Stable carbon and nitrogen isotopic composition of fine mode aerosols (PM_2.5) over the Bay of Bengal: impact of continental sources

Abstract

This study reports on stable carbon (δ¹³C_TC) and nitrogen (δ¹⁵N_TN) isotopic composition of total carbon and nitrogen (TC and TN) in the fine mode aerosols (PM_2.5; N=31) collected over the Bay of Bengal (BoB). The samples represent two distinct wind regimes during the cruise (27 December 2008–28 January 2009); one from the Indo-Gangetic Plain (referred as IGP-outflow) and another from Southeast Asia (SEA-outflow). The PM_2.5 samples from the IGP-outflow show higher δ¹³C_TC (−25.0 to −22.8 ‰; −23.8±0.6 ‰) than those from the SEA-outflow (−27.4 to −24.7 ‰; −25.3±0.9 ‰). Similarly, δ¹⁵N_TN varied from +11.8 to +30.6 ‰ (+20.4±5.4 ‰) and +10.4 to +31.7 ‰ (+19.4±6.1 ‰) for IGP- and SEA-outflows, respectively. Based on the literature data, MODIS-derived fire hotspots and back trajectories, we infer that higher δ¹³C_TC in the IGP-outflow is predominantly associated with fossil fuel and biofuel combustion. In contrast, contribution of primary organic aerosols from the combustion of C₃ plants or secondary organic aerosol (SOA) formation from biomass/biofuel-burning emissions (BBEs) can explain the lower δ¹³C_TC values in the SEA-outflow. This inference is based on the significant linear correlations among δ¹³C_TC, water-soluble organic carbon and non-sea-salt potassium (nss-K⁺, a proxy for BBEs) in the SEA-outflow. A significant linear relationship of δ¹⁵N with and equivalent mass ratio of / is evident in both the continental outflows. Since abundance dominates the TN over the BoB (>90 %), atmospheric processes affecting its concentration in fine mode aerosols can explain the observed large variability of δ¹⁵N_TN.

• stable C- and N-isotopes
• Bay of Bengal
• aerosols
• biomass burning
• fossil-fuel combustion
• South Asia
• South-east Asia
• Indo-Gangetic Plain

To access the supplementary material to this article, please see Supplementary files under ‘Article Tools’.

1. Introduction

The rapidly growing human activities and significant increase in emissions of airborne pollutants from South Asia are of considerable interest owing to their impact on regional climate (Ramanathan et al., 2001a; Lawrence and Lelieveld, 2010). In this context, organic aerosols are considered to be one of the important components of atmospheric particulate matter for their crucial role in regional radiative forcing through direct and indirect effects (Kanakidou et al., 2005; Jimenez et al., 2009). Several studies have attempted to characterize the sources, transport and transformation pathways of organic aerosols over South Asia (Jayaraman et al., 1998; Novakov et al., 2000; Lelieveld et al., 2001; Ramanathan et al., 2001b; Mayol-Bracero et al., 2002a; Rengarajan et al., 2007; Lawrence and Lelieveld, 2010; Ram et al., 2010; Rajput et al., 2011; Srinivas and Sarin, 2013a, 2014). Jimenez et al. (2009) demonstrated that the identified fraction of atmospheric organic aerosols is no more than 20–30 % of total organic matter. These observations emphasize our ability to assess the impact of organic aerosols on climate forcing, mainly due to limited observations. Therefore, the chemical characterization of various organics and their sources from different geographical locations could provide a better means to reassess their realistic aerosol radiative forcing (Jacobson, 2012).

Recent studies on stable isotopic composition of bulk carbon and nitrogen (δ¹³C and δ¹⁵N) of airborne particulatematter have demonstrated their usefulness to apportion the sources and formation pathways of organic aerosols (Kawamura et al., 2004; Aggarwal and Kawamura, 2008; Narukawa et al., 2008; Pavuluri et al., 2011; Aggarwal et al., 2013; Kirillova et al., 2013). Previous studies have used δ¹³C and δ¹⁵N of bulk particulate matter to assess the contribution of biomass/biofuel-burning sources (Martinelli et al., 2002; Kundu et al., 2006; Ometto et al., 2006; Widory, 2007; Aggarwal and Kawamura, 2008; Agnihotri et al., 2011; Aggarwal et al., 2013; Mkoma et al., 2014). Although the stable isotopic composition of OA produced from various sources are significantly different, several processes can affect their δ¹³C and δ¹⁵N, viz., ageing of air masses, condensation/evaporation of volatile organic compounds (VOCs) on/from a pre-existing particles, oxidation or chemical reactions of primary organic aerosols (POA) by O₃ or NO _x (Kirillova et al., 2013 and references therein).

However, studies from South Asia are limited in assessing the stable C- and N-isotopic composition of ambient aerosols (Agnihotri et al., 2011; Pavuluri et al., 2011; Hegde and Kawamura, 2012; Aggarwal et al., 2013). Based on the physicochemical properties of aerosols, earlier studies have demonstrated the continental impact from South and Southeast Asia over the northern Indian Ocean (NIO) (Lelieveld et al., 2001; Ramanathan et al., 2001b; Mayol-Bracero et al., 2002a). Furthermore, the predominant continental influence on the eastern part of the NIO [the Bay of Bengal (BoB)] has been emphasized by subsequent studies compared to that over the Arabian Sea, western side of the NIO (Kumar et al., 2008; Kedia et al., 2010; Sarin et al., 2010; Srinivas et al., 2011a; Srinivas and Sarin, 2012, 2013a). These studies have further suggested that the strength of the continental sources decreases from winter (December–February) to spring–intermonsoon (March–April) (Srinivas et al., 2011b; Srinivas and Sarin, 2013b, 2014).

A recent study by Agnihotri et al. (2011) has assessed stable C- and N-isotopic composition of marine aerosols from the BoB and Arabian Sea during spring to intermonsoon (March–May 2006). However, no single study deals with the assessment of δ¹³C of bulk carbon and δ¹⁵N of total nitrogen in marine aerosols from the BoB during the wintertime when marine atmospheric boundary layer (MABL) is influenced by continental air masses from South and Southeast Asia. Here, we have studied the stable isotopic composition of total carbon and nitrogen (TC and TN) of fine mode aerosols (PM_2.5) collected from the BoB during winter season. We have also made a comparison of our results from the Bay region with those from the nearby continental sites as well as other oceanic regions. The overall objective of this study is to understand the measured stable C- and N-isotopic composition of aerosol-TC and TN, respectively, in terms of their source contribution as well as changes in composition (if any) during atmospheric transport to the oceanic region (BoB) located downwind of pollution sources.

