key: cord-0971903-a8qcxx16
authors: Fu, Shuang; Guo, Meixiu; Luo, Jinmin; Han, Deming; Chen, Xiaojia; Jia, Haohao; Jin, Xiaodan; Liao, Haoxiang; Wang, Xin; Fan, Linping; Cheng, Jinping
title: Improving VOCs control strategies based on source characteristics and chemical reactivity in a typical coastal city of South China through measurement and emission inventory
date: 2020-07-12
journal: Sci Total Environ
DOI: 10.1016/j.scitotenv.2020.140825
sha: 93d67c2cb1e03887f187eae110900ad076fb0f34
doc_id: 971903
cord_uid: a8qcxx16

Abstract In China, the corresponding control directives for volatile organic compounds (VOCs) have been based on primary emissions, rarely considering reactive speciation. To seek more effective VOCs control strategies, we investigated 107 VOC species in a typical coastal city (Beihai) of South China, from August to November 2018. Meanwhile, a high-resolution anthropogenic VOCs monthly emission inventory (EI) was established for 2018. For source apportionments (SAs) reliability, comparisons of source structures derived from positive matrix factorization (PMF) and EI were made mainly in terms of reaction losses, uncertainties and specific ratios. Finally, for the source–end control, a comprehensive reactivity control index (RCI) was established by combing SAs with reactive speciation profiles. Ambient measurements showed that the average concentration of VOCs was 26.38 ppbv, dominated by alkanes (36.7%) and oxygenated volatile organic compounds (OVOCs) (29.4%). VOC reactivity was estimated using ozone formation potential (52.35 ppbv) and propylene-equivalent concentration (4.22 ppbv). EI results displayed that the entire VOC, OFP, and propylene-equivalent emissions were 40.98 Gg, 67.98 Gg, and 105.93 Gg, respectively. Comparisons of source structures indicated that VOC SAs agreed within ±100% between two perspectives. Both PMF and EI results showed that petrochemical industry (24.0% and 33.0%), food processing and associated combustion (19.1% and 29.2%) were the significant contributors of anthropogenic VOCs, followed by other industrial processes (22.2% and 13.3%), transportation (18.9% and 12.0%), and solvent utilization (9.1% and10.5%). Aimed at VOCs abatement according to RCI: for terminal control, fifteen ambient highly reactive species (predominantly alkenes and alkanes) were targeted; for source control, the predominant anthropogenic sources (food industry, solvent usage, petrochemical industry and transportation) and their emitted highly reactive species were determined. Particularly, with low levels of ambient VOC and primary emissions, in this VOC and NOx double-controlled regime, crude disorganized emission from food industry contributed a high RCI.

J o u r n a l P r e -p r o o f 4 increasingly significant to achieve robust SAs by reconciling the inconsistencies between different approaches.

It is essential to scientifically control of VOCs, not only in terms of SAs but also in VOC chemical composition. The current VOC abatement in China is mainly aimed at the primary emissions, rarely taking their reactive speciation into consideration (Li et al., 2018) . The air quality analysis during the COVID-19 lockdown by Li et al. (2020) indicates that extreme reductions in primary emissions cannot fully tackle the current air pollution. To effectively tackle air pollution, a developed integrative VOC control by combining reactive speciation and SAs is necessary and urgent.

Unfortunately, distinguished from the developed megalopolis, the VOCs characteristics research in South China has always been a gap .

With a long-term dominant primary industry, the lack of systematic analysis of VOCs, particularly from disorganized emission sources, has become the main limitation of VOC control in less-developed areas of South China. Beihai, one of the 14 first so-called "coastal open cities" in South China, has consistently good air quality and ranks the forefront in China. The air quality in Beihai probability indicates the future air quality status of Chinese cities, which provides optimistic prospects for future VOC characterizations of Chinese cities. Sources emission reduction during the COVID-19 lockdown has played a significant role in the decrease of PM 2.5 , NO 2 and SO 2 concentrations in YRD , and the concentrations of these pollutants were comparable to those of Beihai in recent years (2016) (2017) (2018) . However, ozone did not show any reduction and increased greatly during the COVID-19 lockdown by Li J o u r n a l P r e -p r o o f 5 et al. (2020) , which was slightly higher than that of Beihai between 2016-2018. Will the VOC study in Beihai provide a reference for effective control of ozone precursors to reduce ozone concentrations?

