Journal of Geophysical Research: Atmospheres

Chemical characterization of submicron aerosol particles collected over the Indian Ocean

Authors


Abstract

[1] Submicron aerosol particles (Dp < 1 μm) were sampled with stacked filter units on the National Center for Atmospheric Research (NCAR) Hercules C-130 aircraft during February–March 1999 as a contribution to the Indian Ocean Experiment (INDOEX). We determined the vertical and spatial distribution of the major aerosol components (NH4+, Na+, K+, Mg2+, Ca2+, methyl sulfonic acid, Cl, NO3, SO42−, oxalate, organic carbon, and black carbon) over the Indian Ocean to examine the role of pollution aerosols on indirect and direct radiative forcing. High pollution levels were observed over the entire northern Indian Ocean down to the Intertropical Convergence Zone (ITCZ) located between the equator and 10°S. In the northern part of the Indian Ocean (5°–15°N, 66°–73°E), high concentrations of carbonaceous aerosol and pollution-derived inorganic species were found in a layer extending from the sea surface to about 3.5 km asl. In this layer, the average mass concentration of all aerosol species detected by our technique ranged between 7 and 34 μg m−3, comparable to pollution levels observed in industrialized regions. In the Southern Hemisphere (1°–9°S, 66°–73°E), the aerosol concentrations rapidly declined to remote background levels of about 2 μg m−3. The concentrations of non-sea-salt sulfate (the main light scattering component) ranged from maximum values of 12.7 μg m−3 in the Northern Hemisphere to 0.2 μg m−3 in the Southern Hemisphere. Carbonaceous aerosol contributes between 40% and 60% to the fine aerosol mass of all determined components. An unusually high fraction of black carbon (up to 16% in the polluted areas) is responsible for its high light absorption coefficient.

1. Introduction

[2] The Indian Ocean Experiment (INDOEX) was an international, multiplatform field campaign to measure long-range transport of air pollution from South–Southeast Asia towards the Indian Ocean during the winter monsoon in February and the beginning of the transition period in March (cf. overview by J. Coakley et al. (General overview of INDOEX, submitted to Journal of Geophysical Research, 2002). This season was selected because the northeasterly winds are persistent while convection over the continental source regions is suppressed by large-scale subsidence, thus limiting upward dispersion of pollution. Fossil fuel combustion and biomass burning cause a large aerosol loading over the region [Lelieveld et al., 2001; Reiner et al., 2001].

[3] One central objective of INDOEX was the study of the chemical and physical properties of aerosols, their radiative effects, and the processes responsible for controlling and modifying their properties as they are transported over the ocean (V. Ramanathan et al., unpublished manuscript, 2000). The composition of the aerosol is one of the key parameters defining its chemical and physical behavior, particularly with regard to the properties important for its climatic effects: light scattering, absorption efficiency, and cloud droplet nucleating ability. Magnitude and sign of the “direct” climate effect is determined by light scattering and absorption of the aerosol. The “indirect” effects, i.e., the climate forcing due to modification of cloud properties depend on the amount of water-soluble material present in a particle. The amount of water-soluble material along with the particle diameter determines at what supersaturation a particle can become activated, grow into a cloud droplet, and thereby contribute to the quantity, lifetime, and brightness of clouds. The composition of the aerosol depends on the source region and changes with the meteorological conditions in the course of a year. Meteorological analysis suggests that high aerosol concentrations over the northern part of the Indian Ocean could be linked to transport from the Indian subcontinent and also from sources in the Middle East and North Africa [Krishnamurti et al., 1998]. Soil dust derived from sources in the Arabian Peninsula and in the Tigris and Euphrates basin, as well as soil in regions of Northwest India and East Africa [Ackerman and Cox, 1989; Prospero, 1981; Savoie et al., 1987] can effect high values of aerosol optical thickness over the northern end of the Arabian Sea (Cautenet et al., Modeling the evolution of some aerosol species during INDOEX 1999 and comparison with experimental data, submitted to Journal of Geophysical Research, in press, 2002, hereinafter referred to as Cautenet et al., submitted manuscript, 2002). A second major source of aerosols mainly affecting the Arabian Sea section of the INDOEX study region is located in India. This aerosol plume appears to be attributable to pollution sources rather than soil dust [Reddy and Venkataraman, 2000; Venkataraman et al., 1999]. For example, Kulshrestha et al. [1998] report on high mixing ratios (3500 ppt SO42− and 8300 ppt NH4+) at a suburban site in northern India (Agra, 27°N, 78°E) during the winter time. Trajectory calculations show that the northeast monsoonal low-level flow can transport sulfates, mineral dust, and other aerosols from the Indian subcontinent to the Intertropical Convergence Zone (ITCZ) within 6 to 7 days and that the transport of mineral dust in the middle troposphere from the Arabian desert can reach as far as 4000 km, with a transit time of 2 to 3 days [Krishnamurti et al., 1998].

[4] In order to understand and model the origin and evolution of sulfate aerosol, the most abundant anthropogenic aerosol species [Andreae, 1995; Tegen et al., 1997], it is essential to know the concentration of its precursor species, SO2. Measurements during INDOEX found areas of high pollution over the Indian Ocean north of the ITCZ during the time of the winter monsoon in February and March 1999. Particularly high levels of pollution were observed in a so-called residual layer in the free troposphere up to altitudes of almost 4000 m [Reiner et al., 2001]. These chemical measurements also revealed the simultaneous presence of different tracer substances characteristic for biomass or biofuel burning and for fossil fuel burning. The different tracer substances were well correlated, indicating thorough mixing of contributions from biomass/biofuel sources as well as from fossil fuel sources. The simultaneous occurrence of emissions from biomass burning and fossil fuel burning [Reiner et al., 2001] together with varying amounts of soil dust in the same air masses creates a situation which differs in many respects from conditions observed in other polluted areas of the world.