2. Methodology

2.1. Cruise track and prevailing meteorology

Aerosol samples (PM_2.5) were collected during a cruise (SK-254) conducted in the BoB as part of Indian national programme on integrated campaign of aerosols, trace gases and radiation budget (ICARB) during 27 December 2008–28 January 2009. During the study period, prevailing winds were mostly north-easterly/westerly. The shallow boundary layer along with the relatively weak winds brings continental air masses to the MABL during the wintertime (December–February). We obtained the meteorological parameters on board (viz., relative humidity, ambient temperature, wind speed and wind direction) every hour, which were averaged for daily means. The daily average of relative humidity, ambient temperature and wind speed during the cruise varied from 49.1 to 79.8 %, 22.3 to 26.8 °C and 1.2 to 6.3 m s⁻¹, respectively (Kumar et al., 2010). Further details on wind regimes and meteorological conditions are described in our earlier publications (Kumar et al., 2010; Srinivas et al., 2011b).

2.2. Aerosol sampling and measurement of chemical constituents

In this study, we used a total number of 31 PM_2.5 samples that were collected using a high volume air sampler (HVS, Thermo-Anderson Tech., flow rate: 1.13 m³ minute⁻¹) on board ORV Sagar Kanya. The sampler was placed on the upper deck in front of the ship's navigation room in order to avoid the contamination from the ship's smoke stack and was operated when the wind direction is from the bow and ship cruising at a speed of more than 10 knots h⁻¹. The sampler was calibrated before and after the cruise, and the variation in the flow rate was within 3 %. On average, the samples were collected for a period of ~20 h, and the volume of filtered air ranged between 1300 and 1400 m³. All PM_2.5 samples were collected on pre-combusted (at 450 °C, for 3–4 h) quartz filters (PALLFLEX^®™, 2500 QAT-UP). After collection, aerosol filters were wrapped in aluminium foils, sealed in zip-lock bags and stored in deep freezer at −19 °C until the chemical analyses.

In PM_2.5, water-soluble inorganic ions (Na⁺, , K⁺, Mg²⁺, Ca²⁺, Cl⁻, and ) were determined withDionex-500 ion chromatograph equipped with suppressed conductivity detector (Srinivas et al., 2011b). Similarly, carbonaceous components (EC/OC) were measured on Sunset EC/OC analyser using NIOSH-5040 method (Srinivas and Sarin, 2013a), whereas water-soluble organic carbon (WSOC) was measured using Schimadzu TOC-5000 analyser (Srinivas and Sarin, 2013a). A part of the chemical composition data (K⁺, , , , EC, OC and WSOC) from earlier publications have been used here to support the inferences related to stable C- and N-isotopic composition of PM_2.5 over the BoB. For more details, see our earlier publications (Rengarajan et al., 2007; Kumar et al., 2010; Srinivas and Sarin, 2014).

To determine the blank levels, one quarter of the filter aliquot (ca. ~100 cm²) was extracted with 50 ml of Milli-Q water (specific resistivity ~18.2 MΩ-cm) like samples and the extracts were measured for water-soluble inorganic constituents. Except for Na⁺, K⁺ and Cl⁻, no signals were detected on ion chromatogram. The detection limits, as defined by three times of standard deviation of the concentrations of procedural filter blanks (N=8) normalized to average filtered volume of air for the PM_2.5 (~1400 m³), were 30 ng m⁻³ for Na⁺, 18 ng m⁻³ for K⁺ and 24 ng m⁻³ for Cl⁻. We have corrected the K⁺ and concentrations in PM_2.5 for the sea-salt contribution to derive the non-sea-salt (nss) K⁺ and , respectively, using measured Na⁺ in aerosols (Keene et al., 1986). All PM_2.5 samples were analysed to assess the mass concentrations of total carbon and nitrogen (TC and TN) along with their stable isotopic composition (δ¹³C_TC and δ¹⁵N_TN) using an elemental analyser (EA) and EA/isotope ratio mass spectrometer as reported in Kawamura et al. (2004) and Kundu et al. (2010).

2.3. Stable C- and N-isotopic composition

For this study, stable isotopic composition of δ¹³C_TC and δ¹⁵N_TN were determined using an EA interfaced online to an isotope ratio mass spectrometer (EA/irMS, model: Carlo Erba NA 1500 EA+Finnigan MAT Delta Plus), and the protocol described in Kawamura et al. (2004). Briefly, the filter aliquot (ca. 2.0 cm in diameter) of aerosol sample was sealed in a tin cup, which is pre-cleaned with acetone to remove organic contaminants and dried prior to use, and introduced into EA where it is oxidized in the combustion column packed with copper oxide (CuO) filings at a temperature of 1020 °C. The evolved gases (CO₂ and NO _x ) were subsequently introduced into a reduction column, where NO _x is converted to molecular N₂ and thereby separated by a gas chromatograph equipped within EA. To measure stable isotopic composition of evolved CO₂ and N₂, an interface (ConFlo II) is used to transfer the gases from EA to irMS. The measured isotopic composition of aerosol-TC and TN was expressed as δ¹³C_TC and δ¹⁵N_TN relative to Pee Dee Beleminite carbon standard and atmospheric N₂, respectively, using the following equations:

2.4. Quality assurance

Prior to the measurements of δ¹³C_TC and δ¹⁵N_TN, instrumental blanks are checked with pre-cleaned tin cups (N=3) and a five-point calibration, having standard amount between 0.2 and 0.6 mg of acetanilide (Thermo Scientific) with known stable carbon and nitrogen isotopic composition as −27.26 ‰ (Cao et al., 2016) and +11.88 ‰ (Kundu et al., 2010), respectively. Filter (lab+field) blanks were also analysed and the analytical signal of samples was suitably corrected for blanks using isotopic mass balance equations described in Turekian et al. (2003). The overall analytical uncertainties in the measurement of δ¹³C_TC and δ¹⁵N_TN were 0.15 ‰ and 0.25 ‰, respectively, as ascertained by repeat analyses of acetanilide and samples, being consistent with those reported in Kawamura et al. (2004). The uncertainties associated with each measurement were propagated and discussed when comparing the significant differences between the Indo-Gangetic Plain (IGP)- and Southeast Asia (SEA)-outflows (Supplementary Table 1).

In this study, TC concentrations (OC+EC) were measured in PM_2.5 soon after their collection on Sunset EC–OC analyser using NIOSH protocol (Rengarajan et al., 2007). These measurements show consistency with total carbon content obtained from the EA-irMS. This comparison clearly demonstrates that artefacts, if any, related to shelf storage of aerosol filters have not caused any error in the isotopic measurements. Similarly, the closure of mass concentration of water-soluble inorganic nitrogen (mainly as -N; >90 %) in the TN (Fig. 2) further establishing the consistency between previous measurements (Srinivas et al., 2011b) and this study. This observation clearly demonstrates the lack of artefacts related to the storage of aerosol filters.

