13C- and 14C-based study of sources and atmospheric processing of water-soluble organic carbon (WSOC) in South Asian aerosols

Authors

  • Elena N. Kirillova,

    Corresponding author
    1. Department of Applied Environmental Science (ITM) and the Bert Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden
    • Corresponding author: E. N. Kirillova and Örjan Gustafsson, Department of Applied Environmental Science (ITM) and the Bert Bolin Centre for Climate Research, Stockholm University, 10691 Stockholm, Sweden. (elena.kirillova@itm.su.se; orjan.gustafsson@itm.su.se)

    Search for more papers by this author
  • August Andersson,

    1. Department of Applied Environmental Science (ITM) and the Bert Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden
    Search for more papers by this author
  • Rebecca J. Sheesley,

    1. Department of Applied Environmental Science (ITM) and the Bert Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden
    2. Current affiliation: Department of Environmental Science, Baylor University, Waco, Texas, USA
    Search for more papers by this author
  • Martin Kruså,

    1. Department of Applied Environmental Science (ITM) and the Bert Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden
    Search for more papers by this author
  • P. S. Praveen,

    1. Maldives Climate Observatory at Hanimaadhoo (MCOH), Republic of the Maldives
    2. Current affiliation: Scripps Institute of Oceanography, Center for Clouds, Chemistry and Climate, University of California-San Diego, La Jolla, California, USA
    Search for more papers by this author
  • Krishnakant Budhavant,

    1. Maldives Climate Observatory at Hanimaadhoo (MCOH), Republic of the Maldives
    2. Indian Institute of Tropical Meteorology, Pune, Maharashtra, India
    3. Vishwakarma Institute of Technology, Pune, Maharashtra, India
    Search for more papers by this author
  • P. D. Safai,

    1. Indian Institute of Tropical Meteorology, Pune, Maharashtra, India
    Search for more papers by this author
  • P. S. P. Rao,

    1. Indian Institute of Tropical Meteorology, Pune, Maharashtra, India
    Search for more papers by this author
  • Örjan Gustafsson

    Corresponding author
    1. Department of Applied Environmental Science (ITM) and the Bert Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden
    • Corresponding author: E. N. Kirillova and Örjan Gustafsson, Department of Applied Environmental Science (ITM) and the Bert Bolin Centre for Climate Research, Stockholm University, 10691 Stockholm, Sweden. (elena.kirillova@itm.su.se; orjan.gustafsson@itm.su.se)

    Search for more papers by this author

Abstract

[1] Water-soluble organic carbon (WSOC) is typically a large component of carbonaceous aerosols with a high propensity for inducing cloud formation. The sources of WSOC, which may be both of primary and secondary origins, are in general poorly constrained. This study assesses the concentrations and dual-carbon isotope (14C and 13C) signatures of South Asian WSOC during a 15-month continuous campaign in 2008–2009. Total suspended particulate matter samples were collected at Sinhagad (SINH) India and at the Maldives Climate Observatory at Hanimaadhoo (MCOH). Monsoon-driven meteorology yields significant WSOC concentration differences between the dry winter season (0.94 ± 0.43 µg m−3 MCOH and 3.6 ± 2.3 µg m−3 SINH) and the summer monsoon season (0.10 ± 0.04 µg m−3 MCOH and 0.35 ± 0.21 µg m−3 SINH). Radiocarbon-based source apportionment of WSOC not only shows the dominance of biogenic/biomass combustion sources but also a substantial anthropogenic fossil-fuel contribution (17 ± 4% MCOH and 23 ± 4% SINH). Aerosols reaching MCOH after long-range over-ocean transport were enriched by 3–4‰ in δ13C-WSOC relative to SINH. This is consistent with particle-phase aging processes influencing the δ13C-WSOC signal in the South Asian regional receptor atmosphere.

1 Introduction

[2] Water-soluble organic carbon (WSOC) is a substantial component of atmospheric particulate organic matter, typically contributing 20–80% of the total organic carbon (TOC) in the South Asian region [Miyazaki et al., 2009; Ram et al., 2010a, 2010b; Pavuluri et al., 2010, 2011; Rengarajan et al., 2011; Khare et al., 2011]. Due to its polar nature, WSOC is expected to act as cloud condensation nuclei (CCN), thus contributing to the indirect aerosol climate effects [e.g., Facchini et al., 1999; Charlson et al., 2001; Svenningsson et al., 2006; Asa-Awuku et al., 2008; Padró et al., 2010]. Together with a range of gaseous and particulate pollutants, such as black carbon (BC), WSOC contributes to the Atmospheric Brown Cloud (ABC) formed over South Asia [Ramanathan and Crutzen, 2003; www.rrcap.unep.org/abc/]. Anthropogenic emissions of TOC are generally high throughout India with a maximum observed over the Indo-Gangetic Plain (IGP) (Figure 1). ABC have considerable impact on regional climate, human health, storm frequency and hydrological cycles [e.g., Ramanathan et al., 2005; Ramanathan and Carmichael, 2008; Bollasina et al., 2011; Evan et al., 2011; Shindell et al., 2012].

Figure 1.

Map depicting emission inventory estimates of anthropogenic emissions of Total Organic Carbon (TOC) estimated as the sum of black carbon (BC) and organic carbon (OC) in tons year−1 per 0.5 × 0.5 degree grid for India in year 2006 (data from Zhang et al., 2009). Locations of the two study sites are marked as SINH (Sinhagad) and MCOH (Maldives Climate Observatory at Hanimaadhoo).

[3] There remains considerable uncertainty regarding the source apportionment of WSOC in the South Asia ABC. Certain primary organic aerosol (POA) emissions are expected to be hydrophilic (e.g., from biomass combustion or marine emissions), but WSOC may stem from a wide range of primary and secondary precursors [e.g., Jimenez et al., 2009; Decesari et al., 2007; Saarikoski et al., 2008; de Gouw et al., 2008]. Oxidation of primary particles during atmospheric transport (i.e., aging) through interactions with, for instance, hydroxyl radicals may make the reacted aerosol more water soluble [e.g., De Gouw et al., 2008]. Secondary organic aerosols (SOA) formed by oxidation of biogenic volatile organic compounds and VOCs from biomass combustion are also considered to be key sources of WSOC [Mayol-Bracero et al., 2002; Decesari et al., 2007; Miyazaki et al., 2009]. However, the relative contributions from different emissions sources of either primary or secondary precursors to WSOC are poorly constrained.

