A proton-transfer-reaction mass spectrometer (PTR-MS) was used for fast-response measurements of volatile organic compounds (VOCs) onboard the NOAA research vessel Ronald H. Brown during leg 2 (4 March–23 March) of the INDOEX 1999 cruise. In this paper, we present a first overview of the distribution of acetonitrile, methanol, acetone, and acetaldehyde over a broad spatial extent of the Indian Ocean (19°N–13°S, 67°E–75°E). The prevailing atmospheric circulation during the winter monsoon transported polluted air from India and the Middle East over the Indian Ocean to meet pristine southern hemispheric air at the intertropical convergence zone (ITCZ). The chemical composition of air parcels changed according to their geographic origin, which was traced by backtrajectory analysis. The relative abundance of acetonitrile, a selective tracer for biomass burning, to that of carbon monoxide, a general tracer for incomplete combustion, reflected the signature of biomass burning or fossil fuel combustion. This indicated a strong biomass burning impact in W-India, mixed pollution sources in NE-India, and the dominance of fossil fuel combustion in the Middle East. Biomass burning impacted air was rich in methanol (0.70–1.60 ppbv), while acetone (0.80–2.40 ppbv) and acetaldehyde (0.25–0.50 ppbv) were elevated in all continental air masses. Pollution levels decreased toward the ITCZ resulting in minima for methanol, acetone, and acetaldehyde of 0.50, 0.45, and 0.12 ppbv, respectively. The observed abundances suggest that there are unidentified sources of acetone and acetaldehyde in biomass burning impacted air masses and in remote marine air.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 A primary objective of the Indian Ocean Experiment (INDOEX) was to characterize the extent and chemical composition of pollution outflow from the Indian subcontinent during the dry winter monsoon season. The dry monsoon is characterized by prevailing northeasterly winds and suppressed convection over the continental source regions due to large-scale subsidence, thus limiting upward dispersion of pollution [Krishnamurti et al., 1997]. Polluted continental air masses are transported over an extended area of the Indian Ocean toward the intertropical convergence zone (ITCZ) to meet pristine air from the Southern Hemisphere. By performing measurements across the ITCZ, heavily polluted air masses originating from the Indian subcontinent, a region rapidly growing in human population and in the exploitation of natural resources, can be contrasted against comparatively clean oceanic air representing natural background conditions.
 According to recent estimates, India's population has already passed the one milliard (billion in US units) mark and at current growth rates, the population will rise to two milliards in the second half of this century [United Nations Population Division (UNPD), 2001, available at http://www.un.org/esa/population/unpop.htm]. Hence the potential for growing pollutant emissions with global impact is large. According to the Emission Database for Global Atmospheric Research (EDGAR) compiled by Olivier et al.  (available at http://www.rivm.nl/env/int/coredata/edgar/index.html) the India region (Bangladesh, Bhutan, India, Maldives, Myanmar, Nepal, Pakistan and Sri Lanka) is already a significant source of global pollution. Estimated carbon monoxide (CO) emissions from the India region are similar to emissions from North America or China and account for 11% of the global anthropogenic CO emissions.
 Several VOCs including acetonitrile (CH3CN), methanol (CH3OH), acetone (CH3COCH3) and acetaldehyde (CH3CHO) were measured during the 1999 Intensive Field Phase (IFP) of INDOEX by a proton-transfer-reaction mass spectrometer (PTR-MS). The PTR-MS system performed airborne measurements aboard the NCAR C-130 aircraft during the research flights (RF) 1 to 6 and on RF 18. An intercomparison of acetonitrile and acetone measurements with an Atmospheric Pressure Chemical Ionization Mass Spectrometer (AP-CIMS) operated by the Max-Planck-Institute for Chemistry was performed during RF 3 and RF 4 [Sprung et al., 2001]. After RF 6 the PTR-MS was installed onboard the NOAA Research Vessel (R/V) Ronald H. Brown. In this paper, we present a first overview of the distribution of acetonitrile, methanol, acetone and acetaldehyde as measured in the Indian Ocean boundary layer during leg 2 of the R/V Ronald H. Brown research cruise.
 The PTR-MS technique developed at the University of Innsbruck has been described in great detail elsewhere [Hansel et al., 1995; Lindinger et al., 1998], thus only the essential points are outlined here. The innovative method for real-time detection of VOCs relies on gas phase proton transfer reactions between H3O+ primary ions and volatile trace gases with a higher proton affinity (PA) than that of water molecules (PAH2O = 165.0 kcal/mol [Hunter and Lias, 1998]). H3O+ primary ions are produced in a hollow cathode ion source and injected into a flow drift tube continuously flushed with ambient air. On each collision between a H3O+ primary ion and an organic molecule the proton H+ is transferred thus charging the reagent. Primary and product ions are mass analyzed in a quadrupole mass spectrometer and detected by a secondary electron multiplier/pulse counting system. The portable instrument (weight: ∼100 kg) has been successfully employed in several field campaigns throughout the world during the past few years: the Cooperative LBA Airborne Regional Experiment (LBA-CLAIRE) in 1998 [Crutzen et al., 2000; Pöschl et al., 2001; Warneke et al., 2001; Williams et al., 2001]; the Southern Oxidants Study (SOS) in 1999 [Hansel and Wisthaler, 2000; Wisthaler and Hansel, 2000]; and the Texas Air Quality Study (TEXAQS) in 2000.