3. Results and discussion

3.1. Temporal variability

During a winter cruise, the PM_2.5 collected over the N-BoB is influenced by the continental outflow from the IGP-outflow (27 December 2008 to 10 January 2009, Fig. 1), whereas those sampled over the S-BoB has influence from SEA-outflow (11–28 January 2009; Fig. 1). This is inferred based on the origin of 7-d isentropic air mass back trajectories (AMBTs) for the sampling days, computed using hybrid single particle Lagrangian-integrated trajectory model, HYSPLIT-4 (Draxler et al., 1999). The impact of IGP- and SEA-outflows on the chemical composition of aerosols over the BoB is a conspicuous seasonal meteorological phenomenon, occurring in winter and spring, documented through large-scale field campaigns (INDOEX and ICARB-I & II) using various optical and chemical properties of aerosols (Kumar et al., 2010; Lawrence and Lelieveld, 2010; Srinivas et al., 2011b).

Fig. 1 Seven-day isentropic air mass back trajectories (AMBTs) computed using hybrid single particle Lagrangian-integrated trajectory model, HYSPLIT-4 (Draxler et al., 1999) for the sampling days (27 December 2008 to 26 January 2009) along the cruise track in the Bay of Bengal. Blue pins and red dots over the Indo-Gangetic Plain represent the locations of coal-fired power plants and MODIS-derived fire spots, respectively.

Organic aerosols in winter over the IGP are mostly emitted from the fossil fuel combustion sources (FF-comb) and biomass/biofuel-burning emissions (BBEs) (wood/crop-residue combustion) (Sudheer and Sarin, 2008; Rajput et al., 2011; Srinivas and Sarin, 2014; Rastogi et al., 2015). However, with the exception of a few studies emphasizing the contribution of organic aerosols from the forest fires over South-east Asia (Narukawa et al., 1999; Streets et al., 2003; Srinivas and Sarin, 2013a), no such detailed information exists in the literature about other pollution sources. To understand the influence of different pollution sources to atmospheric organic aerosols sampled over the BoB (this study), we have superimposed the locations of coal-fired thermal power plants in the IGP (see blue pins in Fig. 1) and MODIS-derived fire count data (red dots in Fig. 1) from South and Southeast Asia on the back trajectory cluster information (Fig. 1).

From Fig. 1, it is implicit that the SEA-outflow has a contribution from intense forest fires occurring over Indonesia, Thailand, Vietnam and Myanmar. However, no such influence from BBEs was observed in the IGP-outflow during winter. On the other hand, the IGP-outflow has a contribution of organic aerosols from coal-fired thermal power plants in the IGP [also referred as ‘coal belt of India’ (Prasad et al., 2006; Nair et al., 2007)]. Besides this, the residential biofuel (wood) combustion emissions (particularly significant in winter) also contribute to organic aerosols in the IGP-outflow (Gustafsson et al., 2009; Ram and Sarin, 2012). Therefore, we try to infer the relative significance of various sources and compositional changes in the continental outflows (consisting of emissions from coal-fired power plants, vehicular traffic and biomass burning) by comparing the observed δ¹³C_TC over the BoB with source-specific signatures and regression analysis between isotopic composition and other chemical tracers measured in the MABL.

TC and TN in PM_2.5 exhibit strong latitudinal variability with relatively high concentrations in the IGP-outflow (close to the continent) compared to those in the SEA-outflow (open ocean samples; Table 1). Similar north (high in the IGP-outflow) to south (low in the SEA-outflow) concentration gradient is also noteworthy for other chemical species in the PM_2.5 (e.g. OC, EC, WSOC and ; Supplementary Fig. 1). The δ¹³C_TC has shown pronounced spatial variability over the BoB (Fig. 3) with significantly higher δ¹³C_TC (t-score: 5.5, df=29; p<0.001) in the IGP-outflow (−25.0 to −22.8 ‰; average −23.8±0.6 ‰) than SEA-outflow (−26.9 to −24.2 ‰; average −25.3±0.9 ‰). This discrepancy is due to varying contributions of aerosol-TC from FF-comb and BBEs over the North and South BoB, respectively. However, no significant differences were observed for δ¹⁵N_TN (t-score: 0.5, df=29, p>0.05) in both the outflows (IGP: +20.4±5.4 ‰ and SEA: +19.4±6.1 ‰; Fig. 3).

Table 1. Statistical description of mass concentrations (in µg m⁻³) of chemical constituents and stable carbon and nitrogen isotopic compositions (δ¹⁵N and δ¹³C) in PM_2.5 collected over the Bay of Bengal during January 2009

Parameter	IGP-outflow (N=15)	SEA-outflow (N=16)	t-Score, df, p-value
^aPM2.5	13.2–76.7 (38.0±20.2)	2.0–35.3 (22.3±9.9)	2.8, 29, <0.05
TN	0.6–8.8 (3.7±2.5)	0.2–3.0 (1.4±0.8)	3.7, 29, <0.05
TC	3.6–15.4 (8.1±4.3)	1.1–6.5 (3.8±1.8)	3.4, 29, <0.05
TC/TN	1.3–9.4 (3.0±2.1)	1.9–5.4 (2.9±0.8)	0.1, 29, >0.05
δ^13CTC (‰)	−25.0 to −22.8 (−23.8±0.6)	−26.9 to −24.2 (−25.3±0.9)	7.3, 29, >0.05
δ^15NTN (‰)	+11.8 to +30.6 (20.4±5.4)	+10.4 to +31.7 (19.4±6.1)	0.5, 29, <0.05
^aOC	1.6–11.6 (5.9±3.7)	0.2–5.3 (2.6±1.9)	3.0, 27, <0.05
^aEC	0.8–5.0 (2.0±1.3)	0.2–1.8 (1.2±0.4)	2.4, 29, <0.05
WSOC	1.5–8.3 (4.4±2.4)	0.5–4.1 (2.3±1.4)	2.9, 29, <0.05
WIOC	0.0–4.5 (1.6±1.4)	0.0–1.8 (0.6±0.5)	2.6, 29, <0.05
^a	0.4–10.1 (4.4±3.0)	0.3–4.1 (2.1±1.2)	2.7, 29, <0.05
^anss-	1.6–28.5 (11.8±7.8)	1.3–12.1 (7.0 ±3.1)	2.3, 29, <0.05
( /nss- )equivalent	0.62–1.12 (0.92±0.15)	0.60–0.93 (0.78±0.11)	3.2, 29, <0.05
nss-K^+/EC	0.19–0.51 (0.29±0.09)	0.23–0.52 (0.39±0.09)	3.0, 29, <0.05
nss- /EC	1.3–14.5 (6.2±3.6)	2.7–10.3 (6.2±2.2)	0.01, 29, >0.05

^aSupporting parameters were obtained from Kumar et al. (2010).

