During the Indian Ocean Experiment (INDOEX) in February-March 1999 the impact of continental outflow to the Indian Ocean was analyzed. On board the R/V Ronald Brown altogether 93 air samples were taken for analysis of nonmethane hydrocarbons (NMHC), methane (CH4), carbon dioxide (CO2), carbon monoxide (CO), and the 14C/12C and 18O/16O isotope ratios of CO. Five types of air masses differing in origin, degree of pollution, and chemical age were identified based on back trajectory analyses and the trace gas data, supported by continuous CO and ozone (O3) observations from other investigators. The Indian Ocean was found to be frequently affected by nearby emissions from the Indian subcontinent and Indochina, but the strongest pollution event (characterized inter alia by high mixing ratios of medium- and long-lived NMHC) was due to long-range advection from the extratropical northern hemisphere. Carbon monoxide 14 showing a distinct meridional profile unequivocally confirms this remote impact. The ratio acetylene/CO was found to be often inadequate as a measure for atmospheric processing, the integrated influence of OH chemistry and mixing. Our data suggest that the influence from fresh continental pollution was less pronounced along the INDOEX R/V Ronald Brown cruise compared to observations made during other tropical campaigns, such as the Pacific Exploratory Mission-West B in the Pacific Ocean.
 The fast industrialization of countries in Asia, in particular India and China, comes along with large and rapidly increasing emissions of various pollutants to the atmosphere. Especially during the winter (northeast) monsoon (January–March), the Indian Ocean becomes a unique region where the temporal and spatial development of emissions from the Indian subcontinent and Southeast Asia can be studied. This particularly implies the mixing of continental outflow of pollutants and aerosols with pristine SH air masses by cross-equatorial monsoonal flow into the Intertropical Convergence Zone (ITCZ). A major intention of the Indian Ocean Experiment (INDOEX) was to improve the understanding of the interactions between aerosols, clouds, chemistry, and climate. Special emphasis was on the radiative forcing effect of anthropogogenic aerosols and on the role of the ITCZ in interhemispheric transport of trace gases and aerosols. The intensive field campaign of INDOEX took place in February-March 1999 and involved several aircrafts, ships, and island stations in the Indian Ocean region. A main goal of the National Oceanic and Atmospheric Administration (NOAA) research vessel Ronald Brown as one of the platforms was to characterize the trace gas and aerosol composition of outflow from the Indian subcontinent and its chemical processing over the Indian Ocean.
 Purpose of the present paper is to characterize the air masses encountered by the R/V Ronald Brown by applying a powerful set of contrasting tracers, in particular nonmethane hydrocarbons. The detected trace gases (NMHC, CH4, CO, CO2, N2O, and SF6) do have lifetimes between days (e.g., propane) and years (e.g., CH4) and diverse sourc es, e.g., biogenic, oceanic, and anthropogenic (see Table 1).
Table 1. Rate Constants and Lifetimes of Selected Trace Gases Under Typical Midlatitude and Tropical Conditions
 In addition to the mixing ratio of CO, its C18O (stable) and 14CO (radioactive) isotope composition was measured. δ18O(CO) expresses the isotope ratio 18O/16O in CO relative to the Vienna-Standard Mean Ocean Water (V-SMOW) standard in per mil. High δ18O(CO) values (∼23‰) are associated with fossil fuel combustion, whereas biomass burning derived CO is less enriched (15 to 18‰), and photochemically generated CO is assumed to show δ18O(CO) of ∼0‰. Note that isotope fractionation during oxidation by OH yields progressive depletion in δ18O(CO). Atmospheric 14CO (given in molecules per cm3 air STP) is mainly produced by cosmic rays, primarily in the midlatitude and high-latitude upper troposphere and lower stratosphere. It has also a small biogenic contribution (“recycled CO”) from the incomplete oxidation of organic matter containing 14C from CO2 assimilation. The main sink of 14CO is, as for CO, oxidation by OH. Note, that CO from fossil fuel combustion is free of 14CO. Observed isotopic composition of CO thus reflects both the signature of the different sources as well as fractionation and dilution occurred during transport [Brenninkmeijer et al., 1999; Jöckel, 2000, and references therein].