 On board the R/V Ronald H. Brown the PTR-MS system was located in the main laboratory. Ambient air was continuously sampled at a flow rate of 15 standard liters per minute (slpm) through a 6.4 mm (OD) tube from the top of a 10 m tower (28 m above the sea surface) located at the bow of the forward deck. Approximately 50 m from the inlet point a variable flow (0.3 to 1 slpm) was drawn through a tee into the PTR-MS inlet system described in detail by Sprung et al. . Sampling air was drawn into the PTR-MS instrument through a second tee at a flow rate of 0.015 slpm. To determine the instrumental background this flow was periodically passed through a heated platinum (Pt, 380°C) catalyst capable of removing VOCs with ≥99.9% efficiency. All wetted surfaces of the inlet system except the scrubber were made out of Teflon® PFA. The total residence time in the inlet system was calculated to be less than 5 s.
 The 50 m long inlet line was tested for artifact generation by gradually reducing the flow rate from 15 slpm to 0.3 slpm and hence changing the residence time in the inlet line from a few seconds to several minutes. At flow rates less than 1 slpm a significant increase (a few hundred pptv; 1 pptv = 10−12 v/v) in acetaldehyde volume mixing ratios (VMRs) was observed implying that under such experimental conditions acetaldehyde is formed in the inlet line. No change in the VMRs of the investigated VOCs was observed at flow rates in the 1-to-15 slpm range indicating that no major positive or negative artifact problems were exhibited at the sampling flow rate of 15 slpm used here.
 In the present study the PTR-MS technique was used for on-line monitoring of methanol (PACH3OH = 180.3 kcal/mol), acetonitrile (PACH3CN = 186.2 kcal/mol), acetaldehyde (PACH3CHO = 183.7 kcal/mol) and acetone (PACH3COCH3 = 194.0 kcal/mol) [Hunter and Lias, 1998], detected at 33, 42, 45 and 59 amu, respectively. The entire background corrected ion signal at these mass-to-charge ratios was converted into VMRs of the given organic trace gases. To date no interferences in ambient air for methanol, acetonitrile and acetaldehyde are known to the authors, and in the case of acetone possible interferences (propanal, glyoxal) are expected to have negligibly low VMRs. Strictly speaking, all reported VMRs must, however, be considered to be upper limits for the given species. Good selectivity for acetaldehyde and acetone was confirmed in an intercomparison study with a gas chromatographic system during SOS 1999 in Nashville, TN [Wisthaler and Hansel, 2000]. This produced very good agreement despite being performed in a highly polluted air matrix. The PTR-MS instrument was calibrated for methanol, acetaldehyde and acetone using a calibration standard (1.1 ± 0.2 ppmv methanol, 1.1 ± 0.2 ppmv acetaldehyde and 1.04 ± 0.2 ppmv acetone in zero air; 1 ppmv = 10−6 v/v) provided by E. Apel (NCAR). A slight humidity dependence of the sensitivity [Sprung et al., 2001] was taken into account for data evaluation. The accuracy of the methanol, acetaldehyde and acetone measurements corresponds to the error in the gas standard of ±20%. No calibration standard was available for acetonitrile, and the PTR-MS sensitivity for acetonitrile was calculated using the procedure outlined in detail by Sprung et al. . An accuracy for the acetonitrile measurements of ±30% was inferred from intercomparison measurements described by Sprung et al. .
 Methanol, acetonitrile, acetaldehyde and acetone (and ∼25 other compounds) were measured on a time shared basis for 10 s respectively, once each 5 min. Time series were plotted as hourly averages of the respective VMRs. The detection limits (S/N = 2) were 100, 10, 80 and 50 pptv for methanol, acetonitrile, acetaldehyde and acetone, respectively. Regressions of VOCs relative to CO measured by Stehr et al.  were calculated using the reduced-major-axis method [Riggs et al., 1978]. The enhancement ratios of the investigated species ΔX/ΔCO (X = CH3CN, CH3OH, CH3COCH3, CH3CHO) in different flow regimes were determined from the slope of the respective regression analysis.
3. Results and Discussion
3.1. Meteorological Conditions and Air Mass Origins
 The data presented in this paper were obtained during leg 2 of the R/V Ronald H. Brown 1999 INDOEX cruise. Figure 1 shows the ship track including the position of the vessel at a particular day of the year (DOY, number of days elapsed since 1 January 1999). The R/V Ronald H. Brown started leg 2 on 4 March (DOY 63) at Male' (Maldives), sailed along the Indian west coast and turned south on 11 March (DOY 70) at the latitude of Bombay (19°N). After passing the Maldives the ship reached its most southerly point (13°S) on 19 March (DOY 78) and turned north, returning to Male' on 23 March (DOY 82).
 As the ship sailed through the Indian Ocean it encountered several different air masses. 3-D 7-day backtrajectories were calculated with the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT 4) model [Draxler and Hess, 1998] at 6-hour intervals arriving at the ship location at an altitude of 500 meters (nominally 950 mbar). The prevailing flow regimes are indicated by clustered backtrajectories also shown in Figure 1. Trajectory clusters should be viewed as indicators of large-scale flow and not as the true path of the air parcels.