[5] In this paper we discuss the results of the measurements of aerosol chemistry performed on board the C-130 aircraft based at Hulule airport, Male, Republic of Maldives (4.2°N, 73.5°E) (cf. C-130 overview paper by A. Clarke et al., An overview of the C-130 flight missions and measurements during INDOEX, submitted to Journal of Geophysical Research, 2000) during INDOEX. Our measurements complement the data obtained on the Citation aircraft [Gouw et al., 2001], the research vessel R. H. Brown (R. Dickerson Overview: The cruise of the research vessel Ronald H. Brown during the Indian Ocean experiment (INDOEX) 1999, submitted to Journal of Geophysical Research, 2002), and at the Kaashidhoo Climate Observatory (KCO) (J. Lobert et al., unpublished manuscript, 2000). The results are analyzed in the context of spatial and vertical distribution of the aerosol and air mass origin.

2. Methods

2.1. Sampling

[6] A total of 18 research flights (RF) were performed with the NCAR C-130 aircraft over the Indian Ocean with an altitude range between 30 and 6500 m asl. Here, we report on results of 15 of these flights (RFs 1, 2, 3, 4, 5, 7, 8, 9, 10, 11, 13, 14, 15, 17 and 18), performed between 16 February and 25 March 1999. The duration of these flights ranged between 6 and 8 hours. 72 samples were taken with sampling times ranging between 15 and 45 min in the marine boundary layer (MBL) and between 1 and 3 hours in the free troposphere (FT). A full description of the different flights is available at the UCAR Internet site (http://www.joss.ucar.edu/indoex/catalog/iop/c130/). The samples of submicron aerosol particles were collected using a system consisting of an isokinetic inlet (community aerosol inlet, CAI) directly coupled to two stacked filter units (SFU) arranged in parallel. The design of the CAI reduces distortion of the pressure field at the nozzle tip and the resulting problems associated with flow separation and turbulence. The aerosol sampling system and procedures are based on a design described in detail previously [Andreae et al., 1988a; Huebert et al., 1998]. For aerosol particles smaller than 1 μm aerodynamic diameter (Dp) the CAI has no significant inlet losses. Even in the moist boundary layer the inlet loss for wet submicron particles is less than 10% [Huebert et al., 2000].

[7] Behind the CAI the air stream was split and conducted over one SFU equipped with filters for soluble ion sampling and a parallel SFU equipped with filters for the sampling of carbonaceous aerosols. To separate the submicron aerosol particles (Dp < 1 μm) from the coarse particles (Dp > 1 μm) we used two-stage SFU that contained two sequential 47-mm diameter filters on polyethylene supports. The first stage held a Nuclepore filter (Corning Costar PC Membrane, nominal pore size 8.0 μm). The sampling system was operated at flow rates that averaged 60 L min−1 (at ambient pressure and temperature) measured with thermal mass flowmeters. At the resulting face velocity of ca. 70 cm s−1 the 50% cutoff aerodynamic diameter (Dp) of the 8-μm Nuclepore filter is 1 μm [John et al., 1983]. The submicron aerosol particles that passed the Nuclepore filter were sampled on the second stage using a PTFE Teflon filter (Pall Gelman Zefluor, nominal pore size 2.0 μm) for the ion analysis (IC) or, in a parallel device, on quartz filters. The analysis of the organic carbon (OC) and black carbon (BC) samples is described in detail by Mayol-Bracero et al. [2002]. Blanks for all filter types were obtained by placing filters in the sampling units in the aircraft and exposing them for a few seconds. For every flight at least 3 blanks were taken in this way and analyzed. The averaged blank values for each species were subtracted from the sample concentrations. For calculation of the uncertainties, the blanks of all flights were averaged. Twice the standard deviation of the blanks together with the analytical error of the IC was used for the error calculation and is given as error bars in the figures.

[8] The coarse particles (between ∼1 μm, the cut-size of the filter pack used, and the cutoff of the CAI of about 3.5 μm) sampled on the Nuclepore filters were analyzed for sodium to obtain a sea-salt tracer for the MBL definition. Since the sampling efficiency of the CAI is much poorer for larger aerosol particles (Dp > 1 μm) [Huebert et al., 2000] we use relative values of coarse sodium.

2.2. Filter Handling and Analysis

[9] The loaded filters were transferred to 30-mL HDPE bottles in the field lab immediately after each flight and stored in a refrigerator until extraction. The Zefluor filters were first wetted with 0.5 mL of methanol. We used 4.5 mL Elga-water (R > 18 MΩ) for the Zefluor filters and 3 mL Elga-water for Nuclepore filters for the extraction of soluble aerosol species. After addition of the extraction solutions, all samples were shaken vigorously for about fifteen minutes.

[10] The first set of samples was analyzed during the campaign in a field laboratory at Hulule airport (Male, Maldives), the rest at the Max Planck Institute for Chemistry (MPIC), Mainz. Anions were determined using a Shimadzu HIC-6A ion chromatograph (IC) with an AS11 (DIONEX) column and ASRS-ultra suppressor (DIONEX) in external water mode. The temperature of the system was 35°C, the flow rate of the eluent (NaOH, using a gradient mode) was 1 mL min−1. A plateau of water at the beginning of the gradient separates the peak produced by the methanol in the sample solution from the aerosol ion peaks. A plateau of high NaOH concentration at the end of the gradient program ensured regeneration of the ion exchange capacity of the column after each analysis. For the cation analysis we used a CS14 (DIONEX) column, an eluent flow rate of 1.2 mL min−1 (8 mM methanesulfonate, isocratic mode), and a CSRS-1 suppressor (DIONEX) in external water mode. The detection limits in standard solutions ranged between 0.09 and 0.13 μmol L−1 for cations and between 0.03 and 0.12 μmol L−1 for anions. Aqueous standards were prepared from high-purity reagents with deionized Elga-water (R > 18 MΩ). We added 10% methanol to the standards to take possible matrix phenomena due to the methanol content of the sample into account for the calibration. All concentrations are reported in molar mixing ratios in air (pmol of analyte per mole of air, parts per trillion (ppt)).

