2006-2007. It was found that all RONO2 increased
between these two sampling periods at all sites. Particularly, the increases of 2-C3H7NO3, 1-
C3H7NO3 and 2-C4H9NO3 at CW, CH3NO3 and 1-C3H7NO3 at TM were significant (p<0.05).
This clearly confirmed the increasing trends of RONO2 in Hong Kong.
55
Figure 4.1 Average mixing ratios of C1-C5 RONO2 and their parent hydrocarbons in Hong
Kong in different sampling campaigns. The orange bars and red lines stand for mixing ratios
of RONO2 and their parent hydrocarbons, respectively. The sampling campaigns are displayed
in chronological order on x axis (see Table 4.1).
Table 4.1 Sites and dates of the RONO2 sampling campaigns in Hong Kong from 3/2001 to
9/2014.
56
No. Site Sampling period 0 HT March 2001-April 2002 1 Tai O August 2001-December 2002 2 TC August 2002-August 2003 3 CW August 2002-August 2003 4 YL August 2002-August 2003 5 TM August 2002-August 2003 6 TC September 2006-July 2007 7 CW September 2006-October 2007 8 TM September 2006-July 2007 9 YL September 2006-July 2007 10 TMS September-November 2010 11 TW September-November 2010 12 MK September 2013-June 2014 13 CWB September 2013-June 2014 14 Central AQMS September 2013-June 2014 15 Cotton Tree Drive September 2013-June 2014 16 Choi Yuen Road September 2013-June 2014 17 30 Sites over the whole territory
of Hong Kong * 27 September 2013, 18 February 2014 and 24 September 2014
* The 30 sites include 24 roadside sites and 6 general sites (see Figure 4.3).
57
Figure 4.2 Time dependent variations of C1-C4 RONO2 (green) and the parent hydrocarbons
(red) in Hong Kong from 2001 to 2014.
It is well documented that oceanic emission, biomass burning and photochemical formation are
the main sources of RONO2. To explore the reasons for the increases of RONO2, the
photochemical formation was firstly examined.
Previous studies [Ou et al., 2015; Wang et al., 2017] in our group based on the long term
observations of VOCs at TC and MK indicated that the light alkanes in the atmosphere of Hong
Kong increased in the past decade, due to the growth of vehicle population and the use of LPG
as the fuels of taxis and public light buses. In this study, we further identified that propane and
n-butane, the parent hydrocarbons of 2-/1-C3H7ONO2 and 2-C4H9ONO2, increased
significantly (p<0.05) at a suburban site (TC) in Hong Kong over the long period between 2005
and 2014, with the increasing rate of 77.3±9.9 pptv/year and 58.4±6.7 pptv/year, respectively.
Further, the levels of parent hydrocarbons at TC, CW, YL and TM were also compared between
two sampling campaigns (i.e., 2002-2003 and 2006-2007). In consistent with the increases of
the parent hydrocarbons, RONO2 observed in 2006-2007 were generally higher than those in
2002-2003, as shown in Table 4.2. For example, propane increased by 52.0±50.3% at TC from
2002-2003 to 2006-2007. Correspondingly, the mixing ratios of 2-C3H7ONO2 and 1-
C3H7ONO2 in 2006-2007 were 24.2±26.0% and 56.5±34.5% higher (p<0.05) than those in
58
2002-2003. Therefore, it was concluded that the increases of RONO2 were partially attributable
to the elevated concentrations of their parent hydrocarbons. We also noticed that the increasing
percentages of C1 and C2 RONO2 were much higher than those of their parent hydrocarbons
(i.e. methane and ethane), which even decreased from 2002-2003 to 2006-2007. On one hand,
the variations of primary emissions of RONO2 (e.g. biomass burning and oceanic emissions)
might be a reason of the discrepancy. However, this cannot be confirmed due to the lack of
tracers indicating the emission strengths of RONO2 from biomass burning and oceanic
emission in the two sampling campaigns. On the other hand, the elevated C1-C2 RONO2 in
2006-2007 were also likely to be caused by the increases of other VOCs like butanes and
carbonyls, as a large fraction of C1-C2 RO2 radicals are produced by these VOCs through
decomposition and isomerization of the larger radicals [Sommariva et al., 2008; Zeng et al.,
2018]. This was confirmed by the modeling of CH3ONO2 at a coastal site (Tai O) in Hong
Kong, which indicated that methane only accounted for 24.2 ± 3.6% of CH3O2 (Chapter 6).
Table 4.2 Increasing percentages of RONO2 and the parent hydrocarbons at TC, CW, YL and
TM during 2006-2007, with 2002-2003 as the base years. Negative values indicate decreases
rather than increases of the species (Unit: %).
TC CW YL TM
CH3ONO2 41.2±15.5 58.3±23.3 25.6±13.6 37.6±11.6
C2H5ONO2 26.3±22.9 28.1±19.8 15.3±21.0 23.6±19.1
2-C3H7ONO2 24.2±26.0 27.6±22.2 17.5±24.1 17.3±22.1
1-C3H7ONO2 56.5±34.5 74.5±30.9 53.0±32.4 55.7±28.5
2-C4H9ONO2 28.0±32.4 34.5±27.4 23.4±30.5 20.8±27.5
Methane 1.0±2.5 4.1±3.4 -2.1±3.1 -0.6±1.5
Ethane 5.7±27.1 -6.1±18.9 0.8±22.9 -2.4±25.4
59
Propane 52.0±50.3 44.9±25.8 34.6±29.4 11.6±30.0
n-Butane 50.0±48.5 66.0±26.2 29.5±23.1 -5.0±23.5
4.2 Spatial distributions of RONO2 in Hong Kong A RONO2 grid study was carried out on September 27, 2013, February 18 and September 24,
2014 to study the spatial patterns of RONO2 in Hong Kong. The canister samples were
collected twice daily at 30 sites covering the whole territory, including 24 roadside sites, 4
general sites and 2 background sites (Table 4.3).
Table 4.3 Information about the 30 sampling sites deployed in RONO2 grid study in Hong
Kong.
Site No. Site Latitude(o) Longitude(o)
1 Mong Kok Air Quality Monitoring Station 22.3225 114.1684
2 Causeway Bay Air Quality Monitoring Station 22.2801 114.1851
3 Central Air Quality Monitoring Station 22.2817 114.1586
4 RS Cotton Tree Drive 22.2790 114.1622
5 RS Choi Yuen Road 22.5010 114.1275
6 * Yuen Long Air Quality Monitoring Station 22.4467 114.0203
7 * Tsuen Wan Air Quality Monitoring Station 22.3733 114.1121
8 # Tap Mun 22.4728 114.3583
9 * Tung Chung Air Quality Monitoring Station 22.2890 113.9440
10 * Central Western Sai Ying Pun 22.2849 114.1445
11 # HKUST 22.3379 114.2675
12 Islands (Tat Tung Road) 22.2900 113.9393
13 Kwai Tsing 22.3576 114.1278
14 Tsuen Wan 22.3690 114.1163
15 Yuen Long 22.4447 114.0296
16 North/Lung Sum Avenue and Lung Wan Street 22.5026 114.1293
17 Tuen Mun 22.3926 113.9751
60
18 Shatin 22.3739 114.1785
19 Tai Po Kwong Fuk Road 22.4486 114.1672
20 Sai Kung 22.3203 114.2585
21 Kwun Tong 22.3137 114.2249
22 Sham Shui Po 22.3319 114.1614
23 Kowloon City 22.3223 114.1876
24 Wong Tai Sin 22.3382 114.1875
25 Eastern King’s Road 22.2908 114.1988
26 Southern Wong Chuk Hang 22.2494 114.1673
27 RS Canal Road Flyover 22.2776 114.1816
28 Lok Ma Chau Road 22.5057 114.0802
29 RS Nam Wan Road 22.4441 114.1676
30 RS Lin Tak Road 22.3148 114.2404 * General sites, # background sites, and the rest are roadside sites.
Figure 4.3 shows the spatial distributions of 7 C1-C5 RONO2 in Hong Kong, which were
acquired by averaging the observed RONO2 on September 27, 2013, February 18 and
September 24, 2014. Generally, higher concentrations of RONO2 were observed in the
northwest part of Hong Kong (e.g., sites 9, 12, 15, 16, 17 and 28), in contrast to the much lower
levels in eastern Hong Kong (e.g., sites 8, 11, 20 and 30). This pattern was consistent for C1 to
C5 RONO2, indicating that primary emission could not be the key factor dominating the spatial
distribution, because the C4 and C5 RONO2 are well known to be mainly derived from
photochemical formation. Therefore, we further looked into the spatial distributions of their
parent hydrocarbons, and found that the mixing ratios of C1-C5 n-alkanes were all lower at the
sites with lower RONO2. For examples, C1-C5 n-alkanes at the background sites 8 (TM) and 11
(HKUST) were always the lowest among all the sites. In contrast, higher concentrations of
parent hydrocarbons were observed at the sites 13, 15, 17, 21, 23 and 27, where the levels of
RONO2 were moderate to high. This implied that the spatial distribution of RONO2 was
partially attributable to the concentrations of parent hydrocarbons and local photochemical
61
formation. However, at the sites 9, 12, 16 and 28, the parent hydrocarbons were relatively low,
inconsistent with the high levels of RONO2. Given that these sites were all adjacent to the
inland PRD region, particularly for sites 9 and 12 which were the downwind sites of the heavily
polluted PRE, the high RONO2 values at these sites were likely caused by regional transport.
Indeed, previous studies confirmed that regional transport was an important contributor to O3
in Hong Kong, and a good correlation was often found between RONO2 and O3 [Rosen et al.,
2004; Perring et al., 2010]. Therefore, it was reasonable that some fractions of RONO2 in
northwest Hong Kong were built up by regional transport. In view of the increased O3 burden
from inland PRD in Hong Kong [Xue et al., 2014; Wang et al., 2017], we infer that regional
transport might also make increased contributions to RONO2 in Hong Kong.
In summary, photochemical formation/regional transport and low concentrations of parent
hydrocarbons were responsible for the relatively high and low RONO2 in northwestern and
eastern Hong Kong, respectively.
62
Figure 4.3 Spatial distributions of (a) CH3NO3, (b) C2H5NO3, (c) 2-C3H7NO3, (d) 1-C3H7NO3,
(e) 2-C4H9NO3, (f) 3-C5H11NO3 and (g) 2-C5H11NO3 in Hong Kong. Numbers on the figures
represent the sampling sites, among which 6, 7, 9 and 10 presented in triangles are general sites,
8 and 11 presented in squares are background sites and the rest presented in circles are roadside
sites.
63
4.3 Impact of LPG program on parent hydrocarbons Starting in 1997, a stepwise program promoting liquefied petroleum gas (LPG) as vehicle fuel
was initiated by the Hong Kong Environmental Protection Department (HKEPD) to improve
the air quality in Hong Kong. By the end of 2010, >99% of the registered taxis (i.e., 17,950
vehicles) and 51% of the registered public and private light buses (i.e., 3,280 vehicles) were
powered by LPG [HKCSD, 2010]. Due to the usage of LPG as vehicle fuel, the ambient VOC
profile in Hong Kong changed substantially with the most abundant species switching from
toluene to propane and n/i-butanes [Ou et al., 2015; Guo et al., 2007; Ho et al., 2004]. In
addition, studies consistently confirmed an increased contribution of LPG-fueled vehicle
emissions to ambient VOC levels, which was 11% at an urban site in 2001 [Guo et al., 2004],
15% at a suburban site during 2002-2003 [Guo et al., 2007], 26.9 ± 0.6% and 40.8 ± 0.8% at
an urban site during 2002-2003 and 2006-2007, respectively [Lau et al., 2010]. Moreover, O3
has been experiencing an increasing trend in Hong Kong over last two decades [Wang et al.,
2009; Guo et al., 2009]. As O3 formation was generally VOC-limited in Hong Kong [Cheng et
al., 2010b; Zhang et al., 2007], the contribution of LPG-fueled vehicle emissions to O3 was not
negligible, in view of the dominance of propane and n/i-butanes in ambient air and the high
contribution of LPG-fueled vehicle emissions to VOCs. In fact, the emissions of LPG fueled
vehicles are very likely an important cause of elevated RONO2 in Hong Kong (see section 4.1),
as propane and n/i-butane are the main components of LPG.
As such, an intervention program aimed to reduce VOCs and NOx emitted by LPG-fueled
vehicles was initiated in September 2013. This subsidy program promoted the replacement of
catalytic converters in LPG-fueled vehicles, and the catalytic converters in ~76% of LPG-
fueled vehicles were renewed by the end of May 2014. The next section (4.3.1) discusses the
64
impact of the LPG program on the exhausts of LPG-fueled vehicles, which accounted for high
fractions of parent hydrocarbons of RONO2, such as propane and n/i-butanes.
As introduced in section 4.2, whole air samples were collected at 30 sites in the RONO2 grid
study. As the LPG program was initiated in September 2013, and only 0.17% of converters
were replaced in that month, samples collected on September 27, 2013 were expected to reflect
the atmospheric VOC composition prior to the program. In contrast, samples collected on
September 24, 2014 were representative of the air profiles after the program, because the
replacement of converters ended in May 2014.
4.3.1 Variations of the observed parent hydrocarbons
Figure 4.4 shows the average geopotential height (HGT) and wind field on 1000 hPa for East
Asia during the two sampling campaigns. The pressure (represented with HGT) over Hong
Kong was comparable between the two campaigns. The wind was northeasterly on 27
September 2013, while it was calm on 24 September 2014 in Hong Kong. The lower wind
speed in the latter campaign was expected to elevate the VOC concentrations. However, the
levels of most VOCs remained similar, whereas those emitted from LPG source decreased
between the two campaigns (see the following sections), indicating that meteorological
parameters did not have substantial influence on the VOC levels of the two campaigns.
65
Figure 4.4 Average geopotential height and wind field on (a) 27 September 2013 and (b) 24
September 2014. The figures are made using NCEP FNL (final) data with a horizontal
resolution of 1°×1°.
Table 4.3 presents the average mixing ratios of VOCs at the 24 roadside sites, 4 general sites
and 2 background sites during the two campaigns. It is noteworthy that the average VOC values
for the sites should reflect the real situation though uncertainties could exist for the samples at
individual sites. It was found that the alkanes dominated the total VOC composition, followed
by aromatics and alkenes, and the mixing ratios at roadside sites were much higher (p<0.05)
than those at general and background sites due to their proximity to the emission sources. From
September 2013 to September 2014, levels of most species remained unchanged except for n/i-
pentanes, which increased at the general sites (p<0.05), indicating possibly increased emission
of gasoline-fueled vehicles. Furthermore, aromatics such as xylenes and propylbenzenes
increased significantly (p<0.05), perhaps due to the increase of solvent usage and/or vehicular
emissions. In contrast, LPG related VOCs (propane and n/i-butanes) remained unchanged,
while propene, the tracer of LPG combustion, even decreased at the roadside sites (p<0.05). In
view of the above fact, to examine the real impact of the LPG program on the parent
hydrocarbons of RONO2, it is necessary to conduct source apportionments to obtain the
emission variations of LPG-fueled vehicles before and after the replacement program.
Table 4.3 Mixing ratio of VOCs collected at the 30 sampling sites during the 2 sampling
campaigns (average±95% confidence interval, pptv)
Species Roadside sites (n=24) a General sites (n=6) a
Sept. 2013 Sept. 2014 Sept. 2013 Sept. 2014
Alkanes Ethane 2518±209 2704±206 1833±179 1884±280
Propane 7723±1872 6996±1039 3631±2478 2849±1076
66
n-Butane 11166±3104 9003±1645 2828±1876 2694±1247
i-Butane 6413±1726 5455±931 1866±1227 1762±799
n-Pentane 773±197 1209±457 331±74 866±321 *
i-Pentane 1331±324 2097±1127 608±112 1372±499 *
n-Hexane 323±92 529±86 * 148±42 415±177 *
2,3-
Dimethylbutane
118±46 186±51 48±22 137±58 *
2-Methylpentane 593±193 847±223 301±137 754±408
3-Methylpentane 315±95 588±153 * 174±85 552±285 *
n-Heptane 437±242 480±122 122±35 226±93 *
2-Methylhexane 444±235 416±82 151±51 255±123
3-Methylhexane 449±243 510±90 145±51 353±156 *
n-Octane 106±37 137±46 47±10 74±24
2,2,4-
Trimethylpentane
255±132 316±81 37±8 109±66 *
Alkenes Ethene 4631±917 3748±669 1097±327 1082±398
Propene 1798±417 * 1084±230 233±70 201±93
1-Butene 197±40 161±32 53±21 48±25
i-Butene 566±145 * 371±87 150±58 94±33
trans-2-Butene 132±36 99±26 20±8 21±8
cis-2-Butene 82±25 56±15 14±4 13±5
1-Pentene 55±11 93±49 19±6 28±11
1,3-Butadiene 149±37 119±24 29±5 17±10
Isoprene 531±90 627±115 500±230 777±285
Alkyne Ethyne 3916±671 3375±391 1978±426 2037±404
67
Aromatics Benzene 886±138 * 662±64 556±67 518±105
Toluene 3270±1751 3371±527 1634±610 2745±1136
Ethylbenzene 643±127 674±87 438±150 703±230
p-Xylene 291±85 420±69 * 130±39 378±163 *
m-Xylene 483±178 703±153 161±50 602±296 *
o-Xylene 337±103 471±79 142±44 428±180 *
i-Propylbenzene 39±13 42±6 17±4 34±10 *
n-Propylbenzene 92±39 79±14 30±9 53±16 *
3-Ethyltoluene 380±210 212±59 66±23 113±47
4-Ethyltoluene 187±112 119±28 37±10 75±30 *
2-Ethyltoluene 164±76 87±22 41±12 53±22
1,3,5-TMB 234±127 * 78±24 40±14 37±15
1,2,4-TMB 821±486 * 288±85 128±54 145±67
1,2,3-TMB 228±102 84±22 56±24 45±19
a Number of sites; * higher mixing ratios compared to those in another sampling campaign at
the confidence level of 95% (p<0.05). TMB refers to the trimethylbenzene isomers hereafter.
4.3.2 Source apportionments of VOCs
Twenty main anthropogenic VOC species quantified in the 64 samples were applied to PMF
for source apportionment for the two campaigns, respectively. The source profiles before and
after the intervention program are similar (Figure 4.5). Three factors were extracted from PMF.
It is noteworthy that most of the samples were collected at the roadside sites, where the sources
of VOCs were relatively uncomplicated, with the dominance of vehicle emissions. In addition,
the exhausts of gasoline and diesel vehicles were allocated into one factor. Increasing factor
numbers led to higher Q value (a function to evaluate the apportionment results, and generally
the resolution with lowest Q value is recommended) and difficulty in explanation of the factors.
68
As such, we accepted the apportionment results of three factors. The first factor was
distinguished by the dominance of propane, n/i-butanes, ethene and propene, representing
LPG-fueled vehicle exhaust. Factor 2 had high percentages of all VOCs and trace gases except
LPG related component and the trimethylbenzene isomers. It was assigned as gasoline and
diesel vehicle exhaust. The third factor was closely associated with solvent usage because of
high loadings of xylenes and trimethylbenzenes. The profiles of factors identified were based
on the results of previous source apportionment studies [Guo et al., 2007, 2011a; Lau et al.,
2010; Ling et al., 2011] and VOCs source emission studies [Borbon et al., 2002; Guo et al.,
2006, 2011b; Ho et al., 2009]. Many different starting seeds were tested and no multiple
solutions were found. In addition, good correlations were found between the observed and
predicted VOC concentrations for the whole dataset (R2 = 0.95 and 0.96, respectively) before
and after the replacement program. Moreover, all of the selected species had scale residuals
normally distributed between -3 and 3, confirming that the measured data were well reproduced
[USEPA, 2008].
69
Figure 4.5 Source profiles of the three sources extracted from PMF in September 2013 (before
the program) and September 2014 (after the program). The standard errors are estimated with
the bootstrap in the PMF model.
To sum up the VOC concentrations in each source, the mass and percentage contributions of
the sources to VOCs are summarized in Table 4.4. Noticeably, the vehicle emissions were the
dominant source of VOCs, with the contribution of 71.1±1.8 μg/m3 (85.5±2.1%) and 77.7±1.3
μg/m3 (92.0±1.6%) before and after the program, respectively. From 2013 to 2014, the VOCs
emitted from gasoline and diesel vehicles increased remarkably (p<0.05), whereas those
originated from LPG vehicle exhaust decreased significantly (p<0.05) from 41.3±1.2 μg/m3
(49.7±1.5%) to 32.8±1.4 μg/m3 (38.8±1.7%).
Table 4.4 Mass concentrations of VOC sources and their percentage contributions to VOCs
before and after the program.
Source Mass concentration
(μg/m3)
Percentage contribution (%)
Before LPG vehicular
exhaust
41.3 ± 1.2 49.7 ± 1.5
Gasoline and diesel
vehicular exhaust
29.8 ± 2.2 35.8 ± 2.6
Solvent usage 12.0 ± 3.0 14.5 ± 3.6
After LPG vehicular
exhaust
32.8 ± 1.4 38.8 ± 1.7
Gasoline and diesel
vehicular exhaust
44.9 ± 1.2 53.2 ± 1.4
Solvent usage 6.8 ± 1.4 8.1 ± 1.7
70
Table 4.5 shows the average concentrations of VOCs and trace gases in LPG vehicle exhaust.
Clearly, CO, ethene, propane, propene, n/i-butanes and trimethylbenzene isomers all reduced
significantly from before to after the replacement program (p<0.05). The emissions of NO and
NO2 from LPG-fueled vehicles were minor, and the decrease of NO was insignificant (p>0.05).
Table 4.6 presents the reductions of VOCs and NO at different sites. The mass and percentage
contribution of LPG vehicle exhaust to VOCs experienced the greatest decrease at the roadside
sites (p<0.05), with the contributions of 54.7±23.2 μg/m3 (54.6±10.0%) before and 25.0±11.2
μg/m3 (30.8±9.9%) after the program, respectively. The effects were much weaker at the urban
and new town sites, where the mass and percentage contribution to VOCs decreased slightly
or even increased (p>0.05). Similarly, NO decreased noticeably (p<0.05) (before: 0.66±0.28
μg/m3; after: 0.04±0.02μg/m3) at roadside sites, while the reductions were not significant at the
urban and new town sites (p>0.05). This inter-site difference was possibly caused by higher
traffic flow and more dense LPG-fueled vehicles (particularly taxis) in the vehicle fleet at
roadside sites.
In conclusion, the parent hydrocarbons of RONO2 (mainly propane and n-butane) decreased
considerably due to the implementation of the LPG program. However, whether RONO2
decreased or not needed to be further identified, because the productions of RONO2 depended
upon not only the abundances of precursors but also the oxidative capacity of the atmosphere.
Table 4.5 Concentrations of VOCs and trace gases emitted from LPG-fueled vehicles before
and after the LPG program.
Species LPG vehicle exhaust
before (μg/m3) after (μg/m3)
71
NO 0.49 ± 0.44 0.03 ± 0.44
NO2 0.00 ± 0.94 0.00 ± 1.05
CO 336.24 ± 12.71 149.5 ± 11.1
Ethane 0.71 ± 0.06 1.03 ± 0.05
Ethene 2.83 ± 0.05 2.00 ± 0.05
Ethyne 0.89 ± 0.06 0.87 ± 0.07
Propane 7.63 ± 0.16 6.07 ± 0.14
Propene 1.76 ± 0.02 0.91 ± 0.02
i-Butane 8.93 ± 0.16 6.98 ± 0.13
n-Butane 16.00 ± 0.26 11.52 ± 0.20
i-Pentane 0.30 ± 0.05 0.75 ± 0.09
n-Pentane 0.14 ± 0.03 0.38 ± 0.06
n-Hexane 0.05 ± 0.02 0.17 ± 0.03
n-Heptane 0.14 ± 0.01 0.20 ± 0.03
Benzene 0.37 ± 0.05 0.44 ± 0.04
Toluene 0.56 ± 0.16 0.94 ± 0.25
Ethylbenzene 0.13 ± 0.05 0.17 ± 0.06
p-Xylene 0.08 ± 0.02 0.10 ± 0.04
m-Xylene 0.14 ± 0.02 0.11 ± 0.05
o-Xylene 0.08 ± 0.02 0.08 ± 0.04
1,3,5-TMB 0.12 ± 0.01 0.00 ± 0.01
1,2,4-TMB 0.32 ± 0.03 0.00 ± 0.03
1,2,3-TMB 0.11 ± 0.01 0.01 ± 0.01
Bolded are the species with significant reduction in LPG vehicle exhaust (p<0.05).
72
Table 4.6 Mass and percentage contribution of LPG vehicle exhaust to VOCs and NO at
different sites before and after the program.
Species Site Mass concentration (μg/m3) Percentage contribution (%)
before after before after
VOCs Urban roadside 54.7 ± 23.2 25.0 ± 11.2 54.6 ± 10.0 30.8 ± 9.9
Urban 28.7 ± 35.8 23.7 ± 11.8 29.3 ± 30.3 31.1 ± 10.5
New town 11.5 ± 9.8 16.5 ± 17.1 27.9 ± 23.6 20.8 ± 20.3
NO Urban roadside 0.66 ± 0.28 0.02 ± 0.01 0.04 ± 0.02 0.005±0.002
Urban 0.34 ± 0.42 0.02 ± 0.02 0.03 ± 0.03 0.005±0.004
New town 0.14 ± 0.12 0.02 ± 0.02 0.01 ± 0.01 0.003±0.004
4.4 Impact of LPG program on oxidative capacity This section evaluates the impact of the LPG program on the atmospheric oxidative capacity,
which determines the oxidation efficiency of hydrocarbons, and subsequently influences the
production of RONO2. It is noteworthy that O3 production was simulated using the PBM-MCM
model, as a measure of the oxidative capacity.
4.4.1 Model validation and O3 simulation
Using the method described in section 3.5.1, the diurnal profiles of individual VOCs were
estimated based on the measured VOCs at 10:00 and 15:00 on each sampling day. Since the
estimated diurnal profiles of VOCs were used to simulate O3, it is necessary to validate the
results with the online measured VOCs. Figure 4.6 shows the estimated and online measured
diurnal patterns of total VOCs at MK, YL and TC, where the real-time VOCs data were available.
26 VOC species were included in the total VOCs for calculation. The diurnal patterns of total
VOCs estimated from the two canister sample data agreed well with the real-time
measurements. Table 4.7 lists the Index of Agreement (IOA) values between the calculated and
73
measured data of the 26 VOC species [Jiang et al., 2010; Wang et al., 2015]. Fair to good
agreement between the calculated and measured profiles of individual VOCs at these three
different types of sites suggested that the proposed method provided a reasonable estimate of
VOC profiles based on the two canister samples. It is noteworthy that the estimated pattern of
VOCs at MK was much smoother than the observed one during 01:00~09:00, perhaps due to
the fact that air pollutants at the roadside site (MK) was significantly influenced by the in-situ
traffic emissions. However, the method to derive the estimation of VOCs diurnal profiles was
based on the emission inventory which documented the average profile of air pollutants over
the entire area of Hong Kong. Despite this discrepancy, it was assumed not to substantially
influence the O3 simulations, because O3 was simulated in daytime hours (07:00~19:00), and
photochemical reactions were often weak in the first three hours (07:00-09:00) when the
variations of total VOCs were not well reproduced by the model at MK.
Figure 4.6 Estimated and real-time measured diurnal profiles of total VOCs at MK, YL and TC.
Table 4.7 IOA values between measured and simulated diurnal profiles of individual VOCs.
Species MK YL TC
Ethane 0.53 0.43 0.41
Propane 0.51 0.48 0.49
74
n-Butane 0.52 0.40 0.67
i-Butane 0.60 0.40 0.67
n-Pentane 0.39 0.30 0.40
i-Pentane 0.42 0.49 0.49
n-Hexane 0.75 0.47 0.46
n-Heptane 0.37 N.D. N.D.
n-Octane N.D. N.D. N.D.
Ethene 0.33 0.55 0.59
Propene 0.76 0.40 0.56
1-Butene 0.47 0.69 N.D.
trans-2-Butene N.D. 0.49 0.59
cis-2-Butene 0.46 N.D. N.D.
1-Pentene N.D. N.D. 0.57
1,3-Butadiene N.D. N.D. 0.64
Isoprene 0.45 0.94 0.55
Ethyne 0.73 0.40 0.45
Benzene 0.30 0.39 0.35
Toluene 0.43 0.57 0.47
Ethylbenzene 0.68 0.25 0.44
m,p-Xylene 0.76 0.20 0.64
o-Xylene 0.76 N.D. 0.56
1,3,5-Trimethylbenzene N.D. N.D. N.D.
1,3,5-Trimethylbenzene 0.57 N.D. N.D.
1,3,5-Trimethylbenzene N.D. N.D. N.D.
N.D.: data are not available since the mixing ratios are below the detection limits.
75
The calculated VOC diurnal profiles were then input into the PBM-MCM model for O3
simulation. Figure 4.7 shows the daytime (07:00~19:00) simulated and observed O3 in 2013
and 2014 at 13 sites where the online data of trace gases were available from the air quality
monitoring network of HKEPD. In general, the simulated O3 agreed well with the observations,
with the consistency of the peaks and troughs. The IOA between the simulated and observed
O3 was 0.7, indicating fairly acceptable performance of the model. In other words, in-situ O3
formation dominated its ambient level at most sites. The difference between model simulation
and observation at some other sites was likely due to the fact that the PBM-MCM model only
considers O3 produced from photochemical reactions while the observed O3 is also influenced
by the downward transport of stratospheric O3, dry deposition and horizontal transport from
other regions/locations [Cheng et al., 2010b; Creilson et al., 2003; Lam et al., 2013; Xue et al.,
2011].
Figure 4.7 Comparison between simulated and observed O3 in (a) September 2013 and (b)
September 2014.
76
4.4.2 Impact of the program on O3 formation
Given the reduction of VOCs and NO emitted from LPG-fueled vehicles, it is interesting to
explore the impact of these changes on O3 formation at different sites.
Sensitivity experiments give the differences in O3 production between the scenarios with and
without the LPG source as input. Through this approach, the O3 produced by LPG source before
and after the program were obtained (Figure 4.8). Since O3 formation was usually limited by
VOCs and suppressed by NO titration, the VOCs and NO in LPG made positive and negative
contributions to O3 production, respectively. Considering the combined effect of VOCs and
NO on O3 formation, LPG generally made a net positive contribution to O3. However, the
contribution of LPG vehicle to O3 formation at roadside site was negative before the program,
mainly due to higher levels of NO emitted from LPG-fueled vehicles (0.66±0.28 μg/m3) than
those at urban (0.34±0.42 μg/m3) and new town sites (0.14±0.12 μg/m3), resulting in higher
NO titration to O3.
Table 4.8 lists the average contributions of LPG vehicle exhaust to O3 at different types of sites
before and after the replacement program. At the roadside site, the contribution of LPG vehicle
turned from O3 destruction (-0.17±0.06 ppbv) before the program to O3 formation
(0.004±0.038 ppbv) after the program, indicating the enhancement of atmospheric oxidative
capacity in the roadside environments. However, this enhancement might not be exclusively
caused by the LPG program. For example, the temperature on September 24, 2014 was 4 °C
higher than that on September 27, 2013, according to the monitoring data at the Hong Kong
International Airport. The high temperature meant more intensive photochemical reactions,
thus higher oxidative capacity. Although the decrease of VOCs and NO was not significant at
the urban site (p>0.05), O3 produced by LPG source decreased significantly (p<0.05),
reflecting nonlinear relationship between O3 and its precursors, and also indicating the
77
effectiveness of the program on O3 production at the urban site. At the new town site, no
significant change in the contribution of LPG vehicle to O3 production was observed (p>0.05).
Figure 4.8 Contribution of LPG vehicle exhaust to O3 production before and after the program.
O3 production by LPG at the roadside site is enlarged in the insert panel.
Table 4.8 Site-dependent average contributions of LPG vehicle exhaust to O3 production (Unit:
ppbv).
Before After
Urban roadside -0.17 ± 0.06 0.004 ± 0.038
Urban 4.19 ± 1.92 0.95 ± 0.38
New town 3.37 ± 1.56 4.47 ± 1.89
4.5 Spatial characteristics of photochemical reactivity Figure 4.9 shows the relative incremental reactivity (RIR) of anthropogenic VOCs (AHC),
biogenic VOCs (BHC), CO and NOx, as a measure of the sensitivity of O3 formation to the
changes of the precursors [Cardelino and Chameides, 1995]. The VOC groups and CO had
positive RIR values, and the RIR values of VOC groups were higher than that of CO, indicating
78
that O3 production was VOC-limited. The RIR values of AHC were mostly the highest,
followed by BHC and CO. In contrast, the average RIR for NOx was negative, suggesting that
cutting NOx led to O3 increase. Different from other sites where O3 formation was limited by
AHC, BHC at the rural site TM was the most predominant reagent limiting O3 formation in
September 2013, whereas the RIR of NOx in September 2014 became positive, same as VOCs
and CO, indicating that O3 formation was limited by both VOCs and NOx. To understand the
dominant VOC groups/species responsible for O3 formation, Table 4.9 shows the average RIR
values of VOC groups/species at different types of sites. The alkenes (6.91) and aromatics (7.01)
had comparable RIR values and were the highest at the roadside sites, indicating that vehicular
emissions were the most important sources of O3 formation at roadside sites. On the other hand,
the aromatics at the urban and new town sites were the most predominant VOCs for O3
formation, with the RIR values of 20.48 and 24.15, respectively. Solvent usage and traffic
emissions were likely the main contributors at these two types of sites. In contrast, isoprene
was responsible for O3 formation at rural site with the RIR of 19.38.
79
Figure 4.9 Spatial characteristics of RIR values for VOC groups, CO and NOx on (a) 27 Sept.
2013 and (b) 24 Sept. 2014.
Table 4.9 Average RIR of VOC groups or species at different sites (Unit: %/%). Bolded fonts
denote the highest RIR at each type of site.
Urban roadside Urban New town Rural
Alkanes 4.22 8.72 9.67 7.98
Alkenes 6.91 10.24 11.35 8.35
Aromatics 7.01 20.48 24.15 15.19
Isoprene 1.31 4.18 8.56 19.38
Ethyne 0.02 0.19 0.31 0.81
4.6 Implication on RONO2 abundances Table 4.10 shows the variations of RONO2, dimethylsulfide (DMS) and methyl chloride
(CH3Cl) between the two sampling campaigns (i.e., one on September 27, 2013 and another on
September 24, 2014). The parent hydrocarbons remained statistically stable (p>0.05) except
for methane and n-pentane, which increased from 2069 ± 27 ppbv before the program to 2316
± 53 ppbv after the program, and from 684.5 ± 164.4 pptv before the program to 1006.6 ± 136.1
pptv after the program, respectively. However, all the RONO2 showed significant increases
from before to after the program (p<0.05). In contrast, DMS, the tracer of oceanic emissions,
decreased remarkably (43.3 ± 5.2 and 16.3 ± 2.6 pptv before and after the program,
respectively). The mixing ratio of CH3Cl before (1040 ± 27 ppbv) and after the program (1046
± 26 pptv) were comparable (p>0.05). This level of CH3Cl is nearly two times the background
level in mid-latitude of north hemisphere, i.e. ~500 pptv [Yokouchi et al., 2000]. However,
there was nearly no correlation between CH3Cl and the biogenic markers (e.g. isoprene and α-
80
/β-pinenes), excluding the possibility of tropical plants emissions as the dominant source of
CH3Cl in Hong Kong [Yokouchi et al., 2000; 2002]. Additionally, CH3Cl correlated poorly
with CHBr3 in Hong Kong, even at a coastal site (HT). Instead, moderate correlation (R within
the range of 0.49 – 0.82) between CH3Cl and CO was found at the non-roadside sites in the
sampling campaigns through 2001 to 2014 (Figure 4.10 (a)). Further, CH3Cl observed at a
mountainous site in Hong Kong (TMS) were also fairly well (R=0.73) correlated with
levoglucosan (a good tracer of biomass burning) in PM2.5 (Figure 4.10 (b)), indicating that
biomass burning might account for the high CH3Cl in Hong Kong. The comparable CH3Cl
between the two sampling campaigns implied that biomass burning was not the cause of the
enhancements of RONO2.
81
Figure 4.10 (a) Correlation between CH3Cl and CO at the non-roadside sites in the sampling
campaigns through 2001 to 2014. (b) Correlation between CH3Cl and levoglucosan in PM2.5 at
a mountainous site (TMS).
Since C5 RONO2 are generally thought to be produced by photochemical reactions, the
correlation analyses between C1-C4 RONO2 and C5 RONO2 (2-C5H11ONO2 + 3- C5H11ONO2)
were done to examine the dominant sources of RONO2. It was found that the correlation
coefficients (R2) of C1 vs C5, C2 vs C5, 2-C3 vs C5, 1-C3 vs C5 and 2-C4 vs C5 RONO2 were 0.34,
0.34, 0.57, 0.60 and 0.60 before the program, which increased to 0.71, 0.93, 0.94, 0.89 and 0.92
after the program, respectively. The better correlations between C1-C4 and C5 RONO2
suggested that primary emissions became less significant in the later sampling campaign. As
such, we inferred that the increases of RONO2 were mainly caused by the photochemical
processes. For C5 RONO2, its parent hydrocarbon (n-pentane) also increased (p<0.05) from
before to after the program at all general sites (see section 4.3.1), which might be partially
responsible for the elevated C5 RONO2. However, the parent hydrocarbons for C2-C4 RONO2
all remained stable between the two campaigns. As indicated in section 4.4.2, the roadside O3
82
production slightly increased induced by the LPG program. Additionally, the average
temperature in the later sampling campaign was ~6 °C higher than the earlier sampling
campaign, implying more intensive photochemical reactions in the later case. These findings
indicated that the enhanced oxidative capacity of the atmosphere, rather than the parent
hydrocarbons and primary emissions, might be the culprit of the increases of C2-C4 RONO2.
In fact, it was clearly identified that propane and n-butane emitted from LPG-fueled vehicles
decreased benefitted from the implementation of the LPG program, which might reduce the
productions of C3-C4 RONO2. However, this effect was most likely reversed by the higher
oxidative capacity of the atmosphere and increased emissions of propane and n-butane from
sources other than LPG-fueled vehicles, in view of the fact that neither propane or n-butane in
the ambient air changed significantly throughout the LPG program.
Table 4.10 Variations of C1-C5 RONO2, DMS and CH3Cl between the two sampling campaigns
(Unit: pptv).
Species September 27, 2013 September 24, 2014
CH3NO3 13.1 ± 1.6 16.8 ± 1.2
C2H5NO3 11.4 ± 0.8 16.2 ± 1.1
2-C3H7NO3 21.4 ± 1.2 36.8 ± 2.7
1-C3H7NO3 2.8 ± 0.3 9.5 ± 0.8
2-C4H9NO3 34.1 ± 2.7 64.0 ± 6.6
3-C5H11NO3 8.1 ± 0.6 11.3 ± 1.1
2-C5H11NO3 10.8 ± 0.8 15.8 ± 0.7
DMS 43.3 ± 5.2 16.3 ± 2.6
CH3Cl 1039.5 ± 27.3 1046.4 ± 25.7
83
4.7 Sub-conclusions This chapter studied the spatiotemporal patterns of RONO2 in Hong Kong, and examined the
impacts of a LPG program on the abundances of RONO2 and their parent hydrocarbons. From
2001 to 2014, RONO2 increased significantly in Hong Kong, which was mainly attributable to
the increases of the parent hydrocarbons and enhanced oxidative capacity of the atmosphere.