 We use labels for flow regimes similar to those of Ball et al., (W. P. Ball et al., Size segregated and bulk aerosol composition observed during INDOEX 1999: Overview of meteorology and continental impacts, submitted to Journal of Geophysical Research, hereinafter referred to as P. Ball et al., submitted manuscript): Northern Hemisphere continental tropical (NHcT) (DOY 63–67), Northern Hemisphere continental extra-tropical (NHcX, DOY 68–70), NHcX mixed with NHcT (NHc-mixed, DOY 71–76), Northern Hemisphere maritime equatorial (NHmE, DOY 77 and DOY 79–81), and Southern Hemisphere maritime equatorial (SHmE, DOY 78). To simplify our discussion the NHcT regime is further divided into two subregimes according to air parcel origin: air originating from the Bay of Bengal is labeled NHcT-Bengal (DOY 63–65) and air advected from the Indian west coast is denoted NHcT-W (DOY 66–67).
 Air from the NHcT-Bengal regime was encountered for half a day after starting our measurement operation on DOY 65.5. The air passed over north-eastern India at 700–950 mbar, transversed the Bay of Bengal and crossed southern India 2–4 days upwind of the sample location. The curved nature of the cluster trajectory through the Bay of Bengal was due to a persistent anti-cyclonic circulation over the Indian subcontinent during the dry season. Along with a strong subsidence of tropospheric air over this region and a suppressed horizontal dispersion [Verver et al., 2002], continental outflow from this region was heavily polluted. Averaged acetonitrile, acetone, methanol, acetaldehyde and CO VMRs by the distinct flow regimes are listed in Table 1. Absolute maximum VMRs of all species were found in the NHcT-Bengal regime.
Table 1. Average (±1σ) Acetonitrile, Methanol, Acetone, Acetaldehyde, and CO VMRs as Observed in Different Flow Regimes
 On DOY 66 and 67 air was advected from the continental boundary layer on the Indian west coast into the Arabian Sea spending several days over the ocean. The NHcT-W regime was also characterized by strong pollution impact from the Indian subcontinent.
 On the way north the subtropical high over India shifted eastward to the Bay of Bengal cutting off the outflow of Indian air with winds changing from north-easterly to northerly flow. This transition from NHcT to NHcX began on DOY 67 and backtrajectories show the origin of air parcels shifting toward the Arabian Peninsula, the Middle East and South East Europe. We will refer to this regime as Arabian air. Backtrajectories for DOY 68 to 70 show general subsidence starting at 200–400 mbar above the Arabian Peninsula 6 days upwind subsiding to 950 mbar just one to two days before reaching the R/V Ronald H. Brown. This flow regime is characterized by low acetonitrile and methanol levels but still elevated acetone, acetaldehyde and CO VMRs (Table 1).
 From DOY 71 to 76 air parcels from the NHcX regime mixed increasingly with the NHcT outflow resulting in gradual increase of pollution levels. For this period backtrajectories indicate that air parcels originated in the Persian Gulf Region at 400–600 mbar, subsided over the Northern Arabian Sea and traveled along the Indian west coast at 950 mbar altitude without any direct contact with the Indian subcontinent for 1 to 5 days before reaching the R/V Ronald H. Brown. The increasing pollution levels may be explained by a local sea breeze circulation (S. Raman et al., Dynamics of the elevated land plume over the Arabian Sea and the northern Indian Ocean during northeasterly monsoon (INDOEX), submitted to Geophysical Research Letters, 2002) that prevented Indian pollution from being directly advected into the marine boundary layer. Instead, the pollution was advected to layers just above the marine boundary layer and subsequent mixing caused increasing pollution levels on the way south along the Indian west coast.
 By DOY 76 the nature of the air mass changed rapidly to NHmE and was characterized by rather low pollution levels. Backtrajectories indicate that the air parcels had spent at least 6 days in the marine boundary layer before encounter. Average concentrations of acetonitrile, acetone, methanol, and CO were substantially lower than encountered in the NHc-mixed regime, but still higher than in the SHmE regime (Table 1). The ITCZ lay well into the SH at about 12°S at this time. On DOY 78 a brief encounter with SHmE air occurred with pollutant levels reaching a local minimum as the ship arrived at the most southerly point of this leg, 13°S. At this location air was sampled in the mixing zone between SH and NH. After DOY 79 during the return passage to Male' NHmE air was encountered again.
 Acetonitrile (methyl cyanide, CH3CN) was found to be ubiquitously present in the Indian Ocean boundary layer. A time plot of acetonitrile VMRs as observed in different meteorological flow regimes is shown in Figure 2a; the spatial acetonitrile distribution as observed in the Indian Ocean boundary layer is shown in Figure 3. Significantly elevated acetonitrile VMRs in the 200-to-320 pptv range were found in all air masses advected from the Indian subcontinent; in air masses with predominantly marine backtrajectories and in Arabian air acetonitrile VMRs were in the 120-to-200 pptv range.