[4] Carbon isotopes, in particular natural abundance radiocarbon (14C), have recently been applied to differentiate between biomass and fossil contributions in the outflow of carbonaceous aerosols from the Indian subcontinent [e.g., Gustafsson et al., 2009; Sheesley et al., 2012]. This methodology offers the advantages of precise estimates (<5% uncertainty) of biomass vs fossil contributions to well-defined sub-fractions of the carbonaceous aerosol such as organic carbon (OC), soot carbon (SC), or elemental carbon (EC). Such information is important for reducing the currently large uncertainties in emission inventories (EI) of OC and black carbon (BC) aerosols [Bond et al., 2004; Venkataraman et al., 2005; Zhao et al., 2011]. Radiocarbon measurements of the WSOC component have only been reported in a few previous studies in Europe and North America and have demonstrated an overwhelming contribution of modern sources both in rural and urban environments [Szidat et al., 2004; Weber et al., 2007; Wozniak et al., 2008, 2012a, 2012b; Kirillova et al., 2010].

[5] In addition to 14C, the 13C/12C ratio provides valuable information concerning both the sources and atmospheric processing of the carbonaceous aerosols (Figure 2). Plants with C3 and C4 metabolism have distinctly different δ13C (Equation (1)), on average −27‰ and −13‰, respectively [e.g., Smith and Epstein, 1971]. In contrast, marine organic aerosol sources like phytoplankton typically have a mean δ13C isotope signature of −22‰ to −18‰ [Miyazaki et al., 2011]. Liquid fossil fuel isotope composition is similar to C3 plants (−28‰ to −25‰ [Huang et al., 2006; Widory, 2006; Lopez-Veneroni, 2009; Agnihotri et al., 2011]), solid fuels such as coal are enriched in δ13C (e.g., −23.6 ± 0.7‰ [Widory, 2006], −21.4 ± 0.1‰ [Agnihotri et al., 2011]) and gaseous fossil fuels are strongly depleted (−39.7‰ to −28.1‰ [Widory, 2006]). Incomplete combustion of fossil fuels yields combustion gases that are depleted in δ13C relative to the starting material by 1.3 ± 0.5‰ for all types of fuels [Widory, 2006] (Figure 2). The δ13C fractionation in primary emitted particles, relative to the starting fuel, differs from nearly zero (coal) to highly positive values (natural gas +11.0±5‰) and depends on combustion conditions [Widory, 2006]. Biomass combustion of C4 plants results in 13C depletion (<0.5 to 7.2‰) in the combustion derived aerosols [Turekian et al., 1998; Das et al., 2010]. While the particulate smoke produced during C3 plant combustion is enriched or depleted by only about 0.5‰ [Turekian et al., 1998, Currie et al., 1999] or demonstrates no significant fractionation compared to plant material [Das et al., 2010]. Primary particulate matter from these combustion sources may impact the δ13C signal of ambient WSOC.

Figure 2.

A simplistic illustration of how different atmospheric processes are expected to affect the δ13C of carbonaceous aerosols. Blue arrows denote depletion in the heavy isotope (lowering of the δ13C), red arrows denote enrichment (increasing δ13C) and green arrows indicate processes that may change in both directions. This view is based on the current relatively limited scientific literature on the subject [Widory, 2006; Turekian et al., 1998; Currie et al., 1999; Das et al., 2010; Rudolph et al., 2002, 2003; Anderson et al., 2004a,2004b; Wang and Kawamura, 2006; Harrington et al., 1999; Aggarwal and Kawamura, 2008; Wang et al., 2010; Pavuluri and Kawamura, 2012]. The process of SOA formation is split into VOC oxidation in gas phase when depletion of the produced oxidized low volatile species (VOCox) happens and further nucleation or partitioning of VOCox to particulate phase with no influence on the isotope composition. Light grey arrows show the equilibrium of VOCs between gas and particle phases without any chemical transformation, hence, no or very little isotope fractionation is expected. It should be emphasized that the presented overall trends may differ depending on which sub-fraction, e.g., WSOC or WINSOC, of the carbonaceous aerosol is examined.

[6] In addition to primary emissions, secondary carbonaceous aerosols, e.g., formed from VOCs in the atmosphere, are also expected to be a significant contribution to ambient WSOC (Figure 2). One of the largest global VOC sources is biogenic emissions (i.e., BVOCs) [e.g., Griffin et al., 1999; Claeys et al., 2004; Tunved et al., 2006]. Aerosols formed from precursors such as VOCs, typically leads to lower δ13C values (13C depletion) relative to the precursors according to both laboratory and field studies [Rudolph and Czuba, 2000; Rudolph et al., 2002, Anderson et al., 2004a, 2004b]. Significant fractionation has been found in the process of the biosynthesis of isoprene (2.6 ± 0.9‰ lighter than leaf carbon) [Rudolph et al., 2003]. The main removal process for BVOCs and anthropogenic volatile organic compounds (AVOCs) is the reaction with OH radicals and ozone. These oxidants preferentially react with lighter isotopes ( inverse kinetic isotope effect, KIE), which results in an enrichment of δ13C in the residual VOCs and a depletion in δ13C of the particulate oxidation products [Rudolph et al., 2002]. These oxidation products comprise a portion of the ambient WSOC.

[7] Aerosol aging, such as interactions with photochemical oxidants (e.g., hydroxyl radical, ozone), also causes fractionation of carbon isotopes in the particulate OC. One example is the oxidation of particle-phase diacids resulting in the removal of CO2/CO through reaction with OH radicals. Recent studies have demonstrated δ13C enrichment in the residual, aged aerosol in field [Aggarwal and Kawamura, 2008; Wang et al., 2010] and laboratory studies [Pavuluri and Kawamura, 2012]. Thus, aging and secondary formation are expected to affect the δ13C-WSOC values in opposite directions.

[8] The overarching objective of this study is a dual isotopic apportionment of the sources of particulate WSOC over South Asia. This study is based on a 15-month continuous aerosol sampling campaign at the international Atmospheric Brown Clouds (ABC) programme Indian Ocean receptor site at the Maldives Climate Observatory, Hanimaadhoo island (MCOH) and at the Indian Institute of Tropical Meteorology hilltop site in West India (Sinhagad; SINH). It builds on the recently reported results for the same campaign on the sources of bulk TOC and SC in aerosols [Sheesley et al., 2012]. The present study adds information on both the sources of the WSOC component and the processes involved in aerosol aging during long-range transport from source areas on the Indian subcontinent. For that purpose, a newly developed method that isolates WSOC [Kirillova et al., 2010] has been applied to this dual isotopic study (13C and 14C) in the south Asian region.

2 Methods

2.1 Sampling Campaigns

[9] High-volume aerosol samples were collected as previously described [Gustafsson et al., 2009; Sheesley et al., 2011, 2012] at two sampling stations of the international ABC project (www.rrcap.unep.org/abc/) in South Asia: the Maldives Climate Observatory at Hanimaadhoo island, Republic of Maldives (MCOH) (6.78°N, 73.18°E) and at the hill-top site of the Indian Institute of Tropical Meteorology located at Sinhagad, west India (SINH) (18.35°N, 73.75°E, 1400 meters above sea level) (Figure 1). Both serve as regional receptor sites. MCOH is an Indian Ocean site, which, during wintertime, is located downwind of the Indian subcontinent and other south Asian source regions. SINH is a regional background site in west-central India. The sampling campaign was conducted from January 2008 to April 2009. Custom built high-volume total suspended particle (TSP) samplers were operated at 14–19 m3 h−1 to collect 140 mm quartz fiber filters (Pall Gelman). The sampling interval at MCOH was roughly 1 week during the non-monsoon periods and 2 weeks during monsoon seasons. The SINH site was generally maintained at a 1-week sample duration. Filter blanks were collected approximately once per month for each site. Quartz fiber filters were pre-baked at 450 °C for 12 h and individually stored in aluminum foil envelopes in double Ziploc bags in the freezer. For courier shipping, samples were encased in a third, sealed plastic bag to minimize contamination. All blanks were shipped, stored, and processed in an identical manner as the samples.