 After a brief description about the analytical techniques applied (section 2), a short overview about the meteorological conditions during INDEX is given (section 3). Thereafter the observed trace gas time series are analyzed (section 4), and the intersected air masses are classified based upon back trajectory analyses and the trace gas data (section 4.1). The influence of long-range transport on the observed trace gas variability is discussed (section 4.2), and the meaning of the acetylene/CO ratio is evaluated (section 4.3). Finally, the entire data set is compared with previous results (section 5), and the conclusions are given (section 6).
2. Experimental Techniques
2.1. Air Sampling and Analyzed Compounds
 Ninety-three air samples (sample locations in Figure 1 were collected in 2.5 L electropolished stainless steel canisters fitted with metal bellow valves (Nupro SS-4H). Using a metal bellows pump (Parker, MB-158-E, operation temperature below 65°C), air was sucked from the top of a bow tower (10 m height) via a 25 m stainless steel tube (ID: 4 mm) through an ice water cold trap (0°C) to reduce the water vapor mixing ratio to ∼0.6%. After flushing several times, the canisters were filled to a final pressure of 3.7 bar leading to sample sizes of ∼9 L (STP) each. With exception of the transport time (∼6 days) to the laboratory (MPI-C, Mainz, Germany) the canisters were kept at −18°C. Twelve hours before gas chromatographic (GC) analysis of NMHC, CH4, CO2, SF6, and N2O the canisters were stored at ambient temperature.
 For measuring CO and its isotope ratios (18O/16O, 14C/12C), fifteen 600 L (STP) air samples (sample locations in Figure 1) were also collected from the bow tower (via a ½ inch PFA tube) by using 5 L aluminum cylinders (Scott-Marrin Inc., Riverside California). An improved version of the clean air compressor described by Mak and Brenninkmeijer  was used. The flushing and filling procedure took ∼60 min.
 On the samples collected during the pre-INDOEX leg 0 (12–19 February 1999, day of year (DOY) 43–50) and INDOEX leg 1 (22–28 February 1999, DOY 53–59), analysis of CH4, CO (including its isotopic composition), CO2, N2O, and SF6 were performed. For leg 2 (5–22 March 1999, DOY 64–81) and leg 3 (26–30 March 1999, DOY 85–89), additionally NMHC were analyzed. Owing to the limited additional information derived, the N2O and SF6 data are not assessed in detail.
2.2. Laboratory-Based NMHC Measurements
 A general overview about gas chromatographic techniques for measuring atmospheric trace species is given by Camel and Caude  and Helmig . Here the 59 samples from leg 2 and 3 were separated on a 50 m, ID: 0.32 mm, 5 μm Al2O3/KCl porous layer open tubular (PLOT) column (Chrompack) by using a gas chromatograph (Hewlett-Packard, HP 6890) with subambient temperature capability connected to a quadrupole mass spectrometer (HP 5973). Sample aliquots of 350 ml (STP) air initially passed two scavenger traps (length: 11 cm, ID: 0.7 cm) filled with anhydrous lithium hydroxide (Merck) and anhydrous magnesium perchlorate (Fluka) to remove CO2 and H2O. Tests indicated insignificant influence of the scavenger traps on the measured NMHC [Matuška et al., 1986; Habram et al., 1998; Kurdziel, 1998; Rasmussen et al., 1996; Lewis et al., 1999]. The condensible compounds (encompassing the NMHC) were cryogenically concentrated in a stainless steel microtrap (−170°C, length: 30 cm, ID: 0.03 inch), packed along 10 cm with porous silica beads (Unibeads 1S, 60/80 mesh, Alltech) over 8 min at a flow of 45 mL min−1 (controlled by a mass flow controller, MKS Type 1179A). Similar techniques were used by Singh et al. , Rudolph et al. , Doskey , Doskey et al. , and Mitra and Yun . During trapping, the sample trap was connected to an evacuated gas reservoir (2.54 L) to determine the sample size volumetrically. Afterwards, the trap was flushed with helium for 5 min to remove traces of oxygen and nitrogen.