 The best-studied loss process of acetonitrile in the troposphere is the reaction with OH radicals. The recommended reaction rate constant kOH298 is 2.2 × 10−14 cm3 molec.−1 s−1 [Atkinson et al., 1999]. Using the European Center Hamburg (ECHAM) chemistry-general circulation model, De Latt et al.  derived a daily average OH mixing ratio of 1.5 × 106 molec. cm−3 along the cruise track, resulting in a lifetime of acetonitrile with respect to reaction with the OH radical of about one year. Hamm et al.  quantified the removal of tropospheric acetonitrile by wet precipitation estimating a residence time due to rainout of 3 ± 1.5 years. Due to the low abundance of Cl atoms in the troposphere [Singh et al., 1996] reaction with Cl atoms is expected to be negligible. According to Junge  the relative standard deviation of VMRs of long-lived atmospheric species is inversely proportional to their atmospheric residence time. Our data reveal that the variability of acetonitrile is similar to that of CO with a known atmospheric lifetime of the order of 1 month. Other removal mechanisms may reduce the atmospheric lifetime of acetonitrile. The main sink for acetonitrile is presently thought to be uptake in the ocean followed by biological degradation [Hamm et al., 1984]. In a recent study Li et al.  concluded from a 3-D model simulation that observations of atmospheric hydrogen cyanide (HCN) can be roughly reproduced by assuming an oceanic uptake of cyanide. A model analysis based on the acetonitrile data presented here has also indicated the presence of an additional sink in the marine boundary layer [De Latt et al., 2001]. The ocean sink hypothesis for HCN and CH3CN, however, remains unconfirmed.
 Regardless of the inconsistencies in the acetonitrile budget we may assume that acetonitrile is a relatively long-lived, selective marker for biomass burning. Thus, the high acetonitrile VMRs observed in air masses advected from India are likely the result of extensive biomass burning activities on the Indian subcontinent. Our finding and interpretation is in agreement with Sprung et al.  who found significantly elevated acetonitrile VMRs (0.1–0.4 ppbv; 1 ppbv = 10−9 v/v) in air parcels advected from India and low values (0.07–0.2 ppbv) in remote marine air. Hamm and Warneck  also came to similar conclusions observing elevated acetonitrile VMRs in the vicinity of the equator during an Atlantic Ocean cruise. The enhancement was attributed to the advection of continental air from Africa, enriched with acetonitrile from biomass burning.
 Results from chemical analyses of aerosols sampled aboard the R/V Ronald H. Brown further support our interpretation. The number concentration of small carbon-containing particles with potassium (K), which are tracers for biomass burning and coal combustion, strongly correlates with acetonitrile VMRs (Guazzotti et al., in preparation).
 The elevated acetonitrile VMRs in the Indian outflow point to extensive biomass burning activities on the Indian subcontinent. Biomass is the major source of energy in India contributing to almost half of the country's total primary energy production [Ravindranath and Hall, 1995]. The dominant biofuel source is fuelwood, although dung and agricultural residues are also widely used. Most of the biomass energy is needed in the domestic sector for cooking and heating water. India also experiences severe forest fires. The actual forest cover of India is about one fifth of the country's geographic area [Forest Survey of India (FSI), 1999, available at http://envfor.nic.in/fsi/sfr99/] and forest fires are reported to be a significant factor in the degradation of Indian forests [Saigal, 1990]. Most of the fires are human-caused (slash-and-burn shifting cultivation, grazing). An incessant heat wave in March and April 1999, the driest months in ten years [India Meteorological Department (IMD), 2000], led to a particularly intense fire season [Sehgal, 1999]. Remote sensing products [World Fire Webhttp://www.gvm.sai.jrc.it/, World Fire Atlashttp://shark1.esrin.esa.it/] providing global maps of vegetation fires show extensive fires in the eastern and north-eastern parts of India and along the Western Ghats in March 1999, exactly the regions transversed by the NHcT-Bengal and the NHcT-W outflow. These large forest fires may have strongly contributed to the observed high levels of biomass burning markers: acetonitrile in the gas-phase and potassium in the particulate phase. We cannot, however, determine the relative strength of different types of biomass burning in India (biofuel consumption versus forest fires) as no measurements were performed outside the fire season.
 Although it is difficult to assess the exact nature of the biomass burning source in India, the observed enhancement ratio of acetonitrile and CO (ΔCH3CN/ΔCO) can be used to estimate the relative strength of the two main combustion-related pollution sources, biomass burning and fossil fuel combustion. Since both gases are emitted by combustion processes and are relatively long-lived, the ΔCH3CN/ΔCO molar ratio is not expected to change significantly during several days of transport over the ocean thus conserving the source characteristic emission ratio. Previous studies [Rhoads et al., 1997; Lobert and Harris, 2002] found that most of the CO variability in the Indian Ocean is a result of varying source regions confirming that the influence of natural CO sources (oceanic emanation) and indirect atmospheric CO sources (reaction of OH with methane and non-methane hydrocarbons (NMHCs)) are minor. In laboratory biomass combustion experiments Lobert et al.  observed that acetonitrile emissions depend strongly upon combustion type (flaming versus smoldering combustion) and the nitrogen content of the fuel, which varies significantly for different types of vegetation. The experiments performed by Lobert et al.  covered a large variety of different types of biofuel, and the mean primary molar emission ratio ΔCH3CN/ΔCO measured was 0.25%. Holzinger et al.  and Hurst et al.  found significantly lower average ΔCH3CN/ΔCO molar emission ratios of 0.13 ± 0.07% and 0.079 ± 0.033%, respectively for different savanna grass fuels. We consider the value reported by Lobert et al. , based on multiple experiments, to be the most representative value for biomass burning of mixed fuel type.