2.2 Sample Handling and Treatment

[10] The analytical method to prepare samples for determination of 13C and 14C composition of WSOC was described previously [Kirillova et al., 2010]. Briefly, the WSOC concentration was measured after 15 min extraction of filter punches/subsamples (1 cm × 1.5 cm size) in 20 mL of Milli-Q water immersed in an ultrasonic bath (Ultrasons, JP Selecta, Abrera, Spain). The amount of filter pieces varied between one and four, as determined by the TOC load. The WSOC extracts were then isolated by centrifugation at 1500 rpm for 10 min (Sigma 4-15, Labex Instrument AB, Helsingborg, Sweden) and filtration of the supernatant using 0.02 µm cutoff aluminum syringe filters (Anotop 10 Plus; Whatman, Maidstone, Kent, UK). The concentration of the WSOC was quantified as total dissolved organic carbon (DOC) in the filtered solutions using a high-temperature catalytic oxidation (HTCO) instrument (Shimadzu-TOC-VCPH analyzer; Shimadzu, Kyoto, Japan) following the non-purgeable organic carbon (NPOC) protocol. The WSOC extraction efficiency of the method is estimated to be 91 ± 6%, and there was no isotope fractionation associated with the combined sonic extraction and freeze-drying [Kirillova et al., 2010].

[11] All WSOC results were blank corrected by subtracting an average field blank value for each site. The WSOC field blanks corresponded to an average 2% (0.20 ± 0.03 s.d., µg cm−2; n = 3) for MCOH and 1% (0.14 ± 0.01 s.d., µg cm−2; n = 3) for SINH of the field sample loadings. Triplicate analysis was done on approximately every tenth filter. The average relative standard deviation of triplicate analysis was 6% for MCOH and 5% for SINH.

2.3 Isotope Analyses and Calculations

[12] A filter area corresponding to at least 250 µg WSOC was composited from subsamples for 13C and 14C analysis. The composites were selected to be identical in time of collection to the isotope measurements of total organic carbon (TOC) and soot carbon (SC) reported on earlier for this campaign [Sheesley et al., 2012]. Due to the difference in carbon loadings roughly 3-week composites for non-monsoon winter season and 6- to 10-week composites for transitional and monsoon season samples were prepared. Each composite included equal amounts of WSOC from the combined samples. The filters were acidified by fumigation in open glass Petri dishes held in a desiccator over 12M (37%) hydrochloric acid for 24 hours to remove carbonates and subsequently dried at 60 °C for 1 hour [Kirillova et al., 2010].

[13] Decarbonated filters were then extracted for WSOC in 10 ml of Milli-Q water using the same ultrasonication method as described for WSOC concentration measurement above. The WSOC solutions were subsequently filtered into 30 ml polycarbonate vials (NALGENE, Rochester, NY, USA). The solutions were dehydrated in a low-carbon background freeze drier (Christ Alpha 2-4, LSC; Vacuum hybrid pump, Vacubrand RC-6; Martin Christ, Labex Instrument AB). The residue material was redissolved in 300 µl of Milli-Q water and transferred into the pre-combusted silver capsules (Säntis Analytical, Uppsala, Sweden). Each sample was divided into two capsules: 100 µl was placed into the capsule for carbon concentration and stable carbon isotope measurements, and the rest into the capsule for radiocarbon measurements. Finally, samples were evaporated in the oven at 60 °C and dried WSOC samples were ready for isotope ratio measurements [Kirillova et al., 2010].

[14] Stable carbon isotope (δ13C-WSOC) measurements were performed at the Stable Isotope Laboratory (SIL) of the Department of Geological Sciences at Stockholm University (Stockholm, Sweden). The instrumental method employed sample combustion with a Carlo Erba NC2500 analyzer connected via a split interface to reduce the gas volume to a Finnigan MAT Delta V mass spectrometer. Results of the analysis are reported as δ13C according to the following equation:

display math(1)

where the standard is the Vienna Pee Dee Belemnite (VPDB).

[15] The radiocarbon measurements of the freeze-dried WSOC isolates were performed at the US National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) facility (Woods Hole, MA, USA) as described previously [McNichol et al., 1992; Pearson et al., 1998; Zencak et al., 2007a; Kirillova et al., 2010].

[16] Radiocarbon data are reported as fraction modern (Fm) calculated from the following equation:

display math(2)

where B, S, and M represent the 14C/12C ratios of the blank, the sample, and the modern reference, respectively. The 14C/12C ratio of the modern reference is defined as 95% of the radiocarbon content of NBS Oxalic Acid I in AD 1950 normalized to δ13C of −25‰.

[17] Due to the very low load of WSOC on field blank filters (0.20 ± 0.032 µg cm−2 for MCOH and 0.14 ± 0.014 µg cm−2 for SINH) there was not sufficient carbon mass for isotope measurements in the WSOC blank samples. To still account for and assess the blank in the isotope data, corrections were performed using the isotopic composition of TOC, in lieu of that for WSOC, reported on earlier for the field blank filters from MCOH and SINH [Sheesley et al., 2012].

[18] The definition of Δ14C by Stuiver and Polach [1977] as the relative difference between the 14C/12C ratios of the absolute international standard (base year 1950) and sample corrected for the age of the sample is described in the following equation:

display math(3)

where λ is 1/(true mean-life) of radiocarbon (corresponding to 1/8267) and x is the year of collection. From these Δ14C values, fractional contributions of contemporary biomass/biogenic sources vs radiocarbon-extinct fossil fuel sources can be determined using the isotopic mass balance equation:

display math(4)

where Δ14Csample is the measured radiocarbon content of a WSOC sample and Δ14Cfossil is −1000‰. The Δ14Cbiomass endmember is between +50‰ and +225‰. The first value corresponds to the Δ14C of contemporary CO2 [Levin et al., 2010; Graven et al., 2012], and thus freshly produced biomass. The second value is for the Δ14C of wood logged in the 1990s to 2000s [Zencak et al., 2007b; Klinedinst and Currie, 1999]. For Indian biomass burning the end-member value of +199‰ have been estimated based on relative contribution from contemporary (meaning one-year plants) biofuel and wood fuel [Gustafsson et al., 2009]. In the absence of such detailed information and since the relative contributions to WSOC from biogenic SOA and from biomass burning primary organic aerosols (POA) and SOA are not known a priori, they were here assumed to be of equal importance. Hence, the biogenic/biomass Δ14C endmember for WSOC was set to +124‰, which is the mean value of +50‰ and +199‰.