In this study, an integrated VOC measurement-SAs-reactivity approach described in the next section including VOCs and other pollutant observations, potential SAs analyses by positive matrix factorization (PMF) model, a high-resolution monthly EI establishment for five levels of sub-sources, and a developed source-end comprehensive reactivity control index (RCI). This study aims to (1) reveal the concentration, compositions and characteristics of ambient VOCs in Beihai, (2) obtain robust SAs based on the comparisons and reconciling the RM and EI results, (3) comprehensively estimate the reactivity of ambient VOCs and their sources, and (4) to determine a developed VOC source-end integrated RCI for future air pollution control combined with VOC SAs and reactivity profiles. It is expected to provide references for VOC control in developing Chinese cities by exploring VOC characterizations and control strategies in Beihai.

J o u r n a l P r e -p r o o f 6 industries play important roles in Guangxi, as well as Beihai .

Besides livelihood enterprises, a few large petrochemical plants, steel plants and numerous small-scale enterprises constitute the simple industrial structure of Beihai.

Furthermore, the characteristics of tourist cities may highlight the contribution of local vehicle exhaust to VOC emissions . It is a typical representative of the underdeveloped areas in South China.

Based on the frequency statistics and levels of ozone pollution, temporal variations in VOC emission, and meteorological factors of 2016-2018 (Table S1 ), we concluded that the late summer and autumn period (from August to November) can basically reflect the year-round ozone conditions in Beihai. To investigate the VOCs concentrations, five sampling locations were established based on automatic air quality monitoring stations, including four national stations (YT, NE, NWL, and BI) and an additional urban station (BJ) (Fig. 1 ). And NWL served as the background monitoring point. Synchronous sampling and analysis were conducted at each site for four workdays per month under fine weather conditions (no rain and low wind speed).

For each sampling day, two ambient VOC samples were collected during 9:00-10:00 and 15:00-16:00, which captured the troughs and peaks of ozone levels based on the photochemical reaction of VOCs in the daytime (Song et al., 2019) . All sampling stations are located 30-50 m above the ground. A total of 160 VOC samples were collected across the sampling events. 

The samples were collected in pre-evacuated stainless steel canisters and subsequently analyzed with the GC-MS/FID system equipped with TH-PKU 300B pre-concentration system. The offline analysis of ambient VOCs is adopted the same procedure used by Zhang et al. (2019) and Simayi et al. (2020) , which is described in Supplement Text 1. The correlation coefficient values of the calibration curve for all target compounds were above 99.5% and the relative mean deviation for the target compounds of five parallel samples was within 15%. In total, 107 VOC species were recognized, which were classified into seven categories (29 alkanes, 11 alkenes, acetylene, 18 aromatics, 35 halocarbons, 12 oxygenated volatile organic compounds [OVOCs] , and carbon disulfide), and their concentrations were quantified. The J o u r n a l P r e -p r o o f 8 concentration, standard deviation, and method detection limit (MDL) of each component are listed in Table S2 .

As previous studies, a PMF 5.0 receptor model was established to identify VOC sources in this study (Li et al., 2017; Mousavi et al., 2018; Simayi et al., 2020; Wu et al., 2016) . Depending on the mass balance rather than the source component spectrum, the model can be optimized according to the standard deviation of the data. Negative constraint alone is an additional advantage of PMF (Dumanoglu et al., 2014) . The mass balance equation of PMF is shown in Eq. (1)

Where x ij is the concentration of species j in sample i; g ik is the concentration of factor k contribution in sample i; f kj is the mass percentage of species j in source k; e ij is other factors for species j in sample i and p is the total number of sources.

