Contribution of VOC sources to photochemical ozone formation and its control policy implication in Hong Kong

environmental science & policy 38 (2014) 180–191

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/envsci

Contribution of VOC sources to photochemical ozone formation and its control policy implication in Hong Kong Z.H. Ling a,b, H. Guo a,* a Air Quality Studies, Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong, China b School of Environmental Science and Engineering, Sun Yat-Sen University, Guangzhou, China

article info

abstract

Article history:

Since volatile organic compounds (VOCs) are important precursors of ozone (O3), identifica-

Received 6 August 2013

tion and quantification of their emission sources are prerequisite for the formulation and

Received in revised form

implementation of O3 pollution control policies. In this study, seven major sources of VOCs

18 December 2013

were identified and their contributions to photochemical O3 formation were evaluated at an

Accepted 19 December 2013

urban site (TW) in Hong Kong by the combined application of the positive matrix factoriza-

Available online 7 January 2014

tion (PMF) model and the observation based model (OBM). The relative incremental reactivity (RIR)-weighted values, which considered both the reactivity and abundance of VOCs,

Keywords:

suggested that paint and sealant solvents, diesel exhaust and liquefied petroleum gas (LPG)

VOC sources

usage were the key contributors to O3 formation. Among the identified sources, xylenes and

O3

ethylbenzene in paint and sealant solvents, toluene in gasoline exhaust, butanes, ethene,

Relative ozone reduction efficiency

propene and propane in LPG usage and n-butane and ethene in diesel exhaust made

Control policy

significant contributions. Sensitivity analysis on the basis of relative O3 reduction efficiency (RORE), using the reduction of VOCs from 10% to 90% as input, indicated that the O3 reduction was the most effective when the identified VOC sources and the major species from these sources were cut by specific percentages. The findings provide valuable insights into the formulation and implementation of VOCs and O3 control strategies in Hong Kong. # 2013 Elsevier Ltd. All rights reserved.

1.

Introduction

With rapid urbanization and industrialization in the past two decades, surface ozone (O3) pollution, which determines the oxidative capacity of the atmosphere, reduces visibility and affects human and vegetation health, has been frequently observed in Hong Kong and the rest of Pearl River Delta (PRD) region (HKEPD, 2012a; Wang et al., 2009; Guo et al., 2009). As key O3 precursors, volatile organic compounds (VOCs) are the most important chemicals contributing to high O3 production rates in the PRD region, where O3 formation is sensitive to

VOCs in urban areas (Ling et al., 2011; Cheng et al., 2010; Zhang et al., 2007, 2008). Therefore, identification of VOC sources and quantification of source contributions are fundamental for the formulation and implementation of O3 pollution control measures. In recent years, with increasing recognition of adverse impact of VOCs on photochemical smog and human health, a series of control measures to reduce VOCs emissions have been implemented in Hong Kong. Among them, most are mass-based, focusing on the control of the weight of total VOC emitted such as the air pollution control (VOCs) regulations (HKPD, 2010). Though measurability and practicality are the

* Corresponding author. Tel.: +852 34003962; fax: +852 23346389. E-mail address: [email protected] (H. Guo). 1462-9011/$ – see front matter # 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.envsci.2013.12.004

environmental science & policy 38 (2014) 180–191

major advantage of the mass-based approach, it does not consider the O3 formation potentials of VOCs, which photochemical O3 formation is more correlated with (Avery, 2006). The O3 pollution would be worse by replacing heavier VOCs with lower photochemical reactivity by lighter VOCs with higher photochemical reactivity, as the more reactive VOCs would increase the photochemical O3 formation (HKPD, 2010; Derwent et al., 2007). Reactivity-based methods using maximum incremental reactivity and O3 formation potential can overcome the limitation by considering the contributions of VOC species to photochemical O3 (Chang et al., 2005; Derwent et al., 1998; Carter, 1994; Chameides et al., 1992); however, the two methods just simply estimate O3 formation under optimum or ideal conditions. As a matter of fact, there is no unique relationship between the competitive reaction rates of a set of organic compounds with hydroxyl radical and their ability to produce O3 in the atmosphere because the latter depends on the subsequent reaction mechanisms of the products of the hydroxyl radical attack. Therefore, a newly reactivity-based method, combining the positive matrix factorization (PMF) model with an observation-based model (OBM), was firstly prompted by Ling et al. (2011), which was only applied at a non-urban site in the inland PRD region. The PMF model is a receptor-oriented source apportionment model, which constrains all the elements in the factor score (source contribution) and the factor loading (source identification) matrix to be positive. Although the fundamental assumption of mass conservation may cause uncertainty in the PMF simulation (Hopke, 2003), Na and Kim (2007) have concluded that the reaction loss does not significantly influence the quantification of source contributions by incorporating the reaction loss of the ambient VOCs in a receptor model. To date, PMF model has been applied extensively and provided robust results in identifying and quantifying the sources of VOCs in different areas in the world, including urban, suburban, rural and background locations (Guo et al., 2011a; Ling et al., 2011; Lauz et al., 2008, 2009; Song et al., 2007; Xie and Bekowitz, 2006). In Hong Kong, Guo et al. (2011a) reported that vehicular emissions and solvent use contributed 48% and 43% to ambient VOCs, respectively, at a suburban site in 2007. Lau et al. (2010) identified 9 sources of VOCs at four sites in Hong Kong in 2002–2003 and 2006–2007 by the same approach, concluding that vehicle and marine vessel related sources and liquefied petroleum gas (LPG) were the most significant local sources. However, most previous studies of the source identification and evaluation regarded each individual VOC as equally important to the O3 formation, without considering actual difference in O3 formation potentials of individual compounds. Hence, the relative importance of potential VOC sources to the O3 formation could be misled. On the other hand, the OBM, which uses observed concentrations of O3 and its precursors (i.e., VOCs, NOx and CO), as well as meteorological data measured as input, is an useful tool to investigate the relationships between O3 and its precursors at given locations based on the carbon bond IV mechanism (Cardelino and Chameides, 1995; Zhang et al., 2007, 2008; Cheng et al., 2010). It should be noted that different from emission-based models, the OBM simulates O3 photochemical production and destruction based on observed ambient concentrations of O3 and its precursors, which can

181

avoid the uncertainties caused by emission inventories and the simulated boundary layer dynamic (Russell and Dennis, 2000; Ling et al., 2011). Though the PMF and OBM models have been used in several studies in the PRD region (Guo et al., 2011a; Lau et al., 2010; Cheng et al., 2010; Zhang et al., 2007, 2008), they are separately adopted to investigate the source apportionment of VOCs and/ or the VOCs/NOx-sensitivity relationships to O3 formation in the region. Neither of them can solely quantify the VOC source contributions to the O3 formation. There is little knowledge about the roles of specific VOC sources and species of individual sources in the formation of O3 in Hong Kong, where different VOC control measures from those in inland PRD have been taken and still, severe O3 pollution is frequently observed (Guo et al., 2009; Wang et al., 2009). As such, the combination of these two models was firstly used to investigate the contributions of different VOC sources and their species to the O3 formation in Hong Kong. In the present study, PMF and OBM were used to identify the major VOC sources and assess the contributions of these sources to photochemical O3 formation by analyzing the VOC data collected at the urban site in Hong Kong. We mainly targeted the following questions: (1) Which sources contributed significantly to ambient VOCs? (2) What were the roles of the VOC sources and the major species in these sources in the photochemical O3 formation? and (3) How much of VOCs should be reduced to attain the highest reduction efficiency of O3 pollution? Based on the answers to the above questions, appropriate control measures were proposed and the difference between existing and proposed VOC control measures was highlighted.