Figure 4a shows the correlation between acetonitrile and CO as observed in different meteorological flow regimes. The slope of the reduced-major-axis regression identifies the enhancement ratio ΔCH3CN/ΔCO. A value of 0.24 ± 0.01 % and a strong correlation (R2 = 0.87) was found in the NHcT-W and NHc-mixed regime. CO is a general marker for incomplete combustion (fossil fuel combustion and biomass burning), but the strong correlation with acetonitrile and the similarity of enhancement ratio to the primary emission ratio observed by Lobert et al.  indicate that biomass burning most likely dominates (>90%) CO emissions in these regions.
 The ΔCH3CN/ΔCO enhancement ratio observed in the NHcT-Bengal regime was 0.11 ± 0.01%. This lower ratio implies that a significant fraction (∼50%) of the CO observed in these air parcels probably does not originate from biomass burning but derives from fossil fuel combustion.
 High VMRs of CO (126 ± 10 ppbv) were observed in the NHcX regime. Acetonitrile VMRs approached southern hemispheric background values and were not correlated with CO. These findings indicate that fossil fuel combustion is very likely the primary source of CO in these air parcels.
 Methanol (CH3OH) was found to be a ubiquitous compound in the Indian Ocean boundary layer with VMRs ranging from 500 pptv to 1600 pptv. Figure 2b shows the time series for methanol as observed in different meteorological flow regimes. All air masses advected from the Indian subcontinent showed significantly elevated methanol levels. The highest average methanol VMRs of 1417 ± 118 pptv were measured in the NHcT-Bengal regime. Values of 1057 ± 163 pptv and 956 ± 157 pptv were found in the NHcT-W and the NHc-mixed flow regime, respectively. Considerably lower VMRs were observed in the NHcX regime (687 ± 106 pptv) and in marine air masses (NHmE: 708 ± 69 pptv and SHmE: 621 ± 62 pptv).
 Although methanol is a ubiquitous component in the atmosphere, the full extent of its sources, occurrence and fate is poorly understood. According to a global source inventory recently published by Singh et al.  methanol has large primary sources from biogenic emissions [MacDonald and Fall, 1993, König et al., 1995, Kirstine et al., 1998] and dead/decaying plant matter [Warneke et al., 1999] and a substantial secondary source from CH4/NMHC oxidation pathways (2 CH3O2 → CH3OH + CH2O + O2). Yokelson et al.  reported that biomass fires may contribute some 10–30% of the global atmospheric methanol budget. They also compiled all published fire-averaged ΔCH3OH/ΔCO emission ratios obtaining an average value of 1.71 ± 0.76%.
 In the NHcT-W and NHc-mixed regime where biomass burning dominated combustion related emissions, we observed a ΔCH3OH/ΔCO enhancement ratio of 1.20 ± 0.08% (Figure 4b). To compare the observed ratio with the reported primary emission ratios the photochemical loss of methanol must be taken into account. The lifetime of methanol with respect to reaction with the OH radical is 8 days ([OH] = 1.5 × 106 molec. cm−3 [De Latt et al., 2001]; kOH298 = 9.3 × 10−13 cm3 molec.−1 s−1 [Atkinson et al., 1999]). The backtrajectory analysis indicates that air parcels in the NHcT-W and NHc-mixed regime had not been affected by continental emissions for about 3–5 days during which 30–45% of the primary methanol is lost due to reaction with OH. Considering also a slow reaction of CO with OH (k298= 2.1 × 10−13 cm3 molec.−1 s−1 [Atkinson et al., 1999]), the primary ΔCH3OH/ΔCO emission ratio of 1.71% [Yokelson et al., 1999] would decrease to 1.10% and 0.82% within 3 and 5 days, respectively. We observed a ΔCH3OH/ΔCO enhancement ratio of 1.20 ± 0.08% implying that the methanol enrichment observed in the NHcT-W and NHc-mixed regime may be entirely attributed to biomass burning emissions.
 In the NHcT-Bengal regime (Figure 4b) we observed a lower ΔCH3OH/ΔCO enhancement ratio (0.74 ± 0.13%) in keeping with the contention that much of the CO from this area emanated from fossil fuel combustion that produces little methanol. In the NHcX regime where fossil fuel burning dominates, methanol VMRs approached southern hemispheric background values.
 The minimum methanol VMRs representing natural background conditions found both in the NHcX and in the SHmE regime were about 500 pptv. This value is in agreement with free tropospheric methanol background VMRs of 400–800 pptv observed by Singh et al. [1995, 2000] over the Pacific and the Atlantic Ocean, but is significantly lower than recently published background values of about 900 pptv in the remote Pacific boundary layer [Singh et al., 2001]. Model calculations by Singh et al.  suggest that ∼160 pptv (global average) of methanol could be synthesized by methane (CH4) oxidation and that the values may approach 500 pptv in the lower tropical troposphere. The small interhemispheric CH4 concentration gradient would explain similar methanol background values observed in continental, northern-hemispheric and marine, southern-hemispheric air masses. We found no evidence for a major oceanic methanol source, although methanol is likely to be formed in the photochemical breakdown of dissolved organic matter [Kieber et al., 1989; World Meteorological Organization (WMO), 1995, available at http://gesamp.imo.org] or the solvolysis of methyl halides [Elliot and Rowland, 1995]. Our observations show that methanol is the most abundant NMHC in the remote marine boundary layer and further investigations are needed to establish possible sources and atmospheric implications.