Based on the uncertainties, the PMF solution minimizes the objective function

(2) Where u ij represents the uncertainty of species j in sample i.

The uncertainty of VOC samples is calculated using Eq. (3-4). If the VOC concentration is below the MDL, Eq. (3) is adopted; otherwise, Eq. (4) is adopted.

J o u r n a l P r e -p r o o f 9 Unc = √ (Error Fraction × concentration) 2 + (0.5×MLD) 2 (4) Where MDL is the detection limit. The error fraction can be 5%-20% (Buzcu and Fraser, 2006; Song et al., 2007) , and it was set to 10% in this study.

To ensure a good PMF solution, we replaced the data below the MDL with values equal to half the MDL. Species that were not found or below the MDL in more than 25% of the samples were excluded ; otherwise, we replaced the missing data with the geometric mean of the detected concentrations, and set their uncertainties to four times the geometric mean (Wu et al., 2016) . Analyzing the PMF running results, all bootstrap runs were performed with correlation coefficient (R 2 ) > 0.8 and almost 98% of the selected species had an absolute scaled residual less than 3, which indicated great observed-predicted correlations.

EI is recommended to estimate the primary emissions, which involves estimation applying the statistics of activity levels and localized emission factors. VOC emissions of each sub-source were calculated separately with the emission factors approach with the Eq. (5)

Where E n refers to total emission for emission source n; EF n,m refers to the emission factor for m th of source n; A n,m refers to the activity levels for m th of source n and η n,m refers to VOCs control efficiency for m th of source n.

A comprehensive high-resolution classification EI of Beihai was first established, based on emission factors and local activity level data in 2018. Specifically, emission J o u r n a l P r e -p r o o f 10 factors were obtained from the corresponding EI Guidebooks, reports, and industrial empirical parameters (Table S4 ). The activity data mainly derived from field investigation and tabulating survey, others were supplemented from the corresponding Environmental Statistics, Yearbook and official bulletins. The survey data were derived from nearly 20 government departments and organizations, and nearly 600 industries related to VOC emissions have been examined in Beihai. Major emission sources, including coal plants, burning boilers and kilns, the petrochemical industry, motor vehicles, shipping, aviation, railway machines, raw chemical manufacturing factories, and biological emissions, were investigated in detail. Additional explanation for the sources and types of activity date using in VOC emission inventory is described in Supplement Text 2. The year-round activity level data and monthly uneven coefficient of five level sub-sources are listed in Table S5 .

VOCs chemical reactivity is not proportional to their concentrations, which exhibits a wide range and act as another important distinguishing characteristic of VOC species. The highly reactive species degrade rapidly, while the less reactive ones are relatively stable, increasing VOC accumulation and exaggerating the potential VOC sources contribution (Ou et al., 2018) . In this study, to adequately estimated the chemical reactivity of VOCs, the OH reactivity and ozone formation potential (OFP)

were of particular concern. Here, OH reactivity level was normalized by

propylene-equivalent ( propy-equiv) concentration, which was evaluated by the Eq. (6) (Goldan and D., 2004; Wu et al., 2016) . Table S2 ).

Based on the maximum incremental reactivity (MIR) factor, OFP was often used to evaluate the contribution of ambient VOCs to ozone formation Vizuete et al., 2008) . Referring to the series of MIR values updated by Carter (2010) (see Table S2 ), OFPs were calculated based on the Eq. (7) OFP

Where OFP i is the OFP for VOC species i and MIR i is the MIR for VOC species i.

Then a balanced reactivity control index (RCI) was defined as the Eq. (8), based on normalized index.

where RCI i is the normalized reactivity control index of VOC species i; k is the weight, and k 1 and k 2 were set to 0.5 in this study; Propy-equiv Min and (Simayi et al., 2020) , much higher than other cities. In general, the short-chain alkanes contributed higher concentrations, and isopentane (2.09 ppbv)

was the most abundant alkanes in Beihai, followed by several C 2 -C 4 alkanes (above 1 ppbv). Additionally, 1,2-dichloropropane (0.92 ppbv), dichloromethane (0.91 ppbv), J o u r n a l P r e -p r o o f 13 and toluene (0.57ppbv) were the common halocarbons and aromatic. The concentration of each VOC species is listed in Table S2 , and the comparisons of VOCs concentration, compositions and top 10 species with other areas are summarized in Table S3 .