2.3. Air Mass Trajectories

[11] Seven-day back trajectories were computed using the HYSPLIT-4 (HYbrid Single-Particle Lagrangian Integrated Trajectory) model [Draxler and Hess, 1997] and the FNL meteorological data set produced by the U.S. National Climatic Data Center (NCDC). The vertical velocities from the FNL data set were used to derive the vertical transport component in the trajectory calculations. Comparison between the trajectories obtained with HYSPLIT-4 and those provided by the Dutch Weather Service (KNMI) during the INDOEX campaign showed good agreement [Verver et al., 2002].

3. Results and Discussion

3.1. Meteorological Setting and Sampling Strategies

[12] The meteorological environment of INDOEX has been discussed in detail by Verver et al. [2002] and Rasch et al. [2001], and only a brief overview will be given here. Specific information on the meteorological situation during the C-130 flights, as well as flight track plots can be found in the C-130 overview paper by Clarke et al., unpublished manuscript, 2000.

[13] The meteorological situation in the region near the Maldives and during the period of the INDOEX intensive field campaign is determined by atmospheric conditions that are crucial for meeting the INDOEX objectives. The atmosphere north of the ITCZ is characterized by continental outflow from the Indian subcontinent and South or Southeast Asia. Northeasterly trade winds are persistent while convection over the continental source regions is suppressed by large-scale subsidence, thus limiting upward dispersion of pollution. Fossil fuel combustion and biomass burning cause large aerosol loadings [Lelieveld et al., 2001; Reiner et al., 2001]. Boundary layer trajectories for this experiment suggest that emissions in India were responsible for most of the pollutant input into this air mass. The three major sources for anthropogenic aerosol to the INDOEX region are (1) a near-surface southward flow near Bombay, (2) a deeper strong plume flowing south and east off Calcutta, and (3) a westward flow originating from Southeast Asia [Rasch et al., 2001]. South of the ITCZ, pristine air from the remote southern Indian Ocean prevails. The effects of continental outflow in the northern Indian Ocean could therefore be contrasted with very clean air from the southern Indian Ocean.

[14] We classified our samples according to 7-day air mass back trajectories terminated at the aircraft location into four types: (a) clean marine air that did not make contact with land over the past 7 days originating in the southern Indian Ocean and sampled south of the equator (SIO); (b) air that had traveled from the Arabian Peninsula across the Arabian sea (ARABIA); (c) air masses with trajectories over the Indian continent, mainly from Pakistan and north India but also air masses which had come across India (INDIA); and (d) air that contained fresh pollution from the Bay of Bengal and/or had traveled from Southeast Asia and was sampled north of the equator (SEA).

[15] For the statistical analysis of our results, we further categorized our samples into vertical layers. We divided the pollution layer into two sublayers: (1) the well-mixed marine boundary layer between the sea surface and the temperature inversion (MBL), and (2) the residual layer between the temperature inversion and the trade inversion around 3500 m (RL). Near the coast, this polluted layer was present between the newly formed polluted MBL and the clean free troposphere (FT). This layer represented a remnant of the continental boundary layer (CBL), formed by more intensive convection over land and preserved as a stable layer above the MBL. The analysis of the ambient temperature and dew point temperature measurements of the C-130 during vertical profiles flown between 30 and 5000 m was used to define the height of the MBL over the Indian Ocean between 800 and 1200 m. Since aerosol sampling was performed during straight level runs, and soundings were not performed during every flight, the concentration of sodium in the coarse fraction was used as sea-salt tracer to determine whether samples had been collected in or above the MBL. The averaged value of coarse sodium above 1200 m was only 10% of the averaged value of coarse sodium below 1200 m. Only samples with concentration lower than 25% (average + standard deviation) of coarse sodium in the MBL (below 1200 m) were assigned to the RL category. The measurements in the MBL were compared with values of other INDOEX platforms. The comparison between aircraft and ship-based as well as ground-based samples collected from the same or similar air masses (air masses with scattering values > 55 Mm−1) showed good agreement (better than 35%) for the fine fraction of the aerosol mass and the main components NH4+ and nss-SO42− (for more details of the intercomparison, see Clarke et al. [2002]).

3.2. Measurements in the Marine Boundary Layer

3.2.1. MBL Composition as a Function of Latitude

[16] Figure 1 shows the latitudinal distribution of fine aerosol in the MBL over the Indian Ocean. In the northern part of the Indian Ocean study region (5°–15°N, 66°–73°E), high concentrations of carbonaceous aerosol and pollution-derived inorganic species were found. The mass concentration of all fine aerosol species detected by our technique reaches values up to 28 μg m−3, comparable to pollution levels observed at the coast of industrialized regions during TARFOX [Hegg et al., 1997] or ACE-2 [Andreae et al., 2000] and in the range of the values (between 13 and 22 μg m−3) of anthropogenic fine aerosols transported from Europe to the Near-East region [Andreae et al., 2002]. As can be seen from the ions/(BC+POM) ratio in Figure 1, carbonaceous aerosol contributes between 40% and 60% to the fine aerosol mass of all determined components, and samples dominated by water soluble ions can be found only over the northern Indian Ocean. Backward trajectory analyses indicate that air masses sampled in the region mainly originate over northern and central India as well as in the Arabian Peninsula (see Table 1). In the Southern Hemisphere (1°–9°S, 66°–73°E), the aerosol concentrations rapidly declined to background levels of about 2 μg m−3. Such a decline of the aerosol mass in the MBL over the eastern Arabian Sea with decreasing latitude was also measured during the r/v Sagar Kanya winter cruise 1996 [Jayaraman et al., 1998].

Figure 1.

Latitudinal cross section of the mass concentration of submicron atmospheric aerosols (± 10%) collected in the marine boundary layer over the Indian Ocean between 66°E and 73°E during INDOEX 1999. Latitude is given in degrees N and S.