2.

Methodology

2.1.

Site description

Field measurements were carried out at the Hong Kong Environmental Protection Department (HKEPD) air quality monitoring station at Tsuen Wan (TW) (Fig. 1). The sampling period was from 06 September to 29 November 2010. A detailed description was provided by Guo et al. (2013). This site represents a typical urban site in Hong Kong, which is adjacent to main traffic roads and surrounded by residential and industrial blocks. The samples were collected on the rooftop of a building with a height of 15–20 m (HKEPD, 2012a).

2.2.

Measurements

2.2.1.

Trace gases analysis

Hourly data of O3, CO, NO–NO2–NOx and meteorological parameters at TW were obtained from the HKEPD (http:// epic.epd.gov.hk/ca/uid/airdata). Detailed information about the measurements, quality assurance and control protocols can be found in the HKEPD report (HKEPD, 2012a).

2.2.2.

Sampling and analysis of VOCs

The VOC samples were collected simultaneously on 20 selected days (Guo et al., 2013). Ambient VOC samples were collected using pre-evacuated 2-L electro-polished stainless

182

environmental science & policy 38 (2014) 180–191

Fig. 1 – The sampling sites and the surrounding environment.

steel canisters. A flow-controlling device was used to allow the canister to be fully filled in 60 min. 7–13 1-h integrated canister samples were collected between 0:00 and 21:00 on the selected days within the sampling period. In total, 183 VOC samples were collected at TW. After sampling, the VOC samples were delivered to the laboratory at University of California, Irvine for chemical analysis. The details on the analytical system with multicolumn gas chromatography were provided in Simpson et al. (2010). Briefly, 41 VOC species were identified and quantified in this study. Those VOC species had a detection limit of 3 pptv. The measurement precision for alkanes, alkenes and alkynes was 3%, while it was 5% for aromatics.

2.2.3.

PMF model simulation

USEPAPMF 3.0 model (EPA, 2008a) was utilized for the source apportionment of the 183 VOC samples. Detailed description of this model can be found in Ling et al. (2011). In general, the budget of ambient VOCs is determined by the emissions from different sources, the deposition of chemical and physical processes. Based on the fundamental assumption of mass conservation of species from the emissions sources to the receptor site, the concentration of one specific VOC could be proportional to its emission amounts from different sources in the certain atmospheric volume. According to the above assumption, a speciated data set in the PMF model is represented as a data matrix X of i by j dimensions, where i number of samples and j chemical species (VOCs) were measured (Eq. (1), Paatero, 2000). The function of the PMF model is to identify the number of emission sources and the species profile of each source, and to attribute the amount of mass from each source to each species in each individual sample by an analyst based on the measured data at the receptor site, which could be presented by Eq. (1). Therefore, two metrics, i.e., factor contributions and factor profiles, were included and exported in the PMF results: xi j ¼

p X gik f k j þ ei j

(1)

k¼1

where xij is the jth species concentration measured in the ith sample, gik is the species contribution of the kth source to the ith sample, fkj is the jth species fraction from the kth source, eij

is the residual for each sample/species, and p is the total number of independent sources (Paatero, 2000). In this study, although 41 species were identified and quantified, it is not necessary to use all of them for the PMF model due to the fundamental assumption of non-reactivity and/or mass conservation of the PMF model. The selection of the VOC species for the input of the PMF model was based on the following principles: (1) species at low concentrations with high uncertainty due to their relatively low abundance and/or high reactivity, i.e., b-pinene, camphene and myrcene, were excluded (Guo et al., 2011a; Lau et al., 2010; EPA, 2008a,b; Paatero, 2000). More than 25% of the samples for these species were below the detection limit; (2) species that are highly reactive (i.e., butenes, pentene and 1,3-butadiene with lifetime of a few hours) were excluded, since they were rapidly consumed in the atmosphere and affected the apportionment results (Zhang et al., 2013; Guo et al., 2011a; Lau et al., 2010; Brown et al., 2007). An exception to this principle was the inclusion of unique species that are important tracers of sources. For example, isoprene is an important biogenic VOC; and (3) species at low concentrations that are not typically tracers of sources were excluded, i.e., ethyltoluenes. In total, 25 major VOCs together with CO were input into the PMF model to explore the sources of observed VOCs. These 25 selected VOCs accounted for 96% (ppbv/ppbv) of the total concentrations of the 41 VOC species. The uncertainties for each species were determined as the sum of 5% of VOC concentration and two times the method detection limit of the species, as suggested by Paatero (2000). For values below the detection limit (DL), they were replaced by half of the DL values and their overall uncertainties were set at 5/6 of the DL values. In addition, due to the complex speciation of VOCs, one VOC may be a tracer of several sources. For instance, benzene, xylenes, hexane and its isomer, C9–C10 alkane species could be emitted not only from vehicular emissions/fuel evaporation, but also from solvent emissions. Another example, though propane and n/i-butanes are typical components of LPG, they are also emitted from vehicles. In this analysis, different numbers of factors were tested, and an optimum solution was determined based on both a good fit to the observed data and the most meaningful results by comparing with previous studies (Guo et al., 2011a; Lau et al., 2010). It is noteworthy that the number and profile of factors (sources) in this study were

183

environmental science & policy 38 (2014) 180–191

determined based on the results from previous receptormodeling studies (Guo et al., 2007, 2011a; Ling et al., 2011; Lau et al., 2010) and VOCs source emission studies (Guo et al., 2011b; Ho et al., 2009; Liu et al., 2008; Tang et al., 2007, 2008; Borbon et al., 2002; Blake and Rowland, 1995). Good correlations were found between the observed and predicted VOC concentrations at TW (R2 = 0.98) after PMF implementation. Moreover, all the selected species had scale residuals normally distributed between 3 and 3, confirming that the measured data were well modeled (EPA, 2008a,b). Furthermore, to estimate the sensitivity of the whole analysis to the selection of 25 species, the results for these two scenarios were compared. The general characteristics for the source profiles on the basis of 41 VOC species input were almost the same as those by using the 25 selected VOC species as input except the attributions of some VOC species, i.e., a/bpinene, camphene, limonene, r-cymene and 1-pentene. Two additional sources were found for the result of 41 species (data not shown). One was the source with relatively high contributions of a/b-pinene, camphene, limonene and rcymene, in addition to the significant contributions of these species to the source of biogenic emissions, while the other was solely dominated by 1-pentene, with some contribution of butenes. However, the contributions of these two sources (2.9  0.2%) and other VOC species excluded in this study (1.0  0.1%) were very small, due to their low abundance. Therefore, it is reasonable to select the 25 species for the PMF simulation in this study.

2.2.4.