In comparison to δ¹³C_TC, δ¹⁵N_TN show large variability in both outflows (IGP: +11.8 to +31.6 ‰; SEA: +10.4 to +31.7 ‰; Table 1). The wide spread combustion of biofuels and agricultural crop residue from South and Southeast Asia is a major source of aerosol-TN over the BoB during winter and spring seasons (Srinivas et al., 2011b). In this regard, Turekian et al. (1998) documented that unlike δ¹³C_TC, δ¹⁵N_TN in particles emitted from the combustion of C₃ and C₄ vegetation overlap within spread of the data. However, their study suggested that changes in combustion temperature during BBEs lead to large variability in the stable N-isotopic composition of aerosols. This is partly due to accessibility of various nitrogenous pools in the source plant at different temperatures during biomass combustion (Turekian et al., 1998). Therefore, observed large variability in the δ¹⁵N_TN over the BoB is attributed to variations in the combustion temperature during BBEs in the continental outflows (i.e. wood burning in the IGP or forest fires in the SEA).

Although wood-burning emissions usually involved with low-temperature combustion processes than forest fires (i.e. high-temperature combustion), however, we observe no significant differences (t-score: 0.5, df=29, p>0.05) in the mean δ¹⁵N_TN between both the outflows (IGP: +20.4±5.4 ‰ and SEA: +19.4±6.1 ‰; Fig. 3). This feature indicates that δ¹⁵N_TN over the BoB in winter season is mostly governed by the atmospheric processes affecting aerosol-TN concentrations during long-range transport rather than contribution from different sources (i.e., δ¹⁵N_TN is ‘process dependent’ over the BoB). From Fig. 2, it is obvious that the concentrations of water-soluble inorganic nitrogen species ( and ) in the MABL account for more than 90 % of the TN mass. Therefore, processes affecting the relative abundances of and in aerosol-TN might control the δ¹⁵N_TN (Section 3.3 for a more detailed discussion).

Fig. 2 (a) Comparison of mass concentrations of total carbon (TC) measured by elemental analyser coupled to isotope ratio mass spectrometer (EA-irMS) versus estimated TC concentration based on the atmospheric abundances of elemental and organic carbon (as EC+OC, measured on Sunset Carbon Analyzer), and (b) correlation plot between total nitrogen (TN) measured by EA-irMS and water-soluble inorganic nitrogen (i.e.

and

) in PM_2.5 samples collected over the Bay of Bengal during January 2009 cruise.

The sudden change (t-score: 5.5, df=29; p<0.001) in wind regimes from the IGP- to SEA-outflow was characterized by distinct δ¹³C_TC over the BoB (−23.8±0.6 ‰ and −25.3±0.9 ‰, respectively). This observation indicates that stable C-isotopic composition of aerosols over the BoB is ‘source dependent’. A significant linear relationship is noteworthy between TN and δ¹⁵N_TN in both the IGP- and SEA-outflows (Fig. 4a). However, such a significant linear relationship between TC and δ¹³C_TC is observed only in the SEA-outflow but not in the IGP-outflow (Fig. 4b). This could be due to variability in the sources of aerosol-TN and TC in the IGP-outflow (wood burning and FF-comb, respectively) than those in the SEA-outflow (forest fires). Thus, a significant depletion of ¹³C in TC for the SEA-outflow compared with that for the IGP-outflow, perhaps, indicates a source-specific contribution from BBEs in Southeast Asia during the cruise.

Overall, ¹³C enrichment and depletion of TC in the IGP- and SEA-outflows, respectively, can be explained based on the differences in the source contribution of carbonaceous species and their transport history to the BoB. A noteworthy feature of the data is evident from a change in the TC concentration in the IGP-outflow with no variability in δ¹³C_TC (Fig. 4b). This feature suggests that dominant contribution of aerosol-TC in the IGP-outflow is derived from point sources in the IGP. Based on the analyses of various aerosol properties (e.g. AOD, Angstrom exponent), derived from Multiangle Imaging SpectroRadiometer (MISR) level 3 remote sensing data, Prasad et al. (2006) have documented that emissions from coal-fired power plant in the IGP is a dominant source of absorbing organic aerosols and other airborne particulate matters in winter season. Furthermore, there exists a notable declining trend in the emission of organic aerosols from the biofuel (wood) combustion and an ongoing rise in coal combustion based on the comparison of energy consumption between 1990 and 2000 (Prasad et al., 2006). Therefore, we attribute coal-fired power plant emissions in the IGP is a major source of organic aerosols sampled over the N-BoB (i.e. in the IGP-outflow).

The forest fires, occurring over Southeast Asia during the cruise in the BoB (January 2009), contribute to large-scale emissions of POA as well as VOCs. Thus, the observed depletion in ¹³C of TC in the SEA-outflow (Fig. 3b) can be explained in terms of contribution of POA from C₃ vegetation burning and/or from the secondary organic aerosol (SOA) formation through the photochemical oxidation of VOCs. Similar to our study, lower values of δ¹³C_TC (−26.6 to −23.2 ‰, average −25.0 ‰) were observed for the wheat residue-burning aerosols sampled over Mt. Tai in East Asia (Fu et al., 2012). Furthermore, Fu et al. (2012) documented a non-linear (exponential) relationship between δ¹³C_TC and levoglucosan (a tracer of BBEs), suggesting that an increased contribution of BBEs causes a significant depletion of ¹³C in aerosols (δ¹³C: −26 to −27 ‰) (Fu et al., 2012). Narukawa et al. (1999) also reported a conspicuous decrease in δ¹³C_TC from −25.5 to −27.5 ‰ for the fine mode aerosols sampled from the Indonesian forest fires, in response to an increased contribution of organics from C₃ vegetation. Also, the fractionation of stable carbon isotopes during the combustion of C₃ plants is not significant (~0.5 ‰ enrichment between unburned plant and aerosols (Turekian et al., 1998).

Fig. 3 Spatial distributions of (a) stable carbon (δ¹³C_TC) and (b) nitrogen (δ¹⁵N_TN) isotopic composition of fine mode (PM_2.5) aerosols collected from the marine atmospheric boundary layer of the Bay of Bengal during a winter cruise (27 December 2008 to 28 January 2009).

Based on the above arguments, the lower δ¹³C_TC in the SEA-outflow can be attributed to contribution of organic aerosols from the combustion of C₃ plants. However, SOAs produced from the photooxidation of biogenic VOCs could also contribute to lower the δ¹³C_TC (Kirillova et al., 2013). A recent laboratory photooxidation/smog chamber experiment of toluene (a VOC emitted from motor vehicles) has demonstrated formation of SOA, whose δ¹³C values were ~6 ‰ lower than that of the parent toluene (Irei et al., 2006, 2011). Furthermore, δ¹³C fractionation between gas and aerosol phases is found to be small during photooxidation (Irei et al., 2006, 2011). Similarly, the laboratory photooxidation on gaseous isoprene and ozone system results in the formation of less volatile methacrolein and methyl vinyl ketone (i.e. known gaseous isoprene oxidation products), whose δ¹³C values are depleted by more than 3.6 and 4.5 ‰, respectively (Iannone et al., 2010). These observations, thus, support the argument that photooxidation of anthropogenic/biogenic VOCs results in not only more oxidized semi-volatile organics for which isotopic fractionation effects are found to be minimal (Irei, 2008) but also lowers the δ¹³C values than those of precursor fuel/vegetation.