Spatially, RONO2 were generally more abundant in northwestern Hong Kong than those in
eastern Hong Kong. In addition to varied concentrations of parent hydrocarbons, regional
transport was an important factor causing this spatial distribution, which might make increased
contributions to RONO2 in Hong Kong. From before to after the LPG program, the parent
hydrocarbons of RONO2 remained stable in the whole air samples. However, source
apportionment revealed that the parent hydrocarbons emitted from LPG-fueled vehicles
significantly decreased at urban roadside site after the program, while they remained
unchanged at urban and new town sites. LPG vehicle exhaust was destructive to O3 formation
at the roadside site before the program, whereas it switched to positive contribution after the
program, suggesting that the program caused the increase of atmospheric oxidative capacity in
roadside environment. Under the combined effect of increasing oxidative capacity and
increasing emissions of parent hydrocarbons from sources other than LPG-fueled vehicles, the
observed RONO2 all increased from before to after the program.
84
Chapter 5 Impacts of replacing catalytic converters in LPG-fueled vehicles on the abundances of RONO2 and their parent hydrocarbons in Hong Kong
As introduced in Chapter 4, an intervention program aimed to reduce the emissions of VOCs
and nitrogen oxides (NOx) from LPG-fueled vehicles was implemented by the HKEPD
between September 2013 and May 2014, for the purpose of improving air quality in Hong
Kong. Chapter 4 also evaluated the impacts of the program on the abundances of RONO2 and
their parent hydrocarbons. However, the evaluation might be accompanied by great uncertainty
due to insufficient sample size.
In this chapter, the variations of RONO2 parent hydrocarbons (propane and n/i-butanes) from
before to during the intervention program were studied. Online measurements of VOCs and
collection of whole air samples were carried out at a roadside site (MK), due to its immediate
vicinity of the vehicle sources and high density of LPG-fueled vehicles, which was appropriate
to reflect the changes of VOC composition in the air caused by the program. Since propane and
n/i-butanes were co-emitted from vehicles with other air pollutants (such as NOx), which were
subject to co-treatment by the newly installed catalytic converters in vehicles, the variations of
NOx and CO were also studied. Long-term real-time measurements at MK indicated that the
program was remarkably effective in reducing propane and n/i-butanes, NOx and nitric oxide
(NO) in the atmosphere. Receptor modeling results further revealed that propane, propene, i-
butane, n-butane and NO in LPG-fueled vehicle exhaust emissions decreased by 40.8±0.1%,
45.7±0.2%, 35.7±0.1%, 47.8±0.05% and 88.6±0.7%, respectively, during the implementation
of the program. In contrast, despite the reduction of VOCs and NOx, O3 following the program
increased by 0.40 ± 0.03 ppbv (~5.6%). The LPG fueled vehicle exhaust was generally
destructive to OH and HO2. However, the destruction effect weakened for OH and it even
turned to positive contribution to HO2 during the program. These changes led to the increases
85
of OH, HO2 and HO2/OH ratio, which might explain the O3 increment. Analysis of O3-VOCs-
NOx sensitivity in ambient air indicated VOC-limited regimes in the O3 formation before and
during the program. Moreover, a maximum reduction percentage of NOx (i.e., 68.9%) and the
lowest reduction ratio of VOCs/NOx (i.e., 1.1) in LPG-fueled vehicle exhaust were determined
to give a zero O3 increment. The findings are of great help to future formulation and
implementation of control strategies on vehicle emissions in Hong Kong, and could be
extended to other regions in China and around the world.
Furthermore, the impact of the LPG program on RONO2 formation was studied. Although the
atmospheric oxidative capacity slightly increased because of the LPG program, RONO2 still
decreased via the implementation of the program, owing to significant reductions of their
parent hydrocarbons from LPG-fueled vehicle emissions. The decreases of RONO2 at MK from
before to during the LPG program, instead of the overall increases in Hong Kong as discussed
in Chapter 4, was likely due to the more proximity of MK to vehicle emissions of RONO2
precursors. In addition, the vehicle density at MK was very high with the average daily traffic
flow in 2014 of ~38,000 (HKTD, 2014). LPG fuelled vehicles constituted a considerable
fraction (30%) of the vehicles in fleet according to our observations during the sampling.
Therefore, the effectiveness of the LPG program on VOC reduction and further reduction of
RONO2 might be more obvious at MK. In addition, as stated in Chapter 4, the two days’
sampling at 30 sites in Hong Kong might have greater uncertainty in representing the average
levels of RONO2 before and after the LPG program.
86
5.1 Variations of LPG-related VOCs and NOx during the intervention
program
5.1.1 Concentrations of primary LPG-related VOCs and NOx
Table 5.1 shows the mixing ratios of the primarily emitted LPG-related VOCs and NOx before
(October 2012-September 2013) and during the program (September 2013-May 2014). The
LPG-related VOCs and NOx were reduced significantly from before to during the program
(p<0.05). To investigate the meteorological influences on these reductions, we looked into the
average geopotential height (HGT) and wind field on 1000 hPa for East Asia for the periods of
“matched” and “during the program”, as shown in Figure 5.1 (the “matched” period (October
2012-May 2013) was defined as the same time span as that “during the program”, but in
different years.) Noticeably, the meteorological conditions were fairly similar between the two
periods, i.e., the pressure decline (indicated by the decrease of HGT) from northern China to
SCS led to the prevailing northeasterly winds in Hong Kong, and the differences of HGT and
wind speed between the two periods for Hong Kong were only 0-2 gpm and less than 1 m/s,
respectively. Ground monitoring data also indicated insignificant differences of temperature
(“matched”: 21.9±0.5 °C; “during”: 21.2±0.7 °C) between the two periods (p>0.05). However,
the mixing ratios of LPG-related VOCs and NOx during the “matched” period were comparable
to those “before” the program, and absolutely higher than those “during” the program (p<0.05)
(see Table 5.1). Given the similar meteorological conditions between the “matched” period and
“during the program”, the significant decreases of LPG-related VOCs and NOx were caused by
the interventional program, rather than meteorological variations.
Table 5.1 Mixing ratios of LPG-related VOCs and NOx during the periods of “before”,
“matched” and “during the program” (Unit: ppbv)
Species “Before”
(Oct. 2012-Sept. 2013)
“During”
(Oct. 2013-May 2014)
“Matched”
(Oct. 2012-May 2013)
87
Propane 8.5±0.1 5.8±0.1 9.1±0.1
i-Butane 6.6±0.1 4.7±0.1 6.9±0.1
n-Butane 13.2±0.2 8.4±0.1 13.7±0.2
NO 201.7±2.1 172.0±2.7 201.9±2.7
NO2 25.9±0.3 23.8±0.4 26.8±0.4
NOx 229.4±2.3 197.7±3.0 231.2±2.9
88
Figure 5.1 Average geopotential height (HGT) and wind field during the periods of (a)
“matched” (1/10/2012-31/05/2013) and (b) “during” (1/10/2013-31/05/2014). (c) Shows the
differences in HGT and wind field between the periods of “matched” and “during”. The red
cycle represents the border line of Hong Kong. The figures were made using NCEP FNL (final)
data with a horizontal resolution of 1°×1°.
To further understand the effectiveness of the program, the monthly reductions of LPG-related
VOCs after the commencement of the converter replacement were calculated and compared
with those before the program (Figure 5.2). Since the program was initiated in October 2013,
the averages of VOC species in September in each year were taken as the baselines for the
calculation of monthly reduction. Briefly, the monthly reductions of LPG-related VOCs were
the differences between their averages in September and those in the other months. Hence,
positive and negative values indicate reductions and increments of the corresponding species,
respectively. It was found that the monthly averages of LPG-related VOCs consistently
decreased from September 2013 to May 2014 except for n-butane in October 2013 when the
program was just initiated. Compared to those before the program (i.e., September 2013), the
mixing ratios of propane, i-butane and n-butane decreased 3.2 ± 0.2, 2.8 ± 0.2 and 4.9 ± 0.2
ppbv by May 2014, respectively, when 99.2% of catalytic converters participating in the
program had been exchanged. Furthermore, the monthly reductions correlated well with the
cumulative converter replacements (R2 = 0.92, 0.93 and 0.89 for propane, i-butane and n-
butane, respectively). In contrast, no consistent reduction was observed from September 2012
to May 2013 for LPG-related VOCs suggesting the effectiveness of the program on the
reduction of LPG-related VOCs. On the other hand, although the average mixing ratios of NOx
decreased significantly during the program as shown in Table 5.1, no consistent reductions
were found for their monthly averages. This might be due to the fact that NOx emitted from
89
LPG-fueled vehicles was minor compared to gasoline- and diesel-fueled vehicles (~ 4.0% from
emission inventory, and 1.1-7.3% from source apportionment. See Section 5.2.2).
Figure 5.2 Monthly reductions of LPG-related VOCs at MK “before” (blue bar) and “during”
(red bar) the program. (The average value of VOC species in September is the baseline. The
values along the dot line were the cumulative numbers of converters replaced. The bars above
and below the X axis refer to the reduction and increase of VOC mixing ratio, respectively.)
5.1.2 Temporal variations of primary LPG-related VOCs and NOx
Figure 5.3 presents the temporal variations of propane and n/i-butanes at MK from June 2013
to May 2014 covering the periods of both before (i.e., June 2013-September 2013) and during
the intervention program (i.e., October 2013-May 2014). The abnormally low alkane levels in
late June 2013 might be caused by the anti-cyclone over southern China and a tropical storm
over the SCS, leading to extremely high temperature on June 18-20 (i.e., favorable for
photochemical reactions), and stronger winds (i.e., conducive to atmospheric dispersion) on
the following days, respectively. In contrast, Hong Kong was strongly influenced by an active
90
ocean flow and a low pressure trough over the SCS in late July 2013. The consequently heavy
rain and low temperature suppressed the photochemical reactions and caused the unusually
high levels of VOCs (see Figure 5.4). Generally, the LPG-related VOCs experienced a
significant (p<0.05) reduction throughout the year, and n-butane had the highest reduction rate
of ~19 pptv/day. Based on the 15 LPG exhaust samples we analyzed in this study, the average
NMHC composition of unburned LPG was primarily n-butane (48%), propane (28%) and i-
butane (23%). The highest fraction of n-butane in LPG exhaust explained the most significant
decrease of n-butane in ambient samples after the replacement of catalytic converters on LPG
fuelled vehicles. The daily changing rates of LPG-related VOCs in two periods, i.e.,
September-December and January-May, in different years at MK are shown in Table 5.2. It is
noteworthy that the two periods were selected based on data availability each year, and the
minimized influence of meteorological parameters in the same month of different years. The
LPG-related VOCs decreased from September to December in 2011 and 2013, but there was
no significant difference in 2012. In general, the levels of VOCs in the atmosphere are
associated with source emissions, photochemical reactions and regional transport. Since the
regional influence was confirmed to be minor at MK (refer to the discussions on propane/CO
ratio and wind speed at MK in Chapter 3), source emissions and photochemical reactions
became the main factors determining the ambient concentrations of VOCs. The decreasing and
unchanged trends in 2011 and 2012 respectively might be related to the integrated influence of
reduced photochemical degradation (which reduces the loss rates of ambient VOCs) and
temperature decrease (which reduces evaporative emissions of VOCs) from September to
December. It is noticeable that the LPG-related VOCs had the highest decreasing rates in
September-December 2013, compared to those in previous years. The higher decreasing rates
of LPG-related VOCs in September-December 2013 implied the possible effectiveness of the
interventional program on VOCs reduction. For the period of January-May, the LPG-related
91
VOCs increased in January-May 2013, except for propane which did not show significant
change from January to May. However, significant decreasing trends (p<0.05) were observed
for propane and n/i-butanes during the same period in 2014, which was likely owing to the
continuous replacement of catalytic converters on LPG-fueled vehicles. By comparison, the
reduction rates of LPG-related VOCs in January-May 2014 were all lower than those in
September-December 2013. This was mainly attributable to the fact that a large portion (i.e.,
~69%) of the converter replacements were completed by the end of December 2013, and the
replaced converters were much fewer (i.e., ~31%) in January-May 2014.
Figure 5.3 Variation trends of the daily average propane and n/i-butanes at MK from June 2013
to May 2014. In the equation label X has units of days.
92
Figure 5.4 Relationships between the LPG-related VOCs with temperature, wind speed and
pressure (a. the high temperature caused by the anti-cyclone over South China (highlighted
with red color); b. the high wind speed caused by the tropical storm over SCS (highlighted with
blue color); c. the low pressure. d. the low temperature caused by an active ocean flow and a
low pressure trough over SCS (highlighted with green color).
Table 5.2 Changing rates of propane and n/i-butanes in September-December and January-May
in different years (ppbv/day)
Time periods September-December January-May
Site Species 2011 2012 2013 2013 2014
MK Propane -0.015 -0.004* -0.017 0.004* -0.014
i-butane -0.010 -0.008* -0.021 0.012 -0.005
n-butane -0.023 -0.009* -0.039 0.016 -0.028
* The changing rate insignificant (p>0.05); the bold numbers are changing rates during the
implementation of the program.
93
Figure 5.5 shows the temporal variations of primary NOx, NO, NO2 and secondary NO2 at MK
from June 2013 to May 2014. The primary NOx and NO mixing ratios decreased significantly
(p<0.05), implying the possible effectiveness of the intervention program on NO/NOx
reduction. However, the secondary NO2 levels increased, while there was no significant change
(p>0.05) for primary NO2. To explore the reasons of NO-NO2-NOx variations, the changing
rates of primary NOx, NO, NO2 and secondary NO2 during the same period (i.e., June in the
previous year to May in the following year) in different years were compared. Table 5.3 shows
the statistics of changing rates of NOx-NO-NO2 at the roadside sites (i.e., MK, CWB and
Central) from June 2011 to May 2014. Taking MK as an example, the primary NOx and NO
decreased much faster (i.e., 91 and 94 pptv/day, respectively) from June 2013 to May 2014
than those during the same period in 2011-2012 and 2012-2013, suggesting the effectiveness
of the program in reducing NOx and NO. Compared to the decreasing rate in June 2012-May
2013 (i.e., 13 pptv/day), no significant change (p>0.05) was observed for primary NO2 from
June 2013 to May 2014. However, it cannot be concluded that the program caused the increase
of primary NO2, in view of the same insignificant variation in June 2011-May 2012 and the
fact that LPG-fueled vehicles emit negligible NO2 (see section 5.2.2). In contrast, secondary
NO2 significantly increased (with a rate of 13 pptv/day) from June 2013 to May 2014, which
was also observed in June 2011-May 2012 (i.e., increasing rate of 20 pptv/day). Since
secondary NO2 is formed by NO reacting with O3, inspection of the O3 production would
provide more comprehensive interpretation on the secondary NO2 increment during the
program. Similar variations were observed at CWB and Central, where primary NOx and NO
showed significant decreasing trends, whereas the variations of primary and secondary NO2
were insignificant (p>0.05).
94
Figure 5.5 Variation trends of the daily average primary NOx, NO and NO2 and secondary NO2
at MK from June 2013 to May 2014. In the equation label x has units of days.
Table 5.3 Changing rates of primary NOx, NO, NO2 and secondary NO2 (ppbv/day).
Site Species Jun. 2011-May 2012 Jun. 2012-May 2013 Jun. 2013-May 2014
MK [NOx]prim -0.004 0.040* -0.091
[NO]prim 0.004* 0.053 -0.094
[NO2]prim 0.008* -0.013 0.003*
[NO2]sec 0.020 -0.002* 0.013
CWB [NOx]prim -0.055* 0.089* -0.121
[NO]prim -0.048* 0.081* -0.127
[NO2]prim -0.010* 0.004* 0.006*
95
[NO2]sec 0.014 -0.001* 0.006*
Central [NOx]prim -0.041* 0.180 -0.187
[NO]prim -0.045* 0.161 -0.188
[NO2]prim 0.005* 0.018 0.002*
[NO2]sec 0.011* -0.004* -0.001*
* The changing rate is insignificant (p>0.05); the bold numbers are changing rates during the
implementation of the program
In summary, the LPG-related VOCs were significantly lowered during the program with
monthly reductions of 3.2 ± 0.2, 2.8 ± 0.2 and 4.9 ± 0.2 ppbv for propane, i-butane and n-
butane by the end of this study, respectively. Continuous decreasing trends were observed for
LPG-related VOCs, and the reduction rates were almost unprecedented, e.g., 39 pptv/day for
n-butane. Furthermore, the mixing ratios of NO and NOx decreased as well during the program.
The reduction rates of NO and NOx during June 2013-May 2014 at the three roadside sites
were much higher than those in previous years. Overall, the field measurement data indicated
that the program was effective in reducing emissions of LPG-related VOCs and NOx.
5.2 Variations of LPG contributions to VOCs and NOx
5.2.1 Source identification
To investigate the change of the contributions of LPG-fueled vehicles to VOCs and NOx, the
online data of 15 VOCs and 3 trace gases at MK “before” (i.e., 8,753 samples during October
2012-September 2013) and during the intervention program (i.e., 5,833 samples during October
2013-May 2014) were separately applied to PMF for source apportionments. It is noteworthy
that the whole-air ambient concentrations of VOCs and trace gases rather than those with the
96
backgrounds being deducted were used here, in order to keep consistency with the simulations
of whole-air ambient O3 and photochemical reactivity. Figure 5.6 shows the source profiles of
the four factors that best reproduced the concentrations of the input species “before” and
“during” the intervention program, respectively. The first factor had high loadings of C3-C5
hydrocarbons and toluene, and was dominated by CO, NO and NO2, which are all main
components of vehicle exhaust [Guo et al., 2011a; Ho et al., 2009]. The n/i-pentanes and
toluene indicated emissions from gasoline vehicles, while the high NOx loading was a signature
of diesel exhaust. Since propane and n/i-butanes were not prominent, this factor was assigned
as gasoline and diesel-fueled vehicle exhaust. Factor 2 was closely associated with LPG-fueled
vehicle exhaust, with the dominance of propane, n/i-butanes and propene, the major
components and combustion product of LPG [Guo et al., 2011b; Blake and Rowland, 1995].
Factor 3 was distinguished by NO2 and the long-lived species, i.e., ethane, ethyne, benzene and
CO. The long-lived species were the indicators of aged air masses, in which NO2 was
accumulated due to photochemical reactions. Hence, this factor represented aged air masses.
Indeed, this profile highly coincided with the aged air in Hong Kong identified by Lau et al.
[2010]. The last factor explained most of the TEX compounds (toluene/ethylbenzene/xylenes),
the tracers of solvent usage [Guo et al., 2007; Borbon et al., 2002]. Therefore, factor 4 was
identified as solvent usage.
97
Figure 5.6 Source profiles resolved by PMF “before” (red bar) and “during” (green bar) the
program. The standard errors are estimated with the bootstrap method in the model.
5.2.2 Source contribution
Based on the VOC loadings in each source, Table 5.4 summarizes the mass and percentage
contributions of the sources to VOCs at MK. LPG-fueled vehicle exhaust was the greatest
contributor to VOCs at MK, with the mass contribution of 114.2 ± 0.1 μg/m3 (51.51 ± 0.05%)
and 64.8 ± 0.1 μg/m3 (39.07 ± 0.05%) before and during the program, respectively. These
contributions were higher than those quantified at other sites in Hong Kong, i.e., suburban TC
(32.6±5.8%) [Ou et al., 2015], urban TW (21±2%) [Ling and Guo, 2014] and suburban YL
(15%) [Guo et al., 2007]. It is noteworthy that factors such as study period, chemical species,
source profiles and models used all influence the source apportionment. Bearing these factors
in mind, the higher contributions of LPG-fueled vehicle exhaust in this study were likely due
to the fact that MK was a roadside site closer to the emission sources. On the other hand, though
gasoline/diesel vehicles emitted considerable VOCs (i.e., 60.5 ± 0.1 and 56.8 ± 0.2 μg/m3
98
before and during the program, respectively), they were significantly lower than those emitted
by LPG-fueled vehicles (p<0.05), particularly before the replacement of the catalytic
converters. While LPG-fueled vehicles accounted for only a small amount of the registered
vehicles (~3.1%) in Hong Kong (see Table 5.5), it was not unreasonable that LPG made the
highest contribution to VOCs, in view of low vapor pressure of the LPG component and high-
density flow of the LPG-fueled vehicles at MK (32%). Apart from vehicular exhaust, the aged
VOCs and solvent usage also contributed ~24% to VOCs at MK.
Table 5.4 Mass concentration and percentage contribution of the four sources to the sum of
VOCs applied in source apportionment at MK “before” and “during the program”.
Source Before the program During the program
Mass (μg/m3) Percentage (%) Mass (μg/m3) Percentage (%)
Gasoline/diesel
vehicle exhaust
60.5 ± 0.1 27.3 ± 0.05 56.8 ± 0.2 34.3 ± 0.1
LPG-fueled
vehicle exhaust
114.2 ± 0.1 51.5 ± 0.05 64.8 ± 0.1 39.1 ± 0.05
Aged air masses 19.8 ± 0.2 8.9 ± 0.1 24.6 ± 0.1 14.8 ± 0.1
Solvent usage 27.3 ± 0.04 12.3 ± 0.01 19.6 ± 0.1 11.8 ± 0.04
Table 5.5 Number of registered vehicles in Hong Kong by the end of 2013 categorized by the
fuel types.
Fuel type Class of vehicles No.
Gasoline Private cars 475,752
Diesel Single decked Public buses 7,162
99
Double decked Public buses 5,619
Public light buses 3,151
Single decked Private buses 511
Double decked Private buses 61
Private light buses 2,757
Light goods vehicles 74,399
Media goods vehicles 37,902
Heavy goods vehicles 4,755
LPG Taxis 17,950
Public light buses 3,280
Electricity Motor cycles (including motor tricycles) 41,766
Total 675,065
Table 5.6 lists the reduction of individual species from LPG-fueled vehicles due to the
intervention program. The reductions were derived from the differences of concentrations
between the period before the LPG program and the period during the LPG program. The
concentrations of VOCs and trace gases attributable to LPG emissions during both periods
were extracted from PMF. Propane (“before”: 21.15 ± 0.02 μg/m3; “during”: 12.53 ± 0.01
μg/m3), n-butane (“before”: 57.96 ± 0.01 μg/m3; “during”: 30.27 ± 0.02 μg/m3) and i-butane
(“before”: 27.20 ± 0.02 μg/m3; “during”: 17.50 ± 0.01 μg/m3) all decreased remarkably
throughout the study period (p<0.05). Meanwhile, the concentrations of CO and NO were also
reduced from 248.4 ± 1.3 and 18.2 ± 0.1 μg/m3 to 228.9 ± 0.6 and 2.08 ± 0.04 μg/m3,
respectively. However, NO2 apportioned in LPG exhaust was extremely minor (i.e., 0.6 ± 0.2
μg/m3 and nil before and during the program, respectively), which might explain the
insignificant decrease or even increase of NO2. During the study period, the LPG-related VOCs
100
and NO increased by ~1.4 and ~2.9 μg/m3, respectively, at the background site, only
respectively accounting for 3.0% and 1.3% of the decreased LPG-related VOCs (~46.1 μg/m3)
and NO (~226.8 μg/m3) in LPG-fueled vehicle exhaust. Therefore, it is believed that the
reductions of VOCs and NO in LPG-fueled vehicle exhaust benefited from the intervention
program.
Table 5.6 VOCs and trace gases emitted from LPG-fueled vehicles before and during the
intervention program (unit: μg/m3 unless otherwise specified). The standard errors are
estimated with the bootstrap method in the PMF model.
Species Before the program During the program
Ethane 1.2 ± 0.001 1.2 ± 0.01
Ethene 1.3 ± 0.005 0.4 ± 0.004
Ethyne 0.0±0.0004 0.0 ± 0.001
Propane 21.2 ± 0.02* 12.5 ± 0.01
Propene 2.9 ± 0.004 1.6 ± 0.001
n-Butane 58.0 ± 0.01 30.3 ± 0.02
i-Butane 27.2 ± 0.02 17.5 ± 0.01
n-Pentane 0.3 ± 0.003 0.04 ± 0.002
i-Pentane 1.0 ± 0.01 0.8 ± 0.002
Benzene 0.0 ± 0.01 0.0 ± 0.001
Toluene 0.7 ± 0.02 0.2 ± 0.01
Ethylbenzene 0.0 ± 0.001 0.0 ± 0.004
m,p-Xylene 0.4 ± 0.001 0.0 ± 0.01
o-Xylene 0.0±0.003 0.01 ± 0.001
CO 248.4 ± 1.3 228.9 ± 0.6
101
NO 18.2 ± 0.1 2.1 ± 0.04
NO2 0.6 ± 0.2 0.0 ± 0.08
Sum of VOCs 114.2 ± 0.1 64.8 ± 0.1
Contribution to VOCs (%) 51.5 ± 0.05 39.1 ± 0.05
*The bold fonts demonstrate significant reductions of LPG related VOCs and NOx.
5.3 Impact of the intervention program on O3 production
5.3.1 O3 Simulation
As VOCs and NOx are key O3 precursors, it is essential to examine the impact of the reduction
of VOCs and NOx during the program on the O3 production. Figure 5.7 compares the observed
and simulated O3 during daytime (07:00-19:00 local time (LT)) in base case.
Figure 5.7 Hourly simulated and observed O3 during daytime hours (07:00-19:00 LT) at MK.
To quantitatively evaluate the performance of the model, the index of agreement (IOA) was
introduced to test the agreement between the simulated and observed O3. In this study, the IOA
reached 0.75, and the accuracy of the simulation was 16.7 ± 2.1%, suggesting good
performance of the model in O3 simulation. Bearing in mind the uncertainty of the model, the
102
good agreement between the simulated and observed O3 implied a minor regional contribution
at this roadside site.
5.3.2 Net O3 production
To understand the impacts of LPG emission constraints on O3 production, the PBM-MCM
model was run with the observed concentrations of air pollutants as input and with the removal
of LPG-related species from the above input, termed as base run and constraint run,
respectively. Three scenarios were included in the constraint run, namely the removal of LPG-
related VOCs, the removal of LPG-related NOx and the removal of both LPG-related VOCs
and LPG-related NOx. The differences of the simulated O3 between the base run and constraint
scenarios represented the contributions of LPG-related species to O3 production. Figure 5.8
shows the net O3 variations induced by VOCs, NOx and VOCs + NOx in LPG-fueled vehicle
exhaust before” and during the program. From “before” to “during” the intervention program,
the daily average O3 variation induced by VOCs decreased significantly (p<0.05), with a rate
of -9.3 × 10-5 ppbv/day, whereas O3 increased significantly at a rate of 1.3 × 10-3 ppbv/day due
to the reduction of NOx titration (p<0.01). As a result, the net contribution to O3 made by the
LPG-fueled vehicle exhaust increased (p<0.01) from -0.47 ± 0.03 ppbv before the program to
-0.06 ± 0.02 ppbv during the program, with a rate of 1.2 × 10-3 ppbv/day. Namely, O3 increased
by 0.40 ± 0.03 ppbv, ~5.6% of the observed O3 at MK (7.14 ± 0.21 ppbv). The simulation
results for the scenarios with and without the input of background concentrations indicated
minor contribution of background concentrations to O3 for both periods of before (0.24 ppbv,
accounting for ~3.5% of measured average) and during (0.27 ppbv; ~3.7%) the program,
suggesting the use of whole-air ambient concentrations without background subtraction for O3
simulation was appropriate. The slight increase of O3 (~0.03 ppbv) caused by the background
variations of O3 precursors from before to during the program constituted only ~7.5% of the
O3 enhancement (0.40 ± 0.03 ppbv) due to the replacement program, further confirming a
103
negligible impact of the background on the assessment of the program. It is noteworthy that
the increase was not a sudden change in reality, but was caused by the segmentation of source
apportionment (i.e., October 2012-September 2013 and October 2013-May 2014). Indeed, the
measurement data also revealed an increasing O3 trend from October 2012 to May 2014, with
a rate of 3.3 pptv/day (p<0.05), higher than the O3 increase caused by LPG. This discrepancy
might be related to the O3 production by other sources and/or other mechanisms not considered
in the above simulations, e.g., alkyl nitrate (RONO2) chemistry.
Figure 5.8 Net O3 variation as a function of VOCs, NOx and VOCs + NOx emitted from LPG-
fueled vehicles
5.4 Photochemical reactivity
5.4.1 OH, HO2 and their formation/loss rates in whole air
As the “detergent” of atmosphere, OH initiates the oxidation of air pollutants including VOCs,
CO and NOx, leading to O3 formation, and the cycling between OH and HO2 accelerates the
propagation of the chain reactions. Thus, the budget of OH and HO2 is an important parameter
of a photochemical system. Figure 5.9 presents the average daytime patterns of OH and HO2
104
before and “during the program”. The simulated OH and HO2 manifest typical bell-shaped
patterns, namely increasing from morning to noon, peaking at 12:00-13:00, and then decreasing
to low levels in the afternoon.
Figure 5.9 Daytime patterns of OH and HO2 (a) “before” and (b) “during the program”.
The average simulated concentrations of daytime (07:00-19:00 LT) OH and HO2 were 1.64 ±
0.78 × 105 and 2.49 ± 1.28 ×105 molecules/cm3 before the program, and 1.80 ± 0.85 × 105 and
4.18 ± 2.03 ×105 molecules/cm3 during the program, respectively. Compared to those modeled
at an urban site in Hong Kong (2.3-3.6 ×106 molecules/cm3 for OH and 3.4-4.4 ×108
molecules/cm3 for HO2) [Ling et al., 2014], and those measured at a VOC-rich site in PRD
(15×106 and 18×108 molecules/cm3 for OH and HO2, respectively) [Lu et al., 2012; Lou et al.,
2010], the OH and HO2 levels were much lower at MK. This is not surprising because much
OH and HO2 were consumed by high levels of VOCs (ppbv) and NOx (ppbv) at the roadside
MK (Figure 5.10 shows the simulated OH and HO2 at different sites with the PBM-MCM
model). In addition, this study covered different seasons of the study years, rather than the O3
episodes in Ling et al. [2014] and summer only in Lu et al. [2012] and Lou et al. [2010] when
photochemical reactivity was usually stronger. Apart from the increase of OH and HO2 from
before to during the program, the ratio of HO2/OH was significantly higher during the program
(i.e., 3.7 ± 0.5) than that before (i.e., 1.9 ± 0.3) (p<0.05), which might partly explain the
105
increase of O3 throughout the study period because the higher HO2/OH ratio means that NO
can be more efficiently transformed to NO2 which further decomposes and generates oxygen
atom for O3 formation (O + O2 → O3). Nevertheless, since the OH, HO2 and HO2/OH were
simulated using the whole-air concentrations of O3 precursors, whether their increases were
mainly caused by the intervention program requires further investigation, as shown below.
106
Figure 5.10 Daytime patterns of OH and HO2 at different sites in Hong Kong simulated by the
PBM-MCM model. TMS, TC, TW and MK are the mountainous, suburban, urban and roadside
site, respectively.
Since the changes in the formation/loss rates from before to during the program were minor for
each pathway of OH and HO2, Figure 5.11 presents their average pathway-specific formation
and loss rates over the whole study period. The total formation/loss rates of OH and HO2 were
5.8 ± 2.4 ×106 and 2.6 ± 1.0 ×106 molecules/cm3/s, respectively. The reaction of HO2 with NO,
HONO photolysis, O3 photolysis and ozonolysis of alkenes were the main sources of OH, with
contributions of 56.7 ± 11.6%, 31.7 ± 10.7%, 6.6 ± 3.3% and 5.0 ± 1.7%, respectively. On the
other hand, OH was consumed by reaction with NO (36.5 ± 5.5%), NO2 (35.1 ± 4.6%), VOCs
(14.8 ± 1.2%) and CO (13.6 ± 1.0%). For HO2 formation, the reaction between RO2 and NO
was the most predominant pathway (54.8 ± 8.8%), followed by the reaction of OH with CO
(23.3 ± 3.9%), HCHO photolysis (13.1 ± 4.6%), ozonolysis of alkenes (7.9 ± 2.1%) and the
reaction of OH with HCHO (0.8 ± 0.2%). Meanwhile, HO2 was almost exclusively consumed
by reacting with NO.
107
Figure 5.11 Average formation and loss pathways of OH and HO2 at MK during the study
period.
5.4.2 Influence of the program on photochemical reactivity
The differences between the simulation outputs of the scenarios with and without LPG VOCs
and NOx inputs determined the contributions of LPG-fueled vehicle exhaust to the
formation/loss rates of OH and HO2, as summarized in Table 5.7. Generally, the formation/loss
rates of OH and HO2 contributed by the LPG source decreased from before to during the
program, which was caused by the reduction of VOCs and NOx in the LPG source. Furthermore,
since the sum of the formation rates were lower than the loss rates for OH for both before and
during the program, but for only HO2 before the program, the net effects of the LPG source to
OH and HO2 were destructive. However, the HO2 formation rate exceeded its loss rate for the
period of during the program, indicating a net production of HO2 by the LPG source.
Table 5.7 Contributions of LPG-fueled vehicle exhaust to the formation and loss of OH and
HO2 “before” and “during” the program.
Reaction Before the program
(molecules/cm3/s)
During the program
(molecules/cm3/s)
OH formation HO2 + NO (1.8 ± 0.8) × 105 (2.8 ± 1.2) × 104
108
O3 + alkenes (3.3 ± 1.2) × 104 (3.1 ± 1.2) × 104
Sum of OH
formation
(2.2 ± 0.9) × 105 (5.2 ± 1.9) × 104
OH loss OH + CO (1.4 ± 0.6) × 104 (1.5 ± 0.6) × 105
OH + NO2 (5.0 ± 2.0) × 103 -
OH + NO (1.9 ± 0.02) × 105 (2.4 ± 1.0) × 104
OH + VOCs (2.3 ± 0.02) × 105 (1.5 ± 0.6) × 105
Sum of OH loss (5.7 ± 1.6) × 105 (3.2 ± 1.4) × 105
Net OH - (3.6 ± 0.8) × 105 - (2.7 ± 1.2) × 105
HO2 formation RO2 + NO (8.1 ± 3.4) × 104 (1.3 ± 0.5) × 104
O3 + alkenes (3.3 ± 1.2) × 104 (3.1 ± 1.2) × 104
Sum of HO2
formation
(1.1 ± 0.4) × 105 (3.6 ± 1.3) × 104
HO2 loss HO2 + NO (1.8 ± 0.8) × 105 (2.8 ± 1.2) × 104
Sum of HO2 loss (1.8 ± 0.8) × 105 (2.8 ± 1.2) × 104
Net HO2 - (7.0 ± 3.1) × 104 (7.8 ± 3.1) × 103
Figure 5.12 (a) shows the net effects of the LPG source on the production of OH and HO2.
From before to during the program, the destruction rate of OH decreased, while the destruction
of HO2 switched to production. These variations led to the increases of OH and HO2 from
before to during the program, as shown in Figure 5.12 (b). Different from the increases of OH
and HO2 in the whole air as shown in Figure 5.9, the increases here were caused by the
intervention program. The OH and HO2 levels increased by 6.9±1.1 ×103 molecules/cm3 and
3.4±1.2 ×104 molecules/cm3, respectively. The higher increase of HO2 than OH led to a higher
ratio of HO2/OH during the program, resulting in a consequent O3 increment.
109
Figure 12 (a) Net production of OH and HO2 by LPG-fueled vehicle exhaust (negative values
indicate net destruction); and (b) increases of OH and HO2 from before to during the program.
5.5 Improvement of the intervention program
5.5.1 O3-VOCs-NOx sensitivity in the whole air
O3-VOCs-NOx sensitivity can reflect the O3 variation relative to the change of VOCs and NOx,
from which VOC-limited regimes and NOx-limited regimes in O3 formation can be indicated.
Figure 5.13 (a) and (b) show the O3-VOCs-NOx sensitivity in the air at MK between 40-100%
and 0.5-40% of the observed average NOx, respectively. The ranges were selected according
to the O3 levels and behaviors responding to the variations of VOCs and NOx (details are given
below). This sensitivity diagram was obtained with the aid of the PBM-MCM model, and based
on the average diurnal profiles of air pollutants in the atmosphere before the intervention
program. The abscissa and longitudinal coordinates represent the percentages of NOx and
VOCs relative to the real average values measured at MK (i.e., 100% NOx = 235.6 ppbv; 100%
VOCs = 51.2 ppbv). In other words, they reflect the reduction percentages. For example, 80%
NOx or VOCs means NOx or VOCs was reduced by 20%. O3 was simulated in 220 cases (i.e.,
10 VOCs × 22 NOx), and the maximum O3 in each case was extracted.
110
It was found that within the range of 0-60% reduction of NOx (Figure 5.13 (a)), O3 increased
with the increase of VOCs and the decrease of NOx, indicating a VOC-limited regime in O3
formation. The black curve, perpendicular to the isopleths of O3, demonstrated the most
sensitive response of O3 to VOCs and NOx. Namely, O3 reduction could be achieved most
efficiently towards the abscissa. Using the absolute value of RIR (i.e.,│RIR│) as a measure of
the O3-VOCs-NOx sensitivity, it increased with the reduction of VOCs and NOx. For example,
│RIR│ for VOCs and NOx increased from 0.09 and 0.22 in the scenario of 90% VOCs and
90% NOx, to 0.25 and 0.90 in the scenario of 50% VOCs and 50% NOx, respectively. From
before to during the program, VOCs and NOx decreased ~12% and ~15%, respectively (i.e.,
from point A to B), causing a slight O3 increment as described in section 5.3.2. The red and
green curves in the lower right corner of Figure 5.13 (a) show the O3 production as a function
of VOCs cut before (NOx=100%) and during the program (NOx=~85%), respectively. With the
reduction of VOCs, O3 decreased. Since O3-VOCs-NOx sensitivity increased during the
program, O3 decreased by 45.5 and 67.6 pptv with 10% cut of VOCs before and during the
program, respectively, which means that O3 reduction could be achieved more efficiently by
further cutting VOCs during the program.
With the continuous reduction of NOx, it was expected that the O3-VOCs-NOx sensitivity might
change substantially due to the dual role of NOx in O3 formation and titration. Figure 5.13 (b)
shows the O3-VOCs-NOx sensitivity in the cutting range of 60-99.5% of NOx. It is noticeable
that in the cutting range of ~60-89.5% of NOx, O3 increased with the increase of VOCs and
decrease of NOx, similar to that in the cutting range of 0-60% of NOx. However, a transition
area appeared when NOx was further cut, where O3 stayed relatively stable with NOx variations,
and decreased with VOC reductions. This transition area changed from ~5.5-10.5% (i.e., VOCs
=100%) to ~2.5-6% of NOx (i.e., VOCs =10%). The appearance of the transition area implied
that the titration of O3 by NOx reached the minimum level, and further cutting of NOx might
111
actually cause O3 reduction. As expected, O3 decreased with the reduction of NOx when NOx
was reduced to lower than ~2.5-5.5% (i.e., ~2.5% and ~5.5% for 10% and 100% VOCs,
respectively), and responded weakly to VOC variations. This means that O3 formation switched
to a NOx-limited regime. It should be emphasized that this is the first attempt on the study of
O3-VOCs-NOx sensitivity at roadsides in Hong Kong, which could be a reference for the
formulation and implementation of future air pollution control strategies in Hong Kong.