Observation-based model

An OBM was applied to assess the contributions of VOC sources and the species in these sources to O3 formation. The detailed method description can be found elsewhere (Ling et al., 2011). Briefly, one of the major functions of the OBM is to investigate the relationship between O3 and its precursors, i.e., the sensitivity of O3 photochemical production to changes in the precursors by calculating the relative incremental reactivity (RIR), which considers the overall reactivity of each precursor that generates elevated O3 (Eq. (2), Cardelino and Chameides, 1995): RIRS ðXÞ ¼

½PSO3 -NO ðXÞ  PSO3 -NO ðX  DXÞ=PSO3 -NO ðXÞ DSðXÞ=SðXÞ

;

(2)

where S(X) means the measured concentration of species X; DX is the change in the concentration of X caused by a hypothetical change DS(X); PSO3 -NO represents the O3 formation potential, which is the net O3 formation and NO consumed during the evaluation period. In addition, the average RIR function for the X species over multiple sampling days (NS, here NS = 20) could be obtained through the following equation (Eq. (3)), PNS RIR ¼

S

½RIRS ðXÞPSO3 -NO ðXÞ PNS S S PO3 -NO ðXÞ

(3)

In addition, to provide a thorough understanding of the roles of VOC sources/species in O3 formation in Hong Kong, a new parameter, i.e., RIR-weighted value, combining overall reactivity (RIR) with the observed concentrations was developed (Eqs. (4) and (5), X represents the VOC species/sources) to

determine the relative contribution of each VOC source and its main component to the photochemical O3 formation: RIR-weighted ¼ RIRðXÞ  concentrationðXÞ Contribution ¼ ½RIRðXÞ  concentrationðXÞ=

(4) X

RIRðXÞ

 concentrationðXÞ

3.

Results and discussion

3.1.

Source profile and source apportionment

(5)

Fig. 2 presents the explained variations of individual apportioned sources and the corresponding major tracers, explaining the contributions of each source to the specific VOCs at TW. Seven sources were identified, including gasoline exhaust, gasoline evaporation, paint and sealant solvent, LPG usage, diesel exhaust, consumer and household products and biogenic emissions. Source 1 was characterized by high percentages of n/ipentanes, n-heptane, benzene and toluene, with considerable presence of 2,3-methylpentane and CO, indicating that it is related to vehicular emissions, likely from gasoline-fueled vehicles as n/i-pentane, 2,3-methylpentane, benzene and toluene were demonstrated good tracers for gasoline exhaust in Hong Kong (Guo et al., 2011a,b; Ho et al., 2009; Tsai et al., 2006). Similarly, source 2 showed a dominance of n/i-pentanes, accounting for 40% of the total VOCs in the source profile, with certain amounts of 2,3-methylpentane, n-heptane and toluene. This source is believed to be gasoline evaporation due the fact that the contributions of other combustion and/or vehicular tracers, i.e., ethane, ethene, benzene and CO were negligible while n/i-pentane levels were relatively high. Source 3 was dominated by high percentages of ethylbenzenes, xylenes and trimethylbenzenes, with aromatics accounting for 70% of the VOC source profile. In addition to vehicular emissions, these species could be from the solvent emissions of paints, inks, sealant, varnish and thinner for architecture and decoration (Liu et al., 2008; Borbon et al., 2002; Seila et al., 2001). The poor correlation among the above aromatics and other combustion tracers, i.e., ethyne, ethene, and CO in this source suggested that combustion and/or vehicular emissions were not the major contributors of ethylbenzenes, xylenes and trimethylbenzenes in Hong Kong. Hence, this source can be identified as paint and sealant solvents. Source 4 was distinguished by high percentages of propane and i/n-butanes, which were typical tracers for LPG (Liu et al., 2008; Blake and Rowland, 1995). In addition, ethene and propene showed high levels in the source profile, indicating that this source could be also related to combustion emissions. Indeed, previous studies on roadside and exhaust samples have demonstrated that LPG fueled vehicular could emit significant amount of ethene and propene (Tang et al., 2007, 2008). In Hong Kong, LPG was used as fuel for taxis and public and private light bus. For example, about 99.9% of the registered taxi, 51.1% of the register public and private buses were powered by LPG by December 2010 in Hong Kong (HKCSD,

184

environmental science & policy 38 (2014) 180–191

Fig. 2 – Explained variations of seven identified sources at TW.

2010). Furthermore, consumer products may also contribute to this source as some of them use LPG as propellant (Lau et al., 2010). Therefore, this source could be assigned as LPG usage. Ethane, ethene, ethyne, benzene, n-decane and CO had a high percentage in source 5, with certain contributions of C3– C4 alkanes and n-nonane. These species are all associated with vehicular emissions, likely from diesel-fueled vehicles as the percentage of C2 species, benzene and n-decane were usually high in the diesel exhaust (Ho et al., 2009; Liu et al., 2008). Source 6 was associated with high percentage of n-hexane and its isomer 2,3-methylpentane, accounting for 62% and 40% of those VOCs measured concentrations. In addition to the possible emissions from combustion processes, these two species could be used in the solvents for household products and consumer products (Guo et al., 2011a; Lau et al., 2010; Kwon et al., 2007). The negligible amount of combustion tracers, i.e., CO, C2 species, in the source profile confirmed that the source could be assigned to consumer and household products.

Fig. 3 – Source apportionments of VOCs at TW.

Source 7 was solely dominated by isoprene, which is the indicator of biogenic emissions (BVOCs) (Tsui et al., 2009; Song et al., 2008). Fig. 3 illustrates the relative contributions of each source to ambient VOCs at TW. Vehicle exhaust made the largest contributions, responsible for about 42% of the ambient VOCs, followed by solvent-related sources (25%), LPG usage (21%), gasoline evaporation (8%) and BVOCs (4%). The relatively higher contribution of vehicle exhaust suggested that though different measures had been implemented for the last decade (HKPD, 2010), the vehicular emissions were still remarkable. To understand variations in VOC sources, we compared the contributions of the identified sources obtained in this work to those from other sites that were previously studied by PMF model in Hong Kong (Table 1). It should be noted that some disagreement may be found for the source apportionment results due to different tracers, sampling locations and sampling time, as well as the influence of substantial changes in transport, urbanization, and chemical intensification of daily life from the last decade. The contribution of vehicular emissions in this study (42  3%) was comparable to those obtained in 2007 at Tung Chung (TC, 48  4%), a suburban site, while it was about twice that estimated at the other four sites in 2002–2003 and 2006–2007 in Hong Kong (Lau et al., 2010). One possible reason for the difference between this study and the study in 2002–2003 and 2006–2007 was the identification of source profiles. It was well documented that vehicular emissions could contribute significantly to the ambient levels of some specific VOCs, i.e., benzene, ethane and ethyne (Guo et al., 2011a,b; Ho et al., 2009; Liu et al., 2008). However, in the study of 2002–2003 and 2006–2007, these species were categorized as aged VOCs due to their low photochemical reactivity, resulting in the lower contributions of vehicle

185

environmental science & policy 38 (2014) 180–191

Table 1 – Comparison of source apportionments at TW with those from other areas in Hong Kong by PMF. Factor Site Vehicle exhaust Diesel Gasoline LPG usage Gasoline evaporation Solvent related sources Biogenic Remarks