Table 1. Fine Aerosol Species (Dp < 1 μm) in the MBL Samples with Different Air Mass Origins (Averages and Standard Deviations, in ppt). nd: Not Detectable, N: Number of Samples
 ARABIAINDIASEASIO
N620137
  • a

    The averaged mass of all measured ionic and carbonaceous aerosols (BC + POM) with a limited number of TC measurements (Arabia, n = 3, India, n = 16, SEA, n = 8). The value of 1.7 was used to convert OC to POM [Mayol-Bracero et al., 2002].

Na+77 ± 41109 ± 11661 ± 103nd
NH4+886 ± 3042160 ± 8951740 ± 932125 ± 57
K+61 ± 43195 ± 83172 ± 113nd
K+ex46 ± 42176 ± 81154 ± 104nd
Mg2+nd6 ± 10ndnd
Ca2+15 ± 1220 ± 2117 ± 22nd
Cl45 ± 4329 ± 3817 ± 31nd
NO363 ± 1941 ± 2933 ± 30nd
SO42−576 ± 2101340 ± 4961070 ± 46161 ± 31
nss-SO42−571 ± 2101330 ± 4951060 ± 45660 ± 31
MSAnd8 ± 35ndnd
Oxalatend25 ± 1623 ± 16nd
NH4+/SO42−1.581.671.632.26
K+ex/nss-SO42−0.070.130.15nd
cations/anions (eq/eq)0.840.920.880.96
Soluble ions (μg m−3)3.7 ± 1.28.3 ± 3.06.6 ± 2.90.5 ± 0.2
BC (μg m−3)0.7 ± 0.82.4 ± 1.41.8 ± 1.2nd
POM (μg m−3)2.8 ± 2.25.1 ± 3.44.7 ± 3.1nd
Ions/(BC + POM)1.11.11.0nd
Total mass (μg m−3)a7.2 ± 4.215.8 ± 7.813.1 ± 7.2nd

[17] Figures 2a and 2b show the latitudinal distribution of fine non-sea-salt sulfate (nss-SO42−) and NH4+. To calculate the amount of nss-SO42−, the fine sodium concentrations from the same filter sample were multiplied with the SO42−/Na+ ratio in bulk seawater, and this value was subtracted from the SO42− fine fraction [Savoie et al., 1987]. Ammonium and sulfate made up most of the inorganic fraction of the fine aerosol (see Table 1). They were present at a molar ratio of 1.6:1, corresponding to a mixture of ammonium bisulfate and ammonium sulfate. The concentrations of non-sea-salt sulfate ranged from 2250 ppt (10 μg m−3) in the Northern Hemisphere to values near the detection limit of 10 ppt (around 0.2 μg m−3) in the Southern Hemisphere. Ammonium declined from 4400 ppt in the north to 50 ppt in the south (Figure 2b). A comparable gradient of NH4+ and nss-SO42− bulk aerosol in the MBL over the eastern Arabian Sea was revealed during the Sagar Kanya winter cruise 1996 with highest values of NH4+ (7300 ppt) and nss-SO42− (1700 ppt) off the Indian coast at 75°E and 15°N [Krishnamurti et al., 1998]. Measurements of submicron aerosol particles on the r/v R.H. Brown during INDOEX 1999 also show this gradient with low values of 90 ppt SO42− and 50 ppt NH4+ south of the ITCZ and high values up to 1300 ppt SO42− and 1700 ppt NH4+ north of the ITCZ, respectively (Norman, personal communication, 2000). High values of potassium up to 350 ppt found in the North indicate a contribution of biomass burning to the aerosol in this region (Figure 2c), since excess particulate phase potassium (present as K+ in the submicron fraction of the aerosol) is considered as a tracer for biomass burning [Andreae, 1983; Andreae et al., 1998; Andreae et al., 1996; Andreae et al., 1988b]. The low potassium values of 5 ppt (near the detection limit) over the southern Indian Ocean indicate the low interhemispheric air exchange over the Intertropical Convergence Zone (ITCZ) in the MBL. The decline of NO3 in the fine fraction (Figure 2d) in the south is less pronounced, indicating a source for NO3 in the Southern Hemisphere possibly via downward transport of NOy from the upper troposphere or stratosphere or production by lightning [Rhoads et al., 1997]; the latter might be especially important in the ITCZ region. Overall, the concentrations of pollutant aerosol species in the MBL are seen to decrease with increasing transport times and distances between pollution sources and the sampling site, and decline rapidly in the region of the ITCZ in the Southern Hemisphere.

Figure 2.

Latitudinal cross section of the concentration of (a) nss-SO42−, (b) NH4+, (c) K+, and (d) NO3 in submicron atmospheric aerosols collected in the marine boundary layer over the Indian Ocean between 66°E and 73°E during INDOEX 1999. Latitude is given in degrees N and S. Error bars shown indicate the total error of the measurement calculated with twice the standard deviation for the blanks together with the analytical error.

3.2.2. MBL Composition as a Function of Air Mass Origin

[18] Table 1 shows the aerosol species in the MBL samples categorized by their different air mass origins. The lowest concentration for all species was found in the samples from the southern Indian Ocean (SIO). The total nss-SO42− concentrations of 60 ppt are in the range of unpolluted air over the North Atlantic [Andreae et al., 2000] or summer levels from the cleanest temperate sites, such as from Cape Grim [Andreae et al., 1999; Ayers et al., 1991; Huebert et al., 1998], and from remote islands in the South Pacific [Savoie and Prospero, 1989]. Lower nss-SO42− values in the MBL (between 16 and 53 ppt) had been found over the Southern Ocean south of Australia [Berresheim et al., 1990]. In sharp contrast to the very clean SIO air masses are the samples collected in the MBL influenced by the Indian pollution plume (INDIA) and by pollution from Southeast Asia (SEA) with an averaged total aerosol mass of 16 and 13 μg m−3, respectively. For these samples, sulfate and ammonium concentrations are more than one order of magnitude above the background values found for the SIO samples and comparable to values for the European plume found over the North Atlantic during ACE-2 [Andreae et al., 2000]. Ammonium and sulfate were present in the MBL at a molar ratio of 1.6:1, corresponding to aerosols with a composition between ammonium bisulfate and ammonium sulfate. We also found in the INDIA and SEA samples very low fine NO3 concentrations of 33–41 ppt, which is in the range (around 40–50 ppt) measured at remote island sites in the South Pacific [Huebert et al., 1998; Savoie and Prospero, 1989] and is comparable to values from island sites in the remote North Pacific [Savoie and Prospero, 1989] and to the background nitrate level at Barbados [Savoie et al., 1989]. However, the analysis of the Nuclepore filters indicates, that most of the nitrate is present in the coarse fraction of the aerosols. According to a recent overview paper on the nitrogen budget for the Arabian Sea [Bange et al., 2000] values of bulk aerosol NO3 over the Arabian Sea range between 155 and 1800 ppt. The lowest values of NO3 reported from the region (155 ppt) were measured by Savoie et al. [1987] over the northwestern Indian Ocean during periods of low dust load.