 The retained compounds were desorbed from the trap by heating from the initial −170°C to +150°C in ∼3 s [Mitra and Yun, 1993] and introduced into the analytical column. The GC column was held at 10°C for 11 min and subsequently heated at 5°C min−1 to 200°C and maintained at 200°C for another 10 min. Helium 6.0 further purified by a three-stage gas purifier and a cryogenic trap (filled with activated charcoal) at liquid nitrogen temperature was used as carrier and flushing gas. The initial column head gauge pressure was 0.81 bar at 10°C, increased during heat up of the column to maintain a constant flow. All connections were made of compression type stainless steel fittings. The sample path and the inlet line are of stainless steel and silcosteel (Restek Corp.), respectively, held at 60°C to avoid condensation of species of low vapor pressure.
 The mass spectrometric detector was operated in Single-Ion Monitoring (SIM) mode to enhance sensitivity compared to SCAN mode. For each compound, m/z values were carefully chosen with respect to background, coeluting compounds and sensitivity. One characteristic m/z value per compound was used to derive its concentration, and up to three other m/z values were used to check identity and to cross-check correct quantification for the (unlikely) case of coelution. Most of the analyzed compounds were clearly separated, except of acetylene (C2H2) and butane. Each chromatogram was integrated by a MSD Software (Hewlett-Packard) and checked manually. The mass spectrometer was tuned weekly. Analyses have been made for light alkanes such as ethane (27), propane (29), butane (43), iso-butane (43), pentane (43), iso-pentane (43), light alkenes such as ethene (27), propene (41), 1-butene (41), trans-2-butene (41), cis-2-butene (41), and acetylene (26); the m/z values used for quantification are in parenthesis.
 A calibration following each fourth sample was made by aliquoting different volumes of a standard gas mixture (1 to 10 mL) leading in a linear calibration line (r2 ≥ 0.99) for each compound. The detection limits (3 σ variation of a blank sample) were 0.2 to 6.8 pptv (ethane, propane, butane, iso-butane, pentane, and iso-pentane), 0.2 to 16.1 pptv (ethene, propene, 1-butene, trans-2-butene, cis- 2-butene), and 14 to 26 pptv (acetylene), and the precision was 5 to 10%.
 A 30 compound reference standard from the National Physical Laboratory (Teddington, United Kingdom) with a certified uncertainty range of 1.2 to 2.2% (95% confidence limit) for each compounds was used for absolute calibration. The calibration is currently cross-checked by measuring samples used during the Nonmethane Hydrocarbon Intercomparison Experiment (NOMHICE) [Apel and Calvert, 1994; Apel et al., 1994, 1999]. First comparisons show an agreement within ∼20% for all species addressed in this paper.
 Long-term laboratory tests (11 months) have revealed significant formation of alkenes in the sample canisters, a problem already discussed by Donahue and Prinn . Even in sample canisters kept at −18°C we detected production rates of ∼4 ppt month−1 for ethene and ∼2 ppt month−1 for propene. No significant changes for alkanes and acetylene were noticed. However, it is worth mentioning that the measured ethene (28 to 128 ppt) and propene (10 to 50 ppt) mixing ratios were highly correlated. This ethene to propene slope of 2.7 ppt / ppt (r2 = 0.90) stands in agreement with values determined by on-line measurements on the Atlantic Ocean of 2.5 in the SH and 2.9 in the NH [Rudolph and Johnen, 1990]. As we are not able to precisely determine how much the INDOEX alkene data were affected by storage artifacts, they are not further discussed.