[19] To incorporate this expected biomass endmember variability into the calculation of the fraction biomass, a Monte Carlo approach was employed [Andersson, 2011]. In those calculations the biomass Δ14C endmember was represented by a normal distribution with mean 124‰. The standard deviation for the normal distribution was set to be one fourth of the expected biomass range = (199 − 50)/4 = 37‰. To test the influence of endmember variability on the computed uncertainty a few different values of normal standard deviations (37, 74 and 148‰) were examined. Little effects on the computed mean values were observed. The uncertainties of the concentration measurements and field blanks were also incorporated in the Monte Carlo calculations by representing the data points with normal distributions, with means and standard deviations defined by the measurement data (Table S3). The results of the Monte Carlo approach were also compared with a sensitivity analysis of the influence of the endmember variability of the computed fraction biomass by using the extreme values (50‰ resp. 199‰) of the biomass endmember (Table S4).

2.4 Calculation of Annual and Seasonal WSOC Concentrations and Source Contributions

[20] To account for the difference in sampling duration between the samples, the calculation of annual and seasonal averages of WSOC concentrations was normalized by sample volume:

display math(5)

where vi is the sampled air volume for sample i, Ci is the concentration of WSOC for sample i, and Cj is the volume-normalized average concentration.

[21] The estimations of average fraction modern (Fmj) for WSOC were similarly calculated by normalizing to both carbon-mass concentration and to the sampling duration for each sample in a composite:

display math(6)

where Fmi is the Fmodern for composite i, Ci denotes the normalized concentration of WSOC, t is the sampling duration and Cj is the normalized average concentration.

[22] The studied receptor sites were each influenced by seasonally-varying air masses with different geographical origins. Backward trajectories (BT) cluster analysis for the air masses arriving at MCOH and SINH during the 2008/2009 sampling campaign revealed the main source regions, as detailed in the preceding accompanying paper of this campaign [Sheesley et al., 2012]. Briefly, 5-day back trajectories were calculated using NOAA HYSPLIT software v4.9 [Draxler and Rolph, 2010]. Trajectories were determined every 6 h for the entire 15-month campaign. Cluster analysis was completed for each individual month at each site. The main source regions for MCOH were Bay of Bengal, South India, West Indian margin, South Indian Ocean and local sources. Generally, outflow from the Bay of Bengal and South Indian source regions dominated during the dry season from December to January, while the West Indian margin is the primary source region for November and February. During the transition seasons (March–April and October) air masses dominantly arrive from the Central and W. Arabian Sea. During monsoon season, the source region is uniformly the South Indian Ocean.

[23] For SINH, the main source regions included North India, East India, North West India and Pakistan, and South Indian Ocean. Early dry season source regions are influenced by North India (October–December); however, this transitions to air mass origins in East India (January) and then NW India and Pakistan later in the dry season (February–March). As with MCOH, transition source regions are dominated by the Central and W. Arabian Sea (March–May and September), while air masses solely originate in the South Indian Ocean during monsoon season (June–August).

[24] BTs were combined with ambient concentrations to determine the contribution from each source area. The mean concentration was computed as the mean of WSOC weighted by the fraction air originating from a certain source region and the sampling duration:

display math(7)

where wji is the fraction of wind originating from source cluster j for sample number i, Ci denotes the concentration of WSOC for sample i, ti is the sampling duration for sample i, and CSRj is the mean concentration for cluster j.

[25] During the summer monsoon season, much of the air at the studied sites came from the Indian Ocean (over the preceding 5 days) and the concentrations of WSOC were very low. Since the primary scope of the study was anthropogenic aerosols from S Asian, the data from the monsoon season were excluded from the analysis. The cut-off for the fractional influence of certain wind directions was applied to avoid small or local variations. For MCOH, the air from Bay of Bengal, South India, West India and Arabic Sea BT clusters was analyzed. For SINH the analysis included BT clusters for East India, North India, Pakistan and the Arabian Sea. To test whether the mean concentration calculated for each source cluster was significantly larger/smaller than the total mean value for either TOC or SC, unpaired one-sided t-tests were performed.

2.5 Calculation of Water-Insoluble Organic Carbon (WINSOC) Concentration and Source Contribution

[26] In order to compare WSOC and water-insoluble organic component (WINSOC) of organic aerosols, the concentrations and isotopic signatures of WINSOC were derived from the WSOC and TOC data. The concentration of WINSOC equals the difference between the TOC and WSOC concentrations. The calculations of the stable carbon isotope composition (δ13C) and fraction modern (Fm) for WINSOC were based on the isotope mass balance equation:

display math(8)

where XTOC, XWSOC and XWINSOC are the stable carbon (δ13C) or fraction modern (Fm) values for TOC, WSOC and WINSOC, CjTOC, CjWSOC and CjWINSOC are the volume-normalized average concentrations of TOC, WSOC and WINSOC for the composite j.

[27] Similarly to above, the fraction biomass (fbiomass) of WINSOC was calculated from Equations (3) and (4).

2.6 Error Analysis and Uncertainty Estimates

[28] The overall precision by which the TOC, WSOC and WINSOC concentrations and isotopic signatures may be estimated depends on uncertainties from several different sources, including: the precision of concentration estimation (estimates from triplicate analysis), mass contributions from field blanks (estimates from several blanks), as well as the effects of positive artifacts during sampling, precision of isotope characterization (instrument precision) and the isotope signature of the field blanks. To obtain the overall precision, these factors need to be combined using an error propagation scheme. Here, this was implemented using a Monte Carlo strategy. The detailed description of the procedure is provided in supplementary information (Text S1).

[29] The average winter-time variability for the estimated fraction fossil/biomass is 3–8% for TOC, WSOC and WINSOC for both SINH and MCOH. The biomass endmember variability discussed in Section 'Isotope Analyses and Calculations' is also included in this estimate. For δ13C, the corresponding winter-time uncertainty range is 0.1–2.9‰. Taken together, this analysis shows that the inter-site (SINH vs MCOH) and inter-species (e.g., WINSOC vs WSOC) variability observed in the present contribution are significantly larger than the estimated errors.

3 Results and Discussion

3.1 Annual Variation of Ambient WSOC Concentration

[30] The monsoon system determines the meteorological conditions in the South Asian region. The organic aerosol concentrations thus show significant seasonal variations [Lawrence and Lelieveld, 2010; Ram et al., 2010b; Pavuluri et al., 2011; Sheesley et al., 2012] with elevated concentrations during the dry winter season (Dec-Feb) and lowest concentration during the summer monsoon season (Jun-Sep). WSOC concentrations exhibited a similar seasonal trend to previously reported concentrations of TOC [Sheesley et al., 2012] (Figure 3, Table 1, Table S1).