[19] Compared to air masses originating over the Arabian peninsula, the high values of excess potassium for the INDIA and SEA samples point to a higher contribution of biomass burning in the Indian pollution plume. To estimate the amount of excess K+, the fine Ca2+ concentrations were multiplied with 0.99 (the molar K+/Ca2+ ratio in seawater and soil), and this value was subtracted from the K+ fine fraction. The average value of 176 ppt fine excess potassium (K+ex) reflects a significant contribution of biomass burning emissions to the aerosol burden originating in India (see Table 1). This is consistent with the findings of Reiner et al., 2001, who found high concentrations of biomass burning tracers in pollution plumes over the Arabian sea during INDOEX. The samples collected in air masses originating over the Arabian peninsula (ARABIA) show a lower aerosol burden. The transit over the Arabian sea to the sampling area may lead to a depletion due to deposition processes, but there is still an obvious difference in the K+ and nss-SO42− concentration detectable. The much lower K+ex/nss-SO42− ratio of 0.07 in aerosol from ARABIA is indicative of a stronger influence of fossil fuel than biomass burning, as compared to K+ex/nss-SO42− ratios of 0.13 and 0.15 for the Indian pollution plume and air masses from Southeast Asia, respectively.

[20] The pollution levels found over the Arabian Sea during INDOEX are in the range of those measured during ACE-2 over the North Atlantic, where fine nss-SO42− reached 2070 ppt [Andreae et al., 2000], comparable to the highest MBL values (2250 ppt) during INDOEX. The neutralization ratio between the anionic and the cationic charges (cations/anions) for the MBL is in the range of 0.84 to 0.92. Since the discrepancy between the measured anionic and cationic charges is most likely due to species that are undetectable by the IC, such as H+, this points to a slightly acid aerosol. Overall, our results from polluted air masses in the MBL over the Indian Ocean indicate a high degree of air pollution in the region, with comparable amounts of soluble ions and carbonaceous species. They are in good agreement with previous observations during ship-based measurements [Krishnamurti et al., 1998; Rhoads et al., 1997].

3.3. Residual continental boundary layers (RL) composition as a function of air mass origin

[21] The typical vertical distribution of the aerosols over the Indian Ocean during INDOEX is shown in Figure 3 for nss-SO42−. High concentrations were found in a layer extending from the sea surface to about 3.5 km. This layer can be divided into the MBL and the residual CBL (RL). When continental air masses are advected over the ocean, especially over relatively cold water, the weaker convective activity over water cannot sustain vertical mixing to the same height as over land. Consequently, a new, shallower MBL evolves at the base of the CBL; and the residual CBL (RL) between the top of this new MBL and the inversion at the top of the CBL becomes disconnected both from the FT and the sea surface [Johnson et al., 2000]. The higher temperature of the Indian subcontinent during March [Venkataraman et al., 1999] could lead to an increase of this phenomenon in March. As can be seen in Figure 3 we found the highest value for nss-SO42− in the RL with a concentration around 3000 ppt measured on 16 March 1999 west of India near 15°N and 71°E. Above 3500 m the concentration of nss-SO42− in the aerosol declined rapidly to values of ca. 20 ppt, typical for the free troposphere. The RL air masses are of particular interest in the context of INDOEX, as they represent samples of the Asian plume that have not interacted to a measurable degree with the marine atmosphere. Evidence for this is the low amount of sea-salt elements in coarse and fine particles. The elevated fine NO3 concentration in the RL of Indian air masses compared to the MBL is consistent with the absence of sea salt, which can act as a NO3 sink (Tables 1 and 2).

Figure 3.

Vertical distribution of nss-SO42− in submicron atmospheric aerosols collected over the Indian Ocean between 66°E and 73°E during INDOEX in the time between 16 February and 19 March 1999. Empty circles: samples taken in the RL and FT (coarse sodium < 25% of MBL average). Filled squares: samples taken in the MBL (coarse sodium > 25% of MBL average). Error bars shown indicate the total error of the measurement calculated with twice the standard deviation for the blanks together with the analytical error.

Table 2. Soluble Ionic Aerosol Species in the Submicron Aerosol from the RL with Different Air Mass Origins (Averages and Standard Deviations, in ppt). nd: not detectable, N: number of samples
 INDIASEA
N44
  • a

    The averaged mass of all measured ionic and carbonaceous aerosols (BC + POM) with a limited number of TC measurements (Arabia, n = 2; SEA, n = 2). The value of 1.7 was used to convert OC to POM [Mayol-Bracero et al., 2002].