Figure 5.13 (a) O3-VOCs-NOx sensitivity within the cutting range of 0-60% of NOx. The black
curve demonstrates the most sensitive response of O3 to VOCs and NOx. Points A and B
represent the O3-VOCs-NOx relationship “before” and “during” the program, respectively. The
red and green curves in the small legend show the O3 production as a function of VOC cuts
before and during the program, respectively.
112
Figure 5.13 (b) O3-VOCs-NOx sensitivity within the cutting range of 60-99.5% of NOx.
5.5.2 O3-VOCs-NOx sensitivity in LPG
Figure 5.14 shows the net O3 increment (i.e., positive and negative values indicate the increase
and decrease of O3, respectively) as a response of the reduction of VOCs and NOx in LPG-
fueled vehicle exhaust. It was found that the net O3 increment uniformly increased with the
increase of NOx reduction and the decrease of VOCs reduction. To ensure that O3 would not
increase during the program, the reduction of VOCs and NOx should be within the highlighted
area. That is, the highest reduction percentage of NOx should be less than 68.9% (i.e., point A).
Furthermore, when the reduction percentage of VOCs increased from 20% to 100%, the
maximum reduction percentage of NOx for zero O3 increment was between 18.2% and 68.9%,
and the reduction ratio of VOCs/NOx increased from 1.1 to 1.45, suggesting that the reduction
ratio of VOCs/NOx should be > 1.1 in order to maintain zero O3 increment. During the program,
VOCs and NOx in LPG-fueled vehicle exhaust were cut by ~43% and ~89% (i.e., point B),
respectively. According to the reduction ratios of VOCs/NOx (~1.45) in the high reduction
range of NOx, the minimum reduction percentage of VOCs should be ~129% when NOx was
cut by ~89%. In other words, O3 would inevitably increase in this case.
113
Figure 5.14 Net O3 increment as a function the reduction percentages of VOCs and NOx in
LPG-fueled vehicle exhaust.
Indeed, NO2 experienced an overall increase from 1999 to 2013 at the roadsides in Hong Kong
[HKEPD, 2016]. According to this study, secondary NO2 might be more responsible for the
increase. Considering nil-emission of NO2 (section 5.2.2) and low emission of NO from LPG
usage (i.e., 4.0% and 1.1-7.3% based on the emission inventory and source apportionment,
respectively), an alternative scheme was proposed. Namely, reasonable cutting ratio of
VOCs/NOx was used to fulfill zero O3 increase when VOC and NOx were cut. Although the O3
production increase was minor (i.e., 0.4 ppbv or 5.6%) in this study, this scheme could be
applicable in future programs. Moreover, since the reactivity and concentration of VOCs
influenced their O3 formation potential, the relative incremental reactivity (RIR) and RIR
(weighted concentration) of LPG-related VOCs were calculated. Table 5.8 shows the RIR and
RIR-weighted concentration of propane, propene and n/i-butanes. Propene had the highest RIR
(i.e., 5.21×10-2), suggesting the highest sensitivity of O3 production to propene. However, n-
butane in LPG was found to have the highest RIR (weighted concentration) (2.61 μg/m3),
indicating that cutting n-butane in LPG source was optimal for O3 pollution control.
Table 5.8 RIR and RIR (weighted concentration) of LPG-related VOCs.
114
Species RIR (%/%×10-2) RIR (weighted concentration) (μg/m3)
Propane 3.47 1.04
Propene 5.21 0.30
i-butane 3.47 1.33
n-butane 3.47 2.61
5.6 Implication on RONO2 formation Apart from the online measurements of VOCs, whole air samples were also collected at MK
on two days per month from September 2013 to June 2014. During the daily sampling, two
canister samples were collected at 10:00 and 15:00 for 3 minutes, respectively. C1-C4 RONO2
and their parent hydrocarbons were quantified in the canister samples.
Figure 5.15 shows the variations of C1-C4 RONO2 throughout the implementation of the LPG
program. It was found that all RONO2 decreased significantly (p<0.1) between September 2013
and June 2014 except for CH3NO3 which remained stable (p=0.57) during this period. To
examine the variations of primary emissions of RONO2, we found that the concentration of
DMS (oceanic tracer) had no significant changes (p=0.84), while CH3Cl (biomass burning
tracer) decreased (p<0.01) with a rate of -40.3 pptv/month. This indicated that the primary
emissions of RONO2 might decrease throughout the program, which was expected to be caused
by the seasonal differences. Specifically, more biomass burning in PRD region was usually
observed in autumn, when the LPG program was initiated. Therefore, the relatively weaker
primary emissions after the program might be partially responsible for the decreases of C2-C4
RONO2. However, the contradiction was that CH3NO3, which should be mostly influenced by
primary emissions, remained stable. This indicated that the variation of primary emissions was
not the cause of the reductions of C2-C4 RONO2. Upon this inference, the photochemical
formation potentials of RONO2 were further discussed.
115
Figure 5.15 Variations of (a) CH3NO3, (b) C2H5NO3, (c) 2-C3H7NO3, (d) 1-C3H7NO3 and (e)
2-C4H9NO3 at MK through the LPG program.
As illustrated in section 5.4, the oxidative capacity of the atmosphere was slightly enhanced
after the program, inconsistent with the decreases of C2-C4 RONO2. However, the parent
hydrocarbons of C2-C4 RONO2 (ethane, propane and n-butane) in the ambient air experienced
remarkable decreases (p<0.1) throughout the program. It is well known that photochemical
formation is the main source of 2-C4H9NO3, and excellent to good correlations of 2-C4H9NO3
with C2H5NO3 (R2=0.69), 2-C3H7NO3 (R2=0.88) and 1-C3H7NO3 (R2=0.93) were found,
implying the dominance of photochemical formation as the sources of C2 and C3 RONO2.
116
Therefore, the reductions of C2-C4 RONO2 were also attributable to the decreases of their
parent hydrocarbons. The source apportionment of VOCs (see Section 5.2.2) already showed
that the reductions of propane and n-butane in ambient air were mainly attributable to the
constrained emissions from LPG-fueled vehicles. This was also confirmed by the chassis
dynamometer tests which indicated that propane and n-butane in LPG fueled vehicle exhausts
were reduced by 91.0-99.3% and 92.6-99.4%, respectively. Therefore, we concluded that the
emission reductions of parent hydrocarbons from LPG-fueled vehicles accounted for the
decreases of C3-C4 RONO2. Although ethane emitted from LPG fueled vehicles remained
unchanged (p>0.05) throughout the LPG program (see Section 5.2.2), the reduction of C2
RONO2 could partially result from the decrease of n-butane, as C2H5O2 can be formed through
the decomposition of the oxidation products of n-butane [Zeng et al., 2018]. In other words,
the implementation of the LPG program also contributed to the reduction of C2 RONO2.
To sum up, although the oxidative capacity increased slightly (O3 increase as a sign) due to the
LPG program, C2-C4 RONO2 decreased substantially due to the emission reductions of parent
hydrocarbons (particularly propane and n-butane) from LPG-fueled vehicles, as well as the
possibly decreased primary emissions.
It is noteworthy that the decreases of RONO2 at MK were contradictory to the increases of
RONO2 observed at 30 sites over the whole territory of Hong Kong (see Chapter 4). This
discrepancy might be caused by the following reasons. Firstly, samples were only collected on
two days before and after the program at the 30 sites as discussed in Chapter 4, thus the samples
were of low representativeness and of high uncertainty. Secondly, the vehicle density at MK
was very high, suggesting that this site was more sensitive to the changes of chemical
compositions in the air caused by the LPG program. In addition, the enhancement of oxidative
capacity of the atmosphere at the roadside site (MK) was expected to be lower than that over
the entire territory, due to much higher concentration of NOx, which might not be able to
117
counteract RONO2 reductions driven by the decreases of parent hydrocarbons. Hence, we
concluded that the LPG program was effective on the reductions of RONO2 and their parent
hydrocarbons in roadside environment of Hong Kong.
5.7 Sub-conclusions In this study, an intervention program, initiated in September 2013 and aimed to reduce
emissions of VOCs and NOx from LPG-fueled vehicles in Hong Kong, was evaluated. The
results indicated that LPG-related VOCs and NOx decreased significantly during the program,
when higher reduction rates were observed for LPG-related VOCs and NOx compared to those
in previous years. Source apportionment revealed that propane, n/i-butanes and NO in LPG-
fueled vehicle emissions were significantly lower during than before the program. It was
therefore concluded that the program was remarkably effective in reducing VOCs and NOx. To
evaluate the program more comprehensively, the variation of O3 production was simulated with
the PBM-MCM model. It was found that LPG-fueled vehicular emission was generally O3
destructive at the roadside MK site, and the O3 destruction decreased from 0.47 to 0.06 ppbv
due to the intervention program, causing an hourly average O3 increase of 0.40 ppbv (~5.6%).
The LPG-fueled vehicle exhaust generally made negative contributions to the production of
OH and HO2. During the program, the destructive effect weakened and even turned to a positive
contribution to HO2 production, resulting in the increases of OH, HO2 and HO2/OH. This was
in line with the fact that O3 increased slightly during the implementation of the program. To
improve the program for future application, an O3-VOCs-NOx sensitivity analysis was
conducted for ambient air that is not partitioned to sources, and the individual source of LPG-
fueled vehicle exhaust. The NOx-limited regime in O3 formation was only found when NOx
was reduced to less than 5.5%. Furthermore, for the emission reductions in LPG-fueled vehicle
exhaust, the maximum NOx cutting percentage of 68.9% and the lowest cutting ratio of
118
VOCs/NOx (i.e., 1.1) were determined to maintain zero O3 increment. Despite the increased
photochemical reactivity, C2-C4 RONO2 experienced significant decreases through the LPG
program, resulting from the emission reductions of related parent hydrocarbons from LPG
fueled vehicles, as well as the probably decreased primary emissions of RONO2. It should be
noted that the findings in this study cannot be directly applied to the other areas in Hong Kong
where the air pollutants profile and atmospheric photochemistry might be different. However,
it can be expected that the impacts of the intervention program on air quality in the other areas
should be less significant because of the lower vehicle flows and lower fractions of the LPG-
fueled vehicles in fleet vehicle stock in non-roadside environments.
119
Chapter 6 Re-examination of C1-C5 alkyl nitrates in Hong Kong using an observation-based model
The previous two chapters (Chapter 4 and Chapter 5) discussed the spatiotemporal variations
of RONO2 and their parent hydrocarbons in Hong Kong, as wells as the impacts of a recent air
quality improvement measure on the abundances of RONO2 and their parent hydrocarbons.
However, the formation processes of RONO2 were not studied, which in fact was expected to
significantly influence the RONO2 levels. In this chapter, the photochemical formation of
RONO2 and their impact on O3 formation at a coastal site in Hong Kong (Tai O) were
investigated using a PBM-MCM model. The model was constrained with field measurement
data collected on selected O3 episode days at Tai O between August 2001 and December 2002.
The in-situ observations showed that the sum of C1-C5 RONO2 varied from 30.7±14.8 pptv in
spring to 120.7±10.4 pptv in autumn, of which 2-butyl nitrate dominated with the highest
concentration of 30.8±2.6 pptv. Source apportionment indicated that the contribution of
photochemical formation increased with the increase in carbon number of alkyl nitrates,
ranging from 44.2±0.7% for methyl nitrate (CH3ONO2) to 68.1±0.9% for 2-pentyl nitrates (2-
C5H11NO2). Conversely, the contribution of oceanic emission to RONO2 decreased from
16.4±0.9% for CH3ONO2 to 0.0±0.7% for 2-C5H11ONO2. Biomass burning was also
responsible for a non-negligible fraction (7.7±0.1% - 25.1±0.6%) of the observed RONO2.
While the emission ratio of CH3ONO2/CO in biomass burning determined in this study was
comparable to those reported in previous studies, the ratios of C2-C4 RONO2/CO were much
higher than the previously reported values, likely due to the differences in combustion
materials/status and in the distance between fires and the sampling sites. Consistent with the
assumption proposed in our previous study [Simpson et al., 2006], the pathway of CH3O
reacting with NO2 made great contribution (93.2±1.1%) to the photochemical production of
methyl nitrate, with the rest (6.8±1.1%) being contributed by the pathway of CH3O2+NO.
120
Model simulation of photochemical O3 influenced by RONO2 chemistry showed that the
formation of methyl-, ethyl-, i-propyl-, n-propyl-, 2-butyl-, 2-pentyl-, and 3-pentyl-nitrates led
to the O3 reduction of 0.05±0.03, 0.05±0.03, 0.06±0.02, 0.02±0.02, 0.18±0.04, 0.09±0.02 and
0.06±0.02 ppbv, respectively, with an average reduction rate of 11.0±3.2 ppbv O3 per 1 ppbv
RONO2 formation. The C1-C5 RONO2 constituted 18.6±1.9% of the entire RONO2, and had a
nitrogen reserve of 4.1±0.2%, implying their potential influence on O3 production in downwind
areas.
6.1 Mixing ratios and seasonal patterns of RONO2 Table 6.1 shows the mean levels of O3, C1-C5 RONO2 and their parent hydrocarbons at Tai O
over the whole sampling period, with 95% confidence intervals. It was found that the average
mixing ratios of RONO2 ranged from 4.0±0.4 pptv for 1-C3H7ONO2 to 30.8±2.6 pptv for 2-
C4H9ONO2. However, n-butane mixing ratio (1893.5±586.0 pptv) was the second lowest
among the parent hydrocarbons, while the parent hydrocarbon of 1-C3H7ONO2 (i.e. propane)
ranked the second highest with the mixing ratio of 2391.7±740.4 pptv. This suggested that
RONO2 not only relates to the abundance, but also the reaction pathways and reactivity of the
parent hydrocarbon [Atlas et al., 1993; Blake et al., 1999, 2003], as well as primary emissions.
Similar seasonal patterns were observed for the sum of C1-C5 RONO2 and O3, i.e., higher in
autumn and lower in spring, suggesting the importance of photochemical formation of RONO2.
The average mixing ratios of the sum of C1-C5 RONO2 were 30.7±14.8, 74.9±23.8, 120.7±10.4
and 91.2±8.8 pptv in spring, summer, autumn and winter, respectively. Similar to RONO2, the
average mixing ratio of O3 also peaked in autumn (68.4 ± 8.0 ppbv). Indeed, the sum of C1-C5
RONO2 correlated well with O3 (R2 = 0.71), and for CH3ONO2 (R2 = 0.62). However,
inconsistent with O3 that had lowest mixing ratios in winter (25.1±5.7 ppbv), RONO2 in winter
was the second highest (91.2±8.8 pptv), implying the possible RONO2 sources of oceanic
121
emissions and/or biomass burning, as well as the reduced losses through photolysis and OH
initiated oxidation. Therefore, the pathway contributions to RONO2 were determined in the
following section.
Table 6.1 Statistics of C1-C5 RONO2 and O3 at Tai O over the whole sampling period (mean±95%
confidence interval) (units: pptv unless otherwise specified).
Species Average
(n* = 188)
Spring
(n = 9)
Summer
(n = 22)
Autumn
(n = 114)
Winter
(n = 43)
O3 (ppbv) 56.5±6.2 31.1±14.7 62.4±17.0 68.4±8.0 25.1±5.7
CH4 (ppmv) 2.1±0.04 1.8±0.03 2.0±0.08 2.1±0.06 2.1±0.07
Ethane 2282.5±344.5 699.8±224.3 809.4±147.8 2405.4±534.2 2980.9±270.9
Propane 2391.7±740.4 161.9±105.1 711.9±290.2 2662.9±1179.2 2964.1±681.7
n-butane 1893.5±586.0 76.4±48.4 615.3±292.0 2131.6±924.8 2277.7±632.9
n-pentane 563.5±236.9 32.3±17.5 218.2±105.9 678.2±383.8 532.0±144.8
CH3ONO2 15.9±1.3 8.6±2.4 17.3±3.7 18.1±1.7 11.0±1.2
C2H5ONO2 13.3±0.9 5.1±1.8 9.4±2.4 15.6±1.2 10.6±1.0
2-
C3H7ONO2 26.0±1.9 7.3±4.5 15.9±5.1 30.2±2.4 23.8±2.2
1-
C3H7ONO2 4.0±0.4 0.8±0.5 2.4±0.9 4.8±0.5 3.5±0.5
2-
C4H9ONO2 30.8±2.6 6.5±4.9 19.7±8.1 35.4±3.4 28.9±3.2
2-
C5H11ONO2 7.7±0.7 1.3±0.8 5.4±2.3 8.9±1.0 6.9±0.7
3-
C5H11ONO2 6.9±0.6 1.2±0.8 4.8±2.0 7.9±0.8 6.5±0.6
122
Sum of C1-
C5 RONO2 104.7±7.9 30.7±14.8 74.9±23.8 120.7±10.4 91.2±8.8
* Number of samples. O3 are averaged over the period matching with canister sampling.
6.2 Construction of the PBM-MCM model
6.2.1 Examination of branching ratios
As introduced in section 3.3.2 “Atmospheric Processes of RONO2 in PBM-MCM”, the
formation pathways of RONO2 include RO2+NO (Reaction 6-1) and RO+NO2 (Reaction 6-2).
The branching ratios for the reactions of RO2+NO were acquired from previous studies
[Lightfoot et al., 1992; Flocke et al., 1998a] or calculated according to formula (6-1) to formula
(6-3) recommended by Carter and Atkinson [1985]. For C1 RONO2, branching ratios of
0.00015, 0.0003, 0.001, 0.003, 0.0041 and 0.005 were examined and considered. However,
since branching ratio data for C2-C4 RONO2 were rather limited, the values calculated using
formula (6-1) to formula (6-3) were used as the branching ratios, which were 0.0094, 0.048,
0.019, 0.085, 0.131 and 0.129 for C2, 2-C3, 1-C3, 2-C4 RONO2, 3-C5H11ONO2 and 2-
C5H11ONO2, respectively. Bearing the model uncertainty in mind, the branching ratios were
accepted only when IOAs between the simulated and observed RONO2 were higher than 0.65.
RO2 + NO → RONO2 (Reaction 6-1)
RO2 + NO → RO + NO2 (Reaction 6-2)
α = [𝑌0
300[𝑀](𝑇
300)
−𝑚0
1+𝑌0
300[𝑀](𝑇
300)
−𝑚0
𝑌∞300(
𝑇300
)−𝑚∞
]𝐹𝑧 (Formula 6-1)
z = {1 + [log𝑌0
300[𝑀](𝑇
300)
−𝑚0
𝑌∞300(
𝑇
300)
−𝑚∞ }]
2
}−1 (Formula 6-2)
𝑌0300 = 𝛽𝑒𝛾𝑛 (Formula 6-3)
123
where T is the temperature (K); M represents the number of molecules (molecules/cm3) and n
is the carbon number in RO2. The values of the constants β (1.95 × 10-22 cm3/molecule), 𝛾
(0.947), 𝑌∞300 (0.435), 𝑚0 (2.99), 𝑚∞ (4.69) and 𝐹 (0.556) are all from Carter and Atkinson
[1985]. On the basis of the calculated results, the branching ratios for the primary and tertiary
RO2 radicals (not available in this study) are calibrated by a factor of 0.4 and 0.25, respectively.
According to the applications of the above formulas in lower stratosphere and lower
troposphere, a correction factor of 3 should be further applied to the branching ratios in lower
stratosphere to obtain the corresponding branching ratios on ground level.
6.2.2 Other settings
In addition to the chemical reactions, many modules were compiled in the PBM-MCM model.
For example, the concentrations of air pollutants can be specified to initiate the model in the
initial concentration module. This is important, because the background RONO2 existed prior
to photochemical reactions are generally non-negligible due to their long lifetimes. In this study,
RONO2 mixing ratios observed at 07:00 (local time, LT) were treated as the initial conditions.
The dry deposition module considers the dry deposition of air pollutants, which are
parameterized as an average deposition rate within the height of the mixed layer (HMIX).
Zhang et al. [2002] indicated that the dry deposition velocity for organic nitrates ranged from
0.03 to 0.56/HMIX cm/s. Within this range, deposition rates of 0.03, 0.13, 0.23, 0.33, 0.43 and
0.53/HMIX cm/s were examined for C1-C4 RONO2 in this study (step=0.1/HMIX cm/s).
Overall, based on the observed mixing ratios of air pollutants, including RONO2 precursors,
the PBM-MCM model simulated RONO2 in different scenarios with changes of branching
ratios and dry deposition rates, and consideration of initial conditions. The results indicated
that the most appropriate branching ratio for C1, C2, 2-C3, 1-C3, 2-C4, 3-C5 and 2-C5 was 0.0003,
0.0094, 0.048, 0.019, 0.085, 0.131 and 0.129, respectively, with the deposition rates of
124
0.13/HMIX cm/s for all RONO2 (detailed test results are provided in Chapter 8). In this chapter,
we only showed the modelling results of RONO2, based on these settings.
Since the PBM-MCM model was constructed by the observed hourly concentrations of air
pollutants (excluding RONO2), the regional transport of RONO2 precursors was considered in
the model, as the observed concentrations of air pollutants were built up by not only local
emissions but also regional transport. In addition, the regional transport of RONO2 at night was
also considered in the simulations, through the settings of the observed RONO2 at 07:00 on
each day as initial conditions of each day’s modelling. However, the regionally transported
RONO2 in daytime were not considered, which might cause the discrepancy between the
simulated and observed RONO2, particularly on the days when northerly winds dominated.
Overall, the model well reproduced the observed RONO2 (see Figure 6.3), indicating that the
daytime regional transport of RONO2 was relatively minor.
The model uncertainty was estimated using the following method. The output uncertainty of
PBM-MCM model derived from two parts, i.e., uncertainties of the input species and
uncertainty inherent to the chemical mechanism. In this study, 61 VOCs, 5 trace gases and 2
meteorological conditions (temperature and relative humidity) were used to construct the
model. The measurement accuracies for VOCs and trace gases ranged from 2% for methane to
20% for dimethyl sulfide and some other compounds. Details about the accuracy for each
species can be found in Simpson et al. [2010].
The uncertainty of temperature, relative humidity and the chemical mechanism (MCM) was
roughly estimated as 5%, 5% and 10%, respectively. Mean root square error (MRSE) was
calculated for the accuracies of the input species and uncertainty of MCM, using formula (6-4)
[Willmott, 1982], which was treated as the uncertainty of the PBM-MCM model.
MRSE = √∑ 𝑋2𝑛
1
𝑛
2 (Equation 6-4)
125
where X represents the individual uncertainty of each component (total of n). According to the
calculation, MRSE equals 13%. Therefore, we estimated the uncertainty of the PBM-MCM
model was around 13%. Since the other factors, such as the height of planetary boundary layer
and the entrainment of free troposphere air, may also bring uncertainties to the model
simulation which are difficult to be quantified, the uncertainty of 13% estimated here is a lower
limit of the model uncertainty.
6.3 Pathways to RONO2
6.3.1 Source apportionment of RONO2
Since PBM-MCM is a chemical mechanism model without the consideration of primary
emissions, the primary emission of RONO2 should be eliminated before the model simulation.
Here, the hourly C1-C5 RONO2 mixing ratios were applied to PMF for source apportionment.
As discussed in Chapter 4, the concentration of CH3Cl in Hong Kong was much higher than
the background level in mid-latitude of north hemisphere (~500 pptv) (see Figure 6.1). The
average mixing ratio of CH3Cl reached 883.7±25.8 pptv in the sampling campaign at Tai O,
with the lowest value of 604 pptv, indicating that sources other than oceanic emission played
important roles in elevating CH3Cl levels at this site. Since there was nearly no correlation
between CH3Cl and the biogenic VOCs, i.e. α-pinene (R=0.14), β-pinene (R=0.11) and
isoprene (R=-0.13), the emission of CH3Cl from tropical plants discovered by Yokouchi et al.
[2000, 2002] cannot be a dominant source of CH3Cl. Instead, we found that CH3Cl exhibited
moderate correlation with CO (R=0.54) (see Figure 4.10), a well-known combustion tracer. As
such, it was anticipated that biomass burning made considerable contributions to CH3Cl in
Hong Kong. Further, CH3Cl observed at a mountainous site in Hong Kong correlated fairly
well levoglucosan in PM2.5 (see Figure 4.10), confirming that biomass burning was a significant
contributor to CH3Cl in Hong Kong. Therefore, we used CH3Cl as a tracer of biomass burning
126
in source apportionment of RONO2. Figure 6.2 shows the linear regressions of individual C1-
C5 RONO2 against CH3Cl in Hong Kong. Moderate correlations were found between CH3Cl
and all the RONO2, with R of 0.10 – 0.64, 0.26 – 0.66, 0.20 – 0.67, 0.36 – 0.66, 0.38 – 0.69,
0.24 – 0.71 and 0.20 – 0.68 for CH3Cl versus C1, C2, 2-C3, 1-C3, 2-C4, 3-C5 and 2-C5,
respectively. This suggested that biomass burning indeed made some contributions to C1-C5
RONO2 in Hong Kong. To identify the contributions of oceanic emission to RONO2, dimethyl
sulfide (DMS), Bromoform (CHBr3) and methyl iodine (CH3I) were selected as the tracers of
oceanic emission. In fact, all of them have other origins except the oceans. For example, DMS
can also be emitted from biomass burning [Friedli et al., 2001] and wastewater treatment plants
[Easter et al., 2005]. CHBr3 is used in manufacturing as a cleaning solvent, thus exhibiting
significant correlations with C2Cl4 in some cases [Blake et al., 2003]. In addition to oceanic
emission, rice paddies, wetlands and biomass burning are also sources of CH3I [Bell et al.,
2002]. Here, the three species are included in source apportionment to examine their suitability
in indicating the oceanic emissions, according to the interpretability and reasonability of the
resolved source profiles.
127
Figure 6.1 Latitudinal distribution of atmospheric CH3Cl cited from Yokouchi et al. [2000].
The grey curve stands for the baseline (background level) of CH3Cl.
128
129
130
Figure 6.2 Correlations between the individual C1-C5 RONO2 and CH3Cl in different sampling
campaigns in Hong Kong through 2001 to 2014.
Figure 6.3 shows the profiles of three sources extracted from PMF. Factor 1 is assigned as
biogenic VOCs, in view of the exclusive dominance of isoprene. A small fraction (≤20%) of
RONO2 are allocated to this factor, which may represent the residues of secondarily formed
RONO2 in the atmosphere, because this factor also has 18.8±0.7% of O3. The coexistence of
isoprene (primary species) and O3 (secondary species) in one factor does not mean that isoprene
is secondarily formed or O3 is primarily emitted. Instead, it represents the background residues,
where some proportions of primary (e.g. isoprene) and secondary air pollutants (e.g. O3) coexist.
Factor 2 is distinguished by high loadings of C3-C5 hydrocarbons, with nearly no CO or O3,
likely representing fuel evaporation. Similar with factor 1, factor 2 also accounts for a small
fraction (≤10%) of RONO2. However, no O3 was allocated to this factor, which therefore most
likely reflects the primary RONO2 residues in the background. Factor 3 has relatively high
loadings of the combustion tracers, including CO (37.3±0.9%), ethane (35.6±1.6%), ethene
131
(45.5±0.9%), ethyne (37.7±0.6%), propene (44.5±0.8%) and benzene (38.0±0.8%). RONO2
attributable to this factor ranged from 7.7±0.1% for CH3ONO2 to 25.1±0.6% for 3-C5 H11ONO2.
In addition, this factor also explains 19.0±1.1% of CH3Cl. Upon the present knowledge on
RONO2 sources, we define this factor as biomass burning. Factor 4 is characterized by
moderate to high loadings of CHBr3 (68.3±0.2%), DMS (55.3±2.9%) and CH3I (24.6±0.4%),
which are typical tracers of marine outflows [Atlas et al., 1993; Nowak et al., 2001]. Therefore,
this factor stands for oceanic emissions, which was responsible for 16.4±0.9% of CH3ONO2.
The percentage contribution decreases with the increase of carbon number, till 0.0±0.7% for 2-
C5H11ONO2. The fifth factor is closely associated with solvent usage, due to the dominance of
TEX (toluene, ethylbenzene and xylenes). Nearly no RONO2 was allocated to this factor.
Lastly, as a tracer of photochemical processes, the high loading (74.2±1.2%) of O3 in factor 6
indicates the photochemical formation of RONO2. Note that ~30 - 40% of the long-lived
species (e.g. CO, ethane, ethyne and benzene) are also attributable to this factor, RONO2 in
this source might also contain the residues of the secondarily formed RONO2 in the background.
132
Figure 6.3 Profiles for the sources of RONO2 at Tai O. The errors were estimated with the
Bootstrap method integrated in PMF.
Table 6.2 lists the mixing ratios of individual C1-C5 RONO2 is association with photochemical
formation, oceanic emission and biomass burning, as well as the ratios of RONO2/CO in the
source of biomass burning in this study and those reported in previous studies. Noticeably,
photochemical formation was the most predominant source of RONO2 with the contribution
increasing from 44.2±0.7% for CH3ONO2 to 68.1±0.9% for 2-C5H11ONO2. Consistent with
previous studies [Bertman et al., 1995; Flocke et al., 1998b], the photochemically formed 2-
C3H7ONO2 and 2-C4H9ONO2 were the highest among all the RONO2, due to the high levels of
propane and n-butane in the atmosphere and the relatively large branching ratios. RONO2
derived from oceanic emission were mainly those with carbon number of 1-3. Though
1.21±0.23 pptv of 2-C4H9ONO2 was emitted by the marine source, it only accounted for
133
4.5±0.9% of the total observed 2-C4H9ONO2. Furthermore, the levels of RONO2 in relation to
oceanic emission were well within the ranges of RONO2 concentrations observed near 20° N
over the Pacific Ocean [Atlas, et al., 1993; Blake et al., 2003]. However, it has been suspected
that RONO2 in the marine atmosphere can be built up by transport, particularly for RONO2
with carbon number larger than 3 [Atlas et al., 1988, 1992, 1993; Blake et al., 2003]. Here, this
impact cannot be eliminated due to the proximity of the sampling site (Tai O) to the urban cores
in Hong Kong. Lastly, biomass burning also made non-negligible contributions to RONO2,
ranging from 0.52±0.005 pptv for 1-C3H7ONO2 to 3.55±0.09 pptv for 2-C3H7ONO2. To
validate the source apportionment results, the ratios of RONO2/CO in the source of biomass
burning were calculated and compared with the reported values in literature. It was found that
RONO2/CO ratios in this study were all higher than those detected in immediate vicinity of
three bushfires in Northern Australia [Simpson et al., 2002]. The ratio of CH3ONO2/CO was
comparable between the present study (2.82±0.05 × 10-6 ppbv/ppbv) and those reported by
Friedli et al. [2001], i.e. 2.9 × 10-6 ppbv/ppbv and 4.38 × 10-6 ppbv/ppbv in temperate forest
fires and sage scrub fires, respectively. However, the ratios of C2-C4 RONO2/CO determined
in this study were 3.3 – 26.2 times higher than the corresponding ratios reported by Friedli et
al. [2001]. On one hand, these discrepancies may be attributable to the differences in
combustion materials (e.g. straw, wood, charcoal and etc.) and combustion states (e.g. flaming
and smoldering). A quick example is the large differences of the RONO2/CO ratios between
those observed by Simpson et al. [2002] and by Friedli et al. [2001]. On the other hand, the
ratios determined in this study are based on the air profiles in the ambient environment, instead
of in the plumes immediately emitted from biomass burning (the case in Simpson et al. [2002]
and Friedli et al. [2001]). The longer distance from the fires to the sampling site means more
time for RONO2 formation under the condition of usually abundant precursors emitted from
biomass burning, which will cause the increase of RONO2/CO ratios. Therefore, the source of
134
biomass burning resolved in this study might include the total RONO2 formed immediately in
biomass burning and those formed during the movement of air masses from the fires to the
sampling site.
Table 6.2 Mixing ratios (pptv) of C1-C5 RONO2 attributable to photochemical formation,
oceanic emission and biomass burning, and the emission ratios (ppbv/ppbv × 10-6) of
RONO2/CO in biomass burning.
RONO2 Photochemical
formation
Oceanic
emission
Biomass
burning
RONO2/CO
CH3ONO2 3.05±0.05 1.13±0.06 0.53±0.01 2.82±0.05 a, 0.68±0.23 b,
2.9 c1, 4.38 c2
C2H5ONO2 4.16±0.07 1.15±0.07 0.96±0.02 5.07±0.11 a, 0.20±0.03 b,
1.28 c1, 1.53 c2
2-C3H7ONO2 12.8±0.20 2.86±0.17 3.55±0.09 18.8±0.49 a, 0.57±0.09 b,
2.05 c1, 1.34 c2
1-C3H7ONO2 2.16±0.03 0.11±0.03 0.52±0.005 2.76±0.03 a, 0.04±0.005 b,
0.43 c1, 0.36 c2
2-C4H9ONO2 17.5±0.24 1.21±0.23 5.42±0.09 28.8±0.47 a, 0.11±0.02 b,
1.13 c1, 1.10 c2
3-C5H11ONO2 3.95±0.06 0.25±0.04 1.53±0.04 8.14±0.21 a
2-C5H11ONO2 4.60±0.06 0.00±0.05 1.69±0.04 8.97±0.19 a
a this study; b Simpson et al. [2002]; c1 temperate forest fires in Colorado [Friedli et al., 2001]; c2 sage scrub fires in California [Friedli et al., 2001].
6.3.2 Photochemical pathways of RONO2
With the exclusion of oceanic sources and the biomass burning, the secondarily/photo-
chemically formed RONO2 was simulated by the PBM-MCM model. In this study, the field
135
observations of 69 parameters consisting of 2 meteorological factors (i.e., temperature and
relative humidity), 5 trace gases (i.e., SO2, CO, O3, NO and NO2), 61 non-methane VOCs and
CH4 were used to constrain the model. Five O3 episode days (i.e., 11th and 25th Oct., 2002 and
06th-08th Nov., 2002) were selected from the whole sampling campaign, in order to study the
photochemical formation of C1-C5 RONO2 and their impacts on net O3 production. Only on
these 5 days were the hourly data required for PBM-MCM input available, and they were
representative of autumn conditions when high photochemical pollution frequently occurred.
The first hourly (i.e., 07:00) data were used to initiate the model simulations. The model was
run for a base case (BC) and two constrained cases (CC1 and CC2). Details about the model
configuration of BC and CC are provided in Table 6.3.
Table 6.3 Model configurations of the base case and constrained cases.
Scenario Case category Description
BC base case Initial and aloft RONO2 were included; the model ran with all
reaction pathways open
CC1 constrained case Initial and aloft RONO2 were included; the pathway of
RO2+NO leading to RONO2 formation was shut down
CC2 constrained case Initial and aloft RONO2 were excluded; all the formation
pathways to RONO2 were shut down
It is noteworthy that NO2 was obtained from the difference between NO and NOy because NOy
detected by the MoO/chemiluminescence analyzer approximately equals NOx when the air
mass is greatly affected by fresh emissions [Xu et al., 2013]. In this study, ~90% air masses
were identified as freshly emitted polluted air with ethyne/CO > 4 pptv/ppbv according to the
method proposed by Symth et al. [1996]. Furthermore, the ratios of propane/ethane (1.3±0.2),
136
ethene/ethane (0.9±0.1) and toluene/benzene (7.4±0.6) at Tai O on the five episode days were
comparable to those at an urban site in Hong Kong (1.0±0.1, 1.1±0.1 and 7.3±1.2 for
propane/ethane, ethene/ethane and toluene/benzene, respectively) [Guo et al., 2007], further
confirming that Tai O was significantly influenced by fresh vehicular emissions during the
episodes.
Figure 6.3 presents the secondary RONO2 resolved by PMF model and simulated by PBM-
MCM model in base case (BC). Here, the simulated results were compared with the
photochemically formed RONO2, rather than the observed ones, as PBM-MCM model only
considered the in situ formation of RONO2. It can be seen that the PBM-MCM model
reasonably simulated the variation trends of the daytime RONO2, i.e., the values increased in
the morning, peaked in early afternoon and decreased in late afternoon. The simulated results
well captured the observed low and high values (standard deviation, SD < ±30%). However,
for most cases, the model did not well track the rapid decrease of RONO2 in the afternoon,
probably due to the fact that the physical processes, i.e., vertical and horizontal dispersion were
not considered in PBM-MCM model [Ling et al., 2011; Lam et al., 2013].
137
Figure 6.3 Secondary RONO2 resolved by PMF and simulated by PBM-MCM at Tai O on
selected O3 episode days.
To further evaluate the model performance, the index of agreement (IOA) was used to examine
the correlation between simulated and observed results. Table 6.4 summarizes the IOAs for the
simulation of C1-C5 RONO2. Good to moderate agreements were found between the simulated
and observed values, indicating the performance of this model was reasonably acceptable.
Table 6.4 IOAs between simulated and observed values of C1-C5 RONO2.
Species IOA
138
CH3ONO2 0.71
C2H5ONO2 0.71
2-C3H7ONO2 0.67
1-C3H7ONO2 0.75
2-C4H9ONO2 0.78
3-C5H11ONO2 0.85
2-C5H11ONO2 0.85
Furthermore, the relative contribution of the two reaction pathways, i.e. “RO2+NO” and
“RO+NO2”, to the secondary RONO2 can be determined with the model running in BC and
CC1 (see model description). Taking CH3ONO2 as an example, the reaction of “CH3O2+NO”
and “CH3O+NO2” contributed 26.2±4.6% and 73.8±12.7% to the secondary CH3ONO2,
respectively. This is in line with the hypothesis proposed by Simpson et al. [2006] that the
reaction of CH3O and NO2 may constitute a major part of CH3ONO2 at Tai O. The branching
ratio (α) of CH3O2 reacting with NO to from CH3ONO2 was 0.0003 in MCM protocol, identical
to that adopted in Simpson et al. [2006]. Despite the consistent conclusions, this study more
comprehensively analyzed the sources and formation routes of CH3ONO2. Firstly, primary
emission was responsible for 24.1±0.9% of the observed CH3ONO2. In addition, although
Simpson et al. [2006] proposed that CH3O reacting with NO2 might be responsible for the
elevated CH3ONO2, the assumption was based on the kinetic calculation which treated the OH-
initiated oxidation of CH4 as the only pathway of CH3O2. Indeed, the sources of CH3O2 are
complicated, e.g., photolysis of acetaldehyde, oxidation of CH4 by chloride and OH. Model
simulation in this study found that OH-initiated oxidation of CH4 only accounted for 24.2 ±
3.6% of CH3O2. Namely, the production of CH3O2 in kinetic calculation was much
underestimated. This might explain why the observed CH3ONO2 were always higher than those
calculated following the photo-oxidation of CH4 even in areas where the reaction of
CH3O+NO2 was suppressed (e.g., urban areas with low ratios of NO2/NO).