Figure 3.

Water-soluble organic carbon (WSOC) concentrations at MCOH - Hanimaadhoo, Maldives (A) and SINH - Sinhagad, India (B) during the study period (January 2008–April 2009) and the ratio of WSOC to previously reported [Sheesley et al., 2012] total organic carbon (TOC) for both sites (C). Vertical bars represent uncertainty, horizontal bars represent sampling duration. The vertically shaded bars indicate approximate transition periods between the different seasons /meteorological regimes (continental SINH is leading the marine MCOH site by up to a month in the transition to the monsoon and dry winter seasons). X-scale tick marks denote the start of a month.

Table 1. Mean Concentrations of WSOC by Geographical Source Regions During the Dry Season (CSRj-WSOC) and Their Ratio to Mean Concentrations of TOC (CSRj-TOC). The p Values Are Also Given for Difference From Dry Season Means (CSRj-WSOC) by Site.
ClusterCj-WSOC (µg m−3)CSRj-WSOC (µg m−3)pSRj-WSOCCj-WSOC/Cj-TOC (%)CSRj-WSOC/CSRj-TOC (%)
MCOH
Dry season mean0.52  25 
Bay of Bengal 1.020.003 26
South India 0.610.25 29
W India margin 0.580.32 25
Central and W Arabian Sea 0.190.003 16
SINH
Dry season mean3.04  35 
North India 3.600.37 40
East India 3.770.15 35
NW India & Pakistan 3.830.15 34
Central and W Arabian Sea 1.630.01 30

3.1.1 WSOC Concentrations at the Maldives Climate Observatory Hanimaadhoo (MCOH)

[31] Substantial variability was observed in WSOC concentrations by season at this South Asian receptor site (Figure 3A, Table S1). The concentration of WSOC at MCOH increased rapidly during the post-monsoon season (Oct-Nov) and was persistently high through the winter until March (Figure 3A). Averaged WSOC concentrations during pre-monsoon season (March–May) in 2008 and 2009 was 0.31 ± 0.15 µg m−3 (Table S1). Very low concentrations of WSOC were observed during the monsoon season (0.10 ± 0.04 µg m−3). Although the average seasonal concentrations were the highest during the dry winter season (0.94 ± 0.43 µg m−3), the WSOC contribution to TOC was similar during the post-monsoon season (29 ± 11%) and the winter season (29 ± 8%). Subsequently, WSOC/TOC ratio decreased during the pre-monsoon season (20 ± 6%) and was at a minimum during the summer monsoon season (15 ± 5%).

3.1.2 WSOC Concentrations at the W. Indian Receptor Site of Sinhagad (SINH)

[32] At the west India continental site the annual average concentration of WSOC was about six times higher than at the Indian Ocean site. This suggests that land-based sources dominate WSOC over S Asia. However, the same seasonal trends were observed with increasing concentrations during the post-monsoon season (2.0±1.4 µg m−3), highest concentrations through the winter season (3.6±2.3 µg m−3) and low levels during the pre-monsoon/transition season (3.1 ± 2.5 µg m−3) (Figure 3B, Table S1). During the summer monsoon season the concentrations of WSOC were low (0.4±0.2 µg m−3), with three samples below the detection limit. The seasonal average WSOC/TOC ratio at SINH was similar for the post-monsoon season (37±13%), the dry winter season (36±3%) and the pre-monsoon season (38±13%).

[33] Previous studies of WSOC abundance in atmospheric aerosols in Indian urban and suburban areas reported very different results for different source regions. The methods used in these studies were all based on extraction of filter samples in ultra pure water by shaking [Khare et al., 2011], soaking [Rengarajan et al., 2007] or ultrasonication from 20 to 30 minutes [Miyazaki et al., 2009, Pavuluri et al., 2011] to 6–8 hours [Ram et al., 2010b, 2012a, 2012b]; all studies measured WSOC concentration of the filtered extracts using TOC analyzers.

[34] WSOC concentrations for TSP in New Delhi [Miyazaki et al., 2009] and PM10 at Jorhat, a suburban site in northeast India [Khare et al., 2011], were very high (more than 20 µg m−3) during winter season. In urban areas of the Indo-Gangetic Plain (IGP), WSOC concentrations in TSP and PM10 were variable. Dry season WSOC concentrations for TSP at two urban sites in the IGP influenced by agricultural crop-residue burning ranged from 9 to 12 µg m−3 and 12–21 µg m−3 [Rengarajan et al., 2007; Ram et al., 2012a]. The studies of PM10 in Kanpur, another urban site in the IGP, showed distinct seasonal variability with higher concentrations of WSOC during post-monsoon 15.6±4.6 µg m−3 (Oct-Nov 2007) [Ram et al., 2010b], 41.5±18.7 µg m−3 (Oct-Nov 2008) [Ram et al., 2012b], and winter seasons 15.3±7.6 µg m−3 (Dec-Feb 2008) [Ram et al., 2010b] and lower concentrations during pre-monsoon and monsoon seasons 6.6±2.0 µg m−3 (Apr-Jun 2007) [Ram et al., 2010b]. WSOC PM10 in mega-city Chennai in East India was reported at 3.8±1.3 µg m−3 [Pavuluri et al., 2011]. Interestingly, the study in Chennai [Pavuluri et al., 2011] did not show significant variation in WSOC concentrations between winter (3.8±1.3 µg m−3) and summer (4.1±0.9 µg m−3). Similarly, only small differences in concentrations between summer (26 µg m−3 for PM10) and winter (29 µg m−3 for PM10) was also reported for Jorhat [Khare et al., 2011]. Taken together, WSOC contribution to TOC or OC in TSP and PM10 in different studies varied 20–77% [Khare et al., 2011; Rengarajan et al., 2007; Miyazaki et al., 2009; Ram et al., 2012a,2012b]. Ram et al. [2010b] reported higher WSOC/OC ratio in summer (50–56%) compared to winter season (32–41%) for the PM10 size fraction in Kanpur, while for the same fraction in Chennai city Fu et al., 2010 did not register a difference with WSOC/OC ratio about 41–42% both during summer and winter seasons. In the fine fraction (PM2.5) the WSOC/OC ratio during the winter season was very similar to the coarse fraction values: in average 48% in Kanpur [Ram et al., 2012b] and 41% in Ahmedabad [Rengarajan et al., 2011]. These reported differences in WSOC are likely due to differences in location, size fraction and season. However, differences in extraction methods cannot be excluded as influence on the reported variability.