Na+23 ± 31nd
NH4+3230 ± 12511950 ± 1773
K+347 ± 158146 ± 108
K+ex338 ± 142140 ± 116
Mg2+ndnd
Ca2+9 ± 18nd
Clndnd
NO3185 ± 9931 ± 49
SO42−1670 ± 8691010 ± 760
nss-SO42−1670 ± 8691010 ± 760
MSAndnd
Oxalate71 ± 3832 ± 29
NH4+/SO42−2.01.9
K+ex/nss-SO42−0.210.14
cations/anions (eq/eq)1.00.97
Total mass (μg m−3)a34.0 ± 10.722.4 ± 10.4

[22] We found the highest aerosol mass in the RL in air masses coming from India with an average value of 34 μg m−3(Table 2). The RL values of the total mass, NH4+, NO3, nss-SO42− and K+ are above the MBL values. This is indicative of a strong impact of the combustion of fossil fuel as well as biomass to this polluted layer, combined with a reduced deposition rate compared to the MBL. The high K+ex concentration and the K+ex/nss-SO42− ratio of 0.21 point to considerable biomass burning contribution to the Indian pollution plume. The RL was also detected by Reiner et al. [2001] between 1000 and 3500 m over the northern Indian Ocean (7.1°–8.3°N and 69.7°–72.1°E) and was more frequently observed at the end of the winter monsoon period during March 1999. Altitude profiles show maximum concentrations of acetone, acetonitrile, SO2, and CO volume mixing ratios in the RL [Reiner et al., 2001]. During 16 March 1999 we performed a slowly flown vertical gradient which enabled the sampling of aerosols (4 samples at different altitudes) in parallel to the gas phase measurements. The good correlation of acetone, acetonitrile, fine K+ and fine SO42− with CO (r2 > 0.9) in all cases measured during this vertical profile flight between the MBL and the FT across the polluted RL west of India (near 15°N and 71°E) is indicative for combustion-derived aerosol [Reiner et al., 2001]. Ammonium and sulfate made up most of the inorganic fraction of the fine aerosol in the RL. They were present at a molar ratio of 2:1, corresponding to ammonium sulfate. High ammonia loadings appear to neutralize the aerosol in the RL completely, which is reflected in a cation/anion ratio of 1. Since Eastern China and India belong to the strongest NH3 source regions of the world [Adams et al., 1999; Bouwman et al., 1997] their NH3 emissions are sufficient to neutralize the acidic aerosols.

3.4. Measurements in the Free Troposphere

[23] During most of the INDOEX flights, samples were collected from the free troposphere (FT) at altitudes between 3500 and 7000 m, often during the transit to or from a study area. Sampling times ranged between ca. 0.5 and 3 hours, to provide enough sample mass at the low aerosol concentrations typical of the FT. Most of the FT air masses sampled came from Southeast Asia at high altitudes. Only 4 samples were taken in air masses originating over Arabia and having crossed Africa within the preceding 10 days. On the basis of the data in Table 3 (ignoring the contribution of species below the detection limit), we estimate FT fine aerosol mass concentrations between 2.9 and 1.0 μg m−3 in the fine fractions for ARABIA and SEA air masses, respectively.

Table 3. Soluble Ionic Aerosol Species in the Submicron Aerosol from the FT with Different Air Mass Origins (Averages and Standard Deviations)a
 ARABIASEA
N414
  • a

    Here, nd: not detectable, N: number of samples. Averages and standard deviations are in ppt.

  • b

    The averaged mass of all measured ionic and carbonaceous aerosols (BC + POM) with a limited number of TC measurements (Arabia, n = 1; SEA, n = 4). The value of 1.7 was used to convert OC to POM [Mayol-Bracero et al., 2002].

Na+43 ± 52nd
NH4+28 ± 1963 ± 72
K+nd14 ± 19
Mg2+ndnd
Ca2+ndnd
Clndnd
NO3ndnd
SO42−27 ± 1217 ± 9
nss-SO42−24 ± 1117 ± 9
MSA19 ± 23nd
Oxalatendnd
NH4+/SO42−1.37nd
cations/anions (eq/eq)1.01.57
Total mass (μg m−3)a2.91.0 ± 0.3

[24] Our FT observations during INDOEX are in good agreement with previous studies in the FT over other marine regions. For comparison, the concentrations measured in the FT of the Southern Hemisphere around Tasmania during ACE-1 were 59 ppt NH4+, 17 ppt NO3, and 48 ppt nss-SO42− [Huebert et al., 1998]. Berresheim et al. [1990] found values of nss-SO42− in the FT over the Southern Ocean south of Australia between 14 and 45 ppt. During ACE-2 values of 107 ppt NH4+, 25 ppt NO3, and 40 ppt nss-SO42− were found over the North Atlantic [Andreae et al., 2000]. In conclusion, we find that our values of 17–24 ppt nss-SO42− and 28–63 ppt ammonium are quite typical for a FT with little or no detectable anthropogenic influence. There is no indication for a strong upward transport of aerosols into the FT owing to the convective processes in the ITCZ up to 7 km. Nevertheless, vertical profiles of CO and O3 mixing ratios measured with the Citation aircraft during INDOEX indicate some interhemispheric exchange, leading to elevated levels of pollutants in the 8–10 km altitude range of the Southern Hemisphere [Williams et al., 2002].

3.5. Characteristic Composition of the Aerosol from Different Source Regions

[25] The composition and amount of aerosol over the northern Indian Ocean varies according to air mass origin and altitude and is strongly affected by both inorganic and organic pollutants, including black carbon (soot). The MBL aerosol in the air masses originating over the Arabian Peninsula contains a total averaged mass concentration of all aerosol species detected by our techniques (consisting mainly of inorganic species, with 34% SO42−) of 7.2 μg m−3. The high contribution of sulfate and nitrate in combination with the low K+ex/nss-SO42− ratio of 0.07 points to dominance of fossil fuel pollution in the source region (Figure 4a).

Figure 4.

Mean composition and mass concentration of submicron atmospheric aerosols over the Indian Ocean between 66°E and 73°E during INDOEX 1999 collected in the marine boundary layer (MBL) in air masses originating over (a) the Arabian Peninsula or (b) northern and central India and (c) collected in the residual layer (RL) in air masses originating over northern and central India.