As shown in Fig 

The average concentrations of O 3 , NOx, and CO during the study period were 67.08, 19.89, and 890 μg/m 3 , respectively. The concentration of NOx was lower than that in many cities . For the diurnal variation of secondary pollution, O 3 revealed a signal peak tendency with minimum and maximum concentrations appearing at 8:00-9:00 and 14:00-16:00, probably affected by solar radiation variations during daytime (Fig. 3) . For the diurnal variations of ozone precursors, NOx showed the opposite tendency and bimodal tendency, corresponding to morning and evening traffic peak; VOCs concentration generally decreased during the intense photochemical reactions about 15:00. However, the VOCs concentration at 15:00 was higher than 9:00 in several sampling days, especially in September and

October. This was likely due to the diurnal accumulation of VOC emissions during the annual peak production period. Similar diurnal variations of NOx were observed, but the occurrence days were not fully synchronized with VOCs. Moreover, associations between VOCs and other air pollutants were shown in Table S6 . A weak positive correlation between VOCs and NOx was performed (r s = 0.100). These results implied that there were some differences between the sources of NOx and .255, p < 0.01) revealed that VOCs may contribute to PM 2.5 generation (Han et al., 2018; Zhang et al., 2019) . CO as a tracer of vehicle exhaust, its significantly correlations with VOCs (p < 0.05) and NOx (p < 0.01) reflected the obvious effects of vehicle emission on ambient ozone precursors (Luo et al., 2010) . The average VOC (ppbC)/NOx ratio (8.38:1) was closed the approximately threshold (8:1) of the transition from VOC-limited to NOx-limited regime (Seinfeld, 1989) . In August and September, VOC/NOx ratios were approximately 7:1, indicating the VOC-limited formation in early autumn. Therefore, in this city with lower concentrations of ozone precursors, ozone formation was more likely to be double-controlled by VOC and NOx. 

Considering the typical tracer species, high concentration species, and reactive identification species, 80 VOC species were selected to input PMF models, Table S7 .

Based on the PMF-resolved results, transportation, other industrial processes, biogenic emissions, and fuel evaporation contributed 16.3%, 19.1%, 13.7%, and 4.6%, respectively. Their identification is described in Supplement Text 3, and other three regional special sources are detailed below.

In factor 2, the petrochemical industry was identified by abundant C 2 -C 4 alkanes, C 2 -C 3 alkenes, ethyne and aromatics, correspond with the emission compositions of refinery and petrochemical processing (Buzcu and Fraser, 2006) . Additionally, 1,2-dichloroethane (67.8%) was dominated in this factor, which is also a significant symbol of the petrochemical industry (Dumanoglu et al., 2014) , followed by 1,2-dichloropropane (65.3%), and methylene chloride (60.5%). Petrochemical industry, a significant industrial VOC emission source in Beihai, was identified as an independent source, contributing highest (20.7%) to VOCs.

Factor 3 had a higher contribution (16.5%) to VOCs, dominated by higher percentages ethene (61.8%), propane (54.5%), propene (50.1%), acetylene (47.9%), n-butane (41.0%), i-butane (37.0%) and some aromatics. VOCs emitted from coal-fired boilers are usually composed of acetylene, C 2 -C 4 alkanes and alkenes (Liu et al., 2008) . Generally regarded to be a typical trace of biomass burning (Liu et al., J o u r n a l P r e -p r o o f the profile was similar to combustion, but the proportion of butane (11.4%) in this factor was higher than that in combustion profiles (2-3%) by Wu and Xie (2017) . In the EI of Beihai, combustion only occupies a small part, which are mostly served for food processing with abundant butane disorganized emission (Wu and Xie, 2017) .