[35] The concentrations of WSOC and the WSOC/TOC ratio at the continental SINH site were higher than those for MCOH for all seasons (Figure 3; Table S1). Aerosols from the same MCOH and SINH sites were investigated by transmission electronic microscopy by Coz and Leck, 2010. At the MCOH site, they primarily found soot aggregates with organic coating in aerosols coming from the Indian subcontinent. The aggregates appear to have hydrophobic behavior, potentially due to free radical condensation (e.g., non-homogenous polymerization) of the organic material in the process of chemical aging [e.g., Kalberer et al., 2004; Zahardis et al., 2006]. Results from another study at MCOH suggest that this kind of soot-containing hydrophobic particles have smaller washout ratio during long-range transport [Granat et al., 2010]. Studies of the concentrations of soot and sulfates in both aerosols and rainwater at MCOH by Granat et al. [2010] showed a longer lifetime of soot during the winter season, which supports the view of preferential scavenging of water-soluble aerosol component compared to that of soot (black carbon).

[36] This possibility of preferential washing out of WSOC during the longer time of transport to the Indian Ocean site is supported by the data for the dry winter season when the precipitation is minimal and the WSOC/TOC ratio for SINH and MCOH indeed does differ less (36±3% and 29±8%, respectively) than during the other seasons (Figure 3C). There is also a possibility of WSOC loss during transport due to atmospheric oxidation that leads to partitioning of the formed short-chain species to the gas phase.

[37] WSOC concentrations may also be compared to the synoptically obtained concentrations of soot carbon (SC) [Sheesley et al., 2012], which represents the most condensed part of the black carbon (BC) combustion continuum [Elmquist et al., 2006]. Consistent with the above, WSOC/SC ratio (Table S1) at MCOH is lower than at SINH during all seasons, which suggests more efficient loss of WSOC during long-range transport compared to soot particles, which have longer atmospheric lifetimes [Granat et al., 2010].

3.2 The Mean Concentrations of WSOC in Air of Different Geographical Regions

[38] The influence of different source regions on the WSOC concentration at the receptor sites was investigated by calculating the mean WSOC concentration for each back trajectory cluster.

[39] This approach has been described in detail in Sheesley et al. [2012], and is summarized here. The main source regions for MCOH were Bay of Bengal, South India, West Indian margin, South Indian Ocean and local sources. For SINH the main source regions included North India, East India, North West India and Pakistan, and South Indian Ocean.

[40] Again, only the dry season was studied, to exclude the expected high influence of wet precipitation from the monsoon period. The dry season mean WSOC concentration and mean values per air mass cluster are presented in Table 1.

[41] At MCOH the highest WSOC concentrations were observed for the air masses coming from the Bay of Bengal (1.02 µg m−3); this was significantly higher than the mean dry season concentration (0.52 µg m−3) (Figure 3). These air masses were ultimately influenced by outflow of pollution from the densely populated Indo-Gangetic Plain and also carried the highest concentrations of TOC [Sheesley et al., 2012]. The proportion of WSOC in TOC for all air masses originating in India was very similar (25–29%). In contrast, the air coming from the Arabian Sea contains the lowest WSOC concentrations (0.19 µg m−3), significantly lower than the mean dry season concentration, and the lowest WSOC/TOC ratio (16%).

[42] At SINH the average concentrations of WSOC from all Indian air masses were very similar during the winter wind conditions (3.77 µg m−3 for East India, 3.60 µg m−3 for North India and 3.83 µg m−3 for Northwest India and Pakistan). The corresponding WSOC/TOC ratios were also similar between air masses from North India (40%), East India (35%) and Northwest India plus Pakistan (34%). Similar to the Indian Ocean receptor site at MCOH, SINH showed the lowest WSOC concentrations (1.63 µg m−3) and WSOC/TOC ratios (30%) with air masses of Arabian Sea origin. However, the contribution of WSOC to TOC at the SINH continental site is still much higher than at MCOH (Table 1); for example, the Arabian Sea BTs pass over Mumbai before reaching SINH. Considering that WSOC concentration was six times lower at MCOH compared to the W India regional background site and that the WSOC concentrations and WSOC/TOC ratio were lowest at both sites when the air masses had the largest marine influence (Arabian Sea; Table 3), any marine source contribution to the WSOC over S Asia is likely to play only a minor role.

3.3 Radiocarbon-Based Source Apportionment

[43] Here we report the first year-round radiocarbon-based source apportionment of WSOC. The temporal dynamics are shown in Figure 4 with the radiocarbon-based calculations of annual and seasonal averages of fraction modern and fraction fossil of WSOC presented in Table 2.

Figure 4.

Natural abundance radiocarbon-based source apportionment (Fraction fossil) for water-soluble organic carbon (WSOC) and total organic carbon (TOC) at Hanimaadhoo (A) and Sinhagad (B). Comparison of fraction fossil for WSOC (C) and water-insoluble organic carbon (WINSOC) (D) between the two sites. Vertical bars represent analytical uncertainty (1 s.d.) for TOC and overall uncertainty for WSOC and WINSOC, horizontal bars represent sampling duration. X-scale tick marks denote the start of a month. *One value for SINH was excluded due to large uncertainties.

Table 2. Radiocarbon-based Estimates of Annual and Seasonal Fossil Contributions to WSOC Weighed by Sampling Duration and Concentration for Both Hanimaadhoo and Sinhagad Sites
 Fms.d.ffossil (%)error
  1. Fm is Fraction modern.

  2. ffossil is fraction fossil in percent (Equation (4)).

  3. s.d. is standard deviation of measurements.

MCOH
2008 annual0.940.053174
Dec–Feb0.910.031193
Mar–May0.990.029134
Jun–Aug1.010.0131112
Sept–Nov0.930.039174
SINH
2008 annual0.870.032244
Nov–Feb0.870.031243
Mar–Apr0.850.004253
May–Aug0.900.0072111
Sept–Oct0.860.055244

3.3.1 14C-WSOC at the Maldives Climate Observatory Hanimaadhoo (MCOH)

[44] The relative contribution from fossil versus contemporary sources to the WSOC observed at the Indian Ocean site showed only small seasonal changes, despite the large variation in the actual concentration. The WSOC at MCOH was dominated by biomass/biogenic contribution (Figure 4A). However, the 14C-WSOC data revealed that there was also a fossil-fuel derived component of the WSOC constituting 17±4% for 2008 (Table 2). During the pre-monsoon and monsoon seasons the fossil contribution was less than average and makes up 13±4% and 11±12% of WSOC, respectively. Precipitation likely washed out most of the particulate matter and the aerosol samples were influenced primarily by marine and local sources. The highest fossil contribution to WSOC was observed during the dry winter season (19±3%) (Figures 4A, 4C). The 5-day back trajectories during these months pointed to the N. Bay of Bengal and were likely influenced ultimately by pollution outflow from the densely populated and industrialized Ganges Valley.

3.3.2 14C-WSOC at the W. Indian Receptor Site of Sinhagad (SINH)

[45] A similar seasonal trend was found in 14C-based WSOC source contributions at the continental SINH site. The average fossil contribution to WSOC at SINH for 2008 was 23±4%. The lowest seasonal fossil contribution was observed during the monsoon season (20±11%). As for MCOH, during this season, the air was coming from the Indian Ocean and the concentrations of WSOC were very low (Figure 3). The contribution of fossil-fuel derived sources to WSOC at SINH was the highest during September 2008 (35%) and February 2009 (29%) (Figure 4B, C). The 5-day back trajectories in September 2008 started over the Arabian Sea but were then passing over the greater Mumbai region prior to arrival at SINH. In February 2009 the air came mostly from the west coast of India and from Pakistan.