Fall, 2010a TW 42  3% 20  2% 22  2% 21  2% 8  1% 25  3% 4  1% PMF

Fall, 2007b TC 48  4% 21  2% 27  3% – – 43  2% – PMF

2002–2003c CW, TC, TM, YL 16–27% – – 15–30% 4–8% 12–19% 2–6% PMF

2006–2007c CW, TC, TM, YL 12–22% – – 16–41% 5–8% 11–15% 1–4% PMF

2010d 24%

60% EIe

TC – Tung Chung, CW – Central/Western, TM – Tap Mum, YL – Yuen Long. This study. b Guo et al. (2011a). c Lau et al. (2010). d HKEPD, http://www.epd.gov.hk/epd/english/environmentinhk/air/data/emissioninve.html. e EI – emission inventory. a

exhaust. In addition, the relatively higher contribution of vehicle exhaust in 2010 may be caused by the increased number of vehicles since last decade. For example, the number of vehicles has been increased from about 580,000 in 2002 to about 660,000 in 2010 on 2090 kilometers of roads in Hong Kong (HKCSD, 2010; HKHYD, 2013). For LPG usage, its contribution in this study was close to that obtained in 2002– 2003 and 2006–2007, indicating that LPG usage has become a remarkable contributor to the ambient VOC loading in Hong Kong. Since 1999, the diesel-fueled taxis and public and private buses were replaced by LPG. These LPG-fueled vehicles would run long hours on roads and generate high mileage, resulting in high emissions of LPG usage (Lau et al., 2010). In addition, the high contribution of LPG usage could be related to the increasing LPG consumption, i.e., from 230,000 tons in 2001 to 400,000 tons in 2010 (HKCSD, 2010). This could be further confirmed by the average concentrations of major tracers of LPG, i.e., propane, n/i-butane at TW, which increased from 4.87, 3.63 and 8.83 in 2001 (Guo et al., 2004) to 6.39, 6.71 and 10.12 mg/m3 in 2010, respectively. For solvent-related sources, its contribution was comparable to that observed in 2002–2003 and 2006–2007 (Lau et al., 2010), while it was much lower than that in fall 2007 (Guo et al., 2011a). Although samples in this study and the study by Guo et al. (2011a,b) were both collected in fall, the variations of the contribution of solvent-related sources might be attributed to the sampling site difference and the influence of urbanization in Hong Kong. Compared to TW, TC is a relatively new town, still under rapid development. Materials for decoration, i.e., paint and varnish, are being applied to residences in each new apartment block, resulting in higher contributions of solvent-related sources (Lam et al., 2013). It is interesting to compare the results of this study (topdown approach) with the current VOC emission inventory (bottom-up approach) in Hong Kong. The contributions of vehicle exhaust and solvent-related emissions were different from those of the emission inventory, which reported that about 24% and 60% of VOCs were from road transport and noncombustion solvent related sources in 2010. The difference between the results of the above two approaches was consistent with previous studies (Lau et al., 2010; Lauz et al., 2008). The discrepancy was related to the different mechanisms and performance in the two approaches. The top-down

approach is about the receptor, while bottom-up method is about the source. For example, the PMF model investigated the source characteristics of VOCs based on the concentrations of VOCs measured at the receptor location, while the emission inventory estimated the strength of individual source emissions based on bottom-up survey (Yu et al., 2011; Zheng et al., 2009b). As such, the PMF results only reflected the contribution of each source at the receptor, which did not necessarily present the actual strength of the emission source (Lau et al., 2010; Zheng et al., 2009b). In addition, the sources extracted from the PMF simulations were different from the emission inventory. In Hong Kong, five anthropogenic source categories are traditionally inventoried (HKEPD, 2012b). Furthermore, different estimation methods were adopted for different emission activities, which could induce significant uncertainties on the VOC emission strengths in the emission inventory (Yu et al., 2011; Zheng et al., 2009b; HKEPD, 2005; EPA, 2004). In summary, the comparison suggests a general characteristic of source apportionments of VOCs in Hong Kong, where vehicle exhaust, LPG usage and solvent-related emissions are the major contributors to ambient VOCs.

3.2.

Roles of VOC sources in photochemical O3 formation

The PMF model could provide the concentration of each VOC in each source directly, defined as PMF extracted concentration. The OBM model was driven on the 20 VOC sampling days by the PMF extracted concentrations. Fig. 4a presents the average RIR values of different VOC sources, while Fig. 4b gives the relative contribution of each VOC source to O3 by considering the reactivity and abundance of VOCs in the function of RIR-weighted value (Eqs. (3) and (4)). It can be found that paint and sealant solvents had the highest RIR value, followed by BVOCs, diesel exhaust, LPG usage, gasoline exhaust, consumer and household products and gasoline evaporation. The relatively higher RIR values for paint and sealant and BVOCs were mostly attributed to the high reactivity of the major VOCs in those sources, i.e., ethylbenzene, trimethylbenzenes, xylenes and isoprene (Simpson et al., 2010). However, after taking into account both RIRs and the emission amount of each VOC source, paint and sealant solvents, diesel exhaust and LPG usage were the main

186

environmental science & policy 38 (2014) 180–191

Fig. 4 – The average RIR values of VOC sources (a) and their relative contributions (b) to photochemical O3 formation.

contributors at TW, with about 78% to the total RIR-weighted value (VOC sources). The result suggested that controlling vehicular- and solvent-related VOCs is essential for effective control of O3 production in Hong Kong. In recent years, a series of control measures have been implemented by the HKEPD to reduce VOC emissions from vehicular- and solvent-related sources. For the vehicular-related emission, tightening vehicle emission standards and vehicle fuel standards, and introducing environmental friendly cars and clean fuel were the major two principles. For the solvent-related emissions, a product labeling system and VOC content limits of different solvent products have been established. All of these measures were conducted to reduce the emissions of VOCs by 55% by 2010 on the basis of the emission levels in 1997. However, the effectiveness of these measures and the influence of the replaced VOC species on O3 formation reduction are rarely unknown. Though the VOC emissions present a gradual decreasing trend in recent years, the O3 levels are still increasing (Wang et al., 2009). One of the possible reasons was that most of the implemented VOC control measures were mass-based, which only limited the mass of total VOC emissions and neglected the reactivity of VOCs. In this study, the RIR-weighted values suggested an optimum O3 control strategy should consider both the reactivity and emission quantity of VOCs. For example, one of the most striking measures of controlling vehicular emissions is the promotion of LPG as a clean substitute fuel. Though the VOC species, i.e., propane and butanes, from LPG usage have relatively lower reactivity, the contribution of this source to O3 formation could be high if considering the quantity of its emission. On the other hand, for the solvent-related emissions, though the emission content of specific solvents used has been limited in different industries, the contributions of solvent emissions to O3 formation were still remarkable, perhaps due to the replacement of heavier VOCs by lighter VOCs without considering their reactivity (HKPD, 2010). Overall, based on the PMF and OBM results, it is recommended that before the formulation and implementation of VOC control strategies, the abundance and reactivity of each VOC in each source should be considered.

