Discussions Published by Copernicus Publications on behalf of the European Geosciences Union.

Abstract. Water-soluble organic carbon (WSOC) is a major component of carbonaceous aerosols. However, the detailed information of WSOC origins is still unclear. In the current study, fine particles (PM2.5) were collected at one sub-urban and two rural sites in the Pearl River Delta (PRD) region, south China during fall–winter 2008 to measure WSOC and organic tracers of biomass burning (BB) and secondary organic aerosols (SOA) from isoprene, monoterpenes, β-caryophyllene, aromatics and 2-ring polycyclic aromatic hydrocarbons (PAHs). WSOC concentrations ranged from 7.63 to 11.5 μg C m−3 and accounted for 38.8–57.9% of organic carbon (OC). Both WSOC and water-insoluble organic carbon (WIOC) exhibited higher levels at the sub-urban site than the rural sites. Subtracting BB-derived WSOC (WSOCBB) from measured WSOC, secondary OC (SOC) and primary OC (POC) were estimated that POC exhibited dominance over SOC and contributed 68–79% to OC. Significant correlation between WSOC and EC was observed, suggesting that BB could have important contributions to ambient WSOC in the PRD region during fall–winter. Organic tracers were applied to do source apportionment of WSOC, which further confirmed that BB was the dominant contributor, accounting for 42–47% of measured WSOC. SOC estimated by SOA tracers totally contributed 22–40% of WSOC, among which anthropogenic SOC (sum of aromatics and 2-ring PAHs, 18–25%) exhibited dominance over biogenic SOC (sum of isoprene, monoterpenes and β-caryophyllene, 4–15%). The unexplained WSOC (18–31%) showed a positive correlation with POC, indicating that this portion might be associated with POC aging.


Introduction
Organic aerosols (OA), originated from both direct emission (primary organic aerosol or POA) and secondary formation (secondary organic aerosol or SOA) (De Gouw and Jimenez, 2009), are the major components of fine particulate matter (PM 2.5 ).As a large Figures

Back Close
Full fraction of organic carbon (OC), water-soluble organic carbon (WSOC) is associated with oxygenated and polar compounds.Due to its potential to modify the hygroscopic and optical properties of atmospheric particles (Jung et al., 2011;Zhang et al., 2011), WSOC has caused increasing concerns about its adverse effects on regional air quality and global climate change.To develop effective control strategies for ambient PM 2.5 , it is important to understand the origins of WSOC.So far, however, there has limited knowledge of the sources of WSOC, although the rough estimate can split it into SOA and biomass burning (BB) (Ding et al., 2008a;Snyder et al., 2009;Stone et al., 2008;Weber et al., 2007).SOA is produced by condensation of oxidation products of volatile organic compounds (VOCs) (Claeys et al., 2004) as well as aerosol aging (Jimenez et al., 2009).
Since SOA is primarily composed of oxygenated compounds that are highly watersoluble (Kanakidou et al., 2005), WSOC is often regarded as an indicator of SOA and an increase in WSOC to OC ratio (WSOC/OC) is considered as the enhancement of SOA formation (Weber et al., 2007).However, the origins of SOA itself are still unclear.
For SOA precursor emissions, biogenic VOCs (terpenes) were dominated over anthropogenic VOCs (e.g.aromatics) on the global scale (Guenther et al., 1995;Piccot et al., 1992); while anthropogenic SOA could play dominant role, especially in the regions with high human activities (Ding et al., 2008b(Ding et al., , 2012;;Spracklen et al., 2011;Volkamer et al., 2006).Moreover, POA aging that significantly improved model prediction of SOA (Robinson et al., 2007) could also enhance anthropogenic contributions to estimated SOA.Recently, SOA aging mechanism (Donahue et al., 2012) was discovered which could further short the gap between model prediction and field observation.At the moment, it is still a challenge to quantitatively identify SOA compositions and sources, which limits the further understanding of WSOC origins.
BB aerosol that contributed more than 90 % of global primary OC (POC) emissions (Bond et al., 2004) also contained a large fraction of highly water-soluble compounds (Lee et al., 2008;Viana et al., 2008).An increase in WSOC/OC ratio is also expected when BB events happen.Therefore, the enhancement of WSOC in the ambient can Introduction