[46] Comparison of the 14C-based source apportionment between the two sites (Figure 4C) showed that the fossil contribution at the continental SINH site was twice as large as that at the ocean receptor MCOH site during the pre-monsoon and monsoon seasons. In September 2008, the difference in sources was the largest with three times higher fossil contribution at the Indian site (35% fossil) than at the Indian Ocean site (12% fossil). In contrast, during the dry winter season the fossil contribution at both sites was very similar (19–23%). The air masses during this high-aerosol loading season were highly influenced by the sources in western India and northern India / Ganges Valley.

[47] Only a limited number of radiocarbon measurements of WSOC have been previously reported with which these South Asian year-round 14C-WSOC results can be compared. A 14C-based study (n=4) in the Atlanta metropolitan area (eastern USA) showed that 19–33% of WSOC in the PM2.5 fraction was fossil during summer [Weber et al., 2007]. WSOC in TSP from the rural areas of Millbrook, NY and Harcum, VA carried 0% to 25% fossil contributions [Wozniak et al., 2008, 2012a, Wozniak et al., 2012b]. In PM10 samples (n = 4) from Zürich, Switzerland, the fossil fraction of WSOC was indirectly estimated based on radiocarbon measurements of OC and WINSOC; these were reported at 4–24% [Szidat et al., 2004]. Kirillova et al. [2010] reported summertime 14C-WSOC results for suburban Stockholm, Sweden of 0–12% fossil consistent with a high production of SOA from BVOC, which has previously been reported for this region [e.g., Tunved et al., 2006].

[48] There are no previous reports of radiocarbon measurements of WSOC for the South Asian region. The present study suggests that biomass burning practices and/or biogenic SOA formation are major contributors to the South Asian WSOC aerosol. Nevertheless, the fossil component of WSOC at both study sites is also considerable (11–22% at MCOH and 21–35% at SINH, Figure 4A, B, C, Table S2).

3.4 Application of Stable Carbon Isotopes to Study Aerosol Sources and Atmospheric Processing

[49] The year-round dynamics in δ13C of TOC and WSOC are shown in Figure 5 and the annual and seasonal averages are presented in Table 3. Given that WSOC has been used as a proxy for SOA, it is valuable to compare the δ13C-WSOC signal with the δ13C-WINSOC component.

Figure 5.

Measured stable carbon isotope composition (δ13C) for water-soluble organic carbon (WSOC) and total organic carbon (TOC), and calculated δ13C for water-insoluble organic carbon (WINSOC) at Hanimaadhoo (A) and Sinhagad (B) sites. The difference in δ13C between WSOC and WINSOC aerosol components at Hanimaadhoo (C) and Sinhagad (D). Vertical bars represent analytical uncertainty (1 s.d.) for TOC and WSOC and overall calculation uncertainty for WINSOC, horizontal bars represent sampling duration. X-scale tick marks denote the start of a month. *One value for SINH was excluded due to large uncertainties.

Table 3. Annual and Seasonal Stable Carbon Averages for WSOC, TOC and WINSOC at Hanimaadhoo and Sinhagad
 δ13CjWSOCs.d.δ13CjTOCs.d.δ13CjWINSOCs.d.
  1. s.d. is standard deviation

MCOH
2008 annual−18.50.5−22.81.2−24.21.7
Dec–Feb−18.40.4−22.61.1−24.11.7
Mar–May−18.40.2−23.51.2−24.21.6
Jun–Aug−19.62.5−23.21.3−23.31.6
Sept–Nov−18.20.4−22.51.2−24.51.9
SINH
2008 annual−20.40.5−22.62.4−23.89.8
Nov–Feb−20.10.3−22.21.3−23.52.2
Mar–Apr−21.50.4−23.21.6−24.42.9
May–Aug−21.33.1−23.111.7−23.568.7
Sept–Oct−20.20.6−22.72.2−24.34.0

[50] The annual average δ13C-WSOC for 2008 at MCOH was −18.4±0.5‰ with a range of 3.3‰ (−20.8‰ to −17.5‰) (Figure 5). The summer monsoon season had the lightest WSOC component (−19.6±2.5‰), with the air coming mostly from the south (Table 3). The following post-monsoon/transition season featured the heaviest δ13C-WSOC values (−18.2±0.4‰). The heaviest δ13C-WSOC values are attributed to the air masses from the West Indian Margin / the Arabian Sea.

[51] The annual average value of δ13C-WSOC for 2008 at Sinhagad was −20.4±0.5‰ with a range of 3.9‰ (−23.7‰ to −19.8‰). As for MCOH, the lightest WSOC component at SINH was observed during the summer monsoon season (−21.3±3.1‰) during the influence of the Indian Ocean air (Table 3). Similarly, the heaviest δ13C-WSOC were during the winter season (−20.1±0.3‰) when the aerosols were transported from different regions of the Indian subcontinent. The δ13C-WSOC at MCOH was on average enriched by 2.1‰ relative to SINH (Figure 5).

[52] The WSOC component was enriched in 13C compared to TOC for both sites (Figure 5). The average enrichment was 4.4‰ at MCOH and 2.0‰ at SINH. This enrichment in δ13C-WSOC vs. δ13C-TOC is consistent with data from a previous study [Fisseha et al., 2009]. It has been interpreted to reflect that oxidized compounds in the atmosphere, that have low vapor pressure, tend to partition to the aerosol phases, which leads to a fractionation effect enriching aerosols with heavy 13C isotopes. Condensation or evaporation-related isotopic fractionation of organic aerosol components has been reported for small molecules [Wang and Kawamura, 2006; Harrington et al., 1999]. Another plausible explanation is that during wood burning WSOC is formed from only certain parts of plants enriched in stable carbon, such as cellulose [Fisseha et al., 2009]. During atmospheric oxidation, compounds may also lose some carbon atoms (preferably the lighter isotope) and thus become enriched in δ13C [Aggarwal and Kawamura, 2008]. These three putative processes are all consistent with the present South Asian aerosol-isotope data set as there are (a) a large contribution from biomass burning in the region, (b) long-range transport and (c) high temperature and sunlight intensity causing considerable aerosol processing. The smallest difference in δ13C values between WSOC and TOC was observed during the monsoon season (3.6‰ for MCOH and 1.2‰ for SINH) (Table 3, Figure 5A, B) when the residence time in the atmosphere is shorter, the influence of local sources higher and the aging processes less prevalent. Seasonal and spatial variability in sources may also contribute to the observed difference between the sites. While some contribution from a marine source cannot be excluded for MCOH the data consistently point to atmospheric processing as a more prevalent contributor to the WSOC.