 A time plot of acetone VMRs as observed in different meteorological flow regimes is shown in Figure 2c. Acetone VMRs ranged from 450 pptv to 2400 pptv. The highest average acetone value of 2081 ± 251 pptv was found in the NHcT-Bengal regime. In the NHcT-W, NHcX and NHcX-mixed regime acetone VMRs were similarly abundant with average values of 1253 ± 190 pptv, 1121 ± 72 pptv and 1111 ± 180 pptv, respectively. Significantly lower VMRs were found in marine air masses (NHmE: 624 ± 63 pptv; SHmE: 515 ± 31 pptv). The observed values are in good agreement with airborne measurements from Sprung et al.  who found 0.8–2.5 ppbv of acetone in the Indian outflow and 0.3–0.8 ppbv of acetone in the pristine Indian Ocean boundary layer.
 Acetone is emitted from natural and industrial as well as biomass burning related sources [Singh et al., 1994, 2000]. It is also photochemically produced in the atmosphere by the oxidation of propane and other NMHCs [Chatfield et al., 1987]. High acetone VMRs > 1 ppbv throughout the northern hemispheric Indian Ocean boundary layer result from various acetone sources in the different flow regimes: mixed continental pollution likely gives rise to strongly increased acetone VMRs in the NHcT-Bengal regime, biomass burning likely enhances acetone levels in the NHcT-W and NHc-mixed regime and fossil fuel combustion likely leads to increased acetone VMRs in the NHcX regime.
 A strong correlation of acetone and CO (R2 = 0.88) was observed in the NHcT-W and NHc-mixed regime (Figure 4c). Our acetonitrile and methanol data strongly imply a predominant biomass burning impact and are not consistent with significant fossil fuel or biogenic emissions in these flow regimes. Acetone has been identified as a direct emission from biomass fires. Andreae and Merlet  have recently provided an updated assessment of emissions of trace gases from various types of biomass burning (savannas and grasslands, tropical forests, extratropical forests, biofuel burning, charcoal making, charcoal burning and burning of agricultural residues) reporting average ΔCH3COCH3/ΔCO primary emission ratios in the range of 0.01% to 0.66%. Holzinger et al.  measured a ΔCH3COCH3/ΔCO primary emission ratio of 0.54 ± 0.27% from laboratory biomass fires. Primary ΔCH3COCH3/ΔCO emission ratios from biomass fires are thus significantly lower than the ΔCH3COCH3/ΔCO enhancement ratio observed in the NHcT-W and NHc-mixed regime, 1.72 ± 0.05% (Figure 4c). An upward bias in the ratio due to photochemical loss of CO with plume aging [Mauzerall et al., 1998] can be ruled out since OH loss rates are comparable ( = 1.9 × 10−13 cm3 molec.−1 s−1; kCO298 = 2.1 × 10−13 cm3 molec.−1 s−1 [Atkinson et al., 1999]).
 The observed acetone enrichment may be the result of secondary acetone formation. Acetone is mainly produced from propane oxidation although higher hydrocarbon (isoalkanes, isoalkenes, alcohols and terpenes) oxidation also results in acetone formation. Singh and Hanst  suggest that nearly 80% of propane on a carbon basis is oxidized by OH radicals to produce acetone. Propane is directly emitted by biomass burning. Andreae and Merlet  report average ΔC3H8/ΔCO primary emission ratios for different fuel types between 0.1 % and 0.8%. Biomass burning is also a significant source of i-butene (ΔC4H8/ΔCO = 0.1–0.5% [Andreae and Merlet, 2002]). Primary emissions ratios of other acetone precursors (e.g. i-butane, i-pentane, terpenes, 2-propanol, 2-methyl-2-butene, 2-methyl-2-pentene) are very low ∼≤0.15% per species [Andreae and Merlet, 2002]). If we assume that all these precursors are converted into acetone with a yield of unity (certainly an overestimation) we can barely account for the observed ΔCH3COCH3/ΔCO ratio. Additional sources of acetone are very likely to be present in biomass burning impacted air parcels. Acetone production mechanisms may involve additional gas-phase precursors and/or heterogeneous acetone formation, e.g., by the interaction of ozone with organic aerosols (D. J. Lary, et al., The potential role of carbonyl production on organic aerosols, submitted to Journal of Geophysical Research, 1999, hereinafter referred to as Lary et al., submitted manuscript, 1999).
 Acetone enrichment in biomass burning plumes has been observed in several previous studies: Singh et al.  report a ΔCH3COCH3/ΔCO enhancement ratio of 2.5% (2.0–3.0%) in processed biomass burning plumes and observed a strong increase in the ΔCH3COCH3/ΔCO ratio with aging. Mauzerall et al.  observed a significant acetone enhancement in some biomass burning plumes studied over the Atlantic Ocean during TRACE A and Warneke  found a ΔCH3COCH3/ΔCO enhancement ratio of 3.3% in a processed biomass burning plume. Andreae et al.  observed a ΔCH3COCH3/ΔCO enhancement ratio of 1.95 ± 0.16% in aged biomass smoke which had been entrained into deep convection and transported to the upper troposphere. Since acetone substantially contributes to the formation of HOx radicals in the upper troposphere [Singh et al., 1995], acetone formation in biomass burning plumes might have a strong impact on upper tropospheric photochemistry.