Conclusions References
Tables Figures

Back Close
Full result from the influence of either SOA or BB.Moreover, intense BB could not only deeply influence SOA formation (Ding et al., 2013) but also lead to significant overestimate of bulk SOC by EC-tracer method (Ding et al., 2012) which is widely used for SOC estimation (Turpin and Huntzicker, 1995).To separate SOA and BB contributions to WSOC and avoid BB influence on SOC estimation, specific organic tracers for SOA and BB are essentially needed.The Pearl River Delta (PRD) is one of the most industrialized and densely populated regions in China.The rapid growth in economy has led to fast increase in anthropogenic emissions of air pollutants (Chan and Yao, 2008).Situated in the subtropical area, biogenic emissions in this region are also expected to be significant during the whole year (Zheng et al., 2010).Previous studies all suggested that air quality during fallwinter was the worst in a year in the PRD region, based on the monitoring of PM 2.5 , O 3 , visibility and other criteria pollutants (Ding et al., 2011;Wang et al., 2012;Zhang et al., 2012).Additionally, open burning of agriculture residuals often happens in the sub-urban and rural areas in the PRD during fall-winter (He et al., 2011).Considering the high VOC emissions, the high atmospheric oxidative capacity and the high BB loading, both SOA and BB aerosol should have significant contributions to WSOC in the PRD during this period.In the current study, 24 h PM 2.5 samples were collected consecutively at three sites in the PRD during fall-winter.WSOC and organic tracers of BB and SOA from isoprene, monoterpenes, β-caryophyllene, aromatics and 2-ring PAHs were measured with the purposes (1) to acquire the spatial distribution of WSOC over the PRD; (2) to estimate bulk SOC in the significant BB-influence season; and (3) to apportion WSOC in detail based on the specific organic tracers.Introduction

Conclusions References
Tables Figures

Back Close
Full PM 2.5 samples were collected using high volume samplers (Tisch Environmental, Inc.) at a flow rate of 1.1 m 3 min −1 at two rural sites: Wangqingsha (WQS) and Dinghushan (DHS) and one sub-urban site: Guangzhou institute of geochemistry (GIG).As Fig. 1 showed, WQS site is located in the central of PRD and surrounded by city clusters (e.g.Hong Kong, Guangzhou, Shenzhen, Foshan and Dongguan) 60 km away and adjacent to the Pearl River estuary.The sampler was put on the rooftop, about 30 m above ground, of a seven-floor building in a high school (22 • 42 N, 113 • 32 E).Since the surrounding terrain is flat with large farmland nearby and rare traffic, this site serves as an ideal location to monitor the regional background levels of air pollutants in the PRD (Ding et al., 2012).Dinghushan Biosphere Reserve ( 23• 09 -23 • 11 N, 112 • 31 -112 • 34 E) located in the northwest of PRD (Fig. 1), is covered by sub-tropical evergreen broadleaf forest and pine forest with monsoon climate (Yi et al., 2007).The sampler was put on the top of a hill (∼ 100 m above ground, 23 • 10 N, 112 • 33 E) at the eastern edge of the reserve.At the foot of the hill are local resident and industry areas and farmland (Fig. 1).Biogenic VOCs (BVOCs) emitted from local vegetations could be mixed and reacted here with anthropogenic pollutants transported from the adjacent region.Observation at DHS site can provide information about the influence of anthropogenic emissions on forest area.GIG is situated 5 km away from downtown Guangzhou and served as a sub-urban site.There are two highways in the south and west, respectively and farmlands in the northeast (Fig. 1).The sampler was put on the rooftop, about 20 m above ground, of a four-floor building in the institute (23 Field campaigns were carried out during fall-winter (November to December) considering the severe air pollution and BB influence (Ding et al., 2011;He et al., 2011).At WQS site, sampling was undertaken from 10 November to 9 December 2008.After Introduction

Conclusions References
Tables Figures

Back Close
Full

Chemical analysis
A punch (1.5 cm × 1.0 cm) of each filter was taken for the measurements of OC and EC using the thermo-optical transmittance (TOT) method (NIOSH, 1999) by OC/EC Analyzer (Sunset Laboratory Inc. ).An additional punch of 2.54 cm diameter was taken from each filter and extracted in 20 mL of 18-Mohm milliQ water and sonicated for 60 min in an ice-water bath (Ding et al., 2008a).After filtered, the extract was analyzed for WSOC using a TOC analyzer (Shimadzu TOC-V CPH ).Ambient OC, EC and WSOC were corrected using field blanks.