2.3. Measurement of Long-lived Trace Gases, CO, δ18O(CO), and 14CO
 Analyses of CH4, CO2, N2O, and SF6 were made with an automated GC system (Hewlett Packard, HP 6890a) with flame ionization detector (FID), electron capture detector (ECD), and a HP 5790 nickel catalyst for reduction of CO2 to CH4 (set up by Atmospheric Environmental Service, Canada) [Bräunlich, 2000]. The absolute uncertainty of the measurements is 3 ppb for CH4, 0.3 ppm for CO2, 2 ppb for N2O, and 0.1 ppt for SF6. CH4, CO2, and N2O are given versus the NOAA scale, SF6 is calibrated on a GC system, described by Maiss et al. .
 The 600 L (STP) samples are analyzed for the concentration of CO and its isotopic composition (δ18O(CO), 14CO). The CO concentration is derived with an absolute measurement method, based on the conversion of CO into CO2 and subsequent volumetric measurement of the derived CO2. The 18O/16O isotope ratio of CO is determined by mass spectrometry, and the 14C content is analyzed via accelerator mass spectrometry [see Brenninkmeijer, 1993; Brenninkmeijer et al., 1999, 2001].
 The continuous measurements of CO and O3 [Stehr et al., 2002] were done with a TECO 48 CO and a TECO 49 O3 analyzer (Thermo Environmental Instruments Inc., Franklin, Massachusetts), respectively, here given as 30 min averages. The agreement between continuous CO and laboratory based CO measurements on the 600 L samples is generally better than 5 ppb.
3. Meteorology During INDOEX
 A short overview about the meteorological conditions encountered along the ship track and the major near-surface airflows over the Indian Ocean is given here.
3.1. Dominating Airflows
 A 10 year climatological study of the Indian Ocean presented in the meteorology overview by Verver et al.  pointed out that four prevailing air flows (F1, F2, F3, F4) advect (mainly continental) air to the Indian Ocean (Figure 1): F1, northeast (NE) trades over the western Arabian Sea, F2, NE-NW flow along the west coast of India, F3, NE trades over the west Bay of Bengal, and F4, NE flow from Southeast Asia. They are mainly driven by pressure difference developing between subtropical high-pressure systems stretched along 20°N (from Arabia to Southeast Asia) and organized clusters of convective systems around the equator (15°S to 10°N). The strength of the four airflows fluctuates on a day-to-day basis, strongly modulated by the actual weather regime.
 During INDOEX, F1 was generally the most active channel, but it was west of the region examined and thus was never passed by the Ronald Brown.F3 was very active up to 6 March (DOY 65) and was most likely responsible for the polluted air masses probed at ±2°N, 70°E during the end of leg 1 (27–28 February, DOY 58–59) and at 5–6°N, 70–73°E the beginning of leg 2a (5–6, March, DOY 64–65). Exceptionally, during the major part of leg 2 (7–22 March, DOY 66–81) and the entire leg 3 (26–30 March, DOY 85–89) the two eastern channels F3 and F4 were nearly absent, but channel F2 was (in contrast to February) quite active. These anomalies had crucial consequences for the trace gas composition probed during leg 2 and 3: (1) during 9–16 March (DOY 68–75), F2 brought midlatitude air (from 20°–45°N) to the measurement site at 2°–19°N, 68°–75°E (Arabian Sea), and (2) during leg 3 (taken place in the Bay of Bengal) polluted but aged continental air (earlier imported by F3 and F4) was intersected.