[26] Back trajectories indicate that India and Southeast Asia are the prevailing source regions for aerosols investigated during INDOEX 1999. In the air masses originating over India, we found an aerosol burden that is twice as high as in the Arabian air mass, with a total carbon (TC) content of 49% and a sulfate content of 36% (Figure 4b). Although there is a comparable composition of the MBL aerosol for sulfate, ammonium, and TC in the Arabian and Indian air masses (Figures 4a and 4b), the elevated K+ concentration in the Indian aerosol points to a higher biomass burning contribution. This could be explained with strong biofuel use and some open vegetation fires in India. For example, Kulshrestha et al. [1998] found high values of K+ in the fine aerosol fraction of the air sampled in the north Indian city Agra. This high K+ concentration was attributed to the fact that about 40% of the population of Agra uses wood as a domestic fuel by the authors. The trace gas measurements on the C-130 aircraft during INDOEX [Reiner et al., 2001] and emission inventories for India Streets and Waldhoff [1999] also support the biomass burning contribution to the Indian aerosol plume. In the air coming from the Arabian peninsula the correlation between the biomass burning tracer acetonitrile and CO is much weaker than in the pollution layers originating over India, and the dCH3CN/dCO values are smaller. However, based on emission inventories for India, fossil fuel combustion accounted for 70 ± 10% of the combustion aerosol emissions [Reddy and Venkataraman, 2000; Lelieveld et al., 2001]. The analysis of the ratios between BC, TC, OC, SO42− and K+ in the fine aerosol sampled on the C-130 during INDOEX suggested that the contribution of fossil fuel burning (by transportation, industrial, and domestic sectors) may be as high as 80% of the total [Novakov et al., 2000]. Comparable aerosol compositions in the MBL were obtained from KCO and from the r/v R.H. Brown, indicating that the aerosol composition was quite uniform over the northern Indian Ocean. The measurements at KCO show that the fine aerosol (dry mass at Dp < 1 μm) was typically characterized by 32% sulfate, 26% organic matter, 14% black carbon (BC), 10% mineral dust, 8% ammonium, 5% fly ash, 2% potassium, 1% sea salt and trace amounts of methyl sulfonic acid (MSA), nitrate, and minor insoluble species [Lelieveld et al., 2001]. Since we were not able to determine the insoluble part of the fly ash and mineral dust, our values for the total mass are slightly underestimated. However, the low amounts of Mg2+ and Ca2+ in the fine aerosol indicate that dust is a minor part of the aerosol in our samples. This is consistent with a low abundance of nonvolatile aerosol observed by Clarke et al. [2002]. Dust storm statistics suggest that there are major sources of mineral dust in northern India, but that there is very little dust deflation in southern India below 15°N at any time of year [Ackerman and Cox, 1989; Ginoux et al., 2001]. During March 1999, the Arabian Peninsula was the main source of dust in the INDOEX region, but air mass trajectories indicate that dust could be advected in a southwesterly direction only on days 5, 15 and 17 of March (Cautenet et al., submitted manuscript, 2002).

[27] Compared to the MBL measurements, the analysis of polluted RL aerosol originating in India shows a relatively high average mass concentration of 34 μg m−3 (Figure 4c) with a strong contribution of carbonaceous aerosol. We found a considerably elevated amount of particulate organic matter (POM) in the RL, which could be explained with a long residence time of carbon-containing compounds in this layer and an ongoing formation of POM due to oxidation and condensation of organic compounds.

[28] The major inorganic species in the RL was SO42− (27%). The lower contribution of nss-SO42− to the total aerosol mass in the RL as compared to the MBL may be related to less efficient oxidation of SO2 in the RL. In order to understand and evaluate the origin and evolution of sulfate aerosol it is essential to know the concentration of its precursor species, SO2. Values of SO2 measured on the C-130 over the northern Indian Ocean during INDOEX ranged between 170 and 1500 ppt with highest values in the RL and lowest values in the FT and the MBL [Reiner et al., 2001] (Sprung, personal communication, 2000). This indicates that SO2 is not completely oxidized when air masses from India are advected over the Indian Ocean in the RL. The average value of nss-SO42− for the RL is 1670 ppt, with maximum values up to 3000 ppt. Even 500 km offshore the southwestern Indian coast there are considerable concentrations of SO2 measurable, which are in the range of polluted continental air. In contrast to the high SO2 concentrations in the RL, the measurements on the r/v R.H. Brown during INDOEX 1999 show low SO2 MBL concentrations (10–70 ppt) north of the ITCZ (Norman, personal communication, 2000). Comparison between particulate and gas-phase sulfur in the ship-based measurements showed that SO2 accounts for less than 4% to the total sulfur in the MBL (Norman, personal communication, 2000). This can be explained with a higher oxidation rate of SO2 in the liquid phase of deliquescent sea-salt aerosol and in the shallow cumulus clouds capping the MBL. Since there are only minor concentrations of MSA detectable in the fine aerosol fraction (Table 1 and Table 2), we can exclude the emission of dimethyl sulfide (DMS) as an important additional source for SO2 and sulfate over the Indian Ocean during INDOEX. The typical nss-SO42− concentrations between 20 and 100 ppt in the MBL measured over clean ocean sites [Andreae et al., 2000; Savoie and Prospero, 1989; Berresheim et al., 1990] are one order of magnitude below the values found in the MBL over the northern Indian Ocean (see Table 1). This confirms the minor importance of biologically derived nss-SO42− in the INDOEX region and indicates the strong influence of the Asian pollution plume during the winter monsoon. According to the EDGAR (Emission Database for Global Atmospheric Research) emission inventory, the SO2 production in India is mainly located along the western coast, with a maximum at Bombay (19°N), and another maximum at the southern tip of India; most eastern coast emissions are located between Madras (13°N) and Calcutta (23°N). Recent estimations of the total emission of sulfur for India range between 2.0 and 2.5 Tg S(SO2) a−1 [Streets and Waldhoff, 1998; Venkataraman et al., 1999; Lelieveld et al., 2001]. They all agree that the vast majority (77–93%) of sulfur emissions are caused by fossil fuel combustion (coal and petroleum) and industrial activities. Ratios of dSO2/dCO of 7.8 × 10−3, and dSO42−/dCO of 16 × 10−3 obtained from CO, SO2 and aerosol sulfate values measured in the polluted residual layer during INDOEX 1999 [Reiner et al., 2001], support this conclusion since values from field measurements in biomass burning plumes gave much lower ratios of dSO2/dCO in the range 1.7–5.2 × 10−3 (M. O. Andreae and P. Merlet, Emission of trace gases and aerosols from biomass burning, submitted to Journal of Geophysical Research, 2001) and d(SO2+SO42−)/dCO values of 2.4–6.5 × 10−3 [Andreae et al., 1988b].