 A striking feature of the acetone/CO correlation plot (Figure 4c) is the flattening out of acetone VMRs at low CO levels. It indicates the presence of additional acetone sources in the maritime regimes maintaining an acetone background VMR of about 400 pptv. This value is in good agreement with previously reported marine background VMRs for acetone: 0.3 ppbv (Indian Ocean [Sprung et al., 2001]), 0.47 ± 0.09 ppbv (Atlantic Ocean [Penkett, 1982]), 0.38 ppbv (Caribbean Sea [Zhou and Mopper, 1993]), 0.35–0.40 ppbv (Frobisher Bay [Singh et al., 1994]), 0.40–0.45 ppbv (Pacific Ocean [Singh et al., 2001]).
 Global three-dimensional chemistry-transport models are unable to correctly simulate acetone background levels of several hundreds of pptv [Poisson et al., 2000; Singh et al., 2001]. A model analysis [De Latt et. al., 2001] of the acetone/CO data presented here indicates the presence of substantial marine acetone sources. A source for acetone in the clean marine boundary layer might be the photochemical oxidation of locally derived, airborne organic matter in the gas- or particulate phase. A net acetone flux from the oceanic surface microlayer to the atmosphere was proposed by Zhou and Mopper  when high concentrations of acetone most likely produced by the photodegradation of aquatic dissolved organic matter [Kieber et al., 1989; WMO, 1995] were observed in the surface microlayer. Marine acetone-producing bacteria have been isolated by Nemecek-Marshall et al. . Ocean biology may be intricately involved in the synthesis of acetone which may contribute to HOx radical formation after being injected convectively into the upper troposphere. In this way the biosphere might exert a direct control over the oxidative power of the atmosphere [Singh et al., 2001].
 Substantial acetaldehyde VMRs in the range from 125 pptv to 500 pptv were found throughout the Indian Ocean boundary layer. The time series of acetaldehyde VMRs as observed in different flow regimes is shown in Figure 2d. The highest average acetaldehyde VMR of 424 ± 41 pptv was measured in the NHcT-Bengal regime. Similarly abundant VMRs of 321 ± 44 pptv, 350 ± 32 pptv and 366 ± 56 pptv were found in the NHcT-W, NHcX and NHc-mixed regime, respectively. A sharp decrease in the acetaldehyde mixing ratio was observed when air with predominantly marine backtrajectories was sampled. The average acetaldehyde VMRs found in the NHmE and SHmE regime were 212 ± 29 pptv and 178 ± 30 pptv, respectively.
 The influence of primary acetaldehyde emissions is, however, restricted to the continental boundary layer. As the high chemical reactivity with OH limits the residence time to hours, long-range transport of acetaldehyde cannot account for several hundreds of pptv of acetaldehyde in remote marine regions. Air parcels with strong biomass burning impact (NHcT-W and NHc-mixed), for example, were analyzed at least 3 days after emission from any continental source. Assuming a photochemical lifetime of acetaldehyde of 11.5 h ([OH] = 1.5 × 106 molec. cm−3 [De Latt et al., 2001]; kOH298 = 1.6 × 10−11 cm3 molec.−1 s−1 [Atkinson et al., 1999]) after 3 days only 0.2% of the primary acetaldehyde is left. The ΔCH3CHO/ΔCO enhancement ratio observed in the outflow from Western India, which was most likely impacted mainly by biomass burning, was 0.44 ± 0.02% with acetaldehyde and CO being distinctly correlated (R2 = 0.72). Primary emission of acetaldehyde from biomass fires may account for a similar acetaldehyde enrichment only if samples were taken in close proximity to the fire. For example, Hurst et al.  measured a ΔCH3CHO/ΔCO ratio of 0.35% when transversing fresh smoke plumes from savanna fires at low altitudes. In our case the short lifetime of acetaldehyde implies that the observed enrichment is the result of a secondary, in-situ production of acetaldehyde. Secondary acetaldehyde formation occurs from the oxidation of NMHCs, e.g., various alkanes, alkenes, alcohols and aromatic compounds [Atkinson, 2000]. Biomass burning is a significant source of most acetaldehyde precursors [Andreae and Merlet, 2002], but only ethane (C2H6) is sufficiently long-lived to reach remote marine regions. Average ethane VMRs in the NHcT-W and NHc-mixed regime were 544.2 ± 86.6 pptv and 926.9 ± 429.0 pptv [Mühle et al., 2002] and from a simple steady state approximation an acetaldehyde VMR of 8.5 ± 1.5 pptv and 14.5 ± 7.0 pptv can be derived. Additional acetaldehyde production mechanisms must be present in biomass burning plumes to account for acetaldehyde VMRs of a few hundred pptv.