3.2. Location of the Intertropical Convergence Zone (ITCZ)
 In February-March 1999 the ITCZ was broken into two convergence zones (CZ) [Verver et al., 2001]. The northern CZ was more active in February at 1°S–7°N, and the southern CZ became dominant in March between 10°S, 60°E and 5°S, 100°E, along the region of highest sea surface temperatures of 28°–29°C. The transition zone between SH air and NH air, unambiguously identified by a sharp jump in several trace species, such as CO from ≤65 ppb (SH) to ≥95 ppb (NH), coincided with the northern CZ on 26 February (end of DOY 57) at ∼1°S and with the southern CZ encountered on 19–20 March (DOY 78–79) south of 9°S (Figure 1).
4. Results and Discussion
 As seen in Figure 1, the 93 sample locations are quite homogeneously distributed over the various legs with an average sample frequency of 2.3 canisters (9 L STP) per day. The time series of medium- and longer-lived trace gases show substantial variability (Figure 2) The highest concentrations were observed on 10 March (DOY 69), except for CO that peaked on 6 March (DOY 64.9). All trace gases minimized during 19–20 March (DOY 78–79) as the ship was south of the chemical ITCZ. These trace gas changes are expected to reflect predominantly the different origin of the air masses since the transport timescale (about 1 week) is shorter than the chemical lifetime for most of the shown trace gases. To support this claim, back trajectories were calculated, and the inferred air mass classification was subsequently compared with the trace gas signature observed in the different types of air masses.
4.1. Trajectory and Trace Gas Composition Based Air Mass Classification
 For each individual sampling location, diabatic 3-D 5-day back trajectories (ending at 950 hPa) were calculated with the Hybrid Single-Particle Langrangian Integrated Trajectories (HYSPLIT) program [Draxler and Hess, 1998] to estimate the origin of the probed air masses. On the basis of the air mass classification proposed by Rhoads et al.  and further specified by W. P. Ball et al. (unpublished manuscript, 2002), altogether five types of air masses were encountered during the INDOEX R/V Ronald Brown cruise. During legs 2 and 3, four air mass types were intersected: Northern Hemisphere continental tropical (NHcT), Northern Hemisphere continental extratropical (NHcX), Northern Hemisphere marine tropical (NHmT), and Southern Hemisphere marine tropical (SHmT). During legs 0 and 1, also Southern Hemisphere maritime extratropical (SHmX) air was temporarily intercepted. The identification of the individual types of air masses was done with respect to the starting area of the back trajectories by applying the following rules:
 NHcX air parcels originated from the continental extratropics (mainly from Arabian Peninsula, Middle East, and Europe) and were advected to the sampling site via the airflow F2 south- eastward along the coast of Pakistan and the Indian west coast. Trajectories that started over the Indian subcontinent were attributed to NHcT air. They reached the Indian Ocean mostly via F3 (e.g., by passing the Bay of Bengal) and to a minor degree via F2. Air parcels without contact to landmasses for more than 5 days were classified as NHmT or SHmT according to their origin (NH or SH). Air masses that originated from the zonal westerly flow of the southern Indian Ocean (south of 30 °S) were classified as SHmX air.
 Because of the known uncertainties of back trajectories [Stohl, 1998], in particular in areas with weak winds and near convective zones, daily stream fields (available for legs 0, 1, and 2 at http://www.joss.ucar.edu/indoex/catalog/) were additionally assessed to better distinguish between SHmT and NHmT air.
 The five types of air masses are assumed to show quite contrasting trace gas compositions, due to their advection from/over very different trace gas source regions. To confirm this claim, trace gas mean values and their variance in the five air mass types were inspected. For legs 0, 1, and 2 the assumption is excellently confirmed (see Table 2 and the following discussion), but for leg 3 (that only covered 5 days of measurements) some differences arose, as discussed later on. Figure 3 shows the source regions of the different types of air masses with typical 5-day back trajectories, which are described in detail as follows.
Table 2. Mixing Ratios (Mean and 1 σ Variation) for the Five Types of Air Masses Encountered by the R/V Ronald Brown During INDOEX
Number of NMHC measurements (only made for legs 2 and 3) are given in parentheses.