These semi-VOCs subsequently partition to SOAs and, hence, can explain their lower δ¹³C signatures. In this regard, lower δ¹³C values in the SEA-outflow can, therefore, be attributed to the contribution of SOAs derived from atmospheric oxidation of VOCs emitted from the BBEs in Southeast Asia. We have also observed a significant linear relationship of δ¹³C_TC with nss-K⁺ (R ²=0.58, p<0.05) and WSOC (R ²=0.60, p<0.05) in the SEA-outflow (Fig. 4c and d) but not in the IGP-outflow (δ¹³C_TC vs. WSOC: R ²=0.01; p>0.05 and δ¹³C_TC vs. nss-K⁺: R ²=0.03; p>0.05). In the literature, the occurrence of nss-K⁺ in the fine mode aerosols and linear relationship with WSOC are used to assess the contribution of carbonaceous aerosols from the BBEs (Andreae, 1983; Andreae and Merlet, 2001; Paris et al., 2010; Pachon et al., 2014; Srinivas and Sarin, 2014). Therefore, a lack of significant linear relationship between nss-K⁺ and δ¹³C_TC in the IGP-outflow sampled over the BoB (Fig. 4c), which is consistent with that of TC and δ¹³C_TC (Fig. 4b), indicate the role of other sources of organic aerosols over the IGP (e.g. coal-fired power plants or vehicular emissions) than BBEs. Based on the literature, SOAs contribute significantly to atmospheric WSOC (Graham et al., 2002; Mayol-Bracero et al., 2002b; Sciare et al., 2008; Ram and Sarin, 2010; Hegde and Kawamura, 2012; Rajput et al., 2014; Srinivas and Sarin, 2014). In this regard, the BBE-derived POA and/or SOA causes ¹³C depletion of TC in the SEA-outflow, which can also be supported by the strong correlations of δ¹³C with nss-K⁺ and WSOC (Fig. 4c and d).

Fig. 4 Correlation plots for (a) TN versus δ¹⁵N_TN, (b) TC versus δ¹³C_TC, (c) nss-K⁺ versus δ¹³C_TC and (d) WSOC versus δ¹³C_TC for fine mode aerosols collected over the Bay of Bengal during a winter cruise (27 December 2008 to 28 January 2009).

The impact of BBEs over FF-comb sources over the BoB can also be inferred based on the relative increase in mass ratio of nss-K⁺ to EC (Table 1) as the latter contains negligible potassium content (Novakov et al., 2000; Mayol-Bracero et al., 2002b). Furthermore, higher nss-K⁺/EC ratios in the SEA-outflow (0.39±0.09) compared with those in the IGP-outflow (0.29±0.09) suggest the dominance of BBEs in the former samples. A significant linear relationship is noteworthy between δ¹³C_TC and δ¹⁵N_TN in the SEA-outflow (slope=− 5.5; R ²=0.63; p<0.05; Supplementary Fig. 3); however, no such correlation is observed for the IGP-outflow (slope=− 3.5; R ²=0.15; p=0.15; Supplementary Fig. 3). The observed weak (or no) and strong correlation of δ¹³C_TC with δ¹⁵N_TN for the IGP- and SEA-outflows, respectively, could be due to variable contribution of TC and TN from FF-comb and BBEs over the BoB. Furthermore, moderate linear correlation is found between nss- and EC for the IGP-outflow samples (R ²=0.57; p<0.05), which indicates an impact of FF-combustion sources over the north BoB. However, rather weak linear correlation was observed for SEA-outflow between nss- and EC (R ²=0.26; p=0.03). This observation emphasizes a significant contribution of TC from FF-comb sources in the IGP over the BoB during the study period.

3.2. Relative contribution of fossil fuel versus biomass/biofuel-burning source

The δ¹³C_TC in PM_2.5 over the BoB during a winter cruise can be explained based on two end-member mixing between FF-comb and BBEs. Since the IGP is a densely populated and industrialized region of India with intense ongoing anthropogenic activities (viz., coal-fired power plants, vehicular emissions, wood/post-harvest crop-residue burning), it is not logical to assume that biogenic emissions mainly contribute to aerosol-TC in the IGP-outflow. The dominant contribution of aerosol-TC in the SEA-outflow samples is from forest fires occurring in Southeast Asia (as inferred from MODIS satellite-derived fire count data; Supplemantary Fig. 3).

Organic compounds, derived from phytoplankton debris and zooplankton excreta in the sea surface microlayer, are coated on sea-salt particles, contributing to marine aerosols collected over the BoB. Therefore, marine phytoplankton-derived organic aerosols have an average δ¹³C value of ca. −22 ‰ (Miyazaki et al., 2011). However, the BoB is not a productive oceanic basin and is perennially influenced by fresh water influx from the Ganga–Brahmaputra and other peninsular rivers. Moreover, the persistence of weak NE-monsoonal winds (<3 m second⁻¹) are not likely to trigger sea-to-air transfer of organics from the surface waters of the BoB during the continental outflow. Therefore, contribution from marine organic matter is unlikely due to prevailing winds during the study period. This is further supported by the relative contribution of sea-salts and hence particulate organic matter from the ocean surface by bubble bursting to ambient aerosols over the BoB is no more than 5 % (Srinivas and Sarin, 2014). Therefore, we believe that a two end-member mixing model is a reasonable first-order pseudo-approximation to assess the relative contribution of aerosol-TC from BBEs and FF-comb over the BoB.

The rationale for the contribution from these two sources is further supported by the observed nss-K⁺/EC and nss- /EC ratios in PM_2.5 over the BoB (for additional information, see Section 3.1). Higher nss-K⁺/EC ratios in the SEA-outflow than those in the IGP-outflow samples indicate a dominance of BBEs over the South BoB (Table 1) (Srinivas and Sarin, 2014). The nss- /EC ratio, a proxy for fossil fuel combustion with higher values documented for coal-fired power plants (~6) (Ramana et al., 2010), showed somewhat consistency between the IGP- and SEA-outflows (Srinivas and Sarin, 2014) (Table 1), indicating a contribution from point sources in the IGP (e.g. coal-fired power plants). Also, the 7-d AMBTs for the PM_2.5 samples collected over the north BoB have the source origin from the IGP (Fig. 1), which intercept the receptor sites dominated by carbonaceous aerosols (and hence TC) derived from coal-fired power plants (see blue pins in Fig. 1).

Widory (2006) and Mori et al. (1999) have documented that δ¹³C of aerosols from coal combustion as −23.6±0.7 ‰ (−22.9 to −24.9 ‰) and −23.4±1.2 ‰, respectively, with almost zero fractionation (or slightly negative). On the other hand, combustion-derived particles from liquid fossil fuels (e.g. unleaded/leaded gasoline; δ¹³C: −26.4 to −28.0 ‰) are depleted in ¹³C relative to those originated from solid fuel (coal: −24 to −21 ‰) (Widory, 2006) but are enriched with ¹³C relative to gaseous fuels (−40 to −28 ‰) (Huang et al., 2006; Widory, 2006; López-Veneroni, 2009; Agnihotri et al., 2011; Kawashima and Haneishi, 2012). A comparison of δ¹³C_TC (−23.8±0.6 ‰) over the north BoB with those representative of various fossil fuels as mentioned above (coal, diesel, gasoline, etc.) suggest a dominant contribution from coal-fired power plant emissions in the IGP (Fig. 5).