Therefore, this factor was primarily attributed to food processing and related fuel combustion.

Solvent utilization was identified in factors 7 and 8. The former was because of high fractions of aromatics (predominantly BTEX), which was distinguished as solvent usage in coating/painting (Cai et al., 2010; Hui et al., 2018; Ling et al., 2011) ; the latter was explained by more than 60% of OVOCs, especially abundant acetone (38.2%) and 2-propanol (19.9%). Previous studies suggested that acetone was mainly emitted from industrial solvents and household solvents (Ou et al., 2015; Zheng et al., 2013) , and 2-butanone was related to the pharmaceutical utilization (Simayi et al., 2020) . They were merged into "solvent utilization", with a total contribution of 9.1%.

To obtain more reliable SAs by comparative analysis, we also investigated the EIs to estimate the primary VOC emissions. As shown in Fig 

In Fig. 6 , the source structures of anthropogenic VOC sources resolved by PMF were compared to that of annual EI. Petrochemical industry contributed the largest to anthropogenic VOCs in both EI and RM, but its proportion in EI (33.0%) was slightly higher than that in RM (24.0%). Similar differences on the food processing and associated combustion between EI (29.2%) and RM (19.1%) were more obvious.

They were consistent with the combustion divergences for the city-center of Chengdu from EI (17.9-59.4%) and RM (12.1-31.0%) by Simayi et al. (2020) . These divergences probably attribute to their active components, such as C 2 -C 4 alkenes and alkyne, which are removed instantly by atmospheric oxidation (Buzcu and Fraser, 2006) . As illustrated in Eq. (9-13), the chemical losses of highly reactive light olefins reduce the source contributions resolved by PMF (Derwent et al., 2001) . But the divergence on the SAs of petrochemical industry was smaller, due to great amounts of inactive alkanes and halocarbons also in petrochemical source profiles. Long-lived inactive components tend to accumulate in the atmosphere (Gouw et al., 2005) , which maybe overestimate the relevant source contributions by PMF. This is one possibility for the SA divergences of other industrial processes between EI (13.3%) and RM (22.2%). The incomplete statistics of other industries may also enhance the underestimation of this source proportion in EI (Ou et al., 2018 

The contribution of transportation with both methods was comparable and significant. The ratios of T/B (1.52) were close to 2, which indicated that vehicle emission is dominant in Beihai (Garzón et al., 2015) . However, the contribution of transportation to EI (12.0%) was slightly lower than that from RM (18.9%). It is worthwhile to note that the kinds of VOCs by measurement in the study (107) The contribution of solvent utilization to EI (9.1%) was comparable to that from RM (10.5%). The low contribution of solvent utilization was consistent with low level ambient aromatics in Beihai. Moreover, the SAs of solvent utilization were both lower than that in Chengdu (26-27%) by Simayi et al. (2020) . These results indicated the SA of solvent utilization in Beihai is low, especially from industrial solvent.

Last but not least, compared to other cities, quite high ratios of X/E (1.61) were observed, suggesting little aging air mass and a lower exposure to the long-distance pollution sources (Garzón et al., 2015; Zhang et al., 2019) . The secondary and aging air masses was excluded from EI, which also was not identified by PMF in this study.

Overall, although there are little differences between RM and EI, the source structures with two methods achieve good consistencies. Furthermore, addressing the impacts of chemical losses by using RMs, obtaining detailed source separation characteristics and supplementing emissions from missing sub-sources will contribute to reconcile the SAs discrepancies between EI and RM. The ratios of T/B and X/E are showed in 

As shown in Fig. 7 , the OFP and propy-equiv concentrations were 52.35 ppbv and 4.22 ppbv, respectively. Although alkene concentrations were lower compared to alkanes, their high reactivity made the largest contribution to ozone formation (30.3%) and propy-equiv concentration (44.1%), confirmed by previous research (Hui et al., 2018; Li et al., 2015; Zhang et al., 2019 (Ou et al., 2018) . The species reactivity based on OH reactivity and OFP was not completely consistent. From the top 10 species based on different scales (Fig.   7) , although isoprene concentration was relatively low, it ranked at first (23.0%) and

fourth (5.1%) in the OH reactivity and OFP, respectively; while ethene contributed the largest to OFP (14.8%), but it contributed little to OH reactivity (6.6%). This is because that propy-equiv method is mainly focused on kinetic activity and probably overestimate the species (like isoprene) with a faster OH reaction rate, while the MIR method is focused on the mechanism reactivity (Zou et al., 2015) .