139
Based on the source apportionment and pathway contributions to photochemical RONO2, the
relative contributions of primary emissions (including oceanic emissions and biomass burning),
“RO2+NO” and “RO+NO2” can be determined, as shown in Table 6.5. The contributions of
residual RONO2 in background are not presented here. It was found that the reaction of
CH3O+NO2 dominated the formation of CH3ONO2, with the contribution of 35.0 ± 6.0% much
higher than the fraction (12.4 ± 2.2%) of CH3ONO2 produced by CH3O2 reacting with NO. In
contrast, RO2 reacting with NO was the main photochemical route to C2-C5 RONO2. The much
higher pathway contribution of “RO+NO2” to CH3ONO2 was attributable to the higher
abundance of CH3O originated from the oxidation of CH4 and decomposition of larger RO2
radicals. The importance of RO2 reacting with NO in RONO2 formation was enhanced with
the increase of carbon number in RO2 radicals except that for 1-C3H7ONO2. This is mainly due
to the fact that the branching ratio leading to RONO2 formation through the pathway of
RO2+NO increased from 0.0003 for CH3ONO2 to 0.131 for 3- C5H11ONO2, and to some extent
related to the abundance of RO2 radicals. For example, the possibility of hydrogen (H)
extraction on the second and third carbon of n-pentane is 0.568/0.349 by OH and 0.558/0.220
by chlorine (Cl), causing a higher contribution of “RO2+NO” to 2- C5H11ONO2 compared to
3- C5H11ONO2, although the branching ratio leading to 2-C5H11ONO2 formation is slightly
lower (0.129).
Table 6.5 Contributions of primary emissions and secondary formation through different
pathways to C1-C5 RONO2 during O3 episodes at Tai O (unit: %). The background
contributions are not listed here.
RONO2 Oceanic emission
+ biomass burning
CH3O2 + NO CH3O + NO2
CH3ONO2 24.1 ± 0.9 12.4 ± 2.2 35.0 ± 6.0
140
C2H5ONO2 28.5 ± 1.0 53.0 ± 6.2 2.9 ± 0.3
2-C3H7ONO2 29.7 ± 0.9 57.3 ± 6.5 0.9 ± 0.1
1-C3H7ONO2 19.5 ± 1.0 67.7 ± 6.7 1.5 ± 0.2
2-C4H9ONO2 24.8 ± 0.9 63.0 ± 6.6 0.3 ± 0.03
3-C5H11ONO2 29.1 ± 0.9 64.3 ± 6.7 -
2-C5H11ONO2 25.0 ± 0.9 67.1 ± 6.8 -
6.4 Impact on O3 formation
6.4.1 Net O3 production
RONO2 chemistry influences the budget of NOx and atmospheric radicals, thus causing the O3
variations. With the model simulation of BC and CC2 (see model description), the net O3
production induced by RONO2 formation was determined, as shown in Figure 6.4. Overall, the
RONO2 formation made negative contributions to O3, with the average reduction of 0.05±0.03,
0.05±0.03, 0.06±0.02, 0.02±0.02, 0.18±0.04, 0.09±0.02 and 0.06±0.02 ppbv for the formation
of CH3ONO2, C2H5ONO2, 2-C3H7ONO2, 1-C3H7ONO2, 2-C4H9ONO2, 2-C5H11ONO2 and 3-
C5H11ONO2, respectively. The average O3 reduction induced by each RONO2 formation
correlated fairly well (R2=0.93) with the concentrations of photo-chemically formed RONO2.
The average O3 reduction rate was -11.0±3.2 ppbv/ppbv. Namely, O3 reduced 11.0±3.2 ppbv
due to per ppbv RONO2 formation. Although RONO2 is generally present as a minor
constituent (i.e., magnitude of pptv), the effect of RONO2 formation on O3 reduction in urban
areas cannot be neglected. For example, on 07 November, 2002, the maximum O3 reduction
caused by the C1-C5 RONO2 reached 2.7 ppbv (~4.4%) at 13:00. On the other hand, not all
RONO2 were considered in this study due to the limitation of analysis technique. Based on the
model simulation, the C1-C5 RONO2 only constituted 18.6±1.9% of the entire RONO2 (i.e.,
141
alkyl and multifunctional nitrates), implying that the potential of RONO2 formation on O3
reduction was far underestimated.
Figure 6.4 Net O3 production induced by C1-C5 RONO2 formation at Tai O.
To explore the mechanism of O3 reduction induced by RONO2 formation, the variation of NOx,
HOx and the O3 were studied in detail. Taking 2-C4H9ONO2 as an example, Figure 6.5 shows
the temporal variation of each species caused by the formation of 2-C4H9ONO2. In general, the
OH, HO2 and O3 decreased with obvious diurnal trends, while no regular variations were found
for NO2 and NO. Moreover, the average O3 reduction rate increased nearly in linear with the
reduction of HOx (R2=0.99) and the increase of NO (R2=0.94), suggesting that O3 formation
was in VOC-limited regime. Generally, OH consumed in the VOC oxidation can be recycled
from the oxidation of RO. However, this pathway was constrained due to the formation of
RONO2, therefore the HO2 and OH generated from the oxidation of RO decreased, and
subsequently O3 was reduced.
142
Figure 6.5 Variations of NO2, NO, HO2, OH and O3 induced by 2-C4H9ONO2 formation.
6.4.2 Nitrogen partitioning
As a temporary reservoir of nitrogen, RONO2 contributes to O3 production in downwind
regions through the release of NO2. Therefore, quantifying the nitrogen budget in RONO2 is
essential to estimate its impact on O3 production in remote areas. Figure 6.6 shows the mixing
ratios of sinks (i.e., HNO3) and temporary reservoirs (i.e., PANs, RONO2, N2O5 and NO3) of
nitrogen, as simulated by the PBM-MCM model. It is noteworthy that the entire RONO2
included alkyl and multifunctional nitrates here, PANs indicated the total peroxyacyl nitrates,
and HONO was not considered as it decomposes quickly in daytime. It was found that PANs,
RONO2 and HNO3 increased from the morning and reached their maximums in the afternoon
or at dusk. As the most important nitrogen reservoirs, PANs, RONO2 and HNO3 accounted for
53.0±4.1%, 29.8±4.9% and 12.8±0.5% of the total oxidized nitrogen, respectively. For NO3
and N2O5, they began to build up from the late afternoon (i.e., 16:00-17:00) due to their poor
stabilities in sunlight, constituting 0.5±0.3% and 3.9±2.0% of the total oxidized nitrogen,
respectively.
143
It is noticeable that the variation trend of the entire RONO2 was consistent with those of C1-C5
RONO2 (see Figure 6.3), indicating the similar formation mechanisms for all RONO2. The C1-
C5 RONO2 accounted for 18.6±1.9% of the entire RONO2, which means that 4.1±0.2%
nitrogen was stored in C1-C5 RONO2. In comparison with PANs, RONO2 manifest to be a
group of more obstinate nitrogen reservoirs, due to their longer lifetimes. While PANs serve as
the main source of NOx in free troposphere of Northern Hemisphere [Moxim, et al., 1996;
Stroud et al., 2003], HNO3 and RONO2 (particularly CH3ONO2) may dominate the nitrogen
sources in remote regions of Southern Hemisphere. This needs further study by undertaking
continuous sampling in both the upwind (i.e., RONO2 formation) and downwind (i.e., RONO2
degradation) regions.
Figure 6.6 Variations of the reserved nitrogen species in daytime simulated by PBM-MCM.
6.5 Sub-conclusions In this chapter, data from a comprehensive field measurement campaign, conducted from
August 2001 to December 2002 at Tai O, a coastal site in Hong Kong, was used to investigate
the pathways leading to C1-C5 RONO2 and their impacts on O3 formation. The sum of C1-C5
RONO2 were the highest in autumn and correlated well with O3, suggesting the importance of
photochemical formation. Evaluation of the Hong Kong PBM-MCM model constrained with
144
the field data gave good-moderate agreement between observed and simulated secondary C1-
C5 RONO2 and O3. The model was further applied to quantify the contribution of each pathway
to C1-C5 RONO2. In consistent with previous suggestions, the reaction of CH3O with NO2 made
considerable contribution to CH3ONO2 formation, accounting for 44.1 ± 0.5% of the total
CH3ONO2. The pathway of CH3O2 reacting with NO and primary emission contributed 3.2 ±
0.5% and 24.1 ± 20.9% to the total CH3ONO2, respectively, with the rest attributable to the
residues in background. This similar results were also reported in Europe, according to the
simulation of CH3ONO2 based on the air pollutants emission inventories in UK (Archibald et
al., 2007). From C2H5ONO2 to PenONO2, the contribution to RONO2 made by photochemical
formation increased, while that of primary emissions decreased. Study on the relationship
between net O3 production and RONO2 formation indicated that the formation of RONO2
limited the O3 formation due to the decrease of OH and HO2. The average O3 reduction rate
was -11.0±3.2 ppbv/ppbv. Although RONO2 is a minor component of NOy, the impact of total
RONO2 on O3 reduction cannot be neglected. Moreover, the nitrogen reserved in RONO2
(4.1±0.2%) may continue to contribute to O3 formation in downwind areas. The findings
enhanced our knowledge on the influence of individual RONO2 species on the O3 production.
145
Chapter 7 New insight into the spatiotemporal variability and source apportionments of C1-C4 alkyl nitrates in Hong Kong
Chapter 6 briefly discussed the sources of RONO2 at a coastal site in Hong Kong. In this chapter,
a phenomenon was observed at two sites in Hong Kong (i.e., a mountain site (TMS) and an
urban site (TW) at the foot of the same mountain) that although the levels of parent
hydrocarbons were much lower at TMS (p<0.05), similar RONO2 levels were found at both
sites regardless of the elevation difference. This suggested various source contributions of
RONO2 at the two sites. As such, the sources of RONO2 were explored at these two sites. Prior
to using PMF model, the data at TW were divided into “meso” and “non-meso” scenarios for
the investigation of source apportionments with the influence of mesoscale circulation and
regional transport, respectively. Secondary formation was the prominent contributor of RONO2
in both the “meso” scenario and “non-meso” scenario. The total C1-C4 RONO2 attributable to
secondary formation increased from 32.9±5.3 pptv in “non-meso” scenario to 55.6±5.3 pptv in
“meso” scenario. Conversely, the total oceanic emission of C1-C4 RONO2 was higher in “non-
meso” scenario (3.5±0.6 pptv) than in “meso” scenario (2.0±0.4 pptv). In contrast to TW, the
RONO2 levels measured at TMS mainly resulted from the photo-oxidation of the parent
hydrocarbons at TW during mesoscale circulation, i.e., valley breezes, corresponding to 52-86%
of the RONO2 levels at TMS. Furthermore, regional transport from the inland PRD region made
significant contributions to the levels of RONO2 (~58-82%) at TMS in the “non-meso” scenario,
resulting in similar levels of RONO2 observed at the two sites.
7.1 Descriptive statistics of RONO2 and their parent hydrocarbons Table 7.1 presents the descriptive statistics of RONO2 and their parent hydrocarbons at TMS
and TW. Figure 7.1 compares the levels of RONO2 measured at TMS and TW with those
146
measured in different environments in previous studies. In general, 2-C3H7ONO2 and 2-
C4H9ONO2 were the most abundant RONO2 at the two sites, consistent with the results
observed in different environments [Blake et al., 2003; Simpson et al., 2006; Russo et al., 2010;
Wang et al., 2013]. The relatively higher levels of 2-C3H7ONO2 and 2-C4H9ONO2 were
associated with the balance between increased branching ratios for photochemical RONO2
formation and the decreased lifetime of both parent alkanes and RONO2 with increasing carbon
number [Arey et al., 2001; Simpson et al., 2006; Russo et al., 2010]. In comparison, the levels
of CH3ONO2, C2H5ONO2 and 2-C3H7ONO2 were slightly higher at TW than at TMS (p < 0.05),
with average values of 12.6 ± 0.5 (mean ± 95% confidence interval), 13.3 ± 0.6 and 26.3 ± 1.2
pptv, respectively, at TW. The average mixing ratios of 1-C3H7ONO2 and 2-C4H9ONO2 were
comparable at the two sites (p > 0.05). The results were contradictory to the fact that the mixing
ratios of their parent hydrocarbons at TMS were much lower than at TW, highlighting the
complexity of sources of RONO2 at both sites.
In comparison with other studies, the average mixing ratios of RONO2 at TMS were much
higher than those measured in forested areas in coastal New England [Russo et al., 2010] and
in tropospheric air influenced by Asian outflow during the airborne TRACE-P mission
[Simpson et al., 2003], where the levels of parent hydrocarbons were also lower. (Note that all
of the UCI data shown in Figure 6.1 were adjusted to UCI’s post-2008 calibration scale of
RONO2 to enable direct comparison [Simpson et al., 2011]. However, the mean mixing ratios
of C1-C3 RONO2 were slightly lower and the 2-C4H9ONO2 mixing ratio was higher at TMS
than at Tai O (Table 7.2), HT and in Karachi, Pakistan [Barletta et al., 2002] (The Karachi
data have also been adjusted to the new UCI alkyl nitrates’ calibration scale). The differences
among TMS, Tai O and HT might result not only from the levels of their parent hydrocarbons,
but also from the influence of air masses with different photochemical ages and sources [Wang
et al., 2003]. Furthermore, the sampling method and sampling period at TMS were different
147
from those at Tai O and HT, where the sampling duration was only 1-min and the sampling
time varied on different sampling days. In particular, many whole air samples were collected
during O3 episodes at Tai O. These could also induce differences in observed levels among the
three sites. At the urban TW site, the mean mixing ratios of RONO2 were lower than those
measured in urban areas in Europe [Worton et al., 2010] and China [Wang et al., 2013].
Compared to the average values of RONO2 at Tai O, the levels of C2H5ONO2, 1-C3H7ONO2
and 2-C4H9ONO2 were slightly higher and the CH3ONO2 and 2-C3H7ONO2 mixing ratio was
lower at TW.
Table 7.1 Descriptive statistics of alkyl nitrates and parent hydrocarbons (pptv) in whole air
samples collected at TMS and TW during the sampling period.
Species TMS TW
Mean* Min Max. 10th# 90th# Mean Min Max. 10th# 90th#
C1 RONO2 10.9±0.4 6.2 21.4 8.1 13.6 12.6±0.5 7.2 26.6 9.2 16.4
C2 RONO2 12.1±0.5 3.2 25.6 7.6 16.5 13.3±0.6 4.0 35.0 8.3 18.1
2-C3 RONO2 24.1±1.1 4.0 51.2 14.8 34.7 26.3±1.2 6.0 49.2 16.2 36.2
1-C3 RONO2 3.8±0.2 0.4 10.6 1.9 5.5 4.0±0.2 0.7 8.1 2.2 6.1
2-C4 RONO2 32.0±1.7 3.1 80.1 18.8 46.6 34.2±1.9 5.1 92.8 20.8 49.2
Methane1 2.0±0.1 1.8 2.2 1.9 2.0 2.0±0.1 1.8 2.5 1.9 2.0
Ethane 1908±78 396 3588 1154 2470 2224±90 717 4315 1359 2906
Propane 1101±75 106 4455 569 1749 3551±41
5
144
3
3380
0
1844 5153
n-Butane 830±91 97 6252 349 1517 4486±48
2
137
2
3470
0
2168 7633
* Average ± 95% confidence interval; # 10th and 90th percentiles; 1 in unit of ppmv.
148
Figure 7.1 Comparison of RONO2 mixing ratios in different locations. Data collected by UCI
before 2008 (PRD and TRACE-P) were adjusted to UCI’s new calibration scale to permit direct
comparison (see text for details about the new calibration. 1. This study, September-November,
2010. 2. Rural site, August 2001-December 2002 [Simpson et al., 2006]. 3. Urban site, 2009-
2011 [Wang et al., 2013]. 4. Urban sites, April-May 2004 [Worton et al., 2010]. 5. Urban sites,
April-May 2004 [Worton et al., 2010]. 6. Coastal site, December 1998-January1999 [Barletta
et al., 2002]. 7. Urban site, August-September 2011 and December 2011-January 2012 [Wang
et al., 2013]. 8. Regional background sites, September 2009 [Wang et al., 2013]. 9. Aircraft
measurement, February-April 2001 [Simpson et al., 2003]. 10. Urban sites, July 2009 [Wang et
al., 2013]. 11. Coastal site, January-February and June-August 2002, July-August 2004 [Russo
et al., 2010]. 12. Regional background site, March 2001-April 2002 (unpublished data).
Table 7.2 Descriptive statistics of RONO2 and parent hydrocarbons in whole air samples
collected at Tai O between 24 August 2001 and 31 December 2002 [Simpson et al., 2006].
Compound Minimum Maximum Median Mean
CH3ONO2 (pptv) 5.5 52.2 13.4 15.9
C2H5ONO2 (pptv) 2.7 34.3 12.1 13.1
1-C3H7ONO2 (pptv) 0.2 14.5 3.5 3.9
149
2-C3H7ONO2 (pptv) 2.4 65.9 24.5 32.6
2-C4H9ONO2 (pptv) 0.8 89.8 27.4 30.7
Methane (ppmv) 1.75 3.70 1.96 2.05
Ethane (ppbv) 0.38 5.05 2.14 2.12
Propane (ppbv) 0.006 13.0 1.54 2.05
n-Butane (ppbv) 0.006 12.8 0.95 1.64
Table 7.3 and Figure 7.2 summarize the synoptic weather conditions and the corresponding
variations of O3 and RONO2 on O3 episode and non-O3 episode days at both sites. In general,
meteorological conditions including temperatures, winds and solar radiation significantly
influenced the levels of air pollutants (Table 7.3). High mixing ratios of O3 and RONO2 were
usually associated with meteorological conditions with high-pressure system and/or stable
conditions, such as high temperatures, intense solar radiation and low wind speeds.
Table 7.3 Summary of synoptic weather conditions and the corresponding variations of air
pollutants on the sampling O3 and non-O3 episode days.
Sampling days
Synoptic weather conditions Variation of pollutants
O3 episode day October 23~24 2010
After the tropical cyclone Megi, the weather was sunny. The temperature (max: 23 and 31 °C at TMS and TW, respectively) and solar radiation levels (max: 843 and 851 W/m2, respectively) increased and remained at high levels. The wind speed decreased and the prevailing wind direction was from the north at TMS. The prevailing winds at TW changed from southeast on October 23 to north on October 24.
O3, NO and SO2 increased clearly and CO increased moderately. O3 reached peaks of 137 ppbv at TMS and 85 ppbv at TW. SO2 reached 10 ppbv at TMS and 14 ppbv at TW. The mixing ratios of RONO2 increased clearly and the diurnal patterns of RONO2 were more significant with peak values observed in the afternoon. The diurnal patterns tracked each other well for C3-C4 RONO2 at TMS and TW.
150
October 29 ~ November 3, 2010
With a continental anticyclone over mainland China, the temperature started to increase and the weather was sunny. A northerly dry monsoon was enhanced at both sites. The solar radiation levels were higher at TMS than that at TW, where their peaks reached 811 and 800 W/m2, respectively. The winds were mostly from the north at TMS, and those at TW were from the southeast, east and northeast.
O3 increased and stayed at high levels at TMS. SO2 and CO at the two sites exhibited an increasing trend and a broad peak. The levels of RONO2 were slightly lower than those on October 23 and 24 at both sites. The diurnal patterns of C1-C2 RONO2 with troughs and peaks during daytime hours were observed on October 29 - November 3 at both sites.
November 9, 2010
A continental anticyclone controlled northwestern China. After rainy days on November 4 ~ 6, the weather was sunny and stable. The temperatures and solar radiation levels increased and the wind speeds decreased.
O3 stayed at a level above 100 ppbv at TMS. NO remained stable and the levels CO and SO2 fluctuated. The levels of RONO2 increased significantly at TMS and TW, with peaks observed in the afternoon.
November 19, 2010
The anticyclone moved over northeastern China and the East China Sea. Although the prevailing direction was from the north at both sites, the wind speeds decreased. The solar radiation levels were higher at TMS than those at TW, with maximum values of 673 and 555 W/m2, respectively.
O3 increased sharply and reached peaks higher than 110 ppb at TMS. CO had a broad peak at both sites. The peak values of RONO2 increased significantly. C3 and C4 RONO2 peaked in the afternoon at the two sites. The maximum CH3ONO2 and C2H5ONO2 levels were observed at midnight at TW and in the afternoon at TMS on some sampling days.
Non-O3 episode days September 28, October 2, 8 and 14, 2010
Low-pressure systems were located in the PRD region and Hainan province on September 28 and October 8. The weather was cloudy in the afternoon on these two days. On October 2 and 14, low-pressure systems (trough) were observed in northern and southern China. The temperatures and solar radiation levels were high on these two days, reaching daily maximum values of 24~27 °C and 775~886 W/m2, respectively. The winds were mostly from the southeast at TW, those at TMS were from the east and northeast at low speeds. Rainfall was observed on the days of September 20-25, October 7, and 9-12.
The levels of O3 and RONO2 were low at TMS and TW on September 28 and October 2. Over the 4 sampling days, the maximum levels of O3 and RONO2 were observed on October 8, with O3 (total RONO2) reaching peaks of 97 ppbv (125 pptv) at TMS and 65 ppbv (129 pptv) at TW.
151
October 18~19, 2010
The tropical cyclones Megi was formed in the South China Sea. The temperature and solar radiation started to decrease from October 18 to 19. The prevailing winds were from the north at TMS and changed from southeast to north from October 18 to 19 at TW.
The daily maximum levels of O3 decreased from 95 ppbv on October 18 to 85 ppbv on October 19 at TMS. Diurnal patterns with maximum values in the afternoon were observed for RONO2 at TMS and TW.
October 27~28, 2010
The tropical cyclone S.T.Chaba was located about 600 km east of Taiwan in the Philippine Sea and moving north. The winds at the two sites were mostly from the north. The temperature started to decrease, with daily maximum temperature reaching 16 and 25 °C at TMS and TW, respectively.
Air pollutants started to accumulate. The maximum levels of O3 reached 80 and 50 ppbv at TMS and TW, respectively. The levels of RONO2 on these two days were lower than those on October 23 and 24, reaching maximum total levels of 95 and 94 pptv at TMS and TW, respectively.
November 20~21, 2010
On the south edge of the high-pressure system located in North China, the weather was sunny. Prevailing southeast winds were observed at TW. The prevailing winds at TMS were from the east. The solar radiation levels were low on November 20, reaching maximum values of 428 and 507 W/m2 at TMS and TW, respectively.
O3 concentrations decreased to low levels at TMS and TW, with maximum hourly average values of 67 and 33 ppbv, respectively. The levels of RONO2 decreased at the two sites on November 19. High RONO2 mixing ratios were observed at midnight at TW.
152
Figure 7.2 Synoptic charts for the sampling days influenced by (a) tropical cyclone and (b)
anticyclone.
Figure 7.3 shows the time series of C1-C4 RONO2 on O3 episode and non-O3 episode days at
both sites, while Figure 7.4 presents the temporal variations of their parent hydrocarbons
accordingly. Although the ranges of RONO2 mixing ratios were similar and maximum values
were observed in the afternoon, the day-to-day variations of individual RONO2 differed during
the sampling period at both sites. The maximum values were comparable and the diurnal
patterns tracked each other for the C3-C4 RONO2 at TMS and TW, especially on the days (24
October to 3 November, 9 and 19 November) with relatively higher O3 mixing ratios (p < 0.05).
The average daytime O3 mixing ratios (0700-1800) on the high O3 days were 77 ± 3 and 38 ±
3 ppbv at TMS and TW, respectively, compared to 58 ± 3 and 23 ± 3 ppbv, on the non-O3
episode days. Typically, the average daytime levels of 2-C3H7ONO2, 1-C3H7ONO2 and 2-
153
C4H9ONO2 on high O3 days at TMS were 27 ± 1 (TW: 28 ± 1), 4.5 ± 0.3 (4.4 ± 0.2) and 37 ± 2
(39 ± 3) pptv, respectively, higher than those on non-O3 episode days (p < 0.05), implying that
secondary formation of RONO2 might be more prominent on O3 episode days. Coincident with
the high C3-C4 RONO2 during high O3 days, their parent hydrocarbons, i.e., propane (0.56-4.46
and 1.55-10.4 ppbv for TMS and TW, respectively) and n-butane (0.28-6.25 and 1.47-16.1 ppbv,
respectively) also showed elevated mixing ratios (Figure 7.4), further suggesting an important
source of C3-C4 RONO2 which was photo-oxidation of the parent hydrocarbons. For the C1-C2
RONO2, the temporal patterns of CH3ONO2 and C2H5ONO2 were different at the two sites,
especially on high-level O3 days. The peaks of CH3ONO2 and C2H5ONO2 were usually
observed between 11 a.m. and 4 p.m. at TMS, except for 14 and 28 October, 1-2, 9, 20-21
November. The peaks of C1-C2 RONO2 corresponded to the high levels of methane and ethane
observed at 11 a.m. to 5 p.m., likely resulted from regional transport [Guo et al., 2009; Jiang
et al., 2010] and/or mesoscale circulations [Gao et al., 2005; Wang et al., 2006] (Section 6.2.3).
At TW, however, besides the maximum concentrations observed in the afternoon, high levels
of CH3ONO2 and C2H5ONO2 were observed from midnight to early morning on 13 out of the
19 sampling days (i.e., 2, 8, 14, 24, 28, 30-31 October, 1-3, 19-21 November), when the
prevailing winds switched to the southeast direction, implying that the high levels of CH3ONO2
and C2H5ONO2 are likely related to marine emissions and aged continental plumes which were
re-circulated from the South China Sea to the coastal urban site at night. Indeed, this
speculation was supported by the source apportionment results at TW, which confirmed that
the high CH3ONO2 and C2H5ONO2 levels from midnight to early morning on the above
sampling days were related to oceanic emissions.
154
Figure 7.3 Time series of CH3ONO2, C2H5ONO2, 1-C3H7ONO2, 2-C3H7ONO2 and 2-
C4H9ONO2 measured at TMS (purple) and TW (red) in 2010. The yellow shading highlights
the O3 episode days.
Figure 7.4 Time series of the parent hydrocarbons of RONO2 at TMS and TW. The yellow
shading highlights the O3 episode days.
Although the levels of the parent hydrocarbons were lower at TMS, similar values of RONO2
were observed at both sites, regardless of the elevation, suggesting the contributions of different
155
sources and/or the influences of different air masses. Hence, the source apportionments of
RONO2, contributions of reaction pathways for the secondary formation of RONO2, and the
relationship between O3 and RONO2 are analyzed in the following sections.
7.2. Sources of RONO2
7.2.1. Photochemical evoluation of RONO2
As photochemical oxidation of parent hydrocarbons is an important source of alkyl nitrates, it
is valuable to study the photochemical evolution of RONO2. To do so, the relationships
between RONO2 and their parent hydrocarbons at the two sites were further examined using a
simplified sequential reaction model developed by Bertman et al. [1995] (Equation 7-1), based
on the assumptions that: (i) the hydrogen abstraction reaction from the parent hydrocarbon was
the rate-limiting step for photochemical production of RONO2, and (ii) the reaction
environment was NOx-rich, making the reaction with NO being the dominant pathway for the
removal of RO2 radicals [Russo et al., 2010]. In this study, the average mixing ratios of NOx at
TMS and TW were 10.7 ± 0.3 and 56.3 ± 1.6 ppbv, respectively, indicating that the environment
was NOx-rich (> 0.1 ppbv, [Roberts et al., 1998]). Hence, reaction with NO was the main
pathway for the removal of RO2 radicals at the two sites. In addition, the results of PBM-MCM
model simulation confirmed that the hydrogen abstraction reaction from the parent
hydrocarbon, namely the reaction of hydrocarbon with OH radical, was indeed the rate-limiting
step for photochemical production of RONO2 at both sites (see Chapter 8 for details).
tkktkk
A
BABA eRH
RONOek
kRH
RONO )(
0
02)(
B
A2
][][)1(
k
(Equation 7-1)
where β = α1α2, kA is the production rate for the formation of RONO2 through the oxidation of
hydrocarbons, RH (kA = k1[OH]), while kB is the destruction rate for RONO2 through photolysis
156
and the reaction with OH (kB = k5[OH] + JRONO2). [RONO2]0 and [RH]0 are the initial
concentrations of RONO2 and the parent hydrocarbons before photochemical processing,
respectively; [OH] is the diurnal average concentration of the OH radical. The relationships of
RONO2 with their parent hydrocarbons derived from the preceding equation are comparatively
independent of the variations of OH and photolysis rates of RONO2 [Roberts et al., 1998; Wang
et al., 2013]. If the initial concentrations of RONO2 and RH are zero, Equation 6-1 can be
expressed as follows (Equation 7-2):
)1( )(
B
A2 tkk
A
BAekk
kRH
RONO
(Equation 7-2)
The relationships between RONO2 and RH are obtained by plotting the measured ratios of
RONO2/RH to a specific ratio, 2-C4H9ONO2/n-butane. The 2-C4H9ONO2/n-butane ratio has
been widely used in the anlysis of RONO2 because n-butane is typically one of the most
abundant hydrocarbons and 2-C4H9ONO2 is the dominant RONO2 [Roberts et al., 1998;
Worton et al., 2010; Wang et al., 2013]. Although some studies have investigated the
relationships between RONO2 and their parent hydrocarbons using zero initial values of
RONO2, more recent studies have used non-zero initial values of RONO2 to evaluate the
influence of background levels on the photochemical evolution of RONO2 [Reeves et al., 2007;
Russo et al., 2010; Wang et al., 2013]. Therefore, in addition to zero initial ratios, non-zero
initial ratios of RONO2/RH, equal to the lowest values from 0000 to 0700 measured at TMS
and TW, respectively, as suggested by Wang et al. [2013], were used to investigate the
relationships between RONO2 and their parent hydrocarbons in this study. The diurnal average
OH mixing ratios were simulated using the PBM-MCM. By providing the values of
photochemical processing time (t), the predicted ratios of RONO2/RH were calculated since
other parameters, i.e., kA, kB, α1, α2 and JRONO2 were obtained from literature [Clemitshaw et al.,
1997; Simpson et al., 2003; Worton et al., 2010; Wang et al., 2013]. In this study, the given
157
photochemical processing time ranged from 30 min to 2 days. The curves generated with zero
initial values were the pure photochemical (PP) curves for the evolution of RONO2, and the
curves with non-zero values, defined as background initial ratio (BIR) curves, were generated
by assuming that both photochemical formation and background levels contributed to the
distribution of RONO2 [Russo et al., 2010; Wang et al., 2013]. Consistent with previous studies
[Russo et al., 2010; Wang et al., 2013], the shapes of the BIR curves were different from those
of PP curves. The BIR curves of C1-C3 RONO2 at both sites were positioned above their PP
curves at shorter processing time (t < 1 d) and converged towards the PP curves at longer
processing times (t = 1.5-2 d) (Figure 7.5), resulting from the decreased influence of the
parameter tkk BAeRH
RONO )(
0
02
][][ on the difference between the two curves as the photochemical
age increased [Wang et al., 2013]. This feature was more pronounced for C3-C4 RONO2 at TW
(Figure 7.6) because of the lower values of [RONO2]0/[RH]0 resulting from the high mixing
ratios of propane and n-butane [Ling and Guo, 2014]. Figure 7.5 presents the relationships of
C1-C3 RONO2/RH to 2-C4H9ONO2/n-butane at TMS. The red dashed curves are pure
photochemical curves, while the blue solid curves are BIR curves with the lowest ratios of
RONO2/RH from 0000 to 0700 LT as the background intial ratio. Similarly, Figure 7.6 shows
the relationships of C1-C3 RONO2/RH to 2-C4H9ONO2/n-butane at TW.
158
Figure 7.5 Relationships of C1-C3 RONO2/RH with 2-C4H9ONO2/n-butane at TMS. The red
dashed curves were obtained based on zero initial concentrations of RH and RONO2 (pure
photochemical curves, PP), while the blue solid curves were obtained based on non-zero initial
levels (background initial ratio curves, BIR), with the lowest ratios of RONO2/RH from 0000
to 0700 LT.
159
Figure 7.6 Relationships of C1-C3 RONO2/RH with 2-C4H9ONO2/n-butane at TW. The red
dashed curves were obtained based on zero initial concentrations of RH and RONO2 (pure
photochemical curves, PP), while the blue solid curves were obtained based on non-zero initial
levels (background initial ratio curves, BIR), with the lowest ratios of RONO2/RH from 0000
to 0700 LT.
At TMS, the measured ratios of CH3ONO2/methane and C2H5ONO2/ethane to 2-C4H9ONO2/n-
butane were much higher than the ratios in the PP curves (Figure 7.5 (c) and (d)), with the
observed ratios larger than their theoretical ratios by factors of 5-25. As expected, the observed
trends approached the PP curves at a longer processing time, suggesting that the measured
ratios of C1-C2 RONO2/RH to 2-C4H9ONO2/n-butane were influenced by aged air masses
resulting from their relatively long atmospheric lifetimes and the slow photochemical reaction
rates of methane and ethane [Worton et al., 2010; Russo et al., 2010]. However, the difference
between the measured ratios and the predicted ratios of C1-C2 RONO2/RH to 2-C4H9ONO2/n-
butane in BIR curves was comparatively smaller, further confirming that there were other
160
sources contributing to ambient C1-C2 alkyl nitrates besides photochemical formation,
including the background levels of C1-C2 alkyl nitrates and their parent hydrocarbons [Wang
et al., 2013]. For example, the average CH3ONO2 and C2H5ONO2 mixing ratios at HT, a PRD
regional background site, were 10.4 ± 0.7 and 9.6 ± 0.7 pptv (see Chapter 4 for details),
respectively.
Regarding the C3 RONO2, the measured ratios of 1- and 2-C3H7ONO2/propane to 2-
C4H9ONO2/n-butane were closer to the ratios of the BIR curve than those of the PP curve at
TMS, further indicating the influence of background C3 RONO2 and their parent hydrocarbons.
However, the evolution of the measured ratios of C3 RONO2/RH to 2-BuON2/n-butane agreed
well with the predicted ratios of BIR and PP curves at TMS, indicating that secondary formation
from propane oxidation contributed significantly to the ambient C3 RONO2, including the
background C3 RONO2. Consistent with previous studies, the slopes of the observed ratios of
C3 RONO2/RH to 2-C4H9ONO2/n-butane were different from those in the PP and BIR curves
[Russo et al., 2010; Wang et al., 2013]. For example, the slopes of the observed ratios of 1- and
2-C3H7ONO2/propane to 2-C4H9ONO2/n-butane were 0.04 ± 0.01 and 0.26 ± 0.02, respectively,
while the slopes for the BIR curves were 0.02 ± 0.01 (PP curve: 0.02 ± 0.01) and 0.12 ± 0.01
(0.10 ± 0.01), respectively. This was reasonable as the difference in the number of samples and
distribution of data between the observed ratios and the ratios of PP and BIR curves,
particularly when the observed ratios were higher than the theoretical ones because of
significant influence of the background levels of RONO2 and RH [Russo et al., 2010; Wang et
al., 2013]. Therefore, to further investigate the influence of secondary formation and backround
mixing ratios on C3 RONO2 at TMS, the ratio of 1-/2-C3H7ONO2 was examined. Previous
studies reported that the theoretical ratio of 1-/2-C3H7ONO2 was the ratio between the yield for
1-C3H7ONO2 and 2-C3H7ONO2 formation, which was equal to the ratio of β1-C3H7ONO2/β2-
C3H7ONO2 (0.21) [Simpson et al., 2003; Wang et al., 2013]. If photochemical production was the
161
dominant source of 1-C3H7ONO2 and 2-C3H7ONO2, the observed ratios should be close to the
theoretical ones. Indeed, the slope of 1-C3H7ONO2 and 2-C3H7ONO2 at TMS was 0.19 (R2 =
0.86, p < 0.05), close to the theoretical ratio (0.21), confirming that photochemical production
from propane, including in-situ photochemical production and transport of photochemically-
formed C3 RONO2 in urban areas and/or during transit from urban areas to TMS, was the
dominant source of ambient C3 RONO2.
At TW, the comparison between the observed ratios of C1-C2 RONO2/RH to 2-C4H9ONO2/n-
butane and the ratios from the PP and BIR curves was consistent with that at TMS. However,
in terms of C3 RONO2, although the evolution of the measured ratios of C3 RONO2/RH to 2-
C4H9ONO2/n-butane followed the trends of the ratios in the PP and BIR curves, the measured
ratios of C3 RONO2/RH to 2-C4H9ONO2/n-butane at TW were further away from the PP/BIR
curves, about 2-3 times the ratios in the PP and BIR curves, implying additional sources of C3
RONO2 [Wang et al., 2013]. High emissions of propane provided sufficient precursors of C3
RONO2, and the lifetimes of 1-C3H7ONO2 and 2-C3H7ONO2 were long enough to sustain
relatively high levels at TW. To further investigate the influence of additional sources on the
distributions of C3 RONO2 at TW, equation 6-1 was used to fit the measured ratios of 1- and 2-
C3H7ONO2/propane to calculate the yield of C3 RONO2 (β). The average yields of 1- and 2-
C3H7ONO2 were 0.032 ± 0.004 and 0.22 ± 0.02, respectively, higher than the laboratory kinetic
values by factors of 4-9 [Kwok and Atkinson, 1995]. This confirms the presence of additional
emissions of C3 RONO2 at TW, including locally-emitted C3 RONO2 and/or secondary
formation other than the production pathway from propane to proxyl radical and PrONO2
[Reeves et al., 2007; Worton et al., 2010]. The slope of 1-C3H7ONO2 to 2-C3H7ONO2 at TW
was 0.15 (R2 = 0.80, p < 0.05), lower than the theoretical ratio of 0.21, further demonstrating
the influence of other significant sources on ambient mixing ratios of C3 RONO2 at TW.
162
7.2.2. Source apportionment of RONO2
As stated in Chapter 6, CH3Cl somewhat indicated biomass burning in Hong Kong, while
CHBr3, DMS and CH3I could be used as the tracers of oceanic emissions. Here, they were also
employed in source apportionment of RONO2 at TW, except for CHBr3 which were not
analyzed in this sampling campaign. Figure 7.7 presents the explained variations of species
(as a percentage of the species total) in the identified sources extracted by the PMF model. The
standard error in Figure 7.7 was obtained from a bootstrap analysis of the PMF model
simulation. The source profiles of the RONO2 and their parent hydrocarbons were altered
resulting from photochemical transformation during transport to the TMS site. Therefore, only
the data collected at the urban site were used for source apportionments of RONO2.
Factor 1 has high loadings of xylene isomers and ethylbenzene, with moderate level of toluene.
These species are known as emissions from solvent products (e.g. paints and coatings).