Conclusions References
Tables Figures

Spatial variations
As summarized in Table 2, WSOC levels were higher at sub-urban site (11.5 ± 4.29 µg C m −3 at GIG) as compared to the rural sites (8.62 ± 3.99 µg C m −3 at WQS and 7.63 ± 2.37 µg C m −3 at, DHS).During fall-winter, the concentrations of WSOC in Chinese megacities were reported as 9.6±3.8µg C m −3 in Guangzhou, 6.7±3.0 µg C m −3 in Shanghai, 7.5 ± 4.9 C µg m −3 in Beijing (Feng et al., 2006) and 20.53-35.58µg C m −3 in Nanjing (Wang et al., 2002).WSOC observed at four sites in the PRD was in the range of 2.13 to 4.00 µg C m −3 during winter (Huang et al., 2012a).Our measurements were comparable with these data in Chinese cities but much higher than those (∼ 1-2 µg C m −3 ) in Europe (Viana et al., 2007) and United States (Ding et al., 2008a;Weber et al., 2007) during the same season.Although WSOC levels at DHS were the lowest among the three sites, the shares of WSOC in OC (WSOC/OC) there, were the highest (57.9 %); and WSOC exhibited significant correlation with OC (Fig. 2a).All these indicated the dominant role of water soluble compounds in OC at the forest site.WQS (38.8 %) and GIG (42.2 %) had the similar OC contributions from WSOC.WSOC/OC observed in this study were comparable with those (20-51 %) over the PRD region (Ho et al., 2006;Huang et al., 2012a), higher than those (∼ 20 %) in Tokyo (Miyazaki et al., 2006) but lower than those (44-72 %) in the southeastern United States (Ding et al., 2008a).
It is worth noting that WSOC was significantly correlated with EC (Fig. 2b).EC is major from diesel exhaust and BB (Ke et al., 2007) and can be regarded as a primary Introduction

Conclusions References
Tables Figures

Back Close
Full tracer; while WSOC mainly comes from SOA and BB (Ding et al., 2008a).If samples were deeply influenced by BB, it is expected that WSOC and EC are correlated well with each other.In fact, fall-winter is the harvest season of crop in the PRD region.Our previous study observed the great enhancement of BB in the PRD during fall-winter (Ding et al., 2012).Since SOA could not contain EC, the positive correlation between WSOC and EC suggested that BB had significant contributions to WSOC in the PRD region during fall-winter.
Water-insoluble organic carbon (WIOC) is mostly from primary emissions (Park and Cho, 2011), such as vehicle exhaust (Miyazaki et al., 2006) and can be calculated as: Among the three sites, the average level of WIOC was the highest at GIG (16.9 ± 7.96 µg C m −3 ), largely due to the more influence of primary emissions near urban Guangzhou.Located in the downwind of the PRD region, WQS site also showed high WIOC concentration (13.9 ± 5.39 µg C m −3 ), probably resulting from the transport of primary pollutants from the adjacent city clusters.The WIOC levels at DHS site (5.74 ± 2.56 µg C m −3 ) were much lower than those at above two sites by a factor of 2-3, indicating the less influence from primary emissions at the forest site.WIOC was dominated in OC at WQS and GIG with the fractions of WIOC in OC (WIOC/OC) 61.2 % and 57.8 %, respectively.Moreover, WIOC exhibited higher correlation coefficient with OC than WSOC at all sites (Fig. 2 a and c).This is more obvious at GIG site that only WIOC was correlated with OC.Since WIOC comes from primary emissions, the higher WIOC/OC and the higher correlation coefficient of WIOC vs. OC as compared to WSOC implied that OC over the PRD was dominantly from primary sources during fall-winter.
Since both WIOC and EC are emitted from primary sources, WIOC exhibited good correlation with EC (Fig. 3d).Miyazaki et al. (Miyazaki et al., 2006) found the ratios of WIOC to EC (WIOC/EC) were very stable (1.0 ± 0.1) during the whole year in urban Tokyo and pointed out that motor vehicle was an important source of WIOC based on 13780 Introduction

Conclusions References
Tables Figures

Back Close
Full the significant correlations of WIOC with EC and carbon monoxide (CO).The WIOC/EC ratios were 3.24±1.03,2.97±1.12 and 2.59±1.13 at WQS, DHS and GIG, respectively.Obviously, WIOC/EC ratios in this study were changing from place to place and about 2-3 times higher than those in urban Tokyo, suggesting additional sources of WIOC other than motor vehicle over the PRD region.