Therefore, VOC reactivity was balanced by equivalent weights for both MIR and propy-equiv methods using Eq. (8). According to the VOCs RCI, 107 ambient VOCs were classified to five control levels (Table S7) , and these fifteen VOCs in level Ⅰ (isoprene, ethene, isopentane, 2-butanone and propene) and level Ⅱ (toluene, i-butane, m/p-xylene, n-pentane, acetone, n-butane, trans-1,3-dichloropropylene, acrolein, 1,2,3-trimethylbenzene and styrene) were selected as active ambient species. The terminal control focused on the reactive VOCs (especially level Ⅰ species) is more targeted to effectively reduce the formation of ozone. Meanwhile, with respect to the VOC abatement, the reactive VOC sources also need to be considered. cannot be ignored, especially from C 4 -C 5 alkanes. The RCI (0.09) of other industrial processes was small, and some attention should be paid to synthetic product and leather manufacturing, as well as metal smelting. In brief, policies to reduce anthropogenic VOC emissions should be mainly aimed at food industry and corresponding combustion by boilers and kilns, solvent usage, petrochemical industry and transportation, and in particular on their highly reactive species (see Tables S9-S10 ). 

In this study, 107 ambient VOCs were continuously measured and their characteristics were explored for the first time in Beihai, providing optimistic prospects for future VOC characterizations of Chinese cities. The correlations between air pollutants and VOC (ppbC)/NOx ratio (8.38:1) revealed ozone formation was more likely to be double-controlled by VOC and NOx in this city with low levels of ozone precursors.

Meanwhile, both SAs and VOC reactivity were synthetically analyzed based on different methods to develop strategies for optimal control of VOC sources and reactive species. SAs were estimated from different perspectives: seven potential sources were identified with the PMF model, and a high-resolution monthly dynamic EI with eight sources (five level sub-sources) was established. Furthermore, robust SAs were obtained by comparing and reconciling the differences of two approaches in terms of chemical reactivity of species, reaction losses, uncertainties, pollutant transmission, T/B and X/E and so on. Both EI and RM results indicated that petrochemical industry, food industry and transportation were significant contributors.

The reactivity of ambient VOCs and sources was evaluated by both MIR and propy-equiv methods. These results by two methods was separately normalized and balanced using equivalent weights. Finally, aimed at the source-end control, a comprehensive VOCs abatement policy was suggested: for end control, the ambient VOC reactivity control index (RCI) was determined, and reducing the emissions of J o u r n a l P r e -p r o o f 32 those fifteen highly reactive VOCs above mentioned (RCI > 0.25) was more effective;

for source control, the predominant anthropogenic sources and their emitted highly reactive VOC species were determined based on SAs and RCI. According to source reactivity, VOCs emission from food industry was the most prominent, which is a significant characteristic in parts of South China. Unorganized emission from the food industry need to be tightly controlled.

The VOCs characterizations and robust SAs in this study provide scientific support to analyze air quality. The RCI of VOCs source-end control established in this study supplemented the consideration of reactive speciation on the basis of VOC emission intensity, which offers a more scientific reference for effectively control future ozone pollution.