3.3. Contribution of major VOC species in different sources to O3 production To further investigate the relative importance of VOC species in different sources, contributions of individual VOCs were calculated using the RIR-weighted values (Section 3.1). Fig. 5 shows the top 15 VOC species with high (a) RIR and (b) RIRweighted values at TW. After taking into account the mass emissions of VOCs in each source, xylenes and ethylbenzene in paint and sealant solvents, toluene in gasoline exhaust, n/ibutane, ethene, propane and propene in LPG usage, toluene in diesel exhaust, toluene in consumer and household product and n-pentane in gasoline evaporation had high RIR-weighted values at TW. It is worthy to note that though the reactivity of n/i-butane and propane may not be as high as that of some aromatics, i.e., toluene and xylenes, their RIR-weighted values were relatively higher, indicating that a VOC with relatively lower reactivity could also have significant impact on O3

Fig. 5 – Contribution of major VOC species in different sources to the O3 production at TW.

environmental science & policy 38 (2014) 180–191

production when its concentration was high. In addition, isoprene contributed significantly to local O3 formation in Hong Kong with relatively higher RIR-weighted values, consistent with previous studies using different methods (Cheng et al., 2010). This indicated that further reduction of anthropogenic VOCs may be less effective due to the significant contribution of BVOCs as most of the practical strategies for controlling O3 pollution in Hong Kong are to effectively reduce anthropogenic VOCs (HKEPD, 2012c). On the other hand, the results in this work were somewhat different from those observed in the inland PRD, where m/p-xylenes and toluene in solvent usage, ethene in biomass/biofuel burning and diesel vehicular emissions contributed significantly to O3 formation. The variations may be attributed to the different industrial and energy structures between Hong Kong and the inland PRD region (HKEPD, 2012a,c). These findings further confirmed that the current policies on controlling vehicular emission and solvent usage should be prioritized to alleviate photochemical O3 production in regional perspective and suggested that different strategies should be implemented in such two close areas (Louie et al., 2012; Zhong et al., 2013). Indeed, in order to effectively control the VOC emissions and to improve the regional air quality in Hong Kong and inland PRD, the Hong Kong Government and the Guangdong Provincial Government set up the Hong Kong/Guangdong Joint Working Group on Sustainable Development and Environmental Protection (the JWGSDEP) in June 2000 (Zhong et al., 2013). Commenced in 2003, a series of emission control measures have been formulated in Hong Kong and the rest PRD region, focusing on the control of vehicular emissions, VOC containing products and the reconstruction of energy consumption. Different from the control measures in Hong Kong, more specific measures were conducted in inland PRD for vehicle emissions including strictly enforcing the vehicle scrapping system, establishing vehicle labeling systems, introducing inspection/maintenance program for in-use vehicles and phasing out high emitting vehicles. For the VOC containing products, limiting the VOC emissions from the solvent usage in different industries and establishing the emission control standards for typical industries were the major measures. Furthermore, measures of adjusting the energy structure, i.e., the domestic fuel structure, promoting the use of cleaner energy and controlling the biomass combustion were gradually implemented. Even so, O3 pollution in Hong Kong and the inland PRD region is continuously worse. This is probably due to the fact that most of control measures in Hong Kong and the inland PRD region only considered the weight/mass of VOC emissions and neglected the reactivity of controlled and replaced VOCs, which reduced the effectiveness of the VOC control measures on the reduction of O3 pollution, in addition to the influence of energy consumption increase, urbanization and industrialization. The findings in this study provided an alternative way to more efficiently alleviate O3 pollution by controlling specific VOCs in certain VOC sources, such as controlling xylenes and ethylbenzene in solvents, toluene in gasoline and diesel vehicles, and propane and propene in LPG usage in Hong Kong, while controlling xylenes and toluene in solvent, ethene in biomass/biofuel burning and diesel vehicles in the inland PRD region.

187

To assess the effectiveness of the control measures, and to devise and revise these measures, it is important to set up local and regional air quality monitoring stations and gain qualityassured criteria data (Louie et al., 2012). Although the PRDHong Kong Regional Air Quality network has been in operation since 2005, the trends of observed VOC values, the responses of ambient VOCs to the implemented control strategies, the contribution of different sources to ambient VOCs and their roles in O3 formation, and the effectiveness of control programs still remained unclear in the inland PRD region due to the lack of data sharing mechanism. In addition, the findings in this study highlighted the importance of monitoring specific VOCs, which provided the base for the revision of regional monitoring network including monitoring locations and specific VOCs concerned.

3.4.

O3 reduction efficiency

From Section 3.2, paint and sealant solvents, diesel exhaust and LPG usage were found to be the major contributors to the O3 formation. Hence, cutting their emissions would be the most effective for the remediation of O3 production. One question is that how much VOC source cut would have the highest O3 reduction efficiency. Here, a new parameter, namely relative O3 reduction efficiency (RORE), was adopted to evaluate the sensitivity of O3 reduction under different scenarios of VOC cut. Additional simulations were run by reducing the original amounts of measured VOCs by 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80% and 90%, while the other measured parameters were unchanged. The RORE was calculated as the relative difference in O3 formation potential ðDPO3 -NO Þ between the base case with original VOCs and the above VOC reduction scenarios divided by the corresponding reduction percentage of VOCs. Fig. 6 presents the average RORE values of each VOC source under different VOC-cut scenarios. It can be seen that paint and sealant solvents (F3) and BVOCs (F7) presented higher RORE values followed by diesel exhaust (F5) and LPG usage (F4), because of the high reactivity of the major VOCs in those sources as described in Section 3.2. It is interesting to note that BVOC source (F7) had relatively higher RORE values (0.20–0.35), implying the high reactivity of its VOC component. Indeed, as its major substituent, isoprene has rather high reactivity with OH radical. Nevertheless, the concentration of the BVOC source was relatively low. In contrast, the source of paint and sealant solvents (F3) also had higher RORE values (0.28–0.44), and remarkably its concentration was higher, suggesting that the most cost-effective approach for O3 reduction was to cut the sources with higher RORE values and higher concentrations. Further inspection found that the RORE value of each source differed under different VOC reduction scenarios, and the scenario with the highest RORE value varied for each source, suggesting that each source had its own VOC cut percentage which would be the most efficient for O3 reduction. For example, the RORE was the highest when 40% of VOCs in paint and sealant solvents were cut, indicating that controlling O3 would be the most effective when the VOC emissions from paint and sealant solvents were reduced by 40%. Similarly, the highest RORE values for the other sources, i.e., gasoline exhaust, gasoline evaporation, LPG usage, diesel exhaust,

188

environmental science & policy 38 (2014) 180–191

Fig. 6 – ROREs of VOC sources under different VOC-cutting scenarios (the ROREs in the Y axis were set between minimum and maximum values for different scenarios in order to present the variations clearly).

consumer and household products and BVOCs were presented in the scenarios of 50%, 40%, 80%, 40%, 10% and 50%, respectively. One of the reasons why each source had its own highest RORE was probably the differences of the VOC composition and their photochemical reactivity among the sources. Different reactivity of VOC sources, caused by the different composition, would lead to different capacities for the O3 formation (Gilman et al., 2009; Zheng et al., 2009a).