Source apportionment of carbonaceous aerosols
Since WSOC is mainly derived from BB and SOC (Weber et al., 2007), SOC can be is estimated as (Ding et al., 2008a): where SOC total is the total amount of SOC, including the known precursor-produced SOC (Kleindienst et al., 2007) and other processes produced SOC, such as the aging of POA (Robinson et al., 2007) and SOA (Donahue et al., 2012) as well as aqueousphase heterogeneous reactions (Ervens et al., 2011).WSOC is the measured value.WSOC BB is BB derived WSOC, which can be estimated using ambient tracer concentration (C tracer ) and the factor, (f tracer/WSOC ) BB , of tracer in WSOC from BB sources (Eq.3).
As a typical BB tracer, levoglucosan is always applied to access the impact of BB (Puxbaum et al., 2007;Simoneit et al., 1999).The factors of levoglucosan in WSOC from BB sources were reported as 0.0994 µg µg C −1 (1210 ng m −3 levoglucosan in 12.17 µg C m −3 WSOC) in a prescribe fire plume (Ding et al., 2008b) and 0.0806 µg µgC −1 (129 ng m −3 levoglucosan in 1.6 µg C m −3 WSOC) in the events of open burning of rice straw residues (Viana et al., 2008).The spatial distribution of BB over the PRD region showed that straw burning was the major type of BB in the central PRD; while both forest fire and straw burning had significant contributions around Introduction

Conclusions References
Tables Figures

Back Close
Full DHS (He et al., 2011).Since our campaign was undertaken in the straw burning season, (f tracer/WSOC ) BB of 0.0806 µg µgC −1 was applied to estimated ambient WSOC BB at WQS and GIG.And the 1 : 1 combination (0.090 µg µgC −1 ) of forest fire and straw burning was applied to estimated ambient WSOC BB at DHS.When WSOC BB is estimated, SOC total and POC can be calculated by Eqs. ( 2) and (4), respectively.
The estimated POC levels were 17.8 ± 7.18 µg C m −3 , 9.13 ± 3.08 µg C m −3 and 22.5 ± 7.73 µg C m −3 at WQS, DHS and GIG, respectively.Such a spatial pattern is similar as those of primary species, EC and WIOC (Table 2).The dominance of POC was observed during the whole campaign (Fig. 3) that the fractions of POC in OC (POC/OC) were 78 ± 11 %, 68 ± 15 % and 79 ± 9 % at WQS, DHS and GIG, respectively.This demonstrated that POC was the dominant contributor to OC during fall-winter over the PRD region.As compared to POC, the estimated SOC total was much lower and the concentrations were 4.78 ± 2.34 µg C m −3 at WQS, 4.18 ± 2.35 µg C m −3 at DHS and 5.97 ± 3.41 µg C m −3 at GIG.Previous studies always considered the increase of OC/EC ratio as the enhancement of SOC (Na et al., 2004;Turpin and Huntzicker, 1995;Zeng and Wang, 2011).This may be true during summer with intense photochemical reactions and rare BB influence.However, under BB impact with high OC/EC ratio (Fine et al., 2002;Zhang et al., 2007), elevated OC/EC ratio could not indicate the enhancement of SOC.This was confirmed that there was no significant correlation observed between OC/EC ratio and SOC/OC at all sites (p > 0.05).
To get insight into the origins of WSOC, SOA tracers were applied to estimate the precursor-produced SOC (SOC precursor ).Such a SOA-tracer method was first proposed by Kleindienst et al. (Kleindienst et al., 2007) with the assumption that in the ambient the mass fractions of tracers in SOC (f tracer/SOC ) remained the same as those from chamber simulations.When determining the concentrations of these tracers in the am-Introduction

Conclusions References
Tables Figures

Back Close
Full where C tracer is the sum of all tracers' concentrations for a certain precursor.Table 2 listed the tracers observed at the three sites and the f tracer/SOC for different precursors.These data were further used to estimate SOC from isoprene (SOC I ), monoterpenes (SOC M ), β-caryophyllene (SOC C ), aromatics (SOC A ) and 2-ring PAHs (SOC P ), respectively.It should be noted that there were nine tracers involved in the source profile of monoterpene SOC (Kleindienst et al., 2007).However, only five of the nine monoterpene SOA tracers were detected in the current study.To lower the uncertainty induced from different tracer compositions, the f tracer/SOC with five monoterpene SOA tracers (Table 2) was calculated based on another chamber simulations (Offenberg et al., 2007) of the same research group.The f tracer/SOC with five monoterpene SOA tracers (0.059 µg µg C −1 ) was one fourth of that with nine tracers (0.231 µg µg C −1 ) (Kleindienst et al., 2007).Figure 4 showed the source apportionment of WSOC over the PRD region during fall-winter.BB was the predominant contributor, accounting for 42-47 % of measured WSOC.And the estimated WSOC BB concentrations were 3.85 ± 2.62 µgC m −3 , 3.39 ± 1.67 µgC m −3 and 5.55± 2.62 µg C m −3 at WQS, DHS and GIG, respectively.In Chinese city clusters, the sub-urban and rural areas are mostly surrounded and occupied by farmland, respectively.Open burning of agriculture residues is very often during harvest season and leads to deteriorating air quality (Huang et al., 2012b;Qu et al., 2012;Wang et al., 2009).Thus, BB is an important primary source to carbonaceous aerosols over the PRD region during fall-winter.SOC precursor contributed 22-40 % of WSOC with the highest at DHS and the lowest at GIG.The shares of SOC A and SOC P in WSOC were 12-18 % and 6-8 %, respectively (Fig. 4).Since both aromatics and PAHs are mainly from anthropogenic emissions, the sum of SOC A and SOC P could be regarded as the SOC from anthropogenic precursors which contributed 18-25 % of measured WSOC.It is interesting to note that the SOC Introduction