Therefore, this factor represents solvent usage. The contributions of this source to RONO2 are
very minor, ranging from 2.1±2.6% for 2-C4H9ONO2 to 3.6±2.7% for CH3ONO2, which can
be treated as residues in background and/or uncertainties in model calculation. The second
factor is distinguished by high loading of DMS, the most abundant biological sulfur compound
emitted by phytoplankton in the oceans [Dacey and Wakeham, 1986]. Hence, we define this
factor as oceanic emission, which is responsible for 13.6±0.9% of the total observed CH3ONO2.
The contribution decreased with the increase of carbon number, specifically 3.6±1.9%,
1.4±2.0%, 0.7±2.2% and 0.6±2.0% for C2, 2-C3, 1-C3 and 2-C4 RONO2, respectively. Factor 3
is assigned as gasoline exhaust, in view of the dominance of n/i-pentanes (see source profile of
gasoline exhaust in Hong Kong shown in Figure 7.8). In contrast, C2-C4 hydrocarbons are
mostly allocated to factor 4, which also contains moderate level of benzene. According to the
source profiles shown in Figure 7.8, propane, propene and n/i-butanes are indicators of LPG
vehicle emission, while diesel exhaust is characterized with high emissions of C2 hydrocarbons
163
and certain amount of benzene. Therefore, factor 4 stands for LPG and diesel exhausts. It is
notable that both factor 3 and factor 4 account for non-negligible fractions of C1-C4 RONO2.
Since vehicular exhausts have never been reported as direct sources of RONO2, the RONO2
apportioned to factor 3 and factor 4 may represent the background RONO2 at the sampling site.
Lastly, as a tracer of photochemical processes, O3 has the highest loading in factor 5, which
means that this factor is in association with the secondary formation of RONO2. In fact, the
proportions of RONO2 in this factor also conform to the characteristics of photochemical
contribution to RONO2. Specifically, the contribution increased from 39.6±2.7% for CH3ONO2
to 58.9±5.1% for 2-C4H9ONO2. Besides, we notice that this factor also explains moderate
levels of CO, ethane, benzene and CH3Cl, which are tracers of biomass burning and may also
exist in aged air masses due to their long lifetimes. Since O3 and RONO2 are formed with the
aging of air masses, it is reasonable that the presence of these long-lived species represents the
aged air masses. However, it cannot be confirmed whether RONO2 in this factor include some
contributions from biomass burning.
164
Figure 7.7 Explained variations of species in the identified sources extracted by the PMF model
for TW.
165
Figure 7.8 Source profiles of solvent usage, diesel exhaust, LPG usage and gasoline exhaust in
Hong Kong.
As mentioned earlier, regional transport and mesoscale circulation had a significant influence
on the distribution of air pollutants at TMS and TW [Guo et al., 2012, 2013]. By using the
Weather Research and Forecasting (WRF) model, air masses affected by mesoscale circulation
were distinguished from those affected by regional transport [Guo et al., 2013]. Nine sampling
days during the entire sampling period (24, 29-31 October, 1-3, 9 and 19 November) were
identified to be affected by mountain-valley breezes (they were also O3 episode days). Hence,
we divided the sampling period into two categories - “meso” and “non-meso” scenarios for
source apportionment analysis. The “meso” scenario included the nine O3 episode days with
apparent mesoscale circulation, while the “non-meso” scenario covered the rest of the sampling
days. Table 7.4 summarizes the mixing ratios of C1-C4 RONO2 attributable to secondary
formation and oceanic emission under the two scenarios (i.e. “meso” and “non-meso”),
respectively. It was found that secondary formation contributed the most to 2-C3H7ONO2 and
2-C4H9ONO2, differing from the oceanic emissions which mainly contains CH3ONO2,
regardless of the “meso” or “non-meso” scenarios.
166
Table 7.4 Mixing ratios of C1-C4 RONO2 in relation to secondary formation and oceanic
emission under “meso” and “non-meso” scenarios. (Unit: pptv).
Species “Meso” “Non-meso”
Secondary
formation
Oceanic
emission
Secondary formation Oceanic
emission
CH3ONO2 5.7±0.5 1.2±0.2 3.4±0.5 2.1±0.4
C2H5ONO2 7.9±0.8 0.3±0.1 4.7±0.8 0.6±0.1
2-
C3H7ONO2
16.1±1.5 0.3±0.1 9.5±1.5 0.5±0.1
1-
C3H7ONO2
2.6±0.3 0.02±0.004 1.6±0.3 0.03±0.01
2-
C4H9ONO2
23.2±2.2 0.2±0.03 13.7±2.2 0.3±0.05
Total C1-C4
RONO2
55.6±5.3 2.0±0.4 32.9±5.3 3.5±0.6
In comparison with those in the “non-meso” scenario, all the C1-C4 RONO2 derived from
photochemical formation increased substantially (p<0.05) in the “meso” scenario. The higher
contribution of secondary formation in the “meso” scenario at TW was mainly associated with
higher degree of photochemical reactions, as reflected from the higher O3 in “meso” scenario
(38.1±3.5 ppbv) than in “non-meso” scenario (23.0±2.8 ppbv). Indeed, the PBM-MCM model
simulation indicated that the average concentration of HOx (HOx = OH + HO2) during daytime
hours (0700-1800 LT) in the “meso” scenario was (2.5 ± 0.7) × 107 molecule/cm3, about twice
that of the “non-meso” scenario. Conversely, oceanic emission made higher contributions to
RONO2 in “non-meso” scenario than in “meso” scenario. To explain the variations of oceanic
167
emission, the meteorological conditions were compared between the two scenarios. The
temperature (27.9±1.0 °C) and relative humidity (67.8±3.1%) in “non-meso” scenario were
significantly (p<0.05) higher than those in “meso” scenario (24.6±0.6 °C and 51.4±3.2%,
respectively), when the southerly winds (90° < wind direction < 270°) were more frequent (60%
in “non-meso” scenario and 51.2% in “meso” scenario). In view of the fact that South China
Sea locates in the south of Hong Kong, the differences in meteorological conditions indicated
that more oceanic flows under higher temperature arrived at the site in “non-meso” scenario,
which explained the increased oceanic emissions of RONO2.
To understand the emission characteristics of RONO2 from biomass burning, linear regressions
were conducted between the observed RONO2 and CO in the nighttime samples collected at
TW, i.e. samples collected at 0:00, 03:00 and 21:00 local time (LT), as shown in Figure 7.9.
Since there was no photochemical reactions at night and oceanic emissions would not lead to
the synchronous variations of CO and RONO2, biomass burning might be a determinant factor
influencing the relationships between RONO2 and CO at night. The slopes, representing the
relative changes of RONO2 to CO, ranged from (3.3 ± 1.0) × 10-6 ppbv/ppbv for 1-C3H7ONO2
to (20.8 ± 7.3) × 10-6 ppbv/ppbv for 2-C4H9ONO2. These ratios were on the same magnitudes
with those in the source of biomass burning at Tai O (refer to Chapter 6), but significantly
higher than those reported in previous studies except for CH3ONO2. Though the nighttime
levels of RONO2 might be influenced by the residual RONO2 left from daytime photochemical
processes, we still cannot eliminate the possibility that the emission ratios of RONO2/CO in
biomass burning in Hong Kong and PRD region were higher than the reported values. More
studies are needed to figure out the emission characteristics of RONO2 in this region.
168
Figure 7.9 Linear regressions between RONO2 and CO in nighttime samples collected at TW.
7.2.3. Contributions of mesoscale circulation, in-situ formation and regional transport to RONO2 at TMS
Valley breezes transported freshly-emitted parent hydrocarbons and RONO2 from the urban
areas at the base of the mountain (TW) to the mountain summit (TMS) during daytime hours,
redistributing the ambient levels of RONO2 at TMS [Guo et al., 2013; Lam et al., 2013]. Except
for CH3ONO2, which had comparable levels in both “meso” and “non-meso” scenarios, the
mixing ratios of daytime C2-C4 RONO2 were all higher in “meso” scenario than those in “non-
169
meso” scenario (p < 0.05), with the average values of 14.21±0.79, 28.73±1.70, 4.67±0.29 and
40.21±2.79 pptv for C2H5ONO2, 2-C3H7ONO2, 1-C3H7ONO2 and 2-C4H9ONO2, respectively.
To quantify the influence of mesoscale circulation on the mixing ratios of RONO2 at TMS, a
moving box model coupled with master chemical mechanism (Mbox) was applied to the data
collected on the days influenced by mesoscale circulation (i.e, “meso” scenario) [Guo et al.,
2013]. The model was developed based on an idealized trajectory movement between TMS and
TW sites, with air pollutants transported from TW to TMS through the valley breeze during
daytime hours (0800-1700 LT) when photochemical formation of RONO2 was occurring,
contributing to their ambient levels at TMS. As such, the model was only constrained with the
observed daytime data at TW. On the other hand, the night-time downslope flow occurred
because of the mountain breeze after sunset until the next morning, and TMS was set as the
center of the box model, which was constrained by the data collected at TMS only for that
period [Lam et al., 2013].
Table 7.5 presents the average concentrations of C1-C4 RONO2 simulated by the Mbox model
at TMS, i.e., the values under the “meso” scenario. It should be noted that the comparison was
only made for daytime RONO2 (0800-1700LT), when the valley breeze occurred. The average
mixing ratios of CH3ONO2, C2H5ONO2, 1-C3H7ONO2, 2-C3H7ONO2 and 2-ButONO2 at
daytime hours estimated using the Mbox model were 9.97 ± 0.85, 7.38 ± 0.44, 3.08 ± 0.16,
18.7 ± 0.77 and 34.7 ± 3.14 pptv, respectively, accounting for 86%, 52%, 66%, 65% and 86%
of the observed values at TMS during the same period, respectively. These results demonstrate
that when there was mesoscale circulation, RONO2 levels at TMS were dominated by the photo-
oxidation of their parent hydrocarbons that originated from the urban site TW. Although the
mixing ratios of the parent hydrocarbons were lower at TMS, this is still one possible
explanation leading to the similar levels of RONO2 measured at the two sites.
170
For the “non-meso” scenario, the simulated levels of in-situ formation of CH3ONO2,
C2H5ONO2, 1-C3H7ONO2, 2-C3H7ONO2 and 2-C4H9ONO2 at TMS were 3.61 ± 0.48, 2.18 ±
0.29, 1.03 ± 0.13, 3.68 ± 0.45 and 10.9 ± 1.31 pptv, respectively, accouting for 18-42% of the
observed C1-C4 RONO2, indicating that other sources rather than local photochemical
formation made significant contributions to ambient levels of RONO2. As stated earlier, TMS
was a mountain site with sparse anthropogenic emissions nearby. However, the prevailing
synoptic northerly winds in “non-meso” scenario suggested possible regional sources of
RONO2 from inland PRD region to the mountain site. The impact of regional transport on the
variations of air pollutants at TMS for the days without mesoscale circulation, especially when
the prevailing winds were from the north with high speeds, was corroborated in Guo et al.
[2013]. By excluding the locally-formed RONO2 from their overall levels, the contribution of
regional sources to RONO2 was determined for TMS. The regional source contributions to
CH3ONO2, C2H5ONO2, 1-C3H7ONO2, 2-C3H7ONO2 and 2-C4H9ONO2 were 7.07 ± 0.50, 8.44
± 0.62, 2.11 ± 0.22, 16.86 ± 1.17, and 15.15 ± 1.49 pptv, respectively, accounting for 58-82%
of RONO2 measured at TMS. It is noteworthy that the regional RONO2 included influences
from all source categories (photochemical formation, biomass burning and oceanic emissions)
for the inland PRD region.
Table 7.5 Mixing ratios of C1-C4 RONO2 influenced by mesoscale circulation (“Meso”), in-
situ formation and regional transport at TMS (unit: pptv).
Scenario CH3ONO2 C2H5ONO2 1-C3H7ONO2 2-C3H7ONO2 2-C4H9ONO2
Meso 9.97 ± 0.85 7.38 ± 0.44 3.08 ± 0.16 18.7 ± 0.77 34.7 ± 3.14
In situ
formation
3.61 ± 0.48 2.18 ± 0.29 1.03 ± 0.13 3.68 ± 0.45 10.9 ± 1.31
171
Regional
transport
7.07 ± 0.50 8.44 ± 0.62 2.11 ± 0.22 16.86 ± 1.17 15.15 ± 1.49
7.3. Relationship between RONO2 and O3 RONO2 are mainly formed through the reaction of peroxy radical (RO2) and NO. However,
NO can be oxidized by RO2 to form NO2, which results in tropospheric O3 formation through
NO2 photolysis. Hence, investigating the relationship between RONO2 and O3 is useful for
evaluating the influence of RONO2 on O3 formation [Simpson et al., 2006]. Since
photochemical formation of O3 and RONO2 occurs during daytime hours, the relationship
between O3 and RONO2 is usually evaluated using the observed daytime data (i.e., 0900-1600
LT). In this study, the “oxidant” Ox (O3 + NO2) was considered to be a better representation of
O3 levels as it takes into account the effect of O3 titration by NO. Figure 7.10 shows the
correlation between Ox and the total RONO2 (∑RONO2) at daytime hours. Good correlations
were found at TMS (R2 = 0.63) and TW (R2 = 0.56) with the slopes of 0.67 and 0.47 ppbv/pptv,
respectively, suggesting that when 1 pptv of total RONO2 were formed from the reaction of
RO2 and NO, 0.67 and 0.47 ppbv of Ox could be simultaneously produced at TMS and TW,
respectively. The lower correlation coefficient at TW might be due to the interference of the
primarily emitted NO2 in Ox, which theoretically had no correlation with RONO2. Besides, the
inhomogeneous distributions of primary sources of RONO2 might also lead to different
correlations between RONO2 and Ox at different sites. The relatively higher slope at TMS than
at TW was owing to higher concentrations of HOx radicals and higher photochemical reactivity
of VOCs at TMS. Another reason was that the more aged air masses at TMS possessed higher
potentials to form O3 than RONO2, because the secondary products (such as aldehydes)
continued to stimulate O3 formation but were not capable of forming RONO2.
172
Figure 7.10 Correlation between Ox (O3 + NO2) and total RONO2 at (a) TMS and (b) TW.
7.4 Sub-conclusion Intensive field measurements of RONO2 and their parent hydrocarbons were conducted
concurrently at a mountain site (TMS) and an urban site (TW) at the base of the same mountain
in Hong Kong from September to November 2010. The levels of CH3ONO2, C2H5ONO2 and
2-C3H7ONO2 were slightly higher at TW than at TMS (p < 0.05), while the average mixing
ratios of 1-C3H7ONO2 and 2-C4H9ONO2 were comparable at the two sites (p > 0.05). However,
the levels of the parent hydrocarbons of RONO2 were lower at TMS, implying the complexity
of sources of RONO2. Receptor model and photochemical box model simulations found that
mesoscale circulation and regional transport had a significant impact on the levels of RONO2
at the two sites. At TW, secondary formation was the dominant contributor to RONO2, which
became more significant under the mesoscale circulations. At TMS, photo-oxidation of the
parent hydrocarbons from TW contributed 52-85% to the ambient levels of RONO2 on the days
with mesoscale circulations between the two sites. On the other hand, RONO2 from the inland
PRD region were responsible for 58-82% of the observed values at TMS on the days with
regional influence. The photo-oxidation of parent hydrocarbons from TW, regional transport
and higher oxidative capacity of the atmosphere at TMS (see Chapter 8) resulted in similar
values of RONO2 observed at the two sites. With regard to the secondarily formed RONO2, the
173
reaction of RO2 and NO was the prominent pathway at both sites. Moreover, the formation of
RONO2 made negative contributions to the O3 formation, with a reduction rate of 4.1 and 4.7
pptv O3 per pptv RONO2 at TMS and TW, respectively. The findings of this study will aid in
understanding the source contributions and photochemical formation pathways of RONO2 in
Hong Kong’s mountainous areas.
174
Chapter 8 Modelling of C1-C4 alkyl nitrate photochemistry and their impacts on O3 production in urban and suburban environments of Hong Kong
In the previous chapter, regional transport and meso-scale circulation were identified to be
important factors leading to comparable RONO2 levels at TMS to those at TW. The source
apportionment indicated that photochemical formation was a significant contributor to RONO2.
Based on these conclusions, in this chapter, a photochemical box model (PBM) incorporating
master chemical mechanism (MCM) was used to further explore the reasons for comparable
RONO2 at these two sites, even under the conditions that the levels of parent hydrocarbons
were quite different. In addition, the relationships between RONO2 and O3 are
comprehensively studied in this chapter. The PBM-MCM model well reproduced the observed
RONO2 at TW and TMS, with index of agreement (IOA) all higher than 0.65. Although levels
of the parent hydrocarbons and nitric oxide (NO) were significantly higher at TW than TMS,
the production of C2-C3 RONO2 was comparable to or even lower than at TMS, due to the lower
photochemical reactivity in the urban environment. Based on the profiles of air pollutants at
TMS, the formation of C2-C4 RONO2 was limited by NOx (VOCs) when TVOCs/NOx was
higher (lower) than 10.0/1 ppbv/ppbv. However, the threshold of this ratio decreased to 8.7/1
ppbv/ppbv at TW. For the formation of C1 RONO2, the NOx limited regime extended the ratio
of TVOCs/NOx to as low as 2.4/1 and 3.1/1 ppbv/ppbv at TMS and TW, respectively. RONO2
formation led to a decrease of simulated O3, with reduction efficiencies (O3 reduction/RONO2
production) of 4-5 pptv/pptv at TMS, and 3-4 pptv/pptv at the urban site. On the other hand,
the variations of simulated O3 induced by RONO2 degradation depended upon the regimes
controlling O3 formation and the relative abundances of TVOCs and NOx.
175
8.1 Modelling of C1-C4 RONO2
8.1.1 Overview of RONO2 sources
As discussed in chapter 7, the air quality at TMS was significantly influenced by regional
transport and mesoscale circulation. Figure 8.1 shows the distributions of wind fields, i.e., wind
direction (WD) and wind speed (WS), and their correlations with RONO2 precursors at TMS
during the sampling campaign. North winds (0°<WD<90° and 270°<WD<360°) were much
more frequent than south winds (90°<WS<180°) at TMS (82.8% and 17.2%, respectively).
While the mixing ratios of NO and NO2 were comparable (p>0.05) for north and south winds,
C1-C4 n-alkanes were more abundant (p<0.05) under north versus south winds, suggesting the
transport of RONO2 precursors from the inland PRD to TMS. During this study the average
wind direction at TMS switched from northeast at night to north during the day [Guo et al.,
2013], implying that regional transport was constant throughout the day or night. Besides
regional transport, our previous paper [Guo et al., 2013] clearly confirmed the mesoscale
circulations between TMS and TW through the good inter-site correlations of SO2 and CO; good
reproduction of O3 at TMS with precursors at TW to initialize the model; and the simulation of
mountain-valley breezes by Weather Research Forecast (WRF). Mesoscale circulations were
identified on the sampling days of October 24 and 29-31, and November 1-3, 9 and 19, when
the valley breeze brought the urban plume at the foot of the mountain to the mountain site in
the daytime, while the mountain breeze drove the mountain air to the urban site at night. In
addition, the air pollution at TMS was also partially dominated by in-situ photochemistry. As
presented in chapter 7, we apportioned the observed RONO2 at TMS to the sources of regional
transport, mesoscale circulation and in-situ formation, as shown in Table 8.1. Although
regional transport and mesoscale circulation made considerable contributions to RONO2 at
TMS, in-situ formation cannot be neglected, which was the main focus of this study.
176
Figure 8.1. Distributions of wind fields and their correlations with precursors of RONO2 at
TMS. (a) methane, (b) ethane, (c) propane, (d) n-butane, (e) NO and (f) NO2.
Table 8.1 Contributions of regional transport, mesoscale circulation and in-situ formation to
individual RONO2 levels at TMS (unit: pptv).
Sources CH3ONO2 C2H5ONO2 2-
C3H7ONO2
1-C3H7ONO2 2-C4H9ONO2
Regional
transport
7.67 ± 0.50 8.44 ± 0.62 16.86 ± 1.17 2.11 ± 0.22 15.15 ± 1.49
177
Mesoscale
circulation
9.97 ± 0.85 7.38 ± 0.44 18.7 ± 0.77 3.08 ± 0.16 34.7 ± 3.14
In-situ
formation
3.61 ± 0.48 2.18 ± 0.29 3.68 ± 0.45 1.03 ± 0.13 10.9 ± 1.31
It is well known that photochemical formation, biomass burning and oceanic emission are the
main sources of RONO2 [Atlas et al., 1993; Bertman et al., 1995; Simpson et al., 2002]. Using
Positive Matrix Factorization (PMF) model, the sources of RONO2 at TW were identified in
Chapter 7. It was found that secondary formation made comparable contributions to RONO2.
However, it is noteworthy that the secondary formation resolved from PMF was not exactly
the same as the in-situ formation, because the secondarily formed RONO2 in background and
regional air masses were also included in the secondary formation source.
8.1.2 Model construction
A PBM-MCM model was developed to simulate RONO2. Master Chemical Mechanism (MCM)
is an explicit chemical mechanism, which has been successfully used in photochemical
simulation in Hong Kong and many other regions of the world [Saunders et al., 2003; Lam et
al., 2013; Ling et al., 2014]. The latest version of MCM (MCM v3.3) includes 17,242 reactions
and 5,836 species [Jenkin et al., 2015]. Reactions of biogenic VOCs including limonene and
myrcene, and RONO2 were added into MCM by our team. In particular, the model describes
reactions of more than 100 RONO2 species. The main formation pathways of RONO2 that are
considered are RO2+NO (reaction 8.1) and RO+NO2 (reaction 8.2), while RONO2 degradation
is presented as photolysis (reaction 8.3) and OH initiated oxidation (reaction 8.4).
RO2 + NO → RONO2, α, k1 (reaction 8.1)
RO + NO2 → RONO2, k2 (reaction 8.2)
RONO2 + hv → RO + NO2, k3 (reaction 8.3)
178
RONO2 + OH → RO + NO2 + products, k4 (reaction 8.4)
The PBM-MCM model simulates the photochemistry within a well-mixed boundary layer air
parcel. The concentrations of air pollutants are presented with ordinary differential equations
(ODEs), as shown in Equation 8.1.
𝑑𝐶𝑖
𝑑𝑡= 𝑃 − 𝐿 · 𝐶𝑖 −
𝑉𝑖𝐶𝑖
ℎ− (𝐶𝑖 − 𝐵𝑖)
𝑙𝑑ℎ
ℎ𝑑𝑡 (Equation 8.1)
Where 𝐶𝑖 denotes the concentration of species (i), 𝑃 is the instantaneous production and 𝐿
represents the loss rate including the chemical consumptions and photolysis. 𝑉𝑖 describes the
dry deposition velocity of species (i) over the mixed layer with the height of ℎ. The mixed layer
height (ℎ) varied from 300m at night to 1300m in daytime. 𝐵𝑖 is the background concentration
in free troposphere. l and d are the length and width of the domain, which are both 7 km in this
study. The FACSIMILE program (http://mcm.leeds.ac.uk/MCM/tutorial_sec1.htt) was used to
solve the ODEs, the integrations of which were output as the simulated concentrations.
The photochemical formation and degradation pathways of RONO2 in the PBM-MCM model
are represented as reactions 8.1-8.4. The reactions between RO2 and NO (reaction 8.1) and RO
and NO2 (reaction 8.2) have been well recognized as the formation pathways of RONO2
[Atkinson et al., 1982b; Arey et al., 2001]. The reaction coefficient (k1) is based on the
recommendations by Eberhard and Howard [1996, 1997], Eberhard et al. [1997] and Atkinson
et al. [1999]. Specifically, it is presented as a function of temperature (T) as 2.3×10-12×e(360/T),
2.55×10-12×e(380/T), 2.7×10-12×e(360/T), 2.9×10-12×e(350/T) and 2.7×10-12×e(360/T) in unit of cm3
molecule-1 s-1 for the formation of C1, C2, 2-C3, 1-C3 and 2-C4 RONO2, respectively. The
branching ratio (α) was an important factor influencing the production of RONO2, which were
acquired from previous studies [Lightfoot et al., 1992; Flocke et al., 1998b] or calculated
according to the formulas recommended by Carter and Atkinson [1985]. For C1 RONO2,
branching ratios of 0.00015, 0.0003, 0.001, 0.003, 0.0041 and 0.005 were examined and
considered. The one that resulted in the highest IOA between the simulated and observed C1
179
RONO2 was adopted. However, since branching ratio data for C2-C4 RONO2 were rather
limited, the values calculated using Equations 8.2-8.4 [Carter and Atkinson, 1985] were used
as the branching ratios, which were 0.0094, 0.048, 0.019 and 0.085 for C2, 2-C3, 1-C3 and 2-
C4 RONO2, respectively.
α = [𝑌0
300[𝑀](𝑇
300)
−𝑚0
1+𝑌0
300[𝑀](𝑇
300)
−𝑚0
𝑌∞300(
𝑇300
)−𝑚∞
]𝐹𝑧 (Equation 8.2)
z = {1 + [log𝑌0
300[𝑀](𝑇
300)
−𝑚0
𝑌∞300(
𝑇
300)
−𝑚∞ }]
2
}−1 (Equation 8.3)
𝑌0300 = 𝛽𝑒𝛾𝑛 (Equation 8.4)
where T is the temperature (K); M represents the number of molecules (molecules/cm3) and n
is the carbon number in RO2. The values of the constants β (1.95 × 10-22 cm3/molecule), 𝛾
(0.947), 𝑌∞300 (0.435), 𝑚0 (2.99), 𝑚∞ (4.69) and 𝐹 (0.556) are all from Carter and Atkinson
[1985]. On the basis of the calculated results, the branching ratios for the primary and tertiary
RO2 radicals are calibrated by a factor of 0.4 and 0.25, respectively.
The reaction coefficient (k2) was set according to the International Union of Pure and Applied
Chemistry (IUPAC) (accessible at http://iupac.pole-
ether.fr/htdocs/show_datasheets.php?category=Gas-phase+organics%3A+RO and
http://iupac.pole-ether.fr/htdocs/summary/vol2_summary.pdf), specifically 1.5×10-11, 2.8×10-
11, 3.4×10-11, 3.6×10-11 and 8.6×10-12×e(400/T) in units of cm3 molecule-1 s-1 for the formation of
C1, C2, 2-C3, 1-C3 and 2-C4 RONO2, respectively. Photolysis (reaction 3) and OH initiated
oxidation (reaction 4) are the main chemical losses of RONO2. The photolysis rates in the
MCM were treated as a function of solar zenith angle, which is fully described in Jenkin et al.
[1997]. The parameterizations of the functions were based on the recommendations or averages
of previous studies, such as Roberts and Fajer [1989], Zhu and Ding [1997], and the IUPAC
recommendations of 2011. The reaction coefficient (k4) was adopted from the IUPAC
180
recommendations of 2011, and the preferred values were set as 4.0×10-13×e(-845/ T), 6.7×10-13×e(-
395/T), 6.2×10-13×e(-230/T), 5.8×10-13 and 8.6×10-13 in units of cm3 molecule-1 s-1 for C1, C2, 2-C3,
1-C3 and 2-C4 RONO2, respectively. The reasonableness of these settings can be found at the
MCM website (http://mcm.leeds.ac.uk/MCM/).
Biogenic chemistry about isoprene, α-pinene and β-pinene in MCM was developed by
Saunders et al. [2003], which has been repeatedly confirmed robust and reliable [Poschl et al.,
2000; Pinho et al., 2005]. In the present version of MCM, the degradations of limonene and
myrcene by the oxidants like OH, O3 and NO3 were updated. The initial reaction rates with
OH, O3 and NO3, and the subsequent degradations of the first generation products are
approximated with the corresponding reactions of the most structurally similar monoterpenes,
because the structurally similar molecules generally have similar chemical properties. In total,
around 450 species and 1,300 reactions were added in the degradation scheme of MCM for the
two species. While the chemistry of limonene and myrcene has not been validated by chamber
experiments or field measurements, MCM with the inclusion of these reactions performed well
in describing O3 formation and the recycling of oxidative radicals in Hong Kong [Ling et al.,
2014]. This suggested that the added reactions of limonene and myrcene were suitable or at
least did not reduce the reliability of the original MCM. In this study, the observed mixing
ratios of limonene and myrcene were in the range of 2-815 pptv and 0.6-62 pptv at TMS, and
in the range of 3-787 pptv and 0.8-51 pptv at TW, respectively. To examine the influence of
limonene and myrcene chemistry on the simulated RONO2, a base scenario with the observed
limonene and myrcene as input and a constrained scenario without the chemical reactions and
the input of limonene and myrcene were simulated for RONO2 production. The impacts of
limonene and myrcene chemistry on RONO2 production were determined from the difference
of the simulated RONO2 between the base scenario and the constrained scenario. It was found
that the observed limonene and myrcene only caused a very minor variations in RONO2
181
production, which was 0.03±0.04% (0.8±0.1%), -0.7±0.2% (0.3±0.04%), -0.4±0.1%
(0.2±0.03%), -0.5±0.1% (0.2±0.03%) and -0.6±0.2% (0.4±0.1%) for CH3ONO2, C2H5ONO2,
2-C3H7ONO2, 1-C3H7ONO2 and 2-C4H9ONO2 at TMS (TW), respectively. Therefore, it can be
concluded that the add limonene and myrcene chemistry made negligible influence on the
modeling of C1-C4 RONO2 in this study.
In addition to the chemical reactions, many modules were compiled in the PBM-MCM model.
For example, the photolysis rate module enables us to calibrate the photolysis rates of many air
pollutants. The Tropospheric Ultraviolet and Visible Radiation (TUVv5) model, which
considers actual location and modelling time periods, is used to calibrate the photolysis rates
[Madronich and Flocke, 1997]. Moreover, the concentrations of air pollutants can be specified
to initiate the model in the initial concentration module. This is important, because the
background RONO2 existed prior to photochemical reactions are generally non-negligible due
to their long lifetimes. The dry deposition module considers the dry deposition of air pollutants,
which are parameterized as an average deposition rate within the height of the mixed layer
(HMIX).
Since this chapter mainly focused on the in situ photochemistry of RONO2, the on-site
observations of 41 hydrocarbons, 10 halocarbons, 6 OVOCs, 5 inorganic trace gases and 2
meteorological parameters (temperature and relative humidity) from 08:00 to 19:00 at each site
were used as the model input to constrain the modeling (the hourly observed RONO2 at 08:00-
19:00 were not input to constrain the simulated RONO2). Moreover, to initiate the modeling,
observed concentrations of the aforementioned species including RONO2 at 07:00 were set as
initial conditions. These initial conditions were expected to be significantly influenced by
regional transport and mesoscale circulation at this mountainous site (because of low on-site
emissions). Note that these species were not constrained at their initial values, but were allowed
to vary over time. Since the daily values of these pollutants varied, different initial
182
concentrations for each pollutant were used for the simulation of each day. The integration of
simulated RONO2 within each hour automatically given by the model was compared with the
observed RONO2 to validate the model performance. The simulations were carried out only
during the daytime hours (07:00-19:00). In terms of dry deposition, Zhang et al. [2002]
indicated that the dry deposition velocity for organic nitrates ranged from 0.03 to 0.56/HMIX
cm/s. Within this range, deposition rates of 0.03, 0.13, 0.23, 0.33, 0.43 and 0.53/HMIX cm/s
were examined for C1-C4 RONO2 in this study (step=0.1/HMIX cm/s). Dry deposition rates
were also available for other species such as O3 and its precursors, in line with the settings in
previous studies [Saunders et al., 2003; Lam et al., 2013]. Overall, based on the observed
mixing ratios of air pollutants, including RONO2 precursors, the PBM-MCM model simulated
RONO2 in different scenarios with changes of branching ratios and dry deposition rates, and
consideration of initial conditions.
The model uncertainty was discussed and roughly estimated with the mean root square error
method [Willmott, 1982]. The output uncertainty of PBM-MCM model derived from two parts,
i.e. uncertainties of the input species and uncertainty inherent to the chemical mechanism. In
this study, 63 VOCs, 5 trace gases and 2 meteorological conditions (temperature and relative
humidity) were used to construct the model. The measurement accuracies for VOCs ranged
from 2% for methane to 20% for dimethyl sulfide and some other compounds. Accuracies for
the measurement of SO2, CO, NO, NO2 and O3 were 10%, 1%, 10%, 10% and 5%, respectively.
Details about the accuracy for each species can be found in Simpson et al. [2010]. The
uncertainty of temperature, relative humidity and the chemical mechanism (MCM) was
roughly estimated as 5%, 5% and 10%, respectively. Mean root square error (MRSE) was
calculated for the accuracies of the input species and uncertainty of MCM, using Equation 8.5
[Willmott, 1982], which was treated as the uncertainty of the PBM-MCM model:
MRSE = √∑ 𝑋2𝑛
1
𝑛
2 (Equation 8.5)
183
where X represents the individual uncertainty of each component (total of n). According to the
calculation, MRSE equals 13%. Therefore, we estimated the uncertainty of the PBM-MCM
model was around 13%.
8.1.3 Modelling of CH3ONO2
This chapter mainly focused on the in-situ formation of RONO2 at TMS and TW. However, the
regional transport and mesoscale circulation were also partially considered based on the
following settings: i) RONO2 and other air pollutants measured at 07:00 on each day were used
to initiate the model. The initial RONO2 at 07:00, which accounted for ~85% of the total
simulated RONO2 (see Figure 8.4), consisted of the regionally transported and locally
circulated RONO2; and ii) the regionally transported and locally circulated fractions of RONO2
precursors were included in the hourly measured data, which were used to construct the model.
Despite the above settings, the impacts of regional transport and mesoscale circulations were
not fully simulated in the model, demonstrated by the fact that the simulated RONO2 mixing
ratios were generally lower than the observed levels during O3 episodes when these impacts
were significant according to the following discussions, most likely due to the insufficient
consideration of regional transport and local circulation. However, the simulations of in-situ
formation of RONO2 should not be significantly influenced.
Figure 8.2(a) shows the average in-situ production of CH3ONO2 at TMS as a function of the
branching ratio, without consideration of initial concentrations and dry deposition. Noticeably,
the CH3ONO2 production linearly increases with increasing branching ratio (CH3ONO2 mixing
ratio in pptv = (4400×branching ratio) + 1.4). A branching ratio of approximately 0.0023 was
determined to match the observed CH3ONO2 (11.3 pptv). This branching ratio was within the
range of 0.00015 to 0.005 as reported earlier [Carter and Atkinson, 1985; Lightfoot et al., 1992;
Flocke et al., 1998b]. However, the initial mixing ratio (8.8 pptv) was not considered in Figure
8.2(a), which should also be a part of the observed CH3ONO2 even though it was subject to
184
degradation and dry deposition. Even taking into account the degradation (OH initiated
oxidation and photolysis) and dry deposition (rate = 0.13/HMIX cm/s), the average residual of
the initial CH3ONO2 was still 7.5 pptv. Based on this value, photochemically formed
CH3ONO2 was about 3.8 pptv, which corresponded to a branching ratio of 0.00055, also within
the range of 0.00015-0.005. However, this value was calculated based on model simulation
rather than laboratory experiment and has never been reported in previous studies. More
importantly, factors other than photochemical reactions (such as regional transport and
mesoscale circulation) might influence the determination of this value. Therefore, 0.00055 was
only treated as a rough estimate of the branching ratio, and we preliminarily accepted a
branching ratio of 0.0003, reported by Flocke et al. [1998b] and adopted by Simpson et al.
[2006], which was the closest to 0.00055 among the examined values.
Figure 8.2 Modelled average CH3ONO2 as a function of (a) branching ratio (no initial or dry
deposition) and (b) deposition rate (branching ratio = 0.0003, and the initial CH3ONO2 was set
as the values measured at 07:00 (LT) at TMS for each day.
To validate the suitability of the branching ratio, we conducted a set of comparative simulations
with the branching ratios tested above. The initial concentrations were taken from the observed
CH3ONO2 at 07:00 on each day, and the mean of 0.03/HMIX-0.53/HMIX was set as the dry
deposition rate. Table 8.2 shows the IOAs between the observed and simulated CH3ONO2 with
different branching ratios. Noticeably, the best agreement was acquired when the branching
185
ratio of 0.0003 was applied. Furthermore, the observed and simulated CH3ONO2 with
branching ratio of 0.0003 were compared on several selected days with the wind speeds less
than 2 m/s and typical patterns of photochemically formed RONO2 (peak observed in earlier
afternoon), as presented in Figure 8.3. It was found that the variations of CH3ONO2 were well
simulated, with IOA reaching 0.75 at TMS and 0.86 at TW. The average positive and negative
biases of the simulations were 18.3% and -12.1% at TMS, and 14.3% and -13.5% at TW,
respectively. Under the assumption that the daytime variation of CH3ONO2 was exclusively
caused by in-situ formation on the selected days, the biases of simulations could be treated as
the errors in branching ratio, because CH3ONO2 production linearly correlated with the
branching ratio (Figure 8.2). By setting the initial concentration and deposition velocity as zero,
the average contribution of photochemical formation to the total simulated CH3ONO2 on the
selected days was determined to be ~44% and ~34% at TMS and TW, respectively. Therefore,
to account for the maximum biases of the simulated CH3ONO2 (18.3% at TMS and 14.3% at
TW), the uncertainty of the branching ratio was ~42% at both sites. Further consideration of
the model uncertainty (~13%) with the mean root square error method resulted in the
uncertainty of the branching ratio of less than 50%. Therefore, 0.0003 was identified as the
most appropriate branching ratio for CH3O2 + NO → CH3ONO2, with the error less than 20%.
It is noticeable that 0.00055 was beyond the range of 0.00015-0.00045 (0.0003 with uncertainty
of 50%). This discrepancy might be due to the fact that the determination of 0.00055 was more
influenced by the factors other than photochemical reactions, while 0.0003 was determined
based on the simulations on the selected days when in situ photochemical formation of
CH3ONO2 was more significant.
Table 8.2 IOAs between the simulated CH3ONO2 with different branching ratios and the
observed CH3ONO2 at TMS and TW.
186
Branching ratio 0.00015 0.0003 0.001 0.003 0.0041 0.005
TMS 0.63 0.66 0.54 0.26 0.20 0.16
TW 0.61 0.65 0.51 0.30 0.24 0.21
Figure 8.3. Comparison between the simulated and observed CH3ONO2 at (a) TMS and (b) TW
on the selected days (branching ratio = 0.0003).
Although 0.0003 was proposed and adopted in some previous studies [Flocke et al., 1998b;
Simpson et al., 2006], other branching ratios over a wide range (0.00015-0.01) have also been
used as the branching ratio of CH3O2 reacting with NO [e.g. Carter and Atkinson, 1989a;
Scholtens et al., 1999]. Here, the determination of the branching ratio of 0.0003, as well as the
uncertainty, may re-constrain the impacts of CH3ONO2 on global oxidative capacity of the
atmosphere.