J o u r n a l P r e -p r o o f

Characteristics and source apportionment of VOCs measured in an industrial area of Nanjing

Atmospheric Degradation of Volatile Organic Compounds

Source identification and apportionment of volatile organic compounds in Houston

Characteristics and source apportionment of VOCs

Development of the SAPRC-07 chemical mechanism

Effects of reactive hydrocarbons on ozone formation in southern Taiwan

Oceanic distributions and air-sea fluxes of biogenic halocarbons in the open ocean

Characterization of the Reactivities of Volatile Organic Compounds Using a Master Chemical Mechanism

Spatial and seasonal variation and source apportionment of volatile organic compounds (VOCs) in a heavily industrialized region

Volatile organic compounds in the atmosphere of Mexico City

Nonmethane hydrocarbon and oxy hydrocarbon measurements during the 2002 New England Air Quality Study

Budget of organic carbon in a polluted atmosphere: Results from the New England Air Quality Study in 2002

Which emission sources are responsible for the volatile organic compounds in the atmosphere of Pearl River Delta?

Do volatile organic compounds (VOCs) emitted from petrochemical industries affect regional PM 2.5 ?

Characteristics, source apportionment and contribution of VOCs to ozone formation in Wuhan, Central China

HTAP_v2.2: a mosaic of regional and global emission grid maps for 2008 and 2010 to study hemispheric transport of air pollution

A comprehensive classification method for VOC emission sources to tackle air pollution based on VOC species reactivity and emission amounts

Verification of anthropogenic VOC emission inventory through ambient measurements and satellite retrievals

Air quality changes during the COVID-19 lockdown over the Yangtze River Delta Region: An insight into the impact of human activity pattern changes on air pollution variation

Insight into the roles of worm reactor on wastewater treatment and sludge reduction in anaerobic-anoxic-oxic membrane bioreactor (A2O-MBR): Performance and mechanism

Characteristics of volatile organic compounds and their role in ground-level ozone formation in the Beijing-Tianjin-Hebei region

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

Sources of ambient volatile organic compounds and their contributions to photochemical ozone formation at a site in the Pearl River Delta, southern China

Emission inventory of atmospheric pollutants from industrial sources and its spatial characteristics in Guangxi

VOC from Vehicular Evaporation Emissions: Status and Control Strategy

Source profiles of volatile organic compounds (VOCs) measured in China: Part I

VOCs and OVOCs distribution and control policy implications in Pearl River Delta region, China

Intake fraction of nonreactive motor vehicle exhaust in Hong Kong

Characterization of non-methane hydrocarbons and their sources in an industrialized coastal city

Diurnal and seasonal trends and source apportionment of redox-active metals in Los Angeles using a novel online metal monitor and Positive Matrix Factorization (PMF)

OVOC and VOC emission inventories and their implications for reactivity-based ozone control strategy in the Pearl River Delta region

Reconciling discrepancies in the source characterization of VOCs between emission inventories and receptor modeling

Analysis of the characteristics of industrial structure evolution in Guangxi

Urban Air Pollution: State of the Science

Estimates of anthropogenic halocarbon emissions based on its measured ratios relative to CO in the Pearl River Delta

Understanding the sources and spatiotemporal characteristics of VOCs in the Chengdu Plain, China, through measurement and emission inventory

Sources and abatement mechanisms of VOCs in southern China

Source Apportionment of Ambient Volatile Organic Compounds in Beijing

Modeling ozone formation from industrial emission events in

Variation of ambient non-methane hydrocarbons in Beijing city in summer

Characteristics and source apportionment of VOCs in the suburban area of Beijing

Review of volatile organic compound source apportionment by chemical mass balance

Characteristics, source apportionment and reactivity of ambient volatile organic compounds at Dinghu Mountain in Guangdong Province

Spatial Distribution of Ozone Formation in China Derived from Emissions of Speciated Volatile Organic Compounds

Ambient volatile organic compounds (VOCs) in two coastal cities in western Canada: Spatiotemporal variation, source apportionment, and health risk assessment

Volatile organic compounds in the Pearl River Delta: Identification of source regions and recommendations for emission-oriented monitoring strategies

Variations of ground-level O 3 and its precursors in Beijing in summertime between

Characteristics, reactivity and source apportionment of ambient volatile organic compounds (VOCs) in a typical tourist city

This study was supported financially by the National Natural Science Foundation