Conclusions References
Tables Figures

Back Close
Full from anthropogenic precursors exhibited little spatial difference that the concentrations were 2.28±1.58µg C m −3 , 2.06±0.97µgC m −3 and 2.13± 1.14 µg C m −3 at WQS, DHS and GIG, respectively.This implied that anthropogenic SOA had regional influence over the whole PRD.As compared to anthropogenic SOC, biogenic SOC (sum of SOC I , SOC M and SOC C ) had minor contributions, only accounting for 7 % and 4 % of WSOC at WQS and GIG, respectively.Significant enhancement of SOC M (11 %) was observed at DHS.This is expected since DHS site is located in the forest area where pine forest, one of the major forest types, could emit large amounts of monoterpenes all the year (Geron et al., 2000).Biogenic SOC contributed 15 % of measured WSOC at DHS.The dominance of anthropogenic SOC over the PRD region was also observed at WQS from summer to winter (Ding et al., 2012).WSOC BB and SOC precursor together explained 69-82 % of measured WSOC.The unexplained fraction ranged from 18 % to 31 % among the three sites.DHS site with low POC exhibited not only the lowest unexplained fraction but also the lowest unexplained WSOC concentration; while GIG site with high POC showed the highest values in the both.Moreover, the unexplained WSOC displayed a significant correlation with POC (Fig. 5).All these suggested that the unexplained WSOC might be associated with POA aging.POA emitted from fossil fuel combustion and BB can form large amounts of SOA during aging processes (Hennigan et al., 2011;Weitkamp et al., 2007).As model predicted, POA aging could contribute one third of global OA (Jathar et al., 2011).Constrained by aerosol mass spectrometer data, the estimated SOA (23 Tg a −1 ) from POA aging contributed 16 % of global SOA (140 Tg a −1 ) (Spracklen et al., 2011).Although it would contain SOC formed by processes uncaptured by the current SOAtracer method and POA emitted from sources other than BB, the unexplained WSOC might be regarded as the upper limit of SOC from POA aging.Introduction

Conclusions References
Tables Figures

Back Close
Full

Conclusions
In this study, spatial variations of WSOC in PM 2.5 were determined over the PRD region during fall-winter and the detailed information about WSOC origins were first reported based on the specific organic tracers of BB and SOA.WSOC exhibited higher levels at the sub-urban site than the rural sites and accounted for 38.8-57.9% of OC.A new method was developed to estimate SOC and POC by subtracting BB-derived WSOC from measured WSOC, which indicated the dominance of POC with the OC contribution exceeding 70 %.The correlation between WSOC and EC implied that BB could have significant contributions to ambient WSOC in the PRD region during fall-winter, which was further confirmed by the source apportionment results that BB accounted for 42-47 % of measured WSOC.The WSOC contributions from anthropogenic SOC were estimated to be 18-25 %; while biogenic SOC had minor contributions to WSOC (4-15 %).The unexplained WSOC (18-31 %) showed a positive correlation with POC, indicating that this portion could be largely due to SOC from POA aging.Introduction

Conclusions References
Tables Figures

Conclusions References
Tables Figures

Back Close
Full  Full  Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | bient, SOC from different precursors can be estimated as: SOC = C tracer /f tracer/SOC(5) Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Weitkamp, E. A., Sage, A. M., Pierce, J. R., Donahue, N. M., and Robinson, A. L.: Organic aerosol formation from photochemical oxidation of diesel exhaust in a smog chamber, Environ.Sci.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Table 1 .
Organic tracers over the PRD region and the conversion factors for WSOC BB and SOC estimate.

Table 2 .
Carbonaceous species at different sites during fall-winter over the PRD (µgC m −3 ).