Hence, the prefect cutting percentage of each VOC source would be subsequently changed. Moreover, it was well recognized that photochemical O3 production is related to the relative concentrations of its precursors, i.e., VOCs and NOx, with a non-linear relationship (Thornton et al., 2002; Jenkin and Clemitshaw, 2000; Sillman, 1999). The reduction of VOCs changed the ratios of VOCs/NOx, resulting in variations of RORE values under different VOC-cutting scenarios. However, the above simulation results were based on the simplified carbon bond IV mechanism which does not look into the chemical reactions of each VOC species in a source. As such, further simulations by a photochemical box model coupled with more explicit mechanisms, i.e., master chemical mechanisms, are needed to better understand detailed processes and pathways of the O3 formation under different VOC-cutting scenarios (Derwent et al., 2007; Jenkin and Clemitshaw, 2000). In order to further evaluate the RORE corresponding to the reduction of major VOC species in each source, additional OBM simulation was conducted with the input of 10–90% of the top 15 VOC species that had relatively higher RIR-weighted values (Fig. 5). Table 2 presents the average RORE values of the top 15 VOC species under different cutting scenarios. It can be found that the cutting percentage for the highest RORE value of the specific species in the sources varied. For example, ethylbenzene, m/o-xylene had the highest RORE values when they were reduced by 10% in the emission of paint and sealant solvents, while the maximum RORE value of p-xylene in the same source was by cutting 40%. Furthermore, n/i-butane, propane and propene in LPG usage, n-butane in diesel exhaust and toluene in consumer and household product displayed the highest RORE values in the scenario with 10% reduction, while toluene in the gasoline exhaust, n-pentane in gasoline evaporation and ethene in diesel exhaust showed maximum RORE value in the scenario of 30%, 30% and 50% reduction, respectively. Interestingly, the RORE values of isoprene in BVOC source (i.e., F7) were relatively higher than those of major VOC species in other sources and reached the

Table 2 – Average RORE values for the main VOC species in each source under different scenarios. Source

F1 F2 F3

F4

F5 F6 F7 a

a

RORE (102) (%/%)

Species

Toluene n-Pentane Ethylbenzene m-Xylene o-Xylene p-Xylene Ethene i-Butane n-Butane Propane Propene n-Butane Ethene Toluene Isoprene

10%

20%

30%

40%

50%

60%

70%

80%

90%

2.83 0.10 3.41 22.66 12.14 8.68 2.89 0.92 2.01 0.61 3.83 0.94 2.55 1.36 12.97

3.25 0.48 2.77 20.59 11.47 9.32 2.93 0.75 1.87 0.38 3.60 0.76 2.59 1.07 12.83

3.83 0.57 2.28 7.85 11.33 9.29 2.97 0.77 1.80 0.33 3.49 0.82 2.62 1.07 12.72

3.68 0.30 2.71 9.28 11.01 9.77 2.97 0.77 1.90 0.34 3.24 0.78 2.64 1.05 16.60

3.66 0.45 2.81 18.22 11.07 8.96 3.18 0.76 1.91 0.33 3.51 0.77 2.73 1.00 19.36

3.64 0.33 2.65 16.55 11.41 9.13 2.78 0.75 1.88 0.32 3.47 0.77 2.45 1.05 16.04

3.68 0.32 2.77 15.82 10.63 8.92 2.90 0.73 1.90 0.32 3.33 0.74 2.56 1.03 18.06

3.62 0.39 2.42 14.35 10.82 8.74 2.64 0.75 1.82 0.29 3.29 0.77 2.33 1.04 12.35

3.67 0.40 2.71 9.01 10.59 8.67 2.85 0.76 1.82 0.33 3.23 0.77 2.52 1.04 13.16

F1–F7 corresponded to the identified sources in Fig. 2.

environmental science & policy 38 (2014) 180–191

maximum in the scenario of 50%, suggesting that reducing BVOC emissions may be more efficient in O3 reduction in terms of single VOC species. Though VOCs emitted from vegetation contribute to the O3 formation, they have also beneficial environment effects such as mitigating the influence of urban heat islands, removing air pollutants from the atmosphere, storing and photolyzing CO2 into O2, and improving quality of life (EPA, 2008b). Retaining proper proportion of vegetation in rapidly and intensively urbanized environments is critical. To achieve the optimal balance, planting low BVOCs emitting trees is more appropriate in the urban green areas. Some tree species, including Acacia confuse, Cinnamomum camphora, and Ficus microcarpa, are the low BVOC emitters and more suitable to be planted in newly-developed urban green areas and roadside vegetation zones (Tsui et al., 2009; Leung et al., 2010; HKEPD, 2012c). Apart from VOCs, other compounds, i.e., NOx were important O3 precursors as well. Therefore, one important issue that the air quality management system should consider is whether controlling VOCs or NOx individually would be more effective than controlling both VOCs and NOx in a specific area, i.e., Hong Kong, given that ambient O3 formation strongly depends on the ratio of VOC/NOx, and O3 formation in the region is generally VOC-limited (Zhang et al., 2007; Cheng et al., 2010; HKEPD, 2012c). To provide scientific support for the formulation of appropriate O3 control strategies, our previous study has conducted an array of simulations constrained with specific VOCs and NOx source removal by a photochemical box model with master chemical mechanism in Hong Kong (Lam et al., 2013). It was found that the peak O3 would be reduced by about 26% when VOCs and NOx were both removed from the specific sources that had significant VOC and NOx emissions, i.e., vehicular emissions, suggesting more stringent policies and regulations for the reduction of both VOCs and NOx from specific sources are needed to exert more effective control on the O3 formation in the region.

4.

Conclusions

A newly developed reactivity-based method, combining positive matrix factorization (PMF) model with observationbased model (OBM), was applied for the first time to better understand the VOC sources and their contributions to O3 formation in Hong Kong. Totally, 7 sources including gasoline exhaust, gasoline evaporation, paint and sealant solvents, LPG usage, diesel exhaust, consumer and household products and BVOCs were identified at TW. Subsequently RIR-weighted values, considering both RIRs and the concentrations of each VOC source, showed that paint and sealant solvents, diesel exhaust and LPG usage were the key contributors to O3 formation at TW, suggesting controlling solvent- and vehicular-related emissions should be the most effective strategy to reduce photochemical O3 formation in Hong Kong. In addition, the RIR-weighted method indicated that m/o/pxylene and ethylbenzene in paint and sealant solvents, toluene in gasoline exhaust, n/i-butane, ethene, propene and propane in LPG usage and n-butane and ethene in diesel exhaust were the significant contributors to the O3 formation at TW. Analysis on the RORE values under varied VOC cutting

189

scenarios of the sources and the major species in these sources indicated that the cutting percentages of the VOC sources and the major species from these sources were different for the most effective O3 reduction in Hong Kong. For instance, by cutting 40% of the dominant VOC source, i.e., paint and sealant solvents, the efficiency of O3 reduction would be the highest for this source. The results provide scientific support to the enhancement and implementation of the existing control policies.