Furthermore, by considering dry deposition, Figure 8.2(b) presents the modelled CH3ONO2
with the branching ratio of 0.0003 and dry deposition velocities of 0.03/HMIX, 0.13/HMIX,
0.23/HMIX, 0.33/HMIX, 0.43/HMIX and 0.53/HMIX cm/s. The modelled CH3ONO2
decreased linearly with increasing dry deposition velocity (CH3ONO2 mixing ratio in pptv = -
4.5 × deposition rate + 11.6). A dry deposition velocity of 0.07/HMIX was determined to best
reproduce the observed CH3ONO2. As such, the branching ratio of 0.0003 and dry deposition
187
velocity of 0.07/HMIX cm/s were treated as the most appropriate settings for CH3ONO2
simulation.
With these settings, the factors influencing the simulated CH3ONO2 were sequentially
considered. Figure 8.4 shows the CH3ONO2 simulated in different scenarios at TMS, i.e., (i)
“reaction”, (ii) “reaction + initial” and (iii) “reaction + initial + dry deposition”. Scenario (i)
only considered the formation and degradation reactions of CH3ONO2, while the initial
concentrations and dry deposition were progressively considered in scenarios (ii) and (iii). The
modeled CH3ONO2 in scenario (i) was typically bell-shaped on a diurnal basis, coincident with
the characteristics of photochemical reactions. However, the mean modelled CH3ONO2 (2.6 ±
0.3 pptv) was much lower than the observed average (11.3 ± 0.3 pptv). By introducing the
initial conditions, the modelled CH3ONO2 in scenario (ii) increased to a comparable level (11.7
± 0.3 pptv) to the measurements, in line with the finding that background initial concentrations
are an important constituent of the observed RONO2 [Ling et al., 2016]. Further consideration
of dry deposition in scenario (iii) resulted in a slight decrease of the modelled CH3ONO2 to
11.0 ± 0.3 pptv, which best agreed with the observed CH3ONO2. By subtracting the modelled
CH3ONO2 in scenario (i) from scenario (ii) and that in scenario (ii) from scenario (iii), the
respective contributions of the processes, including reaction, initial conditions and dry
deposition to the total modelled CH3ONO2, were determined to be 21.5 ± 1.8%, 85.1 ± 2.0%
and -6.6±0.3% (negative contribution means removal of CH3ONO2). It is noteworthy that the
initial concentrations included both primarily emitted RONO2 and residues of secondarily
formed RONO2 in background and regionally transported air masses.
188
Figure 8.4 Comparisons between the measured and modelled CH3ONO2 in different scenarios
at TMS. The O3 episode days are highlighted in red.
8.1.4 Model validation
C2-C4 RONO2 were simulated with the branching ratios calculated according to Carter and
Atkinson [1985, 1989], and the dry deposition velocities identical to that for CH3ONO2. Table
8.3 lists the model settings that best reproduced the magnitudes and patterns of the observed
RONO2 at TMS and TW (shown in Figures 8.5-8.6). It is noteworthy that the dry deposition
velocities set for C2-C4 RONO2 may be higher than the actual values, as studies (Russo et al.,
2009) indicated lower deposition velocities of ≥C2 RONO2 than CH3ONO2. However, due to
the very minor influence of dry deposition on the simulated RONO2 (see Figure 8.5) and the
unavailability of validated deposition velocities for C2-C4 RONO2, we adopted the dry
deposition velocity of 0.07/HMIX for C2-C4 RONO2 in this study, same as for CH3ONO2.
Table 8.3 Model settings for the simulations of C1-C4 RONO2 as well as IOAs between the
simulated and measured RONO2 at TMS and TW.
CH3ONO2 C2H5ONO2 1-C3H7ONO2 2-C3H7ONO2 2-C4H9ONO2
Branching ratio 0.0003 0.0094 0.019 0.048 0.085
Dry deposition 0.07/HMIX 0.07/HMIX 0.07/HMIX 0.07/HMIX 0.07/HMIX
IOA at TMS 0.67 0.72 0.72 0.72 0.72
189
IOA at TW 0.66 0.70 0.69 0.67 0.73
HMIX is the time-dependent mixed layer height, which varies from 300 to 1300 m.
190
Figure 8.5 Comparisons between the observed and simulated C1-C4 RONO2 at TMS, where the
simulated value uses reaction + initial + dry deposition (see Figure 8.4 and Table 8.2). O3
episodes are highlighted in red.
191
Figure 8.6 Comparisons between the observed and simulated C1-C4 RONO2 at TW, where the
simulated value uses reaction + initial + dry deposition (see Figure 8.4 and Table 8.2). O3
episodes are highlighted in red.
192
Overall, the simulated RONO2 agreed well with the measurements (Index of Agreement is
discussed below). However, the morning peaks of RONO2 (e.g. September 28, October 8 and
23, and November 1 and 2) were not well reproduced by the model. Since in situ photochemical
formation could not be the main source of RONO2 in the morning when solar radiation was
weak, the discrepancies between modelling and observation were most likely to be caused by
direct emissions and/or regional transport, which were not considered in the model. In addition,
the modelled RONO2 levels were generally ~50% lower than the observations on O3 episode
days (October 23-24, 29-31, and November 1-3, 9, 19). Methyl chloride (CH3Cl) levels at both
TMS (episode: 1100 ± 33 pptv; non-episode: 926 ± 27 pptv) and TW (episode: 1116 ± 32 pptv;
non-episode: 1031 ± 45 pptv) increased noticeably (p<0.05) during O3 episodes, so did
levoglucosan in fine particles (84.8±27.8 and 31.6±18.5 ng/m3 during episode and non-episode
at TMS, respectively). These suggested emissions of RONO2 from biomass burning.
Furthermore, the frequency of northerly winds was higher during O3 episodes (78% at TMS
and 29% at TW) than during non-O3 episodes (51% at TMS and 21% at TW). In view of severe
photochemical pollution in the adjacent inland PRD cities and increased transport of secondary
pollutants from the inland PRD to Hong Kong during O3 episodes [Lam et al., 2005; Guo et
al., 2009], regional transport might also contribute to the higher observed RONO2 on episode
days. Additionally, it was found that mesoscale circulation made higher contribution to RONO2
at TMS during O3 episodes (chapter 7). An exception was CH3ONO2 at TW on November 19
when the modeled CH3ONO2 remarkably exceeded the measured values (Figure 8.6). This
overestimation was believed to be caused by the abnormally high aromatic levels on that day
(30.2 ± 23.4 ppbv, compared to the average of 4.9 ± 0.6 ppbv over the whole sampling period
excluding that day). Briefly, the photochemical degradation of aromatics generated CH3O2 and
CH3O in the model. Since the box model treated air masses as a steady state without the
consideration of advection and diffusion, the concentrations and reaction times of these
193
precursors of CH3ONO2 were more significantly overestimated on this day than those in normal
periods, leading to overestimation of CH3ONO2.
To quantitatively evaluate the simulations, IOA between the simulated and observed RONO2
was calculated (Table 8.3). The IOA ranged from 0.67 to 0.72 and 0.66 to 0.73 for RONO2
simulations at TMS and TW, respectively. Given that other sources (e.g., biomass burning and
oceanic emission) and regional transport impact were not considered in the model, the IOAs
indicated that the simulations were acceptable.
8.2 Secondary RONO2 formation
8.2.1 RONO2 formation at TMS and TW
Based on the above settings, the in-situ production of RONO2 (referred to as secondary RONO2
hereafter) was simulated without consideration of initial conditions and dry deposition, as
summarized in Table 8.4. Also shown are the concentrations of parent hydrocarbons,
corresponding RO2 radicals, NO and NO2.
Table 8.4 Average mixing ratios of parent hydrocarbons, NOx and secondary RONO2 at TMS
and TW. Error bars represent 95% C.I.s.
TMS TW
CH4 (ppbv) 1950±7 1970±7
C2H6 (pptv) 1848±76 2144±81 *
C3H8 (pptv) 1123±71 3343±331 *
n-C4H10 (pptv) 887±84 4131±361 *
NO (ppbv) 3.5±0.1 26.9±2.9 *
NO2 (ppbv) 8.7±0.8 31.6±3.1 *
CH3O2 (molecules/cm3) (3.1±0.4)×107 * (0.6±0.3)×107
194
C2H5O2 (molecules/cm3) (2.0±0.2)×106 * (0.3±0.1)×106
2-C3H7O2 (molecules/cm3) (4.6±0.5)×105 * (0.7±0.2)×105
1-C3H7O2 (molecules/cm3) (4.0±0.6)×105 * (0.3±0.1)×105
2-C4H9O2 (molecules/cm3) (7.1±0.8)×105 * (1.7 ± 0.5)×105
Secondary CH3ONO2 (pptv) 2.7±0.3 4.8±1.0 *
Secondary C2H5ONO2 (pptv) 4.0±0.4 3.6±0.7
Secondary 2-C3H7ONO2 (pptv) 5.2±0.5 4.5±0.7
Secondary 1-C3H7ONO2 (pptv) 1.1±0.1 * 0.7±0.1
Secondary 2-C4H9ONO2 (pptv) 13.5±1.4 17.6±2.4 *
* Significant difference between the two sites (p<0.05). Bolded are species with observed
values, and the rest are simulated values.
The measured mixing ratios of parent hydrocarbons and NOx (both NO and NO2) were
significantly higher at TW than at TMS (p<0.05). Likewise simulated C1 and 2-C4 RONO2
levels at TW were significantly higher than at TMS (p<0.05), while the simulated C2 and C3
RONO2 levels at TW were comparable to or even lower than those at TMS. To explore the
reasons for these differences, the relative contributions of RO2+NO and RO+NO2 were
quantified (Table 8.5). Briefly, the two pathways were switched off in turn. The simulated
RONO2 was subtracted from that simulated in base scenario with both pathways switched on.
In this way, RONO2 produced by the each pathway was obtained. The pathway of RO2+NO
dominated the formation of C2-C4 RONO2 at both sites. In contrast, the reaction of RO+NO2
made considerable contributions to CH3ONO2 (mean±95% confidence interval (C.I.): 2.7±0.3
pptv or 41.9±5.9% at TMS and 4.8±1.0 pptv or 76.2±15.7% at TW). In addition to higher CH4
levels, the more abundant secondary CH3ONO2 at TW was likely because that NO2 at TW
(31.6±3.1 ppbv) was significantly higher than that at TMS (8.7±0.8 ppbv) (p<0.05). Indeed,
following suggestions that RO+NO2 could be an important pathway for CH3ONO2 formation
195
in polluted environments [Flocke et al., 1998a; Simpson et al., 2006], Archibald et al. [2007]
confirmed that this pathway becomes important at about 10 ppb of NO2, and dominant at about
35 ppb, based on MCM simulations for European conditions.
Table 8.5 Relative contributions (%) of the RO2+NO and RO+NO2 pathways to RONO2 at
TMS and TW. Error bars represent 95% C.I.s.
RONO2 TMS TW
RO2+NO RO+NO2 RO2+NO RO+NO2
CH3ONO2 58.1±6.8 41.9±5.9 23.8±4.8 76.2±15.7
C2H5ONO2 99.0±13.4 1.0±0.2 95.8±24.2 4.2±1.2
2-C3H7ONO2 99.6±12.7 0.4±0.1 98.9±19.4 1.1±0.2
1-C3H7ONO2 99.5±12.4 0.5±0.1 98.1±17.8 1.9±0.4
2-C4H9ONO2 99.9±14.1 0.10±0.02 99.7±18.4 0.3±0.1
For C2-C3 RONO2, although the measured parent hydrocarbons were less abundant at TMS
than at TW, the simulated concentrations of RO2 radicals were all remarkably higher under low
NOx conditions but still in VOC-limited regime (as discussed below), leading to comparable
(for C2H5ONO2 and 2-C3H7ONO2 ) or even higher (for 1-C3H7ONO2) mixing ratios of RONO2
at TMS. The difference in NOx levels was considered to be the main cause of the anti-
correlation between the parent hydrocarbons and related RO2 radicals. As O3 formation is
generally limited by VOCs at both sites [Guo et al., 2013; Ling et al., 2014], the reaction chains
of O3 formation were terminated by NOx reacting with reactive radicals. Figure 8.7 shows that
the simulated OH and HO2 levels were much lower at TW than at TMS. This is because the
higher NOx at TW consumed more oxidative substances (e.g., O3) and radicals (OH and HO2).
Consequently, reactions including the oxidation of parent hydrocarbons at TW were more
suppressed, leading to lower production of RO2 radicals.
196
However, 2-C4H9ONO2 was higher at TW. In addition to the role of NO as the reactant, this
was also attributable to the fact that the concentration of 2-C4H9O2 at TW was lower than at
TMS only by a factor of 4.2, compared to 6.7, 6.6 and 13.3 for C2H5O2, 2-C3H7O2 and 1-C3H7O2,
respectively.
Figure 8.7. Average simulated diurnal patterns of (a) OH at TMS; (b) OH at TW; (c) HO2 at
TMS and (d) HO2 at TW.
8.2.2 Isopleths of RONO2 formation
To further investigate RONO2 formation in different environments, a total of 196 scenarios
were designed for model simulations. The simulations were conducted in daytime hours
(07:00-19:00), in which the concentrations of TVOCs and NOx were allowed to evolve over
time. The simulated hourly RONO2 during 07:00-19:00 in each scenario of the simulations
were averaged to create the corresponding isopleths. RONO2 production was simulated with a
matrix of total VOCs (TVOCs) and NOx ranging from 40-560 ppbv and 5-70 ppbv, with a
consistent scale of 40 ppbv and 5 ppbv, respectively. The ranges of TVOCs and NOx were
197
chosen to include not only the observed TVOCs and NOx in the middle of the sequence, but
also the best representative NOx-limited and VOC-limited regimes as discussed below. Apart
from their mixing ratios, the composition of TVOCs and NOx might also influence the
production of RONO2. Therefore, the ratios between VOC species and NO and NO2 (referred
to as air pollutant profiles hereafter) at TMS and TW were used to distribute VOCs species in
TVOCs, and NO and NO2 in NOx. The mixing ratios of TVOCs and NOx were 24.0±2.2 and
12.2±0.6 ppbv at TMS, and 48.4±7.2 and 50.5±2.3 ppbv at TW, respectively. The ratio of
NO2/NO at TMS (2.5±0.3) was about twice that at TW (1.2±0.2). Further inspection into the
OH reactivity indicated that TVOCs and NOx accounted for ~55% (3.7 s-1) and ~45% (3.0 s-1)
of total OH reactivity at TMS. In contrast, the OH reactivity of NOx at TW (14.3 s-1, ~65%)
dominated over that of TVOCs (7.7 s-1, 35%).
Figure 8.8 shows the isopleths of CH3ONO2 and C2H5ONO2 production with the changes of
TVOCs and NOx based on the air pollutant profiles at TMS (panels (a) and (b)) and TW (panels
(c) and (d)). The isopleth of C2H5ONO2 production was selected as an example of C2-C4
RONO2, which had the same pattern variations in response to the changes of TVOCs and NOx
(see Figure 8.9). It is noteworthy that both formation pathways of RO2+NO and RO+NO2 were
considered for C1-C4 RONO2. Based on Figure 8.8, the NOx limited and VOC limited regimes
in RONO2 formation were identified. Briefly, the points with the lowest TVOCs on each
isopleth line were linked in a straight line (dividing line), and RONO2 formation in the area
below and above the line was limited by VOCs and NOx, respectively. Linear regressions were
carried out for these dividing lines and a TVOC/NOx ratio of approximately 10.0±0.4
ppbv/ppbv (R2=0.97) was obtained for the simulated production of C2-C4 RONO2 based on the
air pollutant profiles at TMS. In other words, when the ratio of TVOCs/NOx was higher (lower)
than 10.0±0.4 ppbv/ppbv, the C2-C4 RONO2 formation was limited by NOx (VOCs). However,
this ratio was significantly (p<0.05) lower based on the air pollutants profiles at TW
198
(TVOCs/NOx=8.7±0.4 ppbv/ppbv, R2=0.96). The differences were likely attributable to the
higher fraction of NO2 in NOx at TMS (70.6±2.4%) than at TW (54.0±4.1%). NO2 reacts with
OH more quickly than NO, which serves as an important sink of OH in the VOC-limited regime.
A higher fraction of NO2 in NOx meant that the system was more NOx-suppressed (VOC-
limited), and a higher ratio of TVOCs/NOx was needed to change the RONO2 formation from
a VOC-limited regime to a NOx-limited regime. When using OH reactivity to present the
dividing ratios, the ratios of 𝑂𝐻 𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑜𝑓 𝑇𝑉𝑂𝐶𝑠
𝑂𝐻 𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑜𝑓 𝑁𝑂𝑥 between TMS (6.4±0.3 s-1/s-1) and TW
(5.6±0.3 s-1/s-1) were still not comparable (p<0.05). This might be due to the fact that O3
formation was not only determined by the reactions between OH and VOCs (NOx). For
example, the reaction of NO and OH cannot be treated as a destruction to O3 in daytime,
because OH can be regenerated through the photolysis of HONO. In addition, the same
reactivity of TVOCs might correspond to different potentials of O3 formation, due to the
different potentials of RO2 in oxidizing NO and regenerating HO2. Analysis of the relationship
between RONO2 production and the TVOC/NOx ratio found that in the NOx limited regime,
increasing NOx stimulated the production of RONO2 (RO2+NO→RONO2). However,
increasing NOx led to a direct or indirect reduction of OH (OH+NO2→HNO3 and
NO+O3→NO2+O2) and subsequent reductions of HO2, RO2 and RO in the VOC limited regime.
Conversely, an increase of TVOCs elevated the production of these radicals. Therefore, in the
VOC limited regime, an increase of TVOCs (NOx) resulted in an increase (decrease) of RO2,
subsequently stimulating (suppressing) RONO2 formation.
199
Figure 8.8 Isopleths of photochemical production (pptv) of (a) CH3ONO2 and (b) C2H5ONO2
(as an example of C2-C4 RONO2) based on the air pollutant profiles at TMS; (c) CH3ONO2 and
(d) C2H5ONO2 (as an example of C2-C4 RONO2) based on the air pollutant profiles at TW. The
black line separates NOx limited regime from VOCs limited regime. The red and blue symbols
in the figure show the daily average observed TVOCs and NOx at TMS (red) and TW (blue),
respectively.
200
Figure 8.9. Isopleths of photochemical production (pptv) of (a) 2-C3H7NO3, (b) 1-C3H7NO3
and (c) 2-C4H9NO3 simulated based on the profiles of air pollutants at TMS; (d) 2-C3H7NO3,
201
(e) 1-C3H7NO3 and (f) 2-C4H9NO3 simulated based on the profiles of air pollutants at TW. The
line with red dots separates the NOx limited regime from VOC limited regime.
8.2.3 Lower thresholds of TVOCs/NOx for CH3ONO2
The threshold ratios of TVOC/NOx turning to VOC-limited were lower for CH3ONO2
formation than for C2-C4 RONO2 at both TMS and TW, which were around 2.4±0.2 ppbv/ppbv
(R2=0.96) and 3.1±0.1 ppbv/ppbv (R2=0.99) based on the air pollutant profiles at TMS and TW,
respectively. In contrast to C2-C4 RONO2, which were mainly generated from RO2 reacting
with NO, CH3ONO2 had two pivotal formation pathways, i.e., RO2+NO and RO+NO2 (Table
8.5). Figure 8.10 (a) and (b) show the respective isopleths of CH3ONO2 produced by the
pathways of CH3O2+NO and CH3O+NO2 based on the air pollutant profiles at TMS (the
isopleths at TW are presented in Figure 8.10 (c) and (d)). The CH3ONO2 generated by
CH3O2+NO (panels (a) and (c)) followed the same patterns as C2-C4 RONO2 (TVOCs/NOx
ratio of 10.0±0.4 and 8.7±0.4 ppbv/ppbv as the threshold between the VOC- and NOx-limited
regimes at TMS and TW, respectively). However, based on the air pollutants profiles at TMS,
the formation of CH3ONO2 from the CH3O+NO2 always increased with increasing NOx,
implying a continuous NOx limited regime. In contrast, NOx facilitated the pathway of
CH3O+NO2→CH3ONO2 when TVOC/NOx was higher than 2.9±0.1 ppbv/ppbv, based on the
air pollutants profiles at TW. The VOC-limited regime under condition of
TVOCs/NOx<2.9±0.1 ppbv/ppbv was due to the inhibition of CH3O formation by NOx. The
production of CH3ONO2 through this pathway depended upon the product of CH3O and NO2.
Since TW had lower fraction of NO2 in NOx, the decrease of CH3O with NOx increasing cannot
be compensated by NO2 increasing, causing a VOC-limited regime for the pathway of
CH3O+NO2. Moreover, for the scenarios with TVOCs ≥ 240 ppbv, CH3ONO2 generated from
CH3O + NO2 continuously increased with increasing NOx (continuous NOx limited regime).
The continuous stimulation effect of NOx on CH3ONO2 formation at low ratios of TVOC/NOx
202
(high NOx) was also identified by Archibald et al. [2007], which might be caused by the
competitiveness of NO2 associating with CH3O relative to the oxidation of CH3O
(CH3O+O2→HCHO+HO2) increasing under high NOx. Since RO reacting with NO2 was an
important pathway for C1 RONO2, a higher fraction of NO2 in NOx at TMS meant a higher
production of C1 RONO2 through this route. Therefore, the NOx-limited regime appeared under
conditions of lower ratios of TVOCs/NOx at TMS.
It should be noted that the aforementioned dividing ratios were obtained based on the average
air pollutants profile at each site. In fact, the dividing ratios should vary in a range, due to the
change of air pollutant profiles. To further confirm the influence of air pollutant profiles on the
dividing ratios, RONO2 formation isopleths were simulated based on the air pollutant profiles
at TMS and TW on November 09, 2010, when the air pollutants profiles at TMS and TW were
obviously different. It was found that the distinctive air pollutant profiles on November 09 led
to completely different dividing ratios of TVOCs/NOx from the average ratios. Table 8.6 lists
the ranges of TVOC/NOx ratios corresponding to the NOx limited and VOCs limited regimes
in RONO2 formation, which were simulated on the basis of the air pollutant profiles at TMS
and TW. Please note: these values were the slopes derived from linear regressions. The
uncertainty of model simulation was roughly estimated by root mean square of the accuracies
of input parameters, which was ~13%.
203
Figure 8.10 Isopleths of CH3ONO2 production (pptv) from the pathway of (a) CH3O2+NO and
(b) CH3O+NO2 based on the air pollutant profiles at TMS; (c) CH3O2+NO and (d) CH3O+NO2
based on the air pollutant profiles at TW. The red and blue symbols in the figure show the daily
average observed TVOCs and NOx at TMS (red) and TW (blue), respectively.
Table 8.6 Ranges of TVOC/NOx ratios corresponding to regimes controlling RONO2 formation
based on the air pollutant profiles at TMS and TW.
TVOCs/NOx
(ppbv/ppbv)
Profiles of air pollutants at TMS Profiles of air pollutants at TW
NOx-limited VOC-imited NOx-limited VOC-limited
204
a CH3ONO2 >2.4±0.2 <2.4±0.2 >3.1±0.1 <3.1±0.1
b CH3ONO2 >10.0±0.4 <10.0±0.4 >8.7±0.4 <8.7±0.4
c CH3ONO2 All ratios None >2.9±0.1 <2.9±0.1
C2-C4 RONO2 >10.0±0.4 <10.0±0.4 >8.7±0.4 <8.7±0.4
a Total CH3ONO2 produced by CH3O2+NO and CH3O+NO2; b CH3ONO2 produced by
CH3O2+NO; c CH3ONO2 produced by CH3O+NO2.
The ratios of TVOCs/NOx that divide NOx- and VOC-limited regimes were verified by the
observed secondary RONO2, which were resolved by Positive Matrix Factorization model.
Details about the source apportionment of RONO2 have been provided in chapter 7. Figure
8.11 shows the relationships between the secondary RONO2 and the ratio of TVOCs/NOx. The
samples were grouped using the dividing ratios of TVOCs/NOx listed in Table 8.6 to categorize
them into the theoretically VOC-limited and NOx-limited regimes (Figure 8.11). C2 RONO2
was selected as an example of C2-C4 RONO2, as they all had similar patterns with changing
TVOCs/NOx. Both C1 and C2 RONO2 increased with increasing TVOCs/NOx in the VOC-
limited regime, and decreased with increasing TVOCs/NOx in the NOx-limited regime (not
applicable to C2-C4 RONO2 since only one sample was in NOx-limited regime). In other words,
the relationships between secondary RONO2 resolved from the observational data and their
precursors followed the patterns predicted by RONO2 formation isopleths.
205
Figure 8.11. Relationships between secondary RONO2 and the ratio of TVOCs/NOx in the
scenarios of (a) VOC-limited regimes for C1 and C2 RONO2 at TMS, (b) NOx-limited regimes
for C1 and C2 RONO2 at TMS, (c) VOC-limited regimes for C1 and C2 RONO2 at TW, and (d)
NOx-limited regimes for C1 and C2 RONO2 at TW. Please refer to Table 4 for details about the
dividing ratios of TVOCs/NOx for VOC-limited and NOx-limited regimes. The C2 RONO2 is
selected as an example of C2-C4 RONO2.
8.3 Impacts on O3 production
8.3.1 During RONO2 formation
To investigate the impacts of RONO2 formation on O3 production, two categories of scenarios,
i.e., a base case and five constrained cases were tested for each sampling day in this study.
Briefly, in the base case all reaction pathways were switched on in the model, while the
formation pathways (RO2+NO and RO+NO2) of each individual RONO2 were switched off in
each corresponding constrained case. The five constrained cases corresponded to five RONO2.
All other settings were identical between the base and constrained cases. The initial
concentrations of C1-C4 RONO2 were set as zero for both the base and constrained cases,
because we focused on the impacts of in-situ RONO2 formation on O3 production. The base
case simulated the secondary production of RONO2. The O3 variations (∆O3) induced by
RONO2 formation were obtained by subtracting O3 in the constrained cases from that in the
base case, as were the variations of NO, NO2, OH and HO2. Figure 8.12 shows the relationship
206
between hourly ∆O3 and secondary RONO2 production at TMS and TW using all measured data
points for the whole sampling period. Each point denotes hourly simulated O3 reduction
induced by RONO2 formation in that hour. Overall, as secondary RONO2 production increased,
O3 levels decreased. The formation of CH3ONO2, C2H5ONO2, 2-C3H7ONO2, 1-C3H7ONO2 and
2-C4H9ONO2 caused an average O3 reduction (mean ± 95% C.I.) of 9.7 ± 1.1, 14.7 ± 1.6, 18.4
± 1.9, 6.9 ± 0.6 and 60.2 ± 6.8 pptv at TMS, and 10.5 ± 3.2, 7.1 ± 2.0, 8.3 ± 2.1, 2.0 ± 0.5 and
40.0 ± 9.8 pptv at TW, respectively. Note that these values were different from those reported
at the same sites in Ling et al. [2016], which considered both RONO2 formation on site and
during mesoscale circulation. O3 reduction was linearly correlated with the production of
secondary RONO2 (0.72 <R2 <0.95 at TMS, 0.77 <R2 <0.84 at TW) for CH3ONO2, C2H5ONO2,
2-C3H7ONO2, 1-C3H7ONO2 and 2-C4H9ONO2, respectively. Furthermore, the daily average O3
reduction correlated well with the reduction of OH (R2= 0.83 and 0.71 at TMS and TW,
respectively) and of HO2 (R2= 0.84 and 0.98 at TMS and TW, respectively), while poor
correlations were found between O3 reduction and the variation of NO or NO2. More
importantly, the O3 reduction efficiencies (∆O3/secondary RONO2) were significantly lower at
TW than at TMS (p<0.05), as reflected from the slopes in Figure 8.12. In view of the higher OH
and HO2 at TMS than at TW, the higher O3 reduction efficiency at TMS suggested that RONO2
formation led to O3 reduction more significantly in the atmosphere with higher oxidative
capacity. This is reasonable because the promotion of OH and HO2 to the propagation of chain
reactions forming O3 is faster in the more oxidative atmosphere. This finding should be
applicable to the regions where O3 formation is limited by VOCs.
207
Figure 8.12 Relationship between O3 reductions and the simulated secondary RONO2
productions at (a) TMS and (b) TW.
8.3.2 During RONO2 degradation
The impacts of RONO2 degradation on O3 production at TMS and TW were studied with two
simulation scenarios, i.e., a base scenario with all the reaction pathways switched on and a
constrained scenario in which OH oxidation and photolysis for all five C1-C4 RONO2 were
switched off. The differences of model output between the two scenarios (base scenario -
constrained scenario) reflected the impact of RONO2 degradation, referred to as “variation”
hereafter. For example, positive variations of O3 meant that O3 increased due to RONO2
degradation. Note that this impact was studied in the form of total C1-C4 RONO2 rather than
individual species, because the O3 variation induced by individual RONO2 was generally minor.
During RONO2 degradation NO2 is released, which decomposes and generates NO and O,
allowing O3 to be formed through the association between O2 and O. On the other hand, the
released NO2 also consumes OH, reducing O3 production subsequently. RO released from
RONO2 degradation fuels O3 formation. Therefore, O3 variation in RONO2 degradation is the
combination of these effects.
Figure 8.13 shows the simulated daily average variations of O3, NO, NO2, OH and HO2 induced
by degradation of the C1-C4 RONO2. The daily average O3 variations ranged from -7.4 pptv to
2.3 pptv at TMS, but increased at TW throughout the sampling campaign (average increase of
208
2.9 ± 0.8 pptv). At TMS, the O3 variation correlated well with the OH and HO2 variations (R2
of 0.86 and 0.85, respectively), but negatively with the variations of NO and NO2 (R2 of 0.89
and 0.76, respectively). This implied that O3 formation at TMS was generally VOC-limited
(NOx-suppressed). When NO and NO2 levels increased at TMS, O3, OH and HO2 decreased.
This might be due to the consumption of OH by NO2 and/or NO titration with O3. Since NO
increases resulted from the decomposition of NO2, we defined this impact of RONO2
degradation on O3 production as NO2 suppressing. In contrast, O3 formation was enhanced by
RONO2 degradation on several days (October 29-31, November 02 and 21), when NO and NO2
decreased (due to increases of O3 and OH) while OH and HO2 increased. That is, NOx led to
an O3 reduction. Conversely, RO released from RONO2 promoted O3 formation. The overall
O3 enhancements indicated that the RO stimulating effect overrode the NO2 suppressing effect
in these cases, causing an O3 increase. Thus, the impact of RONO2 degradation on O3
production manifested as RO stimulating. Consistently, O3, OH and HO2 all increased while
NO decreased at TW, induced by RONO2 degradation. O3 enhancement exhibited moderate to
good correlations with the simulated increase of OH (R2 = 0.50) and HO2 (R2 = 0.81), which
were generated from the evolution of RO in the photochemical reaction chain. As such, the
impact of RONO2 degradation on O3 production at TW was dominated by the effect of RO
stimulating. We found that the ratio of TVOC/NOx in the cases of O3 increase (average: 1.6
ppbv/ppbv) was lower than in the cases of O3 decrease (average ratio: 2.1 ppbv/ppbv) at TMS.
The lower ratio of TVOC/NOx means that O3 formation was more limited by VOCs, which
enabled the added RO to more efficiently stimulate O3 formation and resulted in the increase
of O3. Coincidently, the low ratio of TVOC/NOx (0.9 ppbv/ppbv) at TW also caused O3
increase during RONO2 degradation.
209
Figure 8.13 Simulated variations of (a) O3, (b) NO, (c) NO2, (d) OH and (e) HO2 at TMS and
TW induced by C1-C4 RONO2 degradation.
8.3.3 Impacts on O3 in different environments
To extend the impact of RONO2 degradation on O3 production to different environments, O3
production in a total of 196 base scenarios and 196 constrained scenarios were simulated. The
scenarios were constructed with 14 gradients of TVOCs (from 40 to 560 ppbv with a consistent
step of 40) and 14 gradients of NOx (from 5 to 70 ppbv with a consistent step of 5). Similar to
the simulations of RONO2 formation, these simulations were based on the air pollutant profiles
210
at TMS and TW, respectively. The degradation reactions of C1-C4 RONO2 were switched off in
all the constrained scenarios. Figures 8.14-8.15 show the average differences of O3, NO, NO2,
OH, HO2 and total C1-C4 RONO2 between each base scenario and its corresponding
constrained scenario, based on the air pollutant profiles at TMS and TW, respectively. The
negative values to the right of the y-axis in panel (f) indicates the degradation amounts of total
C1-C4 RONO2 (the difference of simulated RONO2 between the base and constrained
scenarios), and panels (a)-(e) reflect the impact of the RONO2 degradation on the production
of these compounds or radicals. To help understand the variations of these species or radicals,
the NOx-limited and VOC-limited regimes in O3 formation at TMS and TW are shown in Figure
8.16. Notably, the variation patterns of O3 with the changes of TVOCs and NOx were highly
consistent with those of C2-C4 RONO2. Namely, 10.0±0.4 and 8.7±0.4 ppbv/ppbv were the
threshold TVOC/NOx ratios separating the NOx and VOC limited regimes. This is reasonable
because C2-C4 RONO2 and O3 share the same formation pathways.
211
Figure 8.14. Variations of (a) O3 (b) NO (c) NO2 (d) OH (e) HO2 and (f) total C1-C4 RONO2
induced by the degradation of C1-C4 RONO2 following changes of TVOCs and NOx, simulated
based on air pollutant profiles at TMS. The red circles represent the daily averages of the
observed TVOCs and NOx at TMS. Area “R1” (or “R2”) shows the increases (or decreases) of
OH/HO2 in the NOx limited regime. “R3” (or “R4”) are areas with O3 decrease (or increase) in
the VOC limited regime.
212
Figure 8.15. Variations of (a) O3 (b) NO (c) NO2 (d) OH (e) HO2 and (f) total C1-C4 RONO2
induced by the degradation of C1-C4 RONO2 following changes of TVOCs and NOx, simulated
based on air pollutant profiles at TW. The blue squares represent the daily averages of the
observed TVOCs and NOx at TW. Area “R1” (or “R2”) shows the increases (or decreases) of
213
OH/HO2 in the NOx limited regime. “R3” (or “R4”) are areas with O3 decrease (or increase) in
the VOC limited regime.
According to Figures 8.14-8.15, while the dividing ratios of TVOCs/NOx were different
between the simulations based on the air pollutants profiles at TMS and TW, the impacts of
RONO2 degradation on O3 photochemistry were similar and discussed together. With the
degradation of C1-C4 RONO2, O3 generally increased in the NOx limited regime, regardless of
the site. The increase of O3 was always accompanied by increased NO and NO2 (panels (b) and
(c)). However, the oxidative radicals (OH and HO2) could either increase or decrease with
RONO2 degradation in the NOx limited regime. For the convenience of discussion, the areas
with OH or HO2 increases (decreases) were defined as “R1” (“R2”) in panels (d) and (e). The
OH and HO2 increases might be caused by the increase of O3 following by the photolysis. The
added RO from RONO2 degradation also facilitated the production of these oxidative radicals.
However, in areas “R2” where more RONO2 was degraded and more NO2 was released (see
panels (c) and (f)), OH and HO2 decreased, possibly from the higher consumption of OH by
NO2 that was released from RONO2 degradation and/or consumption of OH by RONO2 itself
(RONO2+OH→RO+NO2+products). Since O3 formation was limited by NOx in this regime,
the O3 increase was most likely related to stimulation of O3 formation by NO2 released from
RONO2 degradation. Note that this NO2 stimulating effect on O3 production was not observed
at TMS and TW, where O3 formation was generally VOC limited. In contrast, O3 either
decreased or increased with RONO2 degradation in the VOC limited regime.
In the areas close to the NOx limited regime (defined as area “R3”), O3 generally decreased.
Consistently, both OH and HO2 decreased in this area due to RONO2 degradation, while NO
and NO2 increased. It is widely known that VOCs and NOx favor and inhibit O3 formation in
the VOC limited regime, respectively. Since RONO2 degradation released RO and NO2, the
decrease of O3 implied that the effect of NO2 suppression overrode the effect of RO stimulation
214
on O3 formation. In other words, the net impact of RONO2 degradation on O3 production was
NO2 suppression in this area “R3”. However, RONO2 degradation led to a slight O3 increase
in another area of the VOC limited regimes (defined as area “R4”), where OH and HO2 also
increased. In view of the VOC limited regime controlling O3 formation and the synchronous
increases of O3, OH and HO2, the O3 increase induced by RONO2 degradation was attributable
to the addition of RO and its stimulating effect on O3 formation. Namely, RO stimulation
overrode NO2 suppression in area “R4”. Moreover, O3 variations induced by degradation of
C1-C4 RONO2 at TMS (red symbols in Figure 8.14) and TW (blue symbols in Figure 8.15) also
followed the patterns as indicated in Figure 8.13.
Although the impact of C1-C4 RONO2 degradation on O3 production was generally minor (O3
variations <100 pptv), the mechanisms of RONO2 degradation regulating O3 formation might
apply to a variety of RONO2 species in which more nitrogen and/or more reactive radicals are
temporarily stored. In other words, the impact of RONO2 degradation on O3 production could
be significant when considering more RONO2 species, particularly in remote regions where
fresh emissions of O3 precursors are sparse.
Figure 8.16. Isopleths of photochemical O3 production (ppbv) based on the air pollutant profiles
at (a) TMS and (b) TW. The red and blue symbols in the figure show the daily average observed
TVOCs and NOx at TMS (red) and TW (blue), respectively.
215
8.4 Simulation of ≥C5 RONO2 Due to the difficulty in chemical analysis, RONO2 with higher carbon numbers are less
frequently reported. Isoprene and aromatics are well-known O3 precursors with high O3
formation potentials [Chameides et al., 1988; Lewis et al., 2000]. Therefore, understanding the
photochemistry of isoprene nitrates and aromatics nitrates on O3 production is essential. In this
study, the photochemical production of 5 isoprene nitrates and 17 aromatics nitrates (see Table
8.7) and their relationships with O3 formation were preliminarily studied using the PBM-MCM
model. Also shown in Table 8.7 are the lifetimes of these RONO2, which were calculated with
the consideration of OH oxidation and photolysis as the main sinks of RONO2. The lifetimes
of isoprene nitrates and aromatics nitrates were calculated using the formula: lifetime =
1
𝑂𝐻·𝑘𝑂𝐻+𝑅𝑂𝑁𝑂2+𝑘𝑝ℎ𝑜𝑡𝑜𝑙𝑦𝑠𝑖𝑠 (𝑘𝑂𝐻+𝑅𝑂𝑁𝑂2 and 𝑘𝑝ℎ𝑜𝑡𝑜𝑙𝑦𝑠𝑖𝑠 are the reaction rate constants of RONO2
reacting with OH and RONO2 photolysis, respectively).