[53] In contrast to δ13C-WSOC, the calculated δ13C-WINSOC was more depleted than δ13C-TOC (Figure 5A, B). The difference in δ13C between WSOC and WINSOC varied in the range 3.7‰ to 7.3‰ for MCOH and 2.2‰ to 5.8‰ for SINH and for both sites showed similar annual pattern (Figure 5A, B). Assuming that the WINSOC component better represents primary sources and is less transformed during transport, the larger difference at the MCOH site may thus reflect more aged particles at MCOH compared to aerosols at the continental SINH site.

[54] Due to the relatively long sampling times — 1 week (2 weeks during summer period) — a possible artifact regarding the δ13C values is the potential aging of the aerosols during the sampling. Oxidants present in the air passing through a high-volume sampler may affect collected aerosols. The mean residence time of the aerosols on the filter is 3.5 days (for 1-week samples), which is approximately the same as the time of aerosols transport from a source to SINH and less than transport time to MCOH. Seven (out of 50) samples were 2-week collections (all in period July to November 2008) at MCOH and these were composited to form sufficient quantity for 14C analysis for the monsoon and post-monsoon period. However, since the filters are not exposed to sunlight we expect lower amounts of photooxidation in the filter matrix than during atmospheric transport. In addition, the large difference in the δ13C of WSOC from SINH and MCOH, despite equal sampling times and setup, shows that atmospheric processing is likely to strongly affect the stable isotope ratios.

3.5 Dual Carbon (13C and 14C) Isotope Study of Water-Soluble (WSOC) and Water-Insoluble (WINSOC) Aerosol Components

[55] The integrated effects of source mixing and atmospheric processing of WSOC and WINSOC may be examined by combining the Δ14C and the δ13C data. In Figure 6, Δ14C is plotted against δ13C, separated into WSOC and WINSOC for SINH and MCOH, respectively. Three clusters of the data are observed: 1. WSOC from MCOH, 2. WSOC from SINH and 3. WINSOC from both SINH and MCOH. The fact that the WINSOC data from both SINH and MCOH are overlapping, indicates that the WINSOC arriving at these sites comes from similar sources/have experienced similar processing. Since the aerosols at MCOH are thought to have been transported over substantially longer distances (also indicated by lower concentrations), the absence of any clear shift in δ13C-WINSOC suggest limited atmospheric processing/aging of the WINSOC component.

Figure 6.

Dual isotope (δ13C and Δ14C) analysis of water-soluble organic carbon (WSOC) and water-insoluble organic carbon (WINSOC) at Hanimaadhoo (MCOH) and Sinhagad (SINH) sites including all data (A) and excluding the samples where the 5-day back trajectories clearly point to a marine source (Arabian Sea or to the south of MCOH) (B).

[56] In contrast, there is a clear difference in the δ13C-WSOC between SINH and MCOH. To explore this further, the air mass origins were included in the analysis. To focus on air coming from the Indian subcontinent, data representing 5-day back-trajectories with a marine sector origin (e.g., Indian Ocean / Arabian Sea) were excluded (Figure 6B) as were composites from 2-week sample durations. In this marine-censored data, the three clusters are even more clearly defined. Focusing first on the δ13C dimension, the enrichment in 13C of WSOC compared with WINSOC either indicate effects of atmospheric processing, differences in emission sources, or a combination of both effects. Since SOA formation leads to depletion in δ13C, a substantial influence of secondary formation of WSOC appears less likely (Figure 2). The enrichment of 13C in WSOC at both SINH and MCOH may indicate a contribution from a local source, including marine WSOC emissions, yet, more likely, atmospheric aging of the aerosols. Aerosols arriving at MCOH are expected to be more aged than at SINH, as the distance from emissions sources is expected to be longer. This is consistent with WSOC at MCOH being more enriched in 13C than at SINH. However, the lower δ13C values observed at MCOH is also consistent with expectations from marine emissions (δ13Cmarine ~ −21‰), even though the samples associated with mainly marine air parcels were sorted out in this analysis. However, marine emissions of organic aerosols are generally low, even in oceans with a higher net productivity than the Indian Ocean [Lapina et al., 2011; Ceburnis et al., 2011], and the outflow of TOC from India is expected to be larger than the oceanic background during winter time. Further, the much lower WSOC concentration at MCOH than SINH and in air masses coming from marine sectors (Table 3) suggest a lesser marine contribution to WSOC.

4 Conclusions

[57] This study is the first dual-carbon isotope-based analysis of sources and processing of aerosol water-soluble organic carbon (WSOC) in the South Asian region and it covers a 15-month continuous period in 2008–2009. The distinct seasonality of the monsoon system was reflected in the seasonally-varying WSOC concentrations at the study sites. The highest concentrations of WSOC were associated with the air masses coming from the Bay of Bengal and influenced by the pollution from the industrialized and heavily populated IGP, and the lowest concentrations were attributed to the marine sources of the Arabian Sea and the Indian Ocean. Higher contribution of WSOC to total organic carbon (TOC) at the western Indian site (43%) compared to the Indian Ocean site (26%) may indicate more efficient loss of WSOC compared to more hydrophobic aerosol components, formation of hydrophobic coatings due to free radical condensation of OC during the long-range transport, and possibly addition of some marine organic aerosols with low WSOC content.

[58] The radiocarbon-based source apportionment of WSOC shows the dominance of the biogenic/biomass burning sources similarly as for the TOC and SC components. Nevertheless, there is also a significant contribution of fossil fuel sources (17±4% MCOH and 23±4% SINH) to the WSOC component. The seasonal variation in the fossil contribution at both sites depends on the air mass origins with higher fossil input in the samples impacted by the air masses from industrialized Indian areas and lower fossil signal in the samples influences by Indian Ocean and the Arabian Sea marine sources.

[59] The stable (δ13C) isotope revealed a trend with enrichment in heavy isotopes in particulate WSOC as compared to WINSOC, for SINH, and even more so for MCOH. While the data does not preclude a contribution from a marine source, this trend likely reflects aerosol aging processes during transport from emissions sites to the receptor site, which is longer for MCOH.

Acknowledgments

[60] The authors would like to thank the UNEP-RRC (Bangkok) and the Environmental Protection Agency of the Republic of the Maldives for the operation of MCOH station, and the Indian Institute of Tropical Meteorology in Pune, India for the operation of SINH site. The authors acknowledge financial support from the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS Contract no. 214-2009-970). ÖG also acknowledges financial support as an Academy Researcher from the Swedish Royal Academy of Sciences through a grant from the Knut and Alice Wallenberg Foundation. AA acknowledges financial support from the Knut and Alice Wallenberg Foundation. This study also benefitted from the research environments provided by the Bert Bolin Centre for Climate Research and the Delta Facility (a core facility for compound-specific isotope analysis), both at the Stockholm University, School of Natural Sciences.

Ancillary