Fig. 5 Comparison of δ¹³C of aerosol-TC over the BoB during a winter cruise (January 2009) with literature values of various sources such as combustion/emissions from C₃, C₄ and marine phytoplankton (Turekian et al., 1998; Cao et al., 2013), coal combustion (Mori et al., 1999; Widory, 2006) and vehicular emissions (Widory, 2006; Cao et al., 2013).

Although we have observed a decrease in the number (intensity) of fire counts (derived from MODIS data) over the IGP than those over SEA, the contribution of wood-burning emissions to aerosol-TC (Gustafsson et al., 2009; Ram and Sarin, 2012) cannot be ruled out in the IGP-outflow sampled over the BoB. In this context, laboratory and natural field burning chars from C₃ grasses and trees (wood and leaves) are enriched and/or depleted in ¹³C (e.g. δ¹³C for Eucalyptus wood: −27.2 ‰; Sphagnum: −26.0 ‰; C₃ grasses: −28.2 to −32.4 ‰; Eucalyptus leaf: −28.2 ‰) compared with plant materials (Krull et al., 2003). On a similar note, based on the laboratory-based combustion, Agnihotri et al. (2011) documented that aerosols from typical BBEs in India exhibit a very narrow range of δ¹³C (−27.9±1.1 ‰). However, the observed δ¹³C_TC in the IGP-outflow (−23.8±0.6 ‰) is consistent with that of coal-derived emissions compared with other sources (i.e. gasoline combustion and wood-burning emissions; Fig. 5). Therefore, we have used coal-burning emissions as one end-member (δ¹³C_TC: −23 ‰; Widory, 2006) for explaining the observed stable carbon isotopic signatures of TC in the continental outflow sampled over the BoB during winter. We can further support the significance of coal-fired power plants as a potential source of organic aerosols in the IGP-outflow sampled over the BoB as follows.

As mentioned earlier, FF-comb (coal combustion and vehicular emissions: VEs) and BBEs (wood/crop-residue) in the IGP contribute to organic aerosols over the BoB during the winter period (Rengarajan et al., 2007; Sudheer and Sarin, 2008; Kumar et al., 2010; Srinivas and Sarin, 2013, 2014). In such cases, the δ¹³C_TC of PM_2.5 over the BoB can be expressed as a product of individual fractions of TC from each of these three respective sources times their stable C-isotopic composition as below 1

Based on the ¹⁴C-analysis of EC, Gustafsson et al. (2009) documented near equal contribution of wood-burning emissions and fossil fuel combustion sources to Atmospheric Brown Clouds over South Asia. Therefore, we have assumed 50 % contribution from biofuels (Gustafsson et al. 2009), 30 % from coal-fired thermal power plants and 20 % from vehicular emissions (Prasad et al., 2006) to atmospheric organic aerosols in the IGP-outflow. Agnihotri et al. (2011) documented that source-specific δ¹³C_TC of various biofuels and crop-residue combustion-derived aerosols from South Asia centred on −27.9±1.1 ‰. Similarly, a source-specific δ¹³C_TC of −23.6±0.7 ‰ from coal combustion-derived particles has been documented by Widory (2006). Also, the δ¹³C_TC of particles generated from diesel combustion emissions span from −28 to −26 ‰ (Widory, 2006).

Using the respective source contributions and their corresponding δ¹³C_TC in eq. (1), we have estimated δ¹³C of organic aerosols collected in the IGP-outflow to be −26 to −27 ‰. Based on the narrow spread of δ¹³C_TC in the IGP-outflow (−23.8±0.6 ‰), we estimated δ¹³C_TC from the contribution of above-mentioned three major sources in the IGP. However, it may be argued that atmospheric oxidation, and hence, enrichment in ¹³C could be compensated by SOA formation (with overall ¹³C depletion) during transport. Therefore, the observed overlap of δ¹³C_TC over the N-BoB with coal-fired power plant emissions in the IGP emphasizes the source significance in the IGP-outflow. However, further studies on seasonal variability of the IGP-outflow are needed along with the δ¹³C_TC.

The δ¹³C_TC in the SEA-outflow (−26.9 to −24.2 ‰) overlap with those derived from both vehicular emissions and forest fires (Fig. 5). However, the SEA-outflow samples are collected over the pelagic (deep) waters (i.e. away from the coast and unlike samples from the IGP-outflow, where vehicular emissions could contribute) and, therefore, not likely to have contribution from vehicular emissions. Furthermore, AMBTs originating from Southeast Asia have a significant contribution from forest fires, inferred from the MODIS fire count data (Supplemantary Fig. 2). Therefore, we have considered BBEs as another end-member to account for the variability in δ¹³C_TC in the SEA-outflow.

Aerosol-TC in the SEA-outflow (−25.3±0.9 ‰) is significantly depleted in ¹³C than that derived from C₄ plants (−17 to −9 ‰; average −13 ‰), being somewhat consistent with those originated from C₃ vegetation burning (−32 to −20 ‰; average −28 ‰) (Smith and Epstein, 1971). In a previous study, Turekian et al. (1998) suggested that particles from C₃ plant burning emissions showed a slight enrichment in ¹³C in aerosol-TC (or higher in δ¹³C by only 0.5 ‰ or less) than host plant, while C₄ plants resulted in a depletion of ¹³C (or lower in δ¹³C by 3.5 ‰) (Turekian et al., 1998). According to Turekian et al. (1998), the increased relative proportion of refractory components (e.g. lipids; ¹³C depleted than bulk plant) over labile carbon pools (¹³C enrichment) in aerosols during C₄ plant combustion causes ¹³C depletion of TC; however, no sucheffects were observed for C₃ plant combustion-derived aerosols. A similar result has been reported for the smoke and ash that are derived from the combustion of C₃ plants which show negligible (or no) fractionation (Krull et al., 2003; Das et al., 2010).

Taken together of all these findings, we have assumed −23 and −28 ‰ to be the representative end-member δ¹³C signatures for FF-comb source and BBEs, respectively, based on the two end-member mixing model, which is described as follows: 2 3 4

Here, f_FF-comb and f_BBEs refer to the fractional contribution of TC from fossil fuel combustion and biomass-burning emission sources, respectively. The end-member δ¹³C for FF-comb and BBEs are −23.0 and −28.0 ‰, respectively. Using eq. (4), we estimated the contribution of fossil fuel combustion to both the IGP- and SEA-outflows. In this study, the estimated contribution of FF-comb source to δ¹³C_TC varied from 61 to 100 % (average 84±12 %) for the IGP-outflow samples and 12 to 66 % (average 44±18 %) for the SEA-outflow samples.