Table 8.7 Average lifetimes of isoprene nitrates and aromatics nitrates at TMS and TW. (Unit:
hour)
RH precursor RONO2 a Lifetime at TMS Lifetime at TW
Isoprene NISOPNO3 0.9 ± 0.1 3.6 ± 0.5
Isoprene ISOPANO3 1.1 ± 0.1 4.5 ± 0.6
Isoprene ISOPBNO3 5.1 ± 0.1 16.0 ± 1.7
Isoprene ISOPCNO3 1.1 ± 0.1 4.5 ± 0.6
Isoprene ISOPDNO3 3.0 ± 0.3 11.6 ± 1.4
Benzene BZBIPERNO3 1.0 ± 0.1 4.2 ± 0.5
Toluene C6H5CH2NO3 12.3 ± 1.5 45.8 ± 5.7
Toluene TLBIPERNO3 1.1 ± 0.1 4.3 ± 0.6
o-Xylene OXYLNO3 5.8 ± 0.7 22.7 ± 2.9
o-Xylene OXYBIPENO3 1.0 ± 0.1 4.0 ± 0.5
m-Xylene MXYLNO3 12.3 ± 1.5 45.8 ± 5.6
m-Xylene MXYBIPENO3 1.1 ± 0.1 4.3 ± 0.5
216
p-Xylene PXYLNO3 12.3 ± 1.5 45.8 ± 5.6
p-Xylene PXYBIPENO3 1.1 ± 0.1 4.1 ± 0.5
Ethylbenzene C6H5C2NO3 10.4 ± 1.3 39.3 ± 4.8
Ethylbenzene EBZBPERNO3 1.1 ± 0.1 4.2 ± 0.5
1,2,3-TMB b TM123BNO3 12.3 ± 1.5 45.8 ± 5.5
1,2,3-TMB b TM123BPNO3 1.2 ± 0.1 4.7 ± 0.6
1,2,4-TMB b TM124BNO3 12.3 ± 1.5 45.8 ± 5.5
1,2,4-TMB b TM124BPNO3 1.1 ± 0.1 4.1 ± 0.5
1,3,5-TMB b TMBNO3 12.3 ± 1.5 45.8 ± 5.4
1,3,5-TMB b TM135BPNO3 0.8 ± 0.1 3.3 ± 0.4 a refer to the website of MCM (http://mcm.leeds.ac.uk/MCM/) for the molecular formulas of
RONO2; b TMB represents to trimethylbenzene.
Table 8.8 lists the simulated total productions of isoprene nitrates and aromatics nitrates and
the measured mixing ratios of their parent hydrocarbons during daytime hours (07:00-19:00,
LT) at TMS and TW. Despite higher oxidative capacity at TMS, the sum of isoprene nitrates
(5.7±0.5 pptv; mean±95% confidence interval) at this mountainous site was lower than that at
TW (16.7±1.9 pptv) (p<0.05), mainly due to much lower isoprene levels at TMS (138.6±14.0
and 303.7±35.6 pptv at TMS and TW, respectively). Detailed reasons for the lower isoprene at
TMS can be found in Guo et al. [2013] and Ling et al. [2014]. Briefly, the vegetation above
550 m elevation on the mountain was very sparse and dominated by low-isoprene emitting
trees, while the sampling site (TMS) was located on the level of 640 m. In addition, the
temperature at TMS (19.1±0.5 °C) was significantly (p<0.05) lower than at TW (26.3±0.5 °C),
which might inhibit the biogenic emissions of isoprene. It is noteworthy that isoprene at TW
was also likely elevated by vehicle emissions [Borbon et al., 2001], in view of its highest level
(374±300 pptv) at 07:00. On the other hand, the sum of aromatics nitrates was comparable
between TMS and TW (p > 0.05), even including the observation with extremely high values
of their parent hydrocarbons on November 19 at TW.
217
Table 8.8 Simulated total production of isoprene nitrates and aromatics nitrates and the
measured mixing ratio of parent hydrocarbons at TMS and TW (Unit: pptv; average ± 95%
confidence interval).
TMS TW (including data
on 19/11)
TW (excluding data
on 19/11)
Sum of isoprene nitrates 5.7 ± 0.5 16.7 ± 1.9 16.4 ± 2.0
Sum of aromatics nitrates 30.1 ± 3.7 50.8 ± 18.6 26.9 ± 2.8
Isoprene 138.6 ± 14.0 303.7 ± 35.6 285.4 ± 27.9
Sum of aromatics * 643.3 ± 31.7 722.4 ± 29.0 692.7 ± 25.0
* Aromatics refer to benzene, toluene, xylene isomers, ethylbenzene and trimethylbenzene
isomers.
Similar to C1-C4 RONO2, the formation of isoprene nitrates and aromatics nitrates caused O3
reductions. Figure 8.17 shows the daily maximum O3 reductions (15:00, LT) at TMS and TW.
Following the formation of isoprene nitrates, simulated O3 was reduced by 285 ± 85 and 245
± 204 pptv at TMS and TW, respectively. In comparison, the O3 reduction induced by the
formation of aromatics nitrates was much higher (TMS: 2960 ± 1040 pptv; TW: 2080 ± 2540
pptv). The large uncertainties at TW resulted from the high levels and large uncertainties of
isoprene (633 ± 430 pptv) and aromatics (89910 ± 71080 pptv) on November 19. By excluding
this day, the O3 reductions (114 ± 50 pptv and 764 ± 338 pptv for the formation of isoprene
nitrates and aromatics nitrates, respectively) at TW were significantly lower than those at TMS
(p < 0.05). Based on simulations, the average O3 reduction efficiency due to the formation of
isoprene nitrates and aromatics nitrates was ~31 and ~59 pptv/pptv at TMS and ~9 and ~30
pptv/pptv at TW, respectively. The lower O3 reduction efficiencies at TW coincided with the
weaker photochemical reactivity at the urban site.
218
Figure 8.17 O3 reduction at 15:00 (LT) induced by the formation of isoprene and aromatics
nitrates at (a) TMS and (b) TW.
Figure 8.18 presents the maximum O3 variations (15:00, LT) induced by the degradations of
isoprene nitrates and aromatics nitrates. It was found that O3 was generally reduced at TMS, by
an average of -55.5 ± 20.1 and -302 ±172 pptv for the degradation of isoprene nitrates and
aromatics nitrates, respectively. At TW, O3 production was reduced by 7.0 ± 12.2 pptv due to
the isoprene nitrates degradation. However, the degradation of aromatics nitrates led to an O3
increment of 118 ± 97 pptv. Since O3 formation at both sites was limited by VOCs, the
decreases of O3 induced by RONO2 degradation were mainly caused by the NO2 suppressing
effect. In addition, the degradation of these high carbon-number RONO2 consumed more OH
than C1-C4 RONO2 [Becker and Wirtz, 1989], possibly resulting in the reductions of oxidative
radicals and subsequent reduction of O3. However, the higher level of aromatic nitrates
(compared to isoprene nitrates, see Table 8.9) and the more VOCs limited regime in O3
formation at TW (lower ratios of TVOCs/NOx, see Figure 8.18) enabled RO released from the
degradation of aromatics nitrates to fuel O3 formation. This might explain the O3 increases
induced by the aromatics nitrates degradation at TW. Overall, the photochemistry of isoprene
nitrates and aromatics nitrates has significant impacts on O3 formation, under the assumption
that the present mechanisms of isoprene nitrates and aromatics nitrates in MCM were basically
reliable which however needs further verifications. Therefore, more attentions are required to
219
be paid to RONO2 derived from reactive hydrocarbons (such as isoprene and aromatics) in
future studies.
Figure 8.18 O3 variations at 15:00 (LT) induced by the degradation of (a) isoprene nitrates at
TMS (b) aromatics nitrates at TMS (c) isoprene nitrates at TW and (d) aromatics nitrates at TW.
The red triangles represent the ratios of TVOCs/NOx.
8.5 Sub-conclusions A PBM-MCM model was developed to simulate gas-phase RONO2 measured at an urban and
a mountainous site in Hong Kong. The magnitudes and variations of the observed C1-C4
RONO2 at both sites were well reproduced by the model. The modeling results indicated that
RONO2 formation depended upon not only the abundances of precursors but also the
photochemical reactivity, which was closely related to the levels of VOCs and NOx. Although
220
the precursors of RONO2 at the mountainous site were less abundant than at the urban site, the
higher oxidative capacity of the atmosphere at the mountainous site led to higher production of
RO2 radicals, resulting in comparable or even higher RONO2. The regimes in which the
formation of C2-C4 RONO2 was NOx-limited and VOC-limited were identified namely when
the ratio of TVOCs/NOx was higher and lower than 10.0±0.4/1 ppbv/ppbv, respectively, based
on the air pollutant profiles at mountainous site. However, the dividing ratio was 8.7±0.4/1
ppbv/ppbv at the urban site, due to the different air pollutants profiles at these two sites. For
the formation of CH3ONO2, these simulated thresholds decreased to 2.4±0.2/1 and 3.1±0.1/1
ppbv/ppbv based on the air pollutants profiles at the mountainous and urban site, respectively.
This was mainly because that CH3ONO2 produced from CH3O + NO2 continued to increase
with increasing NOx when the ratios of TVOCs/NOx were relatively low (high NOx). Since O3
formation was generally VOC limited at both sites, and RONO2 formation initially stabilized
RO radicals in RONO2 molecules, O3 production was reduced by RONO2 formation. On the
other hand, the mechanisms of RONO2 degradation influencing O3 production included NO2
stimulating, NO2 suppressing, and RO stimulating processes. At the mountainous site, the
impact of RONO2 degradation on O3 production was dominated by NO2 suppression under the
condition of relatively high ratios of TVOCs/NOx, leading to the decrease of O3, while RO
stimulation occurred at relatively low ratios of TVOCs/NOx, resulting in the increase of O3.
However, the O3 production always increased due to RO stimulation at the urban site.
Furthermore, isoprene nitrates and aromatics nitrates, which are difficult to be chemically
analyzed, may have a non-negligible impact (up to >1 ppbv) on O3 production.
221
Chapter 9 Conclusions and suggestions for future studies 9.1 Conclusions This work studied the spatiotemporal patterns of RONO2 and their parent hydrocarbons in
Hong Kong; the sources and formation mechanisms of RONO2; the impacts of RONO2
photochemistry on O3 production; and a case study on the variations of RONO2 and their parent
hydrocarbons under the influence of a recent air pollution control measure in Hong Kong. The
main findings of the study are concluded as follows.
(i) During the past 15 years, C1-C5 RONO2 all increased significantly in Hong Kong, mainly
as a result of the increases of their parent hydrocarbons and the enhancement of the oxidative
capacity of the atmosphere (O3 increased continuously in Hong Kong [Xue et al., 2014; Wang
et al., 2017]). Higher levels of RONO2 were observed in northwestern Hong Kong, in
comparison with the lower RONO2 in eastern Hong Kong, which was attributable to the
combined effect of the distribution of parent hydrocarbons and regional transport.
(ii) In highly vehicle-populated roadside environments, the LPG program implemented
between September 2013 and May 2014 led to significant decreases of the parent hydrocarbons
of C3-C4 RONO2. Despite the slightly increased oxidative capacity of the atmosphere, C2-C4
RONO2 also decreased due to the effectiveness of the program on emission reductions of VOCs,
particularly propane and n-butane. The decrease of C2 RONO2 was expected to be partially
caused by the constrained emission of n-butane from LPG-fueled vehicles, as the
decomposition of the oxidation products of n-butane are important sources of C2H5O2 and
C2H5O, which form C2 RONO2 through reacting with NOx [Zeng et al., 2018]. In contrast,
RONO2 were found to be increased after the program in most areas of Hong Kong. This
discrepancy might be caused by the differences in vehicle fleet and vehicle compositions, as
well as the different enhancement of atmospheric oxidative capacity. For example, the parent
hydrocarbons of RONO2 might experience greater reductions at the sites with higher density
222
of LPG fueled vehicles (taxis and most public light buses). In addition, the enhancement of
oxidative capacity in roadside environments might not be as significant as in urban and rural
areas, due to the much higher NOx. The combined effect led to the decrease of RONO2 in
roadside environments, while increased in most areas of Hong Kong. It is also noteworthy that
the grid sampling campaign (only one day’s sampling before and after the program,
respectively) over the entire territory might be much less representative, implying that the
increases of RONO2 in most areas of Hong Kong were subject to great uncertainties.
(iii) Although the concentrations of parent hydrocarbons of RONO2 were much lower at the
mountainous site, the observed RONO2 were comparable between the mountainous and urban
sites. Meso-scale circulation and regional transport elevated the RONO2 concentrations at the
mountainous site. In addition, the comparable RONO2 levels were partially attributable to the
higher atmospheric oxidative capacity at the mountainous site. Specifically, OH and HO2 were
much more abundant at the mountainous site, resulting in more efficient oxidation of VOCs
and higher productions of RO2 and RO radicals, both of which were precursors of RONO2. We
also found that the oxidative capacity of the atmosphere closely related to the relative
abundances of VOCs and NOx. In the urban environment, NOx freshly emitted from vehicle
exhausts largely consumed the oxidative radicals, reducing OH and HO2 consequently.
(iv) The observed C1-C4 RONO2 were well simulated by the photochemical box model
incorporating master chemical mechanism. Branching ratio of 0.0003 was identified as the
most appropriate value for CH3O2 reacting with NO, with the uncertainty less than 50%. The
branching ratios calculated according to Carter and Atkinson [1985, 1989] were adopted for
C2-C4 RONO2. The dry deposition velocity of 0.07/HMIX (HMIX: height of mixing layer) was
identified for C1-C4 RONO2, which was also well within the range of 0.03/HMIX-0.56/HMIX
proposed in previous studies [Zhang et al., 2002]. The model with above configurations has
223
been successfully applied in RONO2 simulations at an urban, a mountainous and a coastal site
in Hong Kong.
(v) The pathway of RO2 reacting with NO dominated the formations of C2-C5 RONO2.
However, RO reacting with NO2 also made considerable contribution to CH3NO3, particularly
under the condition of high NO2. In addition to the abundances of RONO2 precursors and the
branching ratios, the atmospheric oxidative capacity significantly influenced the production of
RONO2. NOx-limited and VOC-limited regimes were determined for the formation of C2-C4
RONO2, under the conditions of TVOCs/NOx higher and lower 10.0/1 ppbv/ppbv at the
mountainous site, respectively, and 8.7/1 ppbv/ppbv at the urban site, respectively. Due to the
increased contribution of the pathway of RO+NO2 to CH3NO3 with the increase of NOx, the
threshold between NOx-limited and VOC-limited regimes shifted to 2.4/1 ppbv/ppbv and 3.1/1
ppbv/ppbv at the mountainous and urban site, respectively. It should be noted that the dividing
ratios of TVOCs/NOx between VOC-limited and NOx-limited regimes were strongly dependent
upon the air profiles. Therefore, we do not suggest to directly apply the aforementioned
dividing ratios to other environments. Instead, specific analyses are required to understand the
regimes controlling RONO2 formation and the dividing ratios for different air profiles.
(vi) Generally, RONO2 formation led to reductions of O3 production in urban areas, due to the
stabilization of oxidative radicals in RONO2 molecules. The O3 reductions correlated well with
the decreases of OH and HO2, and the O3 reduction efficiency was higher in the atmosphere
with stronger oxidative capacity. The impacts of RONO2 degradation on O3 production
involved the mechanisms of NO2 stimulating in NOx-limited regime, NO2 suppressing in VOC-
limited regime with higher ratios of TVOCs/NOx and RO stimulating in VOC-limited regime
with lower ratios of TVOCs/NOx. The interferences of C1-C4 RONO2 degradation on O3
production were generally minor (absolute value less than 100 pptv). However, the
mechanisms of C1-C4 RONO2 degradation regulating O3 formation might apply to a variety of
224
RONO2 species in which more nitrogen and/or more reactive radicals are temporarily stored.
In other words, the impact of RONO2 degradation on O3 production could be significant when
considering more RONO2 species, particularly in remote regions where fresh emissions of O3
precursors are sparse.
(vii) In view of the contribution of RONO2 to O3 production in NOx-limited regime through
NO2 stimulation, and the long atmospheric lifetimes of RONO2, it is essential to control the
emissions of RONO2 precursors and inhibit the productions of RONO2 in urban areas, so that
the fuels of O3 formation in remote areas can be reduced. It has been well recognized that China
features intensive emissions of VOCs and NOx, from which high levels of RONO2 may be
formed. It is a reasonable assumption that RONO2 produced in China may increase the burden
to the tropospheric O3, particularly to the global background O3. Therefore, constraints on
emissions of RONO2 precursors may be helpful in improvement of global air quality.
9.2 Suggestions for future studies Although this work filled up some knowledge gaps in the field of RONO2 study, many
questions still need to be addressed. To better understand the sources, evolution and
atmospheric roles of RONO2, the following suggestions are proposed based on the findings of
my study. Firstly, in view of the ubiquity of organics and nitrogen containing species in the
atmosphere, hydrosphere and biosphere, sources except for oceanic emission, biomass burning
and photochemical formation which are presently widely recognized, may exist for RONO2.
Therefore, more efforts need to be made to better understand the sources of RONO2 in the
atmosphere. On the other hand, although some mechanisms were proposed to explain the
formation of RONO2 in water (such as algal activity) and biomass burning (such as the self-
reactions between RO2 to form RO), they require further examination and validation.
225
Laboratory experiments and model simulation may help to verify the pre-existing mechanisms
and to explore the new pathways of RONO2 formation in these media.
Secondly, the evolution of RONO2 in the atmosphere is generally thought to be regulated by
gas phase formation, photolysis, oxidation by OH and dry deposition. However, there may be
other formation and degradation pathways for RONO2, such as the partitioning between gas
and particle phases, and the exchange between troposphere and stratosphere. Actually, it is of
interest and necessity to further probe into the formation pathways of RONO2, even in gas
phase. As repeatedly mentioned in this work, RO2 and RO are the key precursors of RONO2.
However, they are not exclusively formed from the photo-oxidations of their parent
hydrocarbons. Instead, many secondary intermediates, such as aldehydes, ketones and PANs,
can also form RO2 radicals through decomposition, isomerization and other reactions. This
effect is expected to be more significant in more aged air masses, which contain higher
proportions of secondary intermediates. The contributions of VOCs other than the parent
hydrocarbons to RONO2 productions are left to be quantified, and the detailed mechanisms
need to be discovered in different environments. For example, some reactive VOCs (such as
isoprene and aromatics) may influence RONO2 production though regulations on the oxidative
capacity of the atmosphere, rather than the decomposition/isomerization of their oxidation
products. It can be expected that this effect is distinctive at locations with different air
compositions. Additionally, the fate of RONO2 in the region or space remote from
anthropogenic activities, as well as their roles in environment and climate, remains uncertain.
For example, our understanding is nearly blank on the evolution of RONO2 in the stratosphere
where the solar radiation is much stronger. As such, new knowledge about the evolution of
RONO2 is of high necessity to recognize their impacts on environment and climate in the globe.
Thirdly, the present study investigated the impacts of C1-C4 RONO2 photochemistry on O3
production, and tried to shed light on some larger RONO2 (e.g., isoprene nitrates and aromatics
226
nitrates) influencing O3 formation. However, the understandings on these impacts are quite
insufficient, particularly for isoprene nitrates and aromatics nitrates in which the highly active
oxidative radicals are included but the chemical mechanisms have never been verified. On one
hand, the reaction rate constants and branching ratios leading to RONO2 formation and the
degradation reactions of RONO2 can be determined by laboratory experiments and quantum
computation. On the other hand, the mechanisms of larger RONO2 (including isoprene nitrates
and aromatics nitrates) in the present version of MCM and also future developed models need
to be validated by the observation data. This can only be achieved if advanced analysis
techniques for these larger RONO2 are developed, which are now not available. Therefore, to
better understand the roles of RONO2 derived from a wide range of parent hydrocarbons in O3
photochemistry, the appropriate reaction rate constants, branching ratios, reliable observation
data and robust models are needed.
Fourthly, it is definitely necessary to update the explicit, species-based RONO2 mechanisms to
the present chemical transport models (CTMs), such as the Weather Research and Forecasting-
Community Multi-scale Air Quality (WRF-CMAQ) model. To our best knowledge, most CTM
models lack comprehensive descriptions of RONO2 photochemistry. For example, constructed
with the carbon bond mechanism, WRF-CMAQ considers the total RONO2 as the product of
total RO2 and NO, without the speciation of RO2 and RONO2. This hampers the full
consideration of RONO2 photochemistry in CTMs. As we know, CTMs have great advantages
in simulating the atmospheric physical processes and integrating them with photochemical
reactions. With the update of explicit, species-based RONO2 mechanisms in CTMs, the impacts
of RONO2 photochemistry on secondary pollution (such as O3 and SOA) in regional, national
and even global scales can be better understood. Hence, to further develop the chemical
mechanisms of RONO2 and update them in CTMs is a prospective direction of RONO2 study.
227
At last, the regional and even national budgets of RONO2 need to be understood. Hong Kong
adjoins the inland PRD region where photochemical pollution is generally much severer.
Biomass burning is also more intensive in inland PRD. In view of the well-recognized transport
of O3 from inland PRD to Hong Kong, it is most likely that regional transport also makes
positive contributions to RONO2 in Hong Kong. The contradiction is that RONO2 study has
seldom been carried out in mainland China, including in PRD, and the regional transport of
RONO2 to Hong Kong have never been quantified. Figuring out the external load of RONO2
is meaningful in recognizing the sources of RONO2 in Hong Kong. On a larger scale, it remains
unknown whether RONO2 formed in China can be transported to other regions over the world.
China has large quantities of VOCs and NOx emissions. O3, as a representative of atmospheric
oxidative capacity, is also high in China and exhibits an increasing trend. Therefore, it is not
surprising that China may “produce” a great amount of RONO2. Due to the long atmospheric
lifetimes of RONO2, relative to NOx and O3, these RONO2 may be “exported” to other regions
over the world, fueling O3 formation in the downwind areas, particularly in the remote
environments with sparse anthropogenic emissions. All in all, the regional and national budgets
of RONO2 and the global impacts are worthy to be studied.
228
References Ainsworth, E. A., Yendrek, C. R., Sitch, S., Collins, W. J., and Emberson, L. D., 2012. The
effects of tropospheric ozone on net primary productivity and implications for climate change.
Annual review of plant biology, 63, 637-661.
AFCD (Agriculture, Fisheries and Conservation Department): Country Parks and Special
Areas
Distribution Map–Tai Mo Shan Park, available at: http://www.afcd.gov.hk/english/country/
cou_vis/cou_vis_cou/cou_vis_cou.html, last access: 18 August 2015.
Archibald, A. T., Khan, M. A. H., Watson, L. A., Clemitshaw, K. C., Utembe, S. R., Jenkin,
M. E., and Shallcross, D. E., 2007. Comment on ‘Long-term atmospheric measurements of C1-
C5 alkyl nitrates in the Pearl River Delta region of southeast China’by Simpson et al.
Atmospheric Environment, 41(34), 7369-7370.
Arey, J., Aschmann, S. M., Kwok, E. S., and Atkinson, R., 2001. Alkyl Nitrate, Hydroxyalkyl
Nitrate, and Hydroxycarbonyl Formation from the NOx-Air Photooxidations of C5-C8 n-
Alkanes. The Journal of Physical Chemistry A, 105(6), 1020-1027.
Aruffo, E., Di Carlo, P., Dari-Salisburgo, C., Biancofiore, F., Giammaria, F., Busilacchio, M.,
Lee, J., Moller, S., Hopkins, J., Punjabi, S., Bauguitte, S., O’Sullivan, D., Percival, C., Le
Breton, M., Muller, J., Jones, R., Forster, G., Reeves, C., Heard, D., Walker, H., Ingham, T.,
Vaughan, S., and Stone, D., 2014. Aircraft observations of the lower troposphere above a
megacity: Alkyl nitrate and ozone chemistry. Atmospheric Environment, 94, 479-488.
Atherton, C. S., and Penner, J. E., 1988. The transformation of nitrogen oxides in the polluted
troposphere. Tellus B, 40(5), 380-392.
Atkinson, R., 1990. Gas-phase tropospheric chemistry of organic compounds: a
review. Atmospheric Environment. Part A. General Topics, 24(1), 1-41.
229
Atkinson, R., Aschmann, S. M., Carter, W. P., and Winer, A. M., 1982a. Kinetics of the gas-
phase reactions of OH radicals with alkyl nitrates at 299±2 K. International Journal of
Chemical Kinetics, 14(8), 919-926.
Atkinson, R., Aschmann, S. M., Carter, W. P., Winer, A. M., and Pitts Jr, J. N., 1982b. Alkyl
nitrate formation from the NOx/-air photooxidations of C2-C8. J. Phys. Chem.;(United
States), 86(23).
Atkinson, R., Aschmann, S. M., and Winer, A. M., 1987. Alkyl nitrate formation from the
reaction of a series of branched RO2 radicals with NO as a function of temperature and
pressure. Journal of atmospheric chemistry, 5(1), 91-102.
Atkinson, R., Baulch, D. L., Cox, R. A., Hampson Jr, R. F., Kerr, J. A., Rossi, M. J., and Troe,
J., 1999. Evaluated kinetic and photochemical data for atmospheric chemistry, organic species:
Supplement VII. Journal of Physical and chemical reference Data, 28(2), 191-393.
Atlas, E., 1988. Evidence for ≥ C3 alkyl nitrates in rural and remote
atmospheres. Nature, 331(6155), 426-428.
Atlas, E., Pollock, W., Greenberg, J., Heidt, L., and Thompson, A. M., 1993. Alkyl nitrates,
nonmethane hydrocarbons, and halocarbon gases over the equatorial Pacific Ocean during
SAGA 3. Journal of Geophysical Research: Atmospheres, 98(D9), 16933-16947.
Atlas, E. L., Ridley, B. A., Hubler, G., Walega, J. G., Carroll, M. A., Montzka, D. D., Huebert,
B. J., Norton, R. B., Grahek, F. E., and Schauffler, S., 1992. Partitioning and budget of NOy
species during the Mauna Loa Observatory Photochemistry Experiment. Journal of
Geophysical Research: Atmospheres, 97(D10), 10449-10462.
Barletta, B., Meinardi, S., Simpson, I. J., Atlas, E. L., Beyersdorf, A. J., Baker, A. K., Blake,
N. J., Yang, M., Midyett, J. R., Novak, B. J., McKeachie, R. J., Fuelberg, H. E., Sachse, G. W.,
Avery, M. A., Campos, T., Weinheimer, A. J., Rowland, F. S., and Blake, D. R., 2009.
Characterization of volatile organic compounds (VOCs) in Asian and north American pollution
230
plumes during INTEX-B: identification of specific Chinese air mass tracers. Atmospheric
Chemistry and Physics, 9, 5371–5388, doi:10.5194/acp-9-5371-2009.
Barletta, B., Meinardi, S., Simpson, I. J., Khwaja, H. A., Blake, D. R., and Rowland, F. S.,
2002. Mixing ratios of volatile organic compounds (VOCs) in the atmosphere of Karachi,
Pakistan. Atmospheric Environment, 36, 3429-3443.
Bates, D. R., Nicolet, M., 1950. The photochemistry of atmospheric water vapor. Journal of
Geophysical Research, 55(3):301-27.
Becker, K.H., and Wirtz, K., 1989. Gas phase reactions of alkyl nitrates with hydroxyl radicals
under tropospheric conditions in comparison with photolysis. Journal of Atmospheric
Chemistry, 9, 419-433.
Beine, H. J., Jaffe, D. A., Blake, D. R., Atlas, E., and Harris, J., 1996. Measurements of PAN,
alkyl nitrates, ozone, and hydrocarbons during spring in interior Alaska. Journal of
Geophysical Research: Atmospheres, 101(D7), 12613-12619.
Bell, M. L., 2006. The use of ambient air quality modeling to estimate individual and
population exposure for human health research: a case study of ozone in the Northern Georgia
Region of the United States. Environment international, 32(5), 586-593.
Bell, N., Hsu, L., Jacob, D.J., Schultz, M.G., Blake, D.R., Butler, J.H., King, D.B., Lobert, J.M.,
Maier‐Reimer E. Methyl iodide: Atmospheric budget and use as a tracer of marine convection
in global models. Journal of Geophysical Research: Atmospheres, 107(D17).
Bertman, S. B., Roberts, J. M., Parrish, D. D., Buhr, M. P., Goldan, P. D., Kuster, W. C.,
Fehsenfeld, F.C., Montzka, S.A., and Westberg, H., 1995. Evolution of alkyl nitrates with air
mass age. Journal of Geophysical Research: Atmospheres, 100(D11), 22805-22813.
Blake, D. R., and Rowland, F. S., 1995. Urban Leakage of Liquefied Petroleum Gas and Its
Impact on Mexico City Air Quality, Science, 269, 953-956.
231
Blake, N. J., Blake, D. R., Swanson, A. L., Atlas, E., Flocke, F., and Rowland, F. S., 2003.
Latitudinal, vertical, and seasonal variations of C1‐C4 alkyl nitrates in the troposphere over the
Pacific Ocean during PEM‐Tropics A and B: Oceanic and continental sources. Journal of
Geophysical Research: Atmospheres, 108(D2), doi: 10.1029/2001JD001444.
Blake, N. J., Blake, D. R., Wingenter, O. W., Sive, B. C., Kang, C. H., Thornton, D. C., Bandy,
A. R., Atlas, E., Flocke, F., Harris, J. M., and Rowland, F. S., 1999. Aircraft measurements of
the latitudinal, vertical, and seasonal variations of NMHCs, methyl nitrate, methyl halides, and
DMS during the First Aerosol Characterization Experiment (ACE 1). Journal of Geophysical
Research: Atmospheres, 104(D17), 21803-21817.
Borbon, A., Fontaine, H., Veillerot, M., Locoge, N., Galloo, J. C., and Guillermo, R., 2001. An
investigation into the traffic-related fraction of isoprene at an urban location. Atmospheric
Environment, 35(22), 3749-3760.
Borbon, A., Locoge, N., Veillerot, M., Galloo, J. C., and Guillermo, R., 2002. Characterisation
of NMHCs in a French urban atmosphere: overview of the main sources. Science of the Total
Environment, 292: 177-191.
Brown, S. G., Frankel, A., and Hafner, H. R., 2007. Source apportionment of VOCs in the Los
Angeles area using positive matrix factorization. Atmospheric Environment, 41(2), 227-237.
Buhr, M. P., Parrish, D. D., Norton, R. B., Fehsenfeld, F. C., Sievers, R. E., and Roberts, J. M.,
1990. Contribution of organic nitrates to the total reactive nitrogen budget at a rural eastern US
site. Journal of Geophysical Research: Atmospheres, 95(D7), 9809-9816.
Cardelino, C., 1998. Daily Variability of Motor Vehicle Emissions Derived from Traffic
Counter Data. Journal of Air & Waste Management Association, 48, 637-645.
Cardelino, C. A., and Chameides, W. L., 1995. An Observation-Based Model for Analyzing
Ozone Precursor Relationships in the Urban Atmosphere. Journal of Air & Waste Management
Association, 45, 161-180.
232
Carslaw, N., Carpenter, L. J., Plane, J. M. C., Allan, B. J., Burgess, R. A., Clemitshaw, K. C.,
Coe, H., and Penkett, S. A., 1997. Simultaneous observations of nitrate and peroxy radicals in
the marine boundary layer. Journal of Geophysical Research: Atmospheres, 102(D15), 18917-
18933.
Carter, W. P. L., and Atkinson, R., 1985. Atmospheric chemistry of alkanes. Journal of
atmospheric chemistry, 3(3), 377-405.
Carter, W. P. L., and Atkinson, R., 1989a. Alkyl nitrate formation from the atmospheric
photoxidation of alkanes; a revised estimation method. Journal of atmospheric chemistry, 8(2),
165-173.
Carter, W. P. L., and Atkinson, R., 1989b. Computer modeling study of incremental
hydrocarbon reactivity. Environmental Science and Technology, 23, 864-880.
Censtatd. Census and Statistics Department HKG. Geography and climate, Hong Kong.
http://www.censtatd.gov.hk/FileManager/EN/Content_810/geog.pdf retrieved 27 January
2016.
Chan, C. Y., and Chan, L. Y., 2000. Effect of meteorology and air pollutant transport on ozone
episodes at a subtropical coastal Asian city, Hong Kong. Journal of geophysical research:
atmospheres, 105(D16), 20707-20724.
Chan, C.Y., Chan, L.Y., Wang, X.M., Liu, Y.M., Lee, S.C., Zou, S.C., Sheng, G.Y., Fu, J.M.,
2002. Volatile organic compounds in roadside microenvironments of metropolitan Hong Kong.
Atmospheric Environment, 36(12):2039-47.
Chameides, W. L., Lindsay, R. W., Richardson, J., and Kiang, C. S., 1988. The role of biogenic
hydrocarbons in urban photochemical smog: Atlanta as a case study. Science (Washington),
241(4872), 1473-1475.
233
Chan, T. L., Ning, Z., Leung, C. W., Cheung, C. S., Hung, W. T., and Dong, G., 2004. On-road
remote sensing of petrol vehicle emissions measurement and emission factors estimation in
Hong Kong. Atmospheric Environment, 38(14), 2055-2066.
Cheng, H. R., Guo, H., Saunders, S. M., Lam, S. H. M., Jiang, F., Wang, X. M., Simpson, I. J.,
Blake, D. R., Louie, P. K. K., and Wang, T. J., 2010b. Assessing photochemical ozone
formation in the Pearl River Delta with a photochemical trajectory model. Atmospheric
Environment, 44(34), 4199-4208.
Cheng, Y., Lee, S. C., Ho, K. F., Chow, J. C., Watson, J. G., Louie, P. K. K., Cao, J. J., and
Hai, X., 2010a. Chemically-speciated on-road PM2.5 motor vehicle emission factors in Hong
Kong. Science of the Total Environment, 408(7), 1621-1627.
Chuck, A. L., Turner, S. M., and Liss, P. S., 2002. Direct evidence for a marine source of C1
and C2 alkyl nitrates. Science, 297(5584), 1151-1154.
Clemitshaw, K. C., Williams, J., Rattigan, O. V., Shallcross, D. E., Law, K. S., and Cox, R. A.,
1997. Gas-phase ultraviolet absorption cross-sections and atmospheric lifetimes of several C2-
C5 alkyl nitrates. Journal of Photochemistry and Photobiology A: Chemistry, 102(2-3), 117-
126.
Colman, J. J., Swanson, A. L., Meinardi, S., Sive, B. C., Blake, D. R., and Rowland, F. S.,
2001. Description of the analysis of a wide range of volatile organic compounds in whole air
samples collected during PEM-Tropics A and B. Analytical Chemistry, 73(15), 3723-3731.
Cox, R. A., and Roffey, M. J., 1977. Thermal decomposition of peroxyacetylnitrate in the
presence of nitric oxide. Environmental Science & Technology, 11(9), 900-906.
Creasey, D. J., Evans, G. E., Heard, D. E., and Lee, J. D., 2003. Measurements of OH and HO2
concentrations in the Southern Ocean marine boundary layer. Journal of Geophysical
Research-Atmosphere, 108 (D15), doi.org/10.1029/2002JD003206.
234
Creilson, J. K., Fishman, J., Wozniak, A. E., 2003. Intercontinental transport of tropospheric
ozone: a study of its seasonal variability across the North Atlantic utilizing tropospheric ozone
residuals and its relationship to the North Atlantic Oscillation. Atmospheric Chemistry and
Physics, 3, 2053-2066.
Crutzen, P. J., and Kanakidou, M., 1992. Relationship of peroxyacetyl nitrate to active and
total odd nitrogen at northern high latitudes: Influence of reservoir species on NOx and
O3. Journal of Geophysical Research: Atmospheres, 97(D15), 16523-16530.
Dacey, J.W., and Wakeham, S.G., 1986. Oceanic dimethylsulfide: production during
zooplankton grazing on phytoplankton. Science, 233(4770):1314-6.
Dahl, E. E., Yvon‐Lewis, S. A., and Saltzman, E. S., 2005. Saturation anomalies of alkyl
nitrates in the tropical Pacific Ocean. Geophysical research letters, 32(20).
Dahl, E. E., Saltzman, E. S., and De Bruyn, W. J., 2003. The aqueous phase yield of alkyl
nitrates from ROO+NO: Implications for photochemical production in seawater. Geophysical
research letters, 30(6).
Dahl, E. E., Yvon‐Lewis, S. A., and Saltzman, E. S. (2007). Alkyl nitrate (C1-C3) depth
profiles in the tropical Pacific Ocean. Journal of Geophysical Research: Oceans, 112(C1).
Darnall, K. R., Carter, W. P., Winer, A. M., Lloyd, A. C., and Pitts Jr, J. N., 1976. Importance
of RO2 + nitric oxide in alkyl nitrate formation from C4-C6 alkane photooxidations under
simulated atmospheric conditions. The Journal of Physical Chemistry, 80(17), 1948-1950.
Day, D. A., Dillon, M. B., Wooldridge, P. J., Thornton, J. A., Rosen, R. S., Wood, E. C., and
Cohen, R. C., 2003. On alkyl nitrates, O3, and the “missing NOy”. Journal of Geophysical
Research: Atmospheres, 108(D16), doi: 10.1029/2003JD003685.
235
Easter C, Quigley C, Burrowes P, Witherspoon J, and Apgar D, 2005. Odor and air emissions
control using biotechnology for both collection and wastewater treatment systems. Chemical
engineering journal, 113(2-3):93-104.
Eberhard, J., and Howard, C. J., 1996. Temperature-dependent kinetics studies of the reactions
of C2H5O2 and n-C3H7O2 radicals with NO. International journal of chemical kinetics. 28(10),
731-740.
Eberhard, J., and Howard, C. J., 1997. Rate coefficients for the reactions of some C3 to C5
hydrocarbon peroxy radicals with NO. The Journal of Physical Chemistry A, 101(18), 3360-
3366.
Eberhard, J., Villalta, P. W., and Howard, C. J., 1996. Reaction of isopropyl peroxy radicals
with NO over the temperature range 201-401 K. The Journal of Physical Chemistry, 100(3),
993-997.
Farmer, D. K., Perring, A. E., Wooldridge, P. J., Blake, D. R., Baker, A., Meinardi, S., Huey,
L. G., Tanner, D., Vargas, O., and Cohen, R. C., 2011. Impact of organic nitrates on urban
ozone production. Atmospheric Chemistry and Physics, 11(9), 4085-4094.
Fiscus, E. L., Booker, F. L., and Burkey, K. O., 2005. Crop responses to ozone: uptake, modes
of action, carbon assimilation and partitioning. Plant, Cell & Environment, 28(8), 997-1011.
Flocke, F., Atlas, E., Madronich, S., Schauffler, S. M., Aikin, K., Margitan, J. J., and Bui, T.
P., 1998a. Observations of methyl nitrate in the lower stratosphere during STRAT: Implications
for its gas phase production mechanisms. Geophysical research letters, 25(11), 1891-1894.
Flocke, F., Volz-Thomas, A., Buers, H. J., Patz, W., Garthe, H. J., and Kley, D., 1998b. Long-
term measurements of alkyl nitrates in southern Germany: 1. General behavior and seasonal
and diurnal variation. Journal of Geophysical Research: Atmosphere, 103(D5), 5729-5746.
Flocke, F., Volz-Thomas, A., and Kley, D., 1991. Measurements of alkyl nitrates in rural and
polluted air masses. Atmospheric Environment. Part A. General Topics, 25(9), 1951-1960.
236
Friedli, H. R., Atlas, E., Stroud, V. R., Giovanni, L., Campos, T., and Radke, L. F., 2001.