Acknowledgements This study was supported by the Research Grants Council of the Hong Kong Special Administrative Region via grants PolyU5179/09E and N_PolyU545/09. It was partly supported by Central Policy Unit of the Hong Kong Government via the Public Policy Research Scheme (2013.A6.012.13A) and the internal grant of the Hong Kong Polytechnic University (APL65).

references

Avery, R.J., 2006. Reactivity-based VOC control for solvent products: more efficient ozone reduction strategies. Environ. Sci. Technol. 40, 4845–4850. Blake, D.R., Rowland, F.S., 1995. Urban leakage of liquefied petroleum gas and its impact on Mexico City air quality. Science 269, 953–956. Borbon, A., Locoge, N., Veilerot, M., et al., 2002. Characterization of NMHCs in a French urban atmospheric: overview of the main sources. Sci. Total Environ. 292, 177–191. Brown, S.G., Frankel, A., Hafner, H.R., 2007. Source apportionment of VOCs in Los Angeles area using positive matrix factorization. Atmos. Environ. 41, 227–237. Cardelino, C.A., Chameides, W.L., 1995. An observation-based model for analyzing ozone precursor relationships in the urban atmosphere. J. Air Waste Manag. Assoc. 45, 161–180. Carter, W.P.L., 1994. Development of ozone reactivity scales for volatile organic compounds. J. Air Waste Manag. Assoc. 44, 881–899. Chameides, W.L., Fehsenfeld, F., Rodgers, M.O., et al., 1992. Ozone precursor relationships in the ambient atmosphere. J. Geophys. Res. 97, 6037–6055. Chang, C.C., Chen, T.Y., Lin, C.Y., et al., 2005. Effects of reactive hydrocarbons on ozone formation in southern Taiwan. Atmos. Environ. 39, 2867–2878. Cheng, H.R., Guo, H., Wang, X.M., et al., 2010. On the relationship between ozone and its precursors in the Pearl River Delta: application of an observation-based model (OBM). Environ. Sci. Pollut. Res. 17, 547–560. Derwent, R.G., Jenkin, M.E., Passant, N.R., et al., 2007. Reactivity-based strategies for photochemical ozone control in Europe. Environ. Sci. Policy 10, 445–453. Derwent, R.G., Jenkin, M.E., Saunders, S.M., et al., 1998. Photochemical ozone creation potentials for organic compounds in Northwest Europe calculated with a Master Chemical Mechanism. Atmos. Environ. 32, 2429–2441. Environmental Protection Agency (EPA), 2008a. EPA Positive Matrix Factorization (PMF) 3.0 Fundamentals and User Guide. . Environmental Protection Agency (EPA), 2008b. Report on Reducing Urban Heat Islands: Compendium of Strategies,

190

environmental science & policy 38 (2014) 180–191

Trees and Vegetation. http://www.epa.gov/hiri/resources/ compendium.htm. Environmental Protection Agency (EPA), 2004. Emission Inventory Improvement Programs. Technical Report Series. Technology Transfer Network Clearinghouse for Inventories and Emission Factors. http://www.epa.gov/ttn/chief/eiip/ techreport. Gilman, J.B., Kuster, W.C., Goldan, P.D., et al., 2009. Measurements of volatile organic compounds during the 2006 TexAQS/GoMACCS campaign: industrial influences, regional characteristics, and diurnal dependencies of the OH reactivity. J. Geophys. Res. 114, D00F06, http://dx.doi.org/ 10.1029/2008JD011525. Guangdong Provincial Environmental Protection Monitoring Centre (GDEMC), and Environmental Protection Department of Hong Kong (HKEPD), 2005–2012. Pearl River Delta Regional Air Quality Monitoring Network Report. (in Chinese). http:// sc.epd.gov.hk/gb/www.epd.gov.hk/epd/tc_chi/ environmentinhk/air/prob_solutions/air_problems.html. Guo, H., Wang, T., 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. Environ. Pollut. 129, 489–498. Guo, H., So, K.L., Simpson, I.J., et al., 2007. C1 to C8 volatile organic compounds in the atmosphere of Hong Kong: Overview of atmospheric processing and source apportionment. Atmos. Environ. 41, 1456–1472. Guo, H., Cheng, H.R., Ling, Z.H., et al., 2011a. Which emission sources are responsible for the volatile organic compounds in the atmosphere of Pearl River Delta. J. Hazard. Mater. 188, 116–124. Guo, H., Zou, S.C., Tsai, W.Y., et al., 2011b. Emission characteristics of nonmethane hydrocarbons from private cars and taxis at different driving speeds in Hong Kong. Atmos. Environ. 45, 2711–2721. Guo, H., Jiang, F., Cheng, H.R., et al., 2009. Concurrent observations of air pollutants at two sites in the Pearl River Delta and the implication of regional transport. Atmos. Chem. Phys. 9, 7343–7360. Guo, H., Ling, Z.H., Cheung, K., et al., 2013. Characterization of photochemical pollution at different elevations in mountainous areas in Hong Kong. Atmos. Chem. Phys. 13, 3881–3898. HKCSD (Hong Kong Census and Statistics Department), 2010. Hong Kong Energy Statistics: Annual Report. http:// www.censtatd.gov.hk. Ho, K.F., Lee, S.C., Ho, W.K., et al., 2009. Vehicular emission of volatile organic compounds (VOCs) from a tunnel study in Hong Kong. Atmos. Chem. Phys. 9, 7491–7504. HKPD (Hong Kong Protection Department), 2010. Final Report on Review on the Effectiveness of VOC Control Programme in Hong Kong. http://www.epd-asg.gov.hk/english/report/ aqr.html. HKEPD (Hong Kong Protection Department), 2012a. Air Quality in Hong Kong. 2011.http://www.epd-asg.gov.hk/english/ report/aqr.html. HKEPD (Hong Kong Protection Department), 2012b. Hong Kong Air Pollution Emission Inventory. http://www.epd.gov.hk/ epd/english/environmentinhk/air/data/files/ 2010HKEIReportEng.pdf. HKEPD (Hong Kong Protection Department), 2012c. Integrated Data Analysis and Characterization of Photochemical Ozone in Hong Kong. http://www.epd-asg.gov.hk/english/report/ aqr.html. HKHYD (Hong Kong Highways Department), 2013. The Hong Kong Road Network Report. http://www.hyd.gov.hk/en/ road_and_railway/existing/road_network/road.html.