3.3. Elevated δ¹⁵N of TN: transformation of gaseous NH₃ to particulate

In both the IGP- and SEA-outflows, the water-soluble inorganic nitrogen (i.e. -N accounting for 81–100 %) dominates the total soluble nitrogen (Srinivas et al., 2011b). Further, a moderate-to-strong linear correlation of δ¹⁵N_TN with and equivalent mass ratio of / is observed in both continental outflows (Fig. 6). Since dominates aerosol-TN over the BoB (i.e. >90 %; R ²=0.95; Fig. 2), atmospheric processes affecting the concentration of particulate in fine mode aerosols might explain the observed δ¹⁵N_TN. The particulate concentration in the MABL, however, is influenced by the atmospheric reactions of gaseous NH₃ with acidic species (e.g. H₂SO₄, HNO₃) to form NH₄HSO₄, (NH₄)₂SO₄ and NH₄NO₃. Furthermore, NH₄NO₃ is formed mostly under ammonia-rich conditions (i.e. for which to equivalent mass ratio is greater than unity) because the reaction between ammonia and sulphate is preferable. However, NH₄NO₃ is thermodynamically not stable and, therefore, often dissociated into precursor chemical constituents during atmospheric transport, which also affects the ambient concentration. As a result, mainly exists in coarse mode due to the reaction with mineral dust and sea-salt (Jickells, 2006). All these processes/reactions could influence δ¹⁵N_TN in ambient aerosols. It is implicit from Fig. 6 that linear relationship of concentration and the equivalent mass ratios of to with δ¹⁵N_TN can explain the observed enrichment of ¹⁵N in the PM_2.5 sampled over the BoB.

Fig. 6 Linear regression analysis between (a) concentration of

and δ¹⁵N, and (b) equivalent mass ratio of

to

and δ¹⁵N in PM_2.5 collected over the Bay of Bengal during a winter cruise (January 2009).

The mean equivalent mass ratio of / in the samples from IGP- and SEA-outflow corresponds to 0.92±0.15 and 0.77±0.12, respectively, suggesting an ammonia-deficit (poor) condition. Moreover, the observed ambient temperature during the cruise averages around 26.7 °C (IGP-outflow: 24.4–27.4 °C; SEA-outflow: 25.9–29.0 °C) and high nss- concentrations facilitate the dissociation of NH₄NO₃. Interestingly, Srinivas et al. (2011b) have documented the predominant occurrence of in coarse mode and in fine mode aerosols during the January 2009 cruise. This implies that particulate in the MABL mainly occur due to the formation of more stable NH₄HSO₄ or (NH₄)₂SO₄. In this regard, we observed a very strong correlation between and (Srinivas et al., 2011b), suggesting the formation (NH₄)₂SO₄ and NH₄HSO₄ over the BoB during the study period. Therefore, we believe that atmospheric transformation of gaseous NH₃ to particulate might be responsible for the observed large variability in δ¹⁵N of TN over the BoB.

Similar to our study, Pavuluri et al. (2010) reported an elevated δ¹⁵N of TN and its significant positive linear correlation with sulphate to ammonium molar ratio (or negative correlation with that of / molar ratio) for the urban coastal aerosols from Chennai, India, during the late winter (January). Further, they observed a positive relationship between ambient temperature and δ¹⁵N; in particular, higher δ¹⁵N are noteworthy for those sampling days with temperatures >20 °C (Pavuluri et al., 2010). The ambient temperature range over the BoB in this study is higher than that documented in Pavuluri et al. (2010). During a laboratory experiment, an exposure of gaseous NH₃ to aerosol filters loaded with sulphuric acid results in the formation of (NH₄)₂SO₄ (i.e. ascertained through stoichiometric equilibrium between and ), which is enriched in ¹⁵N with isotope enrichment factor of +33 ‰ (Heaton et al., 1997). Combining all the above arguments, we believe that gas to particle formation of NH₃ to particulate in the MABL can explain the enrichment of ¹⁵N in the PM_2.5 sampled over the BoB during winter.

The mean value of δ¹⁵N_TN in the IGP-outflow sampled over the BoB during a winter cruise (this study) is higher than those obtained during a spring–intermonsoon cruise [+10.6±2.7 ‰ (Agnihotri et al., 2011)]. Although the continental outflow from the IGP persists during spring intermonsoon, the observed depletion of ¹⁵N could be explained based on the decrease in source strength/variability. The weakening of the IGP-outflow during spring–intermonsoon, compared with winter months (Table 1), is inferred by a decrease in the atmospheric concentrations of (0.01–2.3 µg m⁻³; 0.9±0.6 µg m⁻³) and nss- (2.5–10.3 µg m⁻³; 10.3±2.5 µg m⁻³), and equivalent mass ratios of to nss- (0.01–0.66; average 0.41±0.20) over the BoB. Therefore, we believe that the observed decrease in particulate in the marine aerosols during spring intermonsoon might indicate the non-availability of gaseous NH₃ to form NH₄HSO₄ and (NH₄)₂SO₄ aerosols and, thus, explain the observed lower δ¹⁵N_TN.

Similar to our study, higher δ¹⁵N values (+13.4 to +22.1 ‰) have been reported for fine mode aerosols sampled over Tanzania during dry season, which is attributed to a significant ¹⁵N enrichment by isotopic exchange reaction of N between gaseous NH₃ and aerosol during atmospheric transport (Mkoma et al., 2014). Similarly, Pavuluri et al. (2010) have also reported similar δ¹⁵N values in PM₁₀ aerosols, ranging from +18.0 to +27.8 ‰ and +18.6 to +25.7 ‰ in day and night samples, respectively, collected during winter from a coastal site in India (Chennai). Therefore, observed variability in δ¹⁵N_TN for the IGP-outflow samples (+11.8 to +30.6 ‰) can be attributed to pronounced continental impact of anthropogenic sources (reflected from their higher ambient abundances of and over the north BoB) to TN and also the exchange reactions between gaseous NH₃ and particulate during transport. It is important to emphasize that δ¹³C_TC and δ¹⁵N_TN showed a good correlation for the SEA-outflow (influenced by BBEs; see Fig. 1), but not in the IGP-outflow. Similar to this study, Agnihotri et al. (2011) have documented a strong correlation between δ¹³C and δ¹⁵N over the BoB during spring–intermonsoon cruise, which is also influenced by the BBEs. In a previous study, Turekian et al. (1998) documented that various stages of biomass combustion resulted in a large spread in stable nitrogen isotopic signatures of aerosol-TN. Therefore, variability in δ¹⁵N_TN in the SEA-outflow (+10.4to +30.7 ‰; average +19.4±6.1 ‰) can be explained by the highly variable temperature conditions that prevail during the forest fires in Southeast Asia.