Volatile organic trace gases emitted from North American wildfires. Global biogeochemical
cycles, 15(2), 435-452.
Fry, J. L., Kiendler-Scharr, A., Rollins, A. W., Wooldridge, P. J., Brown, S. S., Fuchs, H.,
Dubé, W., Mensah, A., dal Maso, M., Tillmann, R., Dorn, H. P., Brauers, T., and Cohen, R.C.,
2009. Organic nitrate and secondary organic aerosol yield from NO3 oxidation of β-pinene
evaluated using a gas-phase kinetics/aerosol partitioning model. Atmospheric Chemistry and
Physics, 9(4), 1431-1449.
Gao, J., Wang, T., Ding, A. J., and Liu, C. B., 2005. Observation study of ozone and carbon
monoxide at the summit of mount Tai (1534 m a.s.l.) in central-eastern China. Atmospheric
Environment, 39, 4779-4791.
Guenther, A. L. E. X., 1999. Modeling biogenic volatile organic compound emissions to the
atmosphere. Reactive hydrocarbons in the atmosphere, 97-118.
Guenther, A. B., Zimmerman, P. R., Harley, P. C., Monson, R. K., Fall, R., 1993. Isoprene and
monoterpene emission rate variability: Model evaluations and sensitivity analyses. Journal of
Geophysical Research-Atmospheres, 98, 12609-12617.
Guo, H., Cheng, H. R., Ling, Z. H., Louie, P. K. K., Ayoko, G. A., 2011b. Which emission
sources are responsible for the volatile organic compounds in the atmosphere of Pearl River
Delta? Journal of Hazardous Materials, 188: 116-124.
Guo, H., Jiang, F., Cheng, H. R., Simpson, I. J., Wang, X. M., Ding, A. J., Wang, T. J., Saunders,
S. M., Wang, T., Lam, S. H. M., Blake, D. R., Zhang, Y. L., and Xie, M., 2009. Concurrent
observations of air pollutants at two sites in the Pearl River Delta and the implication of
regional transport. Atmospheric Chemistry and Physics, 9(19), 7343-7360.
Guo, H., Ling, Z. H., Cheung, K., Jiang, F., Wang, D. W., Simpson, I. J., Barletta, B., Meinardi,
S., Wang, T. J., Wang, X. M., Saunders, S. M., and Blake, D. R., 2013. Characterization of
237
photochemical pollution at different elevations in mountainous areas in Hong
Kong. Atmospheric Chemistry and Physics, 13(8), 3881-3898.
Guo, H., Ling, Z. H., Simpson, I. J., Blake, D. R., and Wang, D. W., 2012. Observations of
isoprene, methacrolein (MAC) and methyl vinyl ketone (MVK) at a mountain site in Hong
Kong. Journal of Geophysics Research, 117, D19303, doi:10.1029/2012JD017750.
Guo, H., So, K. L., Simpson, I. J., Barletta, B., Meinardi, S., and Blake, D. R., 2007. C1-C8
volatile organic compounds in the atmosphere of Hong Kong: overview of atmospheric
processing and source apportionment. Atmospheric Environment, 41(7), 1456-1472.
Guo, H., Wang, T., Blake, D. R., Simpson, I. J., Kwok, Y. H., and Li, Y. S., 2006. Regional
and local contributions to ambient non-methane volatile organic compounds at a polluted
rural/coastal site in Pearl River Delta, China. Atmospheric Environment, 40, 2345-2359.
Guo, H., Wang, T., and Louie, P. K. K., 2004. Source apportionment of ambient non-methane
hydrocarbons in Hong Kong: application of a principal component analysis/absolute principal
component scores (PCA/APCS) receptor model. Environmental Pollution, 129(3), 489-498.
Guo, H., Zou, S. C., Tsai, W. Y., Chan, L. Y., and Blake, D. R., 2011a. Emission characteristics
of nonmethane hydrocarbons from private cars and taxis at different driving speeds in Hong
Kong. Atmospheric environment, 45(16), 2711-2721.
HKCSD (Hong Kong Census and Statistics Department), 2010. Hong Kong Energy Statistics:
Annual Report. http:// www.censtatd.gov.hk.
HKEPD. Air Quality in Hong Kong 2016. http://www.aqhi.gov.hk/en/download/air-quality-
reportse469.html?showall=&start=1.
HKEPD. Hong Kong Air Pollutant Emission Inventory. http://www.epd.gov.hk/epd/english
/environmentinhk/air/data/emission_inve.html 2015b.
HKEPD. Pearl River Delta Regional Air Quality Monitoring Network Report
http://www.epd.gov.hk/epd/english/resources_pub/publications/m_report.html 2015a.
238
Ho, K. F., Lee, S. C., Guo, H., and Tsai, W.Y., 2004. Seasonal and diurnal variations of volatile
organic compounds (VOCs) in the atmosphere of Hong Kong. Science of The Total
Environment, 322, 155-166.
Horii, C. V., Munger, J. W., Wofsy, S. C., Zahniser, M., Nelson, D., and McManus, J. B., 2004.
Fluxes of nitrogen oxides over a temperate deciduous forest. Journal of Geophysical Research:
Atmospheres, 109(D8).
Holzworth, G. C., 1967. Mixing Depths, Wind Speeds and Air Pollution Potential for Selected
Locations in the United States. Journal of Applied Meteorology, 6, 1039-1044.
Horowitz, L. W., Fiore, A. M., Milly, G. P., Cohen, R. C. , Perring, A., Wooldridge, P. J., Hess,
P. G., Emmons, L. K., and Lamarque, J. F., 2007. Observational constraints on the chemistry
of isoprene nitrates over the eastern United States. Journal of Geophysical Research:
Atmosphere, 112, D12S08, doi:10.1029/2006JD007747.
Huang, J. P., Fung, J. C., Lau, A. K., and Qin, Y., 2005. Numerical simulation and process
analysis of typhoon-related ozone episodes in Hong Kong. Journal of Geophysical Research:
Atmospheres, 110(D5).
Jacobson, M. Z., 2005. Fundamentals of Atmospheric Modelling: Cambridge University Press.
Jenkin, M. E., Saunders, S. M., and Pilling, M. J., 1997. The tropospheric degradation of
volatile organic compounds: a protocol for mechanism development. Atmospheric
Environment, 31(1), 81-104.
Jenkin, M. E., Young, J. C., and Rickard, A. R., 2015. The MCM v3.3.1 degradation scheme
for isoprene. Atmospheric Chemistry and Physics, 15(20), 11433-11459.
Jiang, F., Guo, H., Wang, T. J., Cheng, H. R., Wang, X. M., Simpson, I. J., Ding, A. J., Saunders,
S. M., Lam, S. H. M., Blake, D. R., 2010. An O3 episode in the Pearl River Delta: Field
observation and model simulation. Journal of Geophysical Research, 115,
http//dx.doi.org/10.1029/2009JD013583.
239
Jiang, F., Wang, T., Wang, T., Xie, M., and Zhao, H., 2008. Numerical modeling of a
continuous photochemical pollution episode in Hong Kong using WRF-chem. Atmospheric
Environment, 42(38), 8717-8727.
Kastler, J., Jarman, W., and Ballschmiter, K., 2000. Multifunctional organic nitrates as
constituents in European and US urban photo-smog. Fresenius' journal of analytical chemistry,
368(2), 244-249.
Katzenstein, A. S., Doezema, L. A., Simpson, I. J., Blake, D. R., and Rowland, F. S., 2003.
Extensive regional atmospheric hydrocarbon pollution in the southwestern United
States. Proceedings of the National Academy of Sciences, 100(21), 11975-11979.
Kwok, E. S. C. and Atkinson, R., 1995. Estimation of hydroxyl radical reaction-rate constants
for gas-phase organic-compounds using a structure-reactivity relationship-an update.
Atmospheric Environment, 29, 1685-1695.
Lam, K. S., Wang, T. J., Wu, C. L., and Li, Y. S., 2005. Study on an ozone episode in hot
season in Hong Kong and transboundary air pollution over Pearl River Delta region of China.
Atmospheric Environment, 39, 1967-1977.
Lam, S. H. M., Saunders, S. M., Guo, H., Ling, Z. H., Jiang, F., Wang, X. M., and Wang, T. J.,
2013. Modelling VOC source impacts on high ozone episode days observed at a mountain
summit in Hong Kong under the influence of mountain-valley breezes. Atmospheric
Environment, 81, 166-176.
Lam, W. H. K., Tang, Y. F., Chan, K. S., Tam, M. L., 2006. Short-term hourly traffic forecasts
using Hong Kong Annual Traffic Census. Transportation, 33, 291-310.
Lau, A. K. H., Yuan, Z., Yu, J. Z., and Louie, P. K., 2010. Source apportionment of ambient
volatile organic compounds in Hong Kong. Science of the total environment, 408(19), 4138-
4149.
240
Lee, E., Chan, C. K., and Paatero, P., 1999. Application of positive matrix factorization in
source apportionment of particulate pollutants in Hong Kong. Atmospheric
Environment, 33(19), 3201-3212.
Lee, S.C., Chiu, M.Y., Ho, K.F., Zou, S.C., Wang, X., 2002. Volatile organic compounds
(VOCs) in urban atmosphere of Hong Kong. Chemosphere, 48(3):375-82.
Lefer, B. L., Talbot, R. W., and Munger, J. W., 1999. Nitric acid and ammonia at a rural
northeastern US site. Journal of Geophysical Research: Atmospheres, 104(D1), 1645-1661.
Legates, D. R., and McCabe, G. J., 1999. Evaluating the use of “goodness-of-fit” measures in
hydrologic and hydroclimatic model validation. Water resources research, 35(1), 233-241.
Lewis, A. C., Carslaw, N., Marriott, P. J., Kinghorn, R. M., Morrison, P., Lee, A. L., Bartle, K.
D. and Pilling, M. J., 2000. A larger pool of ozone-forming carbon compounds in urban
atmospheres. Nature, 405(6788), 778-781.
Li, X., Poon, C. S., and Liu, P. S., 2001. Heavy metal contamination of urban soils and street
dusts in Hong Kong. Applied geochemistry, 16(11), 1361-1368.
Lightfoot, P. D., Cox, R. A., Crowley, J. N., Destriau, M., Hayman, G. D., Jenkin, M. E.,
Moortgat, G. K., and Zabel, F., 1992. Organic peroxy radicals: kinetics, spectroscopy and
tropospheric chemistry. Atmospheric Environment Part A. General Topics, 26(10), 1805-1961.
Lin, J., Pan, D., Davis, S. J., Zhang, Q., He, K., Wang, C., Streets, D. G., Wuebbles, D. J., and
Guan, D., 2014. China’s international trade and air pollution in the United States. Proceedings
of the National Academy of Sciences, 111(5), 1736-1741.
Ling, Z. H., and Guo, H., 2014. Contribution of VOC sources to photochemical ozone
formation and its control policy implication in Hong Kong. Environmental Science & Policy,
38, 180-191.
241
Ling, Z. H., Guo, H., Cheng, H. R., and Yu, Y. F., 2011. Sources of ambient volatile organic
compounds and their contributions to photochemical ozone formation at a site in the Pearl
River Delta, southern China. Environmental Pollution, 159, 2310-2319.
Ling, Z. H., Guo, H., Lam, S. H. M., Saunders, S. M., and Wang, T., 2014. Atmospheric
photochemical reactivity and ozone production at two sites in Hong Kong: Application of a
master chemical mechanism–photochemical box model. Journal of Geophysical Research:
Atmospheres, 119(17), 10567-10582.
Ling, Z., Guo, H., Simpson, I.J., Saunders, S.M., Lam, S.H., Lyu, X., Blake, D.R., 2016. New
insight into the spatiotemporal variability and source apportionments of C 1–C 4 alkyl nitrates
in Hong Kong. Atmospheric Chemistry and Physics, 16(13):8141-56.
Ling, Z. H., Guo, H., Zheng, J. Y., Louie, P. K. K., Cheng, H. R., Jiang, F., Cheung, K., Wong,
L.C., and Feng, X. Q., 2013. Establishing a conceptual model for photochemical ozone
pollution in subtropical Hong Kong. Atmospheric environment, 76, 208-220.
Lobert, J. M., Scharffe, D. H., Hao, W., Kuhlbusch, T. A., Seuwen, R., Warneck, P., and
Crutzen, P. J., 1991. Experimental evaluation of biomass burning emissions: Nitrogen and
carbon containing compounds. In Global biomass burning. Atmospheric, climatic, and
biospheric implications.
Lou, S., Holland, F., Rohrer F., Lu, K., Bohn, B., Brauers, T., Chang, C. C., Fuchs, H., Haseler,
R., Kita, K., Kondo, Y., Li, X., Shao, M., Zeng, L., Wahner, A., Zhang, Y., Wang, W., and
Hofzumahaus, A., 2010. Atmospheric OH reactivities in the Pearl River Delta-China in
summer 2006: measurement and model results. Atmospheric Chemistry and Physics, 10,
11243-11260.
Louie, P. K., Watson, J. G., Chow, J. C., Chen, A., Sin, D. W., and Lau, A. K., 2005. Seasonal
characteristics and regional transport of PM 2.5 in Hong Kong. Atmospheric
Environment, 39(9), 1695-1710.
242
Lyu, X., Guo, H., Simpson, I.J., Meinardi, S., Louie, P.K., Ling, Z., Wang, Y., Liu, M., Luk,
C.W., Wang, N., Blake, D.R., 2016. Effectiveness of replacing catalytic converters in LPG-
fueled vehicles in Hong Kong. Atmospheric Chemistry and Physics, 16(10):6609-26.
Lyu, X.P., Liu, M., Guo, H., Ling, Z.H., Wang, Y., Louie, P.K., Luk, C.W., 2016.
Spatiotemporal variation of ozone precursors and ozone formation in Hong Kong: Grid field
measurement and modelling study. Science of the Total Environment, 569:1341-9.
Lyu, X.P., Zeng, L.W., Guo, H., Simpson, I.J., Ling, Z.H., Wang, Y., Murray, F., Louie, P.K.,
Saunders, S.M., Lam, S.H., Blake, D.R., 2017. Evaluation of the effectiveness of air pollution
control measures in Hong Kong. Environmental Pollution, 220:87-94.
Lu, K. D., Rohrer F., Holland, F., Fuchs, H., Bohn, B., Brauers, T., Chang, C. C., Haseler, R.,
Hu, M., Kita, K., Kondo, Y., Li, X., Lou, S. R., Nehr, S., Shao, M., Zeng, L. M., Wahner, A.,
Zhang, Y. H., and Hofzumahaus, A., 2012. Observation and modelling of OH and HO2
concentrations in the Pearl River Delta 2006: a missing OH source in a VOC rich atmosphere.
Atmospheric Chemistry and Physics, 12, 1541-1569.
Luke, W. T., Dickerson, R. R., and Nunnermacker, L. J., 1989. Direct measurements of the
photolysis rate coefficients and Henry's law constants of several alkyl nitrates. Journal of
Geophysical Research: Atmospheres, 94(D12), 14905-14921.
Madronich, S., and Flocke, S.: Theoretical estimation of biologically effective UV radiation at
the Earth’s surface, In: Zerefos, C. (Ed.), Solar Ultraviolet Radiation-Modeling, Measurements
and Effects, NATO ASI Series, vol. I52. Springer-Verlag, Berlin, 1997.
Marandino, C. A., Tegtmeier, S., Krüger, K., Zindler, C., Atlas, E. L., Moore, F., and Bange,
H. W., 2013. Dimethylsulphide (DMS) emissions from the western Pacific Ocean: a potential
marine source for stratospheric sulphur?, Atmospheric Chemistry and Physics, 13, 8427-8437,
doi:10.5194/acp-13-8427-2013.
243
Martien, P. T., Harley, R. A., Milford, J. B., and Russell, A. G., 2003. Evaluation of incremental
reactivity and its uncertainty in Southern California. Environmental Science and Technology,
37, 1598-1608.
Ma, Z., Hu, X., Sayer, A. M., Levy, R., Zhang, Q., Xue, Y., Tong, S., Bi, J., Huang, L., and
Liu, Y., 2016. Satellite-based spatiotemporal trends in PM2.5 concentrations: China, 2004-2013.
Environmental Health Perspectives (Online), 124(2), 184.
Michelsen, H..A., Salawitch, R. J., Wennberg, P. O., Anderson, J. G., 1994. Production of O
(¹D) from photolysis of O3. Geophysical research letters. 21(20):2227-30.
Moxim, W. J., Levy, H., and Kasibhatla, P. S., 1996. Simulated global tropospheric PAN: Its
transport and impact on NOx. Journal of Geophysical Research: Atmospheres, 101(D7),
12621-12638.
Murphy, J. G., Day, D. A., Cleary, P. A., Wooldridge, P. J., and Cohen, R. C., 2006.
Observations of the diurnal and seasonal trends in nitrogen oxides in the western Sierra
Nevada. Atmospheric Chemistry and Physics, 6(12), 5321-5338.
NAAQS, 2012. National Ambient Air Quality Standard of China, GB3095-2012.
Norris, G., Duvall, R., Brown, S., and Bai, S., 2014. EPA Positive Matrix Factorization (PMF)
5.0 Fundamentals and User Guide. U.S. Environmental Protection Agency, Washington, DC,
EPA/600/R-14/108 (NTIS PB2015-105147).
Neff, J. C., Holland, E. A., Dentener, F. J., McDowell, W. H., and Russell, K. M., 2002. The
origin, composition and rates of organic nitrogen deposition: A missing piece of the nitrogen
cycle?. In The nitrogen cycle at regional to global scales (pp. 99-136). Springer Netherlands.
Neu, J. L., Lawler, M. J., Prather, M. J., and Saltzman, E. S., 2008. Oceanic alkyl nitrates as a
natural source of tropospheric ozone. Geophysical Research Letters, 35(13), L13814,
doi:10.1029/2008GL034189.
244
Nielsen, O. J., Sidebottom, H. W., Donlon, M., and Treacy, J., 1991. An absolute-and relative-
rate study of the gas-phase reaction of OH radicals and Cl atoms with n-alkyl nitrates. Chemical
physics letters, 178(2-3), 163-170.
Nowak, J. B., Davis, D. D., Chen, G., Eisele, F. L., Mauldin III, R. L., Tanner, D. J., Cantrell,
C., Kosciuch, E., Bandy, A., Thornton, D., Clarke, A., 2001. Airborne observations of DMSO,
DMS, and OH at marine tropical latitudes. Geophysical Research Letter, 28, 2201-2204.
O'Brien, J. M., Shepson, P. B., Muthuramu, K., Hao, C., Niki, H., Hastie, D. R., Taylor, R.,
and Roussel, P. B., 1995. Measurements of alkyl and multifunctional organic nitrates at a rural
site in Ontario. Journal of Geophysical Research: Atmospheres, 100(D11), 22795-22804.
Ou, J., Guo, H., Zheng, J., Cheung, K., Louie, P. K., Ling, Z., and Wang, D., 2015.
Concentrations and sources of non-methane hydrocarbons (NMHCs) from 2005 to 2013 in
Hong Kong: A multi-year real-time data analysis. Atmospheric Environment, 103, 196-206.
Paatero, P., 1997. Least squares formulation of robust non-negative factor analysis.
Chemometrics and intelligent laboratory systems, 37(1), 23-35.
Paatero, P., and Tapper, U., 1994. Positive matrix factorization: A non‐negative factor model
with optimal utilization of error estimates of data values. Environmetrics, 5(2), 111-126.
Perring, A. E., Bertram, T. H., Farmer, D. K., Wooldridge, P. J., Dibb, J., Blake, N. J., Blake,
D. R., Singh, H. B., Fuelberg, H., Diskin, G., and Sachse, G., 2010. The production and
persistence of ΣRONO2 in the Mexico City plume. Atmospheric Chemistry and Physics, 10(15),
7215-7229.
Perring, A. E., Pusede, S. E., and Cohen, R. C., 2013. An observational perspective on the
atmospheric impacts of alkyl and multifunctional nitrates on ozone and secondary organic
aerosol. Chemical Reviews, 113(8), 5848-5870.
245
Pinho, P. G., Pio, C. A., and Jenkin, M. E., 2005. Evaluation of isoprene degradation in the
detailed tropospheric chemical mechanism, MCM v3, using environmental chamber data.
Atmospheric Environment, 39(7), 1303-1322.
Poschl, U., Von Kuhlmann, R., Poisson, N., and Crutzen, P. J., 2000. Development and
intercomparison of condensed isoprene oxidation mechanisms for global atmospheric
modeling. Journal of Atmospheric Chemistry, 37(1), 29-52.
Reeves, C. E., Slemr, J., Oram, D. E., Worton, D., Penkett, S. A., Stewart, D. J., Purvis, R.,
Watson, N., Hopkins, J., Lewis, A., Methven, J., Blake, D. R., and Atlas, E., 2007. Alkyl
nitrates in out- flow from North America over the North Atlantic during intercontinental
transport of ozone and precursors 2004. Journal of Geophysical Research, 112, D10S037,
doi:10.1029/2006JD007567.
Ridley, B. A., Shetter, J. D., Grandrud, B. W., Salas, L. J., Singh, H. B., Carroll, M. A., Hubler,
G., Albritton, D. L., Hastie, D. R., Schiff, H. I., Mackay, G. I., Karecki, D. R., Davis, D. D.,
Bradshaw, J. D., Rodgers, M. O., Sandholm, S. T., Torres, A. L., Condon, E. P., Gregory, G.
L., and Beck, S. M., 1990. Ratios of peroxyacetyl nitrate to active nitrogen observed during
aircraft flights over the eastern Pacific Ocean and continental United States. Journal of
Geophysical Research: Atmospheres, 95(D7), 10179-10192.
Roberts, J. M., Bertman, S. B., Parrish, D. D., Fehsenfeld, F. C., Jobson, B. T., and Niki, H.,
1998. Measurement of alkyl nitrates at Chebogue Point, Nova Scotia during the 1993 North
Atlantic Regional Experiment (NARE) intensive. Journal of Geophysical Research:
Atmospheres, 103(D11), 13569-13580.
Roberts, J. M., and Fajer, R. W., 1989. UV absorption cross sections of organic nitrates of
potential atmospheric importance and estimation of atmospheric lifetimes. Environmental
Science and Technology, 23(8), 945-951.
246
Rollins, A. W., Pusede, S., Wooldridge, P., Min, K. E., Gentner, D. R., Goldstein, A. H., Liu,
S., Day. D. A., Russell, L. M., Rubitschun, C. L., Surratt, J. D., and Cohen, R. C., 2013.
Gas/particle partitioning of total alkyl nitrates observed with TD-LIF in Bakersfield. Journal
of Geophysical Research: Atmospheres, 118(12), 6651-6662.
Rosen, R. S., Wood, E. C., Wooldridge, P. J., Thornton, J. A., Day, D. A., Kuster, W., Williams,
E. J., Jobson, B. T., and Cohen, R. C., 2004. Observations of total alkyl nitrates during Texas
Air Quality Study 2000: Implications for O3 and alkyl nitrate photochemistry. Journal of
Geophysical Research: Atmosphere (1984–2012), 109(D7). doi: 10.1029/2003JD004227.
Russo, R. S., Zhou, Y., Haase, K. B., Wingenter, O. W., Frinak, E. K., Mao, H., Talbot, R. W.,
and Sive, B. C., 2010. Temporal variability, sources, and sinks of C1-C5 alkyl nitrates in coastal
New England. Atmospheric Chemistry and Physics, 10(4), 1865-1883.
Saunders, S. M., Jenkin, M. E., Derwent, R. G., and Pilling, M. J., 2003. Protocol for the
development of the Master Chemical Mechanism, MCM v3 (Part A): tropospheric degradation
of non-aromatic volatile organic compounds. Atmospheric Chemistry and Physics, 3(1), 161-
180.
Scholtens, K. W., Messer, B. M., Cappa, C. D., and Elrod, M. J., 1999. Kinetics of the CH3O2+
NO reaction: temperature dependence of the overall rate constant and an improved upper limit
for the CH3ONO2 branching channel. Journal of Physical Chemistry A, 103(22), 4378-4384.
Seinfeld, J. H., Pandis, S. N., 1997. Atmospheric Chemistry and Physics: From Air Pollution
to Climate Change. John Wiley & Sons, Inc..
Sharma, P., Kuniyal, J. C., Chand, K., Guleria, R. P., Dhyani, P. P., and Chauhan, C., 2013.
Surface ozone concentration and its behaviour with aerosols in the northwestern Himalaya,
India. Atmospheric environment, 71, 44-53.
247
Shepson, P. B., Anlauf, K. G., Bottenheim, J. W., Wiebe, H. A., Gao, N., Muthuramu, K., and
Mackay, G. I., 1993. Alkyl nitrates and their contribution to reactive nitrogen at a rural site in
Ontario. Atmospheric Environment. Part A. General Topics, 27(5), 749-757.
Simpson, I. J., Akagi, S. K., Barletta, B., Blake, N. J., Choi, Y., Diskin, G. S., Fried, A.,
Fuelberg, H. E., Meinardi, S., Rowland, F. S., Vay, S. A., Weinheimer, A. J., Wennberg, P. O.,
Wiebring, P., Wisthaler, A., Yang, M., Yokelson, R. J., and Blake, D. R., 2011. Boreal forest
fire emissions in fresh Canadian smoke plumes: C1-C10 volatile organic compounds (VOCs),
CO2, CO, NO2, NO, HCN and CH3CN. Atmospheric Chemistry and Physics, 11(13), 6445-
6463.
Simpson, I. J., Blake, N. J., Barletta, B., Diskin, G. S., Fuelberg, H. E., Gorham, K., Huey, L.
G., Meinardi, S., Rowland, F. S., Vay, S. A., Weinheimer, A. J., Yang, M., and Blake, D. R.,
2010. Characterization of trace gases measured over Alberta oil sands mining operations: 76
speciated C2-C10 volatile organic compounds (VOCs), CO2, CH4, CO, NO, NO2, NOy, O3 and
SO2. Atmospheric Chemistry and Physics, 10(23), 11931-11954.
Simpson, I. J., Blake, N. J., Blake, D. R., Atlas, E., Flocke, F., Crawford, J. H., Fuelberg, H.
E., Kiley, C. M., Meinardi, S., and Rowland, F. S., 2003. Photochemical production and
evolution of selected C2-C5 alkyl nitrates in tropospheric air influenced by Asian
outflow. Journal of Geophysical Research: Atmospheres, 108(D20).
Simpson, I. J., Wang, T., Guo, H., Kwok, Y. H., Flocke, F., Atlas, E., Meinardi, S., Rowland,
F.S., and Blake, D. R., 2006. Long-term atmospheric measurements of C1-C5 alkyl nitrates in
the Pearl River Delta region of southeast China. Atmospheric Environment, 40(9), 1619-1632.
Simpson, I. J., Meinardi, S., Blake, D. R., Blake, N. J., Rowland, F. S., Atlas, E., and Flocke,
F., 2002. A biomass burning source of C1-C4 alkyl nitrates. Geophysical Research
Letters, 29(24).
248
Sin, D.W., Wong, Y.C., Louie, P.K., 2000. Monitoring of ambient volatile organic compounds
at two urban sites in Hong Kong from 1997 to 1998. Indoor and Built Environment, 9(3-4):216-
27.
Singh, H. B., and Hanst, P. L., 1981. Peroxyacetyl nitrate (PAN) in the unpolluted atmosphere:
An important reservoir for nitrogen oxides. Geophysical Research Letters, 8(8), 941-944.
Singh, H. B., Herlth, D., O'hara, D., Zahnle, K., Bradshaw, J. D., Sandholm, S. T., Talbot, R.,
Cox, R. A., and Roffey, M. J., 1977. Thermal decomposition of peroxyacetyl nitrate in the
presence of nitric oxide. Environmental Science & Technology, 11(9), 900-906.
Singh, H.B., O'Hara, D., Herlth, D., Bradshaw, J.D., Sandholm, S.T., Gregory, G.L., Sachse,
G.W., Blake, D.R., Crutzen, P.J., Kanakidou, M.A., 1992. Atmospheric measurements of
peroxyacetyl nitrate and other organic nitrates at high latitudes: Possible sources and sinks.
Journal of Geophysical Research: Atmospheres, 97(D15):16511-22.
Singh, H. B., Viezee, W., Chen, Y., Thakur, A. N., Kondo, Y., Talbot, R. W., Gregory, G. L.,
Sachse, G. W., Blake, D. R., Bradshaw, J. D., Wang, Y., and Jacob, D. J., 1998. Latitudinal
distribution of reactive nitrogen in the free troposphere over the Pacific Ocean in late
winter/early spring. Journal of Geophysical Research: Atmospheres, 103(D21), 28237-28246.
So, K.L., Wang, T., 2004. C3–C12 non-methane hydrocarbons in subtropical Hong Kong:
spatial–temporal variations, source–receptor relationships and photochemical reactivity.
Science of the Total Environment, 328(1-3):161-74.
So, K. L., and Wang, T., 2003. On the local and regional influence on ground-level ozone
concentrations in Hong Kong. Environmental Pollution, 123(2), 307-317.
Sommariva, R., Trainer, M., de Gouw, J. A., Roberts, J. M., Warneke, C., Atlas, E., Flocke, F.,
Goldan, P. D., Kuster, W. C., Swanson, A. L., and Fehsenfeld, F. C., 2008. A study of organic
249
nitrates formation in an urban plume using a Master Chemical Mechanism. Atmospheric
Environment, 42(23), 5771-5786.
Stroud, C., Madronich, S., Atlas, E., Ridley, B., Flocke, F., Weinheimer, A., Talbot, B., Fried,
A., Wert, B., Shetter, R., Lefer, B., 2003. Photochemistry in the arctic free troposphere: NOx
budget and the role of odd nitrogen reservoir recycling. Atmospheric Environment,
37(24):3351-64.
Symth, S., Bradshaw, J., Sandholm, S., Liu, S., Mckeen, S., Gregory, G., Anderson, B., Talbot,
R., Blake, D., Rowland, S., Browell, E., Fenn, M., Merrill, J., Backmeier, S., Sachse, G.,
Collins, J., Thornton, D., Davis, D., Singh, H., 1996. Comparison of free trospheric western
Pacific air mass classification schemes for the PEM-West A experiment. Journal of
Geophysical Research, 101, 1743-1762.
Takekawa, H., Chatani, S., and Akiyoshi, I., 2013. A new approach for estimation of the effect
of NOx emission reduction on roadside NO2 concentration in Tokyo. Atmospheric Environment,
68, 92-102.
Talbot, R., Dibb, J., Scheuer, E., Seid, G., Russo, R., Sandholm, S., Tan, D., Singh, H., Blake,
D., Atlas, E., Sachse, G., Jordan, C., and Avery, M., 2003. Reactive nitrogen in Asian
continental outflow over the western Pacific: Results from the NASA Transport and Chemical
Evolution over the Pacific (TRACE‐P) airborne mission. Journal of Geophysical Research:
Atmospheres, 108(D20).
Tang, J. H., Chan, L. Y., Chan, C. Y., Li, Y. S., Chang, C. C., Liu, S. C., Wu, D., and Li, Y.
D., 2007. Characteristics and diurnal variations of NMHCs at urban, suburban, and rural sites
in the Pearl River Delta and a remote site in South China. Atmospheric Environment, 41(38),
8620-8632.
Thornberry, T., Carroll, M. A., Keeler, G. J., Sillman, S., Bertman, S. B., Pippin, M. R., Ostling,
K., Grossenbacher, J. W., Shepson, P. B., Cooper, O. R., Moody, J. L., and Stockwell, W. R.,
250
2001. Observations of reactive oxidized nitrogen and speciation of NOy during the PROPHET
summer 1998 intensive. Journal of Geophysical Research: Atmospheres, 106(D20), 24359-
24386.
USEPA. EPA Positive Matrix Factorization (PMF) 3.0 Fundamentals & User Guide.
www.epa.gov 2008.
USEPA. User's Guide for the Industrial Source Complex (ISC3) Dispersion Models, Volume
II Description of Model Algorithms. EPA-454/B-95-003b. U.S. Environmental Protection
Agency, Research Triangle Park, NC. 1985.
Walega, J. G., Ridley, B. A., Madronich, S., Grahek, F. E., Shetter, J. D., Sauvain, T. D., Hahn,
C. J., Merill, J. T., Bodhaine, B. A., and Robinson, E., 1992. Observations of peroxyacetyl
nitrate, peroxypropionyl nitrate, methyl nitrate and ozone during the Mauna Loa Observatory
Photochemistry Experiment. Journal of Geophysical Research: Atmospheres, 97(D10), 10311-
10330.
Wang, M., Shao, M., Chen, W. T., Lu, S. H., Wang, C., Huang, D. K., Yuan, B., Zeng, L. M.,
and Zhao, Y., 2013. Measurements of C1–C4 alkyl nitrates and their relationships with carbonyl
compounds and O3 in Chinese cities. Atmospheric Environment, 81, 389-398.
Wang, N., Guo, H., Jiang, F., Ling, Z. H., Wang, T., 2015. Simulation of O3 formation at
different elevations in mountainous area of Hong Kong using WRF-CMAQ model. Science of
the Total Environment, 505, 939-951.
Wang, T., Poon, C. N., Kwok, Y. H., and Li, Y. S., 2003. Characterizing the temporal
variability and emission patterns of pollution plumes in the Pearl River Delta of China.
Atmospheric Environment, 37, 3539-3550.
Wang, T., Wei, X. L., Ding, A. J., Poon, S. C., Lam, K. S., Li, Y. S., Chan, L. Y., and Anson,
M., 2009. Increasing surface ozone concentrations in the background atmosphere of Southern
China, 1994-2007. Atmospheric Chemistry and Physics, 9, 6217-6227.
251
Wang, T., Wong, H. L. A., Tang, J., Ding, A., Wu, W. S., and Zhang, X. C., 2006. On the
origin of surface ozone and reactive nitrogen observed at a remote site in the northeastern
Qinghai Tibetan Plateau, western China. Journal of Geophysical Research, 111, D08303,
doi:10.1029/2005JD006527, 2006.
Wang, Y., Wang, H., Guo, H., Lyu, X. P., Cheng, H. R., Ling, Z. H., Louie, P. K. K., Simpson,
I. J., Meinardi, S., Blake, D. R., 2017. Long-term O3-precursor relationships in Hong Kong:
field observation and model simulation. Atmospheric Chemistry and Physics, 17, 10919-10935.
Williams, J. E., Le Bras, G., Kukui, A., Ziereis, H., and Brenninkmeijer, C. A. M., 2014. The
impact of the chemical production of methyl nitrate from the NO+CH3O2 reaction on the global
distributions of alkyl nitrates, nitrogen oxides and tropospheric ozone: a global modelling
study. Atmospheric Chemistry and Physics, 14(5), 2363-2382.
Willmott, C. J., 1982. Some comments on the evaluation of model performance. Bulletin of the
American Meteorological Society, 63(11), 1309-1313.
Willmott, C. J., Ackleson, S. G., Davis, R. E., Feddema, J. J., Klink, K. M., Legates, D. R.,
O'Donnell, J., and Rowe, C. M., 1985. Statistics for the evaluation and comparison of models.
Journal of Geophysical Research, 90, 8995-9005.
Worton, D. R., Reeves, C. E., Penkett, S. A., Sturges, W. T., Slemr, J., Oram, D. E., Bandy, B.
J., Bloss, W. J., Carslaw, N., Davey, J., Emmerson, K.M., Gravestoc, T. J., Hamilton, J. F.,
Heard, D. E., Hopkins, J. R., Hulse, A., Ingram, T., Jacob, M. J., Lee, J. D., Leigh, R. J., Lewis,
A. C., Monks, P. S., and Smith, S. C., 2010. Alkyl nitrate photochemistry during the
tropospheric organic chemistry experiment. Atmospheric Environment, 44(6), 773-785.
Xia, L., Shao, Y., 2005. Modelling of traffic flow and air pollution emission with application
to Hong Kong Island. Environmental Modelling & Software, 20, 1175-1188.
Xu, Z., Wang, T., Xue, L. K., Louie, P. K. K., Luk, C. W. Y., Gao, J., Wang, S. L., Chai, F. H.,
Wang, W. X., 2013. Evaluating the uncertainties of thermal catalytic conversion in measuring
252
atmospheric nitrogen dioxide at four differently polluted sites in China. Atmospheric
Environment, 76, 221-226.
Xue, L., Wang, T., Louie, P. K., Luk, C. W., Blake, D. R., and Xu, Z., 2014. Increasing external
effects negate local efforts to control ozone air pollution: a case study of Hong Kong and
implications for other Chinese cities. Environmental science & technology, 48(18), 10769-
10775.
Xue, L. K., Wang, T., Zhang, J. M., Zhang, X. C., Poon, D. C. N., Ding, A. J., Zhou, X. H.,
Wu, W. S., Tang, J., Zhang, Q. Z., and Wang, W. X., 2011. Source of surface ozone and
reactive nitrogen speciation at Mount Waliguan in western China: New insights from the 2006
summer study. Journal of Geophysical Research-Atmospheres, 2011; 116: D07306.
Yokouchi, Y., Ikeda, M., Inuzuka, Y., Yukawa, T., 2002. Strong emission of methyl chloride
from tropical plants. Nature, 416(6877):163.
Yokouchi, Y., Noijiri, Y., Barrie, L.A., Toom-Sauntry, D., Machida, T., Inuzuka, Y., Akimoto,
H., Li, H.J., Fujinuma, Y., Aoki, S., 2000. A strong source of methyl chloride to the atmosphere
from tropical coastal land. Nature, 403(6767):295.
Yuan, B., Liu, Y., Shao, M., Lu, S. H., and Streets, D. G., Biomass burning contributions to
ambient VOCs species at a receptor site in the Pearl River Delta (PRD), China. Environmental
Science & Technology, 44, 4577-4582.
Zeng, L., Lyu, X., Guo, H., Zou, S., and Ling, Z., 2018. Photochemical formation of C1-C5
alkyl nitrates in suburban Hong Kong and over South China Sea. Environmental science &
technology, doi: 10.1021/acs.est.8b00256.
Zhang, J., Wang, T., Chameides, W. L., Cardelino, C., Kwok, J., Blake, D. R., Ding, A., So,
K. L., 2007. Ozone production and hydrocarbon reactivity in Hong Kong, Southern China.
Atmospheric Chemistry and Physics, 7, 557-573.
253
Zhang, L., Moran, M. D., Makar, P. A., Brook, J. R., and Gong, S., 2002. Modelling gaseous
dry deposition in AURAMS: a unified regional air-quality modelling system. Atmospheric
Environment, 36(3), 537-560.
Zhang, Y. H., Su, H., Zhong, L. J., Cheng, Y. F., Zeng, L. M., Wang, X. S., Xiang, Y. R., Wang,
J. L., Gao, D. F., Shao, M., Fan, S. J., and Liu, S. C., 2008. Regional ozone pollution and
observation-based approach for analyzing ozone–precursor relationship during the PRIDE-
PRD2004 campaign. Atmospheric Environment, 42(25), 6203-6218.
Zhu, L., and Ding, C. F., 1997. Temperature dependence of the near UV absorption spectra and
photolysis products of ethyl nitrate. Chemical physics letters, 265(1), 177-184.