[29] A strong pollution source for SO2 in combination with low deposition and/or fast transport seems to be responsible for the high nss-SO42− values. Since clouds are much more prevalent over ocean than land during the winter monsoon, SO2 is transported in the cloud-free regions over land and is undergoing relative slow gas-phase transformation to sulfate. Assuming a 24-hour average OH concentration of 2 × 106 cm−3, the reaction of OH with SO2 leads to an SO2 lifetime of about 6 days, which is of similar magnitude as the transport time from India to the measurement locations over the Indian Ocean. As SO2 encounters clouds capping the MBL over the Indian Ocean, it is more rapidly transformed to sulfate in the liquid phase in cloud droplets or in deliquescent sea-salt aerosol. Many of the clouds during INDOEX were nonprecipitating tradewind cumuli, which are not efficient at removing aerosol by precipitation scavenging. Therefore, in the INDOEX study region the residence time of sulfate aerosol can reach 7 days [Rasch et al., 2001], which is still above the values (between 4 and 5 days for February and March, respectively) estimated for India [Venkataraman et al., 1999]. The high emissions of SO2 at the Indian coast in combination with the low wet and dry deposition of sulfate and SO2, especially in the RL, and the continuous low-level offshore airflow with a measured wind speed around 6–7 m s−1 lead to the widespread high nss-SO42− aerosol concentrations in the aerosol layer observed over the northern Indian Ocean.

[30] Overall, the aerosol concentrations measured over the Indian Ocean are comparable to urban air pollution in North America and Europe, with a high contribution of sulfate and POM. Moreover, the high BC concentration gives the aerosol a strong light absorbing character.

4. Summary and Conclusions

[31] During the winter monsoon in February and March 1999 we found areas of high aerosol pollution over the Indian Ocean north of the ITCZ region up to altitudes of almost 3500 m. The highest aerosol concentration was found in a residual layer originating from the Indian continental boundary layer. Trajectory analyses verified that the air in the boundary layer north of the ITCZ was strongly influenced by polluted air masses originating over India or Southeast Asia as well as the Arabian peninsula. The chemical characterization of the aerosol showed the simultaneous presence of different tracer substances characteristic for biomass or biofuel burning (excess potassium) and for fossil fuel burning (nss-SO42− and nitrate). The K+ex/nss-SO42− ratio of 0.13 in the MBL and 0.21 in the RL for India and 0.15 for Southeast Asia, respectively, is closer to the ratio for aerosols from biomass burning than the ratio found in fossil fuel combustion. Because of the high impact of anthropogenic pollutants over the northern Indian Ocean, biogenic nss-SO42− that was oxidized from DMS makes a very minor contribution to the total aerosol atmospheric loading. In contrast, during the SW-monsoon, nss-SO42− of biogenic origin can reach values up to 75% of the total nss-SO42− [Johansen et al., 1999].

[32] The interhemispheric transect measurements suggest that the ITCZ serves as an efficient isolator in blocking the transport of pollutants from the Indian subcontinent to the Southern Hemisphere. The differences in the aerosol concentrations between the northern and the southern Indian Ocean are much larger during the northeast monsoon than the concentration differences found during the monsoonal transition [Rhoads et al., 1997]. Acetone and acetonitrile mixing ratios measured during the C-130 flights with two different mass spectrometric techniques also displayed a strong north–south gradient in the boundary layer [Sprung et al., 2001]. However, interhemispheric transport could be affected by eddies that wrap around the ITCZ; these eddies bring clean Southern Hemisphere air to about 10°N in the Indian Ocean and carry polluted continental air into the Southern Hemisphere. The rapid changes in the ITCZ and the steep northwest–southeast slope of the ITCZ could also lead to a significant interhemispheric transport during the northeast monsoon [Krishnamurti et al., 1998]. Nevertheless, back trajectory analysis for our INDOEX samples reveals that air masses with low aerosol concentration north of the ITCZ region mainly originate over the western Arabian Sea and not south of the equator.

[33] The most important finding of our INDOEX measurements was the large loading of aerosols that scatter and absorb sunlight. Averaged over the northern Indian Ocean, the aerosol burden detected during INDOEX can reduce the winter-time solar heating of the ocean by 29 W m−2 and enhance the atmospheric solar heating by 18 W m−2 (V. Ramanathan et al., unpublished manuscript, 2000). This radiative effect of the pollution-derived aerosols over the Indian Ocean can influence the hydrological cycle, the monsoonal circulation, cloudiness, and possibly ecosystem functioning. Our measurements indicate that the strength of these climatic effects depends on the characteristic composition of the aerosols. The aerosols originating over the Arabian Peninsula have a more scattering character whereas the aerosols originating over India have a stronger absorbing character due to their high BC content. Warming or cooling effects may therefore be influenced by the prevailing source regions, depending on the meteorological conditions. In concert with the rapid economic development on the Indian subcontinent and Southeast Asia we might expect aerosol concentrations over the Indian Ocean and the associated effects to increase in the future.

Acknowledgments

[34] We thank all participants of the INDOEX project, in particular the C-130 team and the staffs of NCAR-RAF and the MPIC mechanical workshop for technical help and Carol Strametz for editorial support. We gratefully acknowledge the government and people of the Republic of Maldives for their hospitality. The National Center for Atmospheric Research is sponsored by the National Science Foundation and managed by the University Corporation for Atmospheric Research. The MPIC part of this study was funded by the German Max Planck Society, with additional support for the flight operations by the U.S. National Science foundation.

Ancillary