Hopke, P.K., 2003. Recent developments in receptor modeling. Chemometr. Intell. Lab. 17, 255–265. Jenkin, M.E., Clemitshaw, K.C., 2000. Ozone and other secondary photochemical pollutants: chemical processes governing their formation in the planetary boundary layer. Atmos. Environ. 34, 2499–2527. Kwon, K.D., Jo, W.K., Lim, H.J., et al., 2007. Characterization of emissions composition for selected household products available in Korea. J. Hazard. Mater. 148, 192–198. Lam, S.H.M., Ling, Z.H., Guo, H., et al., 2013. Modeling VOC source impacts on high ozone episode days observed at a mountain summit in Hong Kong under the influence of mountain-valley breezes. Atmos. Environ. 81, 166–176. Lau, A.K.H., Yuan, Z.B., Yu, J.Z., et al., 2010. Source apportionment of ambient volatile organic compounds in Hong Kong. Sci. Total Environ. 408, 4138–4149. Lauz, V.A., Henne, S., Staehelin, J., et al., 2009. Statistical analysis of anthropogenic non-methane VOC variability at a European background location (Jungfraujoch, Switzerland). Atmos. Chem. Phys. 9, 3445–3459. Lauz, V.A., Hueglin, C., Buchmann, B., et al., 2008. Receptor modeling of C2–C7 hydrocarbon sources at an urban background site in Zurich, Switzerland: changes between 1993–1994 and 2005–2006. Atmos. Chem. Phys. 8, 2313–2332. Leung, D.Y.C., Wong, P., Cheung, B.K.H., et al., 2010. Improved land cover and emission factors for modeling biogenic volatile organic compounds emissions from Hong Kong. Atmos. Environ. 44, 1456–1468. Ling, Z.H., Guo, H., Cheng, H.R., et al., 2011. Sources of ambient volatile organic compounds and their contributions to photochemical ozone formation at a site in the Pearl River Delta, southern China. Environ. Pollut. 159, 2310–2319. Liu, Y., Shao, M., Lu, S.H., et al., 2008. Source apportionment of ambient volatile organic compounds in the Pearl River Delta, China: Part II. Atmos. Environ. 42, 6261–6274. Louie, P.K.K., Ho, J.W.K., Tsang, R.C.W., et al., 2012. VOCs and OVOCs distribution and control policy implications in Pearl River Delta region, China. Atmos. Environ., http://dx.doi.org/ 10.1016/j.atmosenv.2012.08.058. Na, K., Kim, Y.P., 2007. Chemical mass balance receptor model applied to ambient C2–C9VOC concentration in Seoul, Korea: effect of chemical reaction losses. Atmos. Environ. 41, 6715– 6728. Paatero, P., 2000. User’s Guide for Positive Matrix Factorization Programs PMF2 and PMF3, Part 1: Tutorial. University of Helsinki, Finland. Russell, A., Dennis, R., 2000. NARSTO critical review of photochemical models and modelling. Atmos. Environ. 34, 22832324. Seila, R.L., Main, H.H., Arriaga, J.L., et al., 2001. Atmospheric volatile organic compounds measurements during the 1996 Paso del Norte ozone study. Sci. Total Environ. 276, 153–169. Sillman, S., 1999. The relation between ozone, NOx and hydrocarbons in urban and polluted rural environments. Atmos. Environ. 33, 1821–1845. Simpson, I.J., Blake, N.J., Barletta, B., et al., 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. Atmos. Chem. Phys. 10, 11931–11954. Song, Y., Shao, M., Liu, Y., et al., 2007. Source apportionment of ambient volatile organic compounds in Beijing. Environ. Sci. Technol. 41, 4348–4353. Song, Y., Dai, W., Shao, M., et al., 2008. Comparison of receptor models for source apportionment of volatile organic compounds in Beijing, China. Environ. Pollut. 156, 174–183. Tang, J.H., Chan, L.Y., Chan, C.Y., et al., 2007. Characteristics and diurnal variations of NMHCs at urban, suburban, and

environmental science & policy 38 (2014) 180–191

rural sites in the Pearl River Delta and a remote site in South China. Atmos. Environ. 41, 8620–8632. Tang, J.H., Chan, L.Y., Chan, C.Y., et al., 2008. Implications of changing urban and rural emissions on non-methane hydrocarbons in the Pearl River Delta region of China. Atmos. Environ. 42, 3780–3794. Thornton, J.A., Wooldridge, P.J., Cohen, R.C., et al., 2002. Ozone production rates as a function of NOx abundances and Hox production rates in the Nashville urban plume. J. Geophys. Res. 107 (D12) 4146, http://dx.doi.org/10.1029/2001JD000932. Tsai, W.Y., Chan, L.Y., Blake, D.R., et al., 2006. Vehicular fuel composition and atmospheric emissions in South China: Hong Kong, Macau, Guangzhou, and Zhuhai. Atmos. Chem. Phys. 6, 3281–3288. Tsui, J.K.Y., Guenther, A., Yip, W.K., et al., 2009. A biogenic volatile organic compound emission inventory for Hong Kong. Atmos. Environ. 43, 6442–6448. Wang, T., Wei, X.L., Ding, A.J., et al., 2009. Increasing surface ozone concentrations in the background atmosphere of Southern China 1994–2007. Atmos. Chem. Phys. 9, 6217–6227. Xie, Y.L., Bekowitz, C.M., 2006. The use of positive matrix factorization with conditional probability functions in air quality studies: an application to hydrocarbon emissions in Houston, Texas. Atmos. Environ. 40, 3070–3091. Yu, Y.F., Lu, Q., Zheng, J.Y., Zhong, L.J., 2011. VOC emission inventory and its uncertainty from the key VOC-related industries in the Pearl River Delta Region. China Environ. Sci. 31, 195–201 (in Chinese). Zhang, J., Wang, T., Chameides, W.L., et al., 2007. Ozone production and hydrocarbon reactivity in Hong Kong, Southern China. Atmos. Chem. Phys. 7, 557–573. Zhang, Y.H., Su, H., Zhong, L.J., et al., 2008. Regional ozone pollution and observation-based approach for analyzing ozone-precursor relationship during the PRIDE-PRD2004 campaign. Atmos. Environ. 42, 6203–6218. Zhang, Y., Wang, X., Barletta, B., et al., 2013. Source attributions of hazardous aromatic hydrocarbons in urban, suburban and

191

rural areas in the Pearl River Delta (PRD) region. J. Hazard. Mater. 250–251, 403–411. Zheng, J.Y., Shao, M., Che, W.W., et al., 2009a. Speciated VOC emission inventory and spatial patterns of ozone formation potential in the Pearl River Delta, China. Environ. Sci. Technol. 43, 8580–8586. Zheng, J.Y., Zhang, L.J., Che, W.W., et al., 2009b. A highly resolved temporal and spatial air pollutant emission inventory for the Pearl River Delta region, China and its uncertainty assessment. Atmos. Environ. 43, 5112–5122. Zhong, L., Louie, P.K.K., Zheng, J., et al., 2013. Science-policy interplay: air quality management in the Pearl River Delta region and Hong Kong. Atmos. Environ., http://dx.doi.org/ 10.1016/j.atmosenv.2013.03.012. Dr. Hai Guo is an Associate Professor in the Department of Civil and Environmental Engineering of the Hong Kong Polytechnic University. He is also an adjunct Professor at Queensland University of Technology. His research focuses on atmospheric chemistry, regional and urban air pollution, source apportionment, photochemistry, acidic ultrafine particles and new particle formation, with over 132 papers published in high-quality international journals and conference proceedings with over 1410 SCI citations to date. He is an editorial Board member for four international journals and an active reviewer for over 35 journals, including Atmospheric Chemistry and Physics, Environmental Science and Technology, Atmospheric Environment, Journal of Geophysical Research, etc. Mr. Zhenhao Ling is currently a Ph.D. student in Dr. Hai Guo’s group of the Hong Kong Polytechnic University. His research interests include outdoor air pollution, atmospheric chemistry, regional and urban air pollution, source apportionment and photochemistry. He has published several papers in international journals, i.e., Atmospheric Chemistry and Physics, Environmental Pollution and Atmospheric Environment.