3.4. Comparison with studies from the Northern Indian Ocean

The stable isotopic composition of TC and TN from this study (January 2009) was compared with those from earlier cruises conducted in the NIO (Agnihotri et al., 2011) and other sampling sites from South Asia. A significant difference in the δ¹³C_TC is noteworthy during the cruises conducted in the BoB in January 2009 (Table 2) and March–April 2006 [range: −26.6 to −24.1 ‰; average −25.6±0.6 ‰ for TSP samples (Agnihotri et al., 2011)]. Although δ¹³C of TC for both cruises refers to different size fractions of aerosols, it is relevant to state that during long-range transport the mass size distribution of atmospheric particulates changes from coarse to fine mode due to the settling of coarser particles (Duce et al., 1991). This is most relevant for our sampling from the MABL of the BoB and the chemical composition mostly comprised dominant fine particles (Srinivas et al., 2011b). Although source strength of carbonaceous aerosols in the IGP-outflow is varied in between the cruises (i.e. higher abundances in winter than spring–intermonsoon) over the BoB, their sources remain the same, as inferred from the diagnostic mass ratios of chemical composition of aerosols (Sarin et al., 2010; Srinivas and Sarin, 2012).

Table 2. Comparison of stable carbon and nitrogen isotopic compositions of aerosols from this study with those documented in the literature over South Asia

Region	Season	Sample type	δ^13C (‰)	δ ^15N (‰)	Reference
Bay of Bengal	Winter (IGP-outflow)	PM2.5	−23.8±0.6	20.4±5.4	This study
Bay of Bengal	Winter (SEA-outflow)	PM2.5	−25.8±0.9	19.4±6.1	This study
Bay of Bengal	Spring–intermonsoon	TSP	−25.6±0.6	10.6±2.7	Agnihotri et al. (2011)
Arabian Sea	Spring–intermonsoon	TSP	−26.5±0.8	1.4±3.2	Agnihotri et al. (2011)
Mumbai	Summer	TSP	−26.5±0.3	20.2±1.2	Aggarwal et al. (2013)
Mumbai	Winter	TSP	−25.9±0.3	22.8±1.4	Aggarwal et al. (2013)
Chennai	Summer and winter	PM10	−25.0±0.6	–	Pavuluri et al. (2011)
Chennai	Winter (day)	PM10	–	25.5±2.4	Pavuluri et al. (2010)
Chennai	Winter (night)	PM10	–	23.0±2.3	Pavuluri et al. (2010)
Chennai	Summer (day)	PM10	–	22.2±4.3	Pavuluri et al. (2010)
Chennai	Summer (night)	PM10	–	24.3±3.6	Pavuluri et al. (2010)
Africa	Summer	PM10	−25.9±0.3	13.7±2.2	Mkoma et al. (2014)

The δ¹³C_TC in the IGP-outflow from a winter cruise (January 2009) in the BoB (this study) is significantly higher than that obtained during spring–intermonsoon (Agnihotri et al., 2011) (Table 2). However, δ¹³C_TC in the SEA-outflow sampled over the BoB in winter (11–28 January, 2009) is somewhat similar with that from spring–intermonsoon cruise (Agnihotri et al., 2011). The observed differences in δ¹³C_TC during the two cruises could be related to temporal shift in the source contributions (anthropogenic aerosols accounting for more than 80 % in winter vis-à-vis near equal contribution of dust and anthropogenic species during spring–intermonsoon, March–April 2006) (Srinivas and Sarin, 2012). Analysis of MODIS fire images (Supplementary Fig. 3) clearly showed hotspots over the IGP during spring–intermonsoon (March–April 2006) and Southeast Asia during winter (January 2009; Fig. 1). Therefore, ¹³C depleted values of TC in the IGP-outflow samples in spring–intermonsoon and the SEA-outflow samples in winter can be explained by a significant contribution of carbonaceous species from BBEs. The impact of BBEs in the IGP-outflow sampled over the BoB during spring–intermonsoon and the SEA-outflow during a winter cruise was also reflected through consistent nss-K⁺/EC ratios (0.35±0.11and 0.39±0.09, respectively).

We have also observed consistency in the δ¹³C_TC from this study with those documented for wintertime aerosols (−25.9±0.3 ‰) (Aggarwal et al., 2013) from Mumbai (a continental site in India). The biofuel/biomass burning is a dominant source of TC over Mumbai (Venkataraman et al., 2002, 2005) and also over the IGP during winter (Rengarajan et al., 2007; Ram and Sarin, 2012; Rajput et al., 2014; Srinivas and Sarin, 2014). Therefore, the overlapping δ¹³C_TC in the SEA-outflow samples in winter with those of the IGP-outflow samples from the BoB during spring–intermonsoon and of continental aerosols from Mumbai might correspond to biofuel/biomass-burning emissions.

4. Conclusions

This study reports the stable C- and N-isotopic composition of wintertime aerosols (PM_2.5) over the BoB. During the study period (December 2008–January 2009), the MABL is influenced by the continental outflow from the IGP- and SEA-outflows, respectively, as inferred from the 7-d AMBTs over the BoB. The δ¹³C_TC in the IGP-outflow (−23.8±0.6 ‰) is distinctly different than that in the SEA-outflow (−25.3±0.9 ‰). A significant linear relationship of δ¹³C_TC with TC, nss-K⁺ and WSOC is noteworthy in the SEA-outflow but not in the IGP-outflow, which indicates the impact of BBEs in the former samples (which is also supported by the MODIS fire count data). However, no such influence from BBEs was observed for the IGP-outflow. Besides, the temporal variability of δ¹³C_TC in the IGP-outflow is rather small unlike TC concentration and overlaps with those derived from coal combustion sources (−23.6±0.7 ‰) in the IGP. The observed ¹³C-depletion of aerosol-TC in the SEA-outflow is attributed to the contribution of POA directly emitted from the BBEs or due to SOAs formed by the oxidation of BVOCs from the forest fires. We observe no significant differences in the δ¹⁵N_TN between IGP- and SEA-outflows. Moreover, mass concentration of dominates the aerosol-TN over the BoB, showing significant linear relationship with δ¹⁵N_TN. Thus, we infer that δ¹⁵N_TN in aerosols over the BoB during winter is governed by the atmospheric processes affecting the concentrations of aerosol water-soluble inorganic nitrogen species (e.g. , ) during transport.

5. Acknowledgements

The authors thank all the crewmembers of ORV Sagar Kanya for their help during aerosol collection. The authors acknowledge the partial financial support provided from the ISRO-Geosphere Biosphere Programme. The logistic support provided by Drs. K. Krishnamoorthy and C.B.S. Dutt is thankfully acknowledged. The financial support provided by the Japanese Society for the Promotion of Science (JSPS) through grant-in-aid no. 24221001 is acknowledged for carrying out C and N isotope analyses at Hokkaido University.

Notes

To access the supplementary material to this article, please see Supplementary files under ‘Article Tools’.