 The enrichment of both acetone and acetaldehyde in biomass burning plumes points to the presence of substantial, currently unknown pathways for carbonyl formation. Additional gas-phase precursors and/or heterogeneous production on organic aerosols have been discussed in Section 3.4 as possible carbonyl sources. In contrast to acetone, acetaldehyde is highly reactive and enriched acetaldehyde VMRs imply strong local production. The PTR-MS technique is sensitive to most VOCs and mass spectra in the 20–150 amu range, in which most volatile species are typically found, and our data do not indicate the presence of any gas-phase precursors. Since high aerosol concentrations were found throughout the northern hemispheric Indian Ocean boundary layer (P. Ball et al., submitted manuscript), photochemical reactions on organic aerosols (Lary et al., submitted manuscript, 1999) should be investigated in detail as possible sources of atmospheric carbonyls.
 Our observations have strong implications for the fate of HOx radicals in the lower and upper troposphere. At low altitudes the reaction of acetaldehyde with OH radicals may contribute about 10–15% of the total OH loss rate if the reaction with CO and CH4 are considered as the only additional OH sinks. On the other hand, Lary and Shallcross  and Müller and Brasseur  emphasize the importance of acetaldehyde in upper tropospheric HOx formation. Thus, convective injection into the upper troposphere of biomass burning plumes with enriched acetaldehyde VMRs could significantly increase HOx production at high altitudes.
 Minimum acetaldehyde VMRs of about 125 pptv were found in the SHmE regime. This value is in good agreement with previously reported marine background VMRs for acetaldehyde: 190 pptv (Atlantic Ocean [Schubert et al., 1988]), 380 ± 100 pptv (Caribbean Sea [Zhou and Mopper, 1993]), 50–60 pptv (Southern Pacific Ocean [Arlander et al., 1995]), 90–180 pptv (Atlantic Ocean west of Africa [Arlander et al., 1995]), 140 pptv (Northern Atlantic Ocean [Tanner et al., 1996]), 110 pptv (Northern Pacific Ocean [Singh et al., 2001]), 80 pptv (Southern Pacific Ocean [Singh et al., 2001]). The Harvard global, 3-D chemistry-transport model used to simulate Singh's experimental data underestimates acetaldehyde VMRs over the South Pacific by a factor of 10 [Singh et al., 2001]. We derived an acetaldehyde VMR from ethane oxidation of about 3 pptv using a simple steady state approximation and an ethane VMR of 193.5 ± 17.9 pptv in the SHmE regime [Mühle et al., 2002]. Other significant acetaldehyde sources must be present in the remote marine boundary layer. Zhou and Mopper  found high concentrations of acetaldehyde in the surface microlayer most likely produced by the photodegradation of aquatic dissolved organic matter [Kieber et al., 1989; WMO, 1995] and proposed a net flux from the ocean surface to the atmosphere. A source for acetaldehyde in the clean marine boundary layer may also be the photochemical oxidation of locally derived airborne organic matter in the gas- or particulate phase. High abundances of both acetone and acetaldehyde in the remote marine boundary layer indicate the presence of large marine carbonyl sources that need to be studied in more detail especially with regards to possible effects on the oxidative power of the atmosphere. In this context a rigorous field intercomparison of carbonyl measurement techniques should also be carried out (in view of potential interferences or artifacts as described in Section 2).
4. Summary and Conclusions
 The Innsbruck PTR-MS system was used for fast-response VOC measurements onboard the NOAA R/V Ronald H. Brown during leg 2 of the INDOEX 1999 cruise. The cruise covered a broad spatial extent of the Indian Ocean and a trajectory analysis revealed that continental outflows from NE-India, W-India and Arabia and maritime air masses from the Northern and Southern Hemisphere were encountered. The chemical composition of the sampled air changed according to different air mass origins. Acetonitrile, a gaseous tracer for biomass burning, was significantly increased in the Indian outflow most likely due to large-scale vegetation fires and/or widespread domestic biofuel use. The relative abundance of acetonitrile to that of CO was highest in air from W-India pointing to a predominant biomass burning impact in these air parcels. Strong evidence was found that the outflow from NE-India was polluted both by biomass and fossil fuel burning, while air from Arabia was only polluted by fossil fuel combustion. Methanol was elevated in biomass burning impacted air masses, while acetone and acetaldehyde were elevated in all continental air masses. Pollution levels decreased toward the ITCZ and reached local minima of 0.50 ppbv methanol, 0.45 ppbv acetone and 0.12 ppbv acetaldehyde in the Southern Hemisphere. The observed abundances suggest that there are unidentified sources of acetone and acetaldehyde in biomass burning impacted air masses and in remote marine air masses.
 This paper is dedicated to the memory of our teacher, colleague and friend, Werner Lindinger, who passed away in February 2001.
 Our thanks go the coordinators and all participants of the exciting INDOEX campaign. We are grateful for the support provided by the captain and crew of NOAA's Ronald H. Brown. Trajectory analysis was provided by NOAA's Pacific Marine Environmental Laboratory (PMEL) and can be accessed at http://saga.pmel.noaa.gov/indoex/traject/. Special thanks should be referred to T. D. Märk from the University of Innsbruck for his support. Financial support from the Austrian Bundesministerium für Wissenschaft und Verkehr is gratefully acknowledged. The NSF supported CO measurements. Finally, we would like to thank anonymous reviewers for their constructive comments and suggestions.