 NHmT (DOY 76.2–77.7 and 79.8–81.2) and SHmT (DOY 78.6–79.2) maritime air masses stayed over the Indian Ocean for at least 5 days (see example trajectories d and e in Figure 3), and the concentrations of most compounds were correspondingly low (Table 2). The transition from NHmT air to SHmT air (DOY 78–79), i.e., the crossing of the chemical ITCZ, is manifested by a distinct drop of all trace gases except O3. Although the transition between SHmT and NHmT during leg 2b is not illustrated by the back trajectories, it is seen from the stream fields that the ship was under strong influence of southern hemispheric air in the period classified as SHmT. The SHmT air probed during leg 0 (between Mauritius and the equator) were of more pristine origin than during leg 2 (DOY 78–79), demonstrated inter alia by the lower δ18O(CO) values.
 In NHcX (DOY 67.7–75.7) air masses most trace gases such as several hydrocarbons, CH4, O3, δ18O(CO), and 14CO showed the highest concentrations detected during INDOEX. They originated from the midlatitude free troposphere (<550 hPa) (Arabian Peninsula, Middle East, and Europe), only subsided near the coastline (trajectory c), and reached the sampling locations via F2. Up to 1866 ppt ethane, 325 ppt acetylene, 1821 ppb CH4, 53 ppb O3 (DOY 69.2), 4.8‰ δ18O(CO), and 17.2 14CO molecules cm-3 (STP, DOY 69.7) were observed (Figure 2). This event was not due to biomass burning, since the typical marker acetonitrile minimized [Wisthaler et al., 2002]. Most trace gas mixing ratios decreased while the distance to the coast lengthens (∼DOY 70–75.7). The isolated acetylene and CO peaks (DOY 73–75) were probably related to input from combustion processes [Wisthaler et al., 2002], supported by the trajectories which were tangent to the coastline. Compared to the relative uniform trace gas composition in the maritime regimes and the characteristic air mass origin of the NHcX regime, the NHcT air mass type composed of several quite different air parcels and a more detailed distinction is made (Table 3).
Table 3. Mixing Ratios (Mean and 1 σ Variation) for Air Masses Classified as NHcT
West Coast India
Bay of Bengal
Number of NMHC measurements are given in parentheses.
 During the NHcT “Calcutta” event (DOY 64.2–65.2) the air parcels that crossed the ship track had passed the highly populated area of Calcutta (trajectory a, via F3), and thus were influenced by combustion processes, manifested by high levels of CO, 18O(CO), C2H2, CO2, and CH4 (Figure 2). During the NHcT “west coast” event (DOY 65.7–67.2), weaker pollution was seen as the air parcels reached the sampling site via the west coast of India (trajectory b). Ethane and propane were comparable during both periods.
4.1.2. Leg 3
 The trajectory-based air mass classification attributed all air masses probed during leg 3 (occurred in the Bay of Bengal) to NHmT air since the trajectories stayed over the ocean for at least 5 days before encounter. The trace gas data, however, revealed a quite high degree of pollution, CO varied between 100 and 170 ppb, and acetylene varied between 60 ppt and 150 ppt. Meteorological analyses indicated synoptical disturbances with small-scale local winds and thus less reliable back trajectories for this time. The trace gas data suggest that the air masses were imported from southeast Asia via channels F3 (and F4) before examination. Conclusively, all air masses during leg 3 were assigned to NHcT air (although the back trajectories suggest NHmT air).
 For NHcT, “high NMHC” event (DOY 87.3–88.1), compared to NHmT air, acetylene and ethane were strongly elevated concurrent with slightly enhanced levels of CO and O3 (Tables 2 and 3), which most likely points to an aged biomass burning plume [Mauzerall et al., 1998; Singh et al., 2000]. A thunderstorm on early DOY 87 may have mixed several air masses, making back trajectory analysis unreliable (trajectory f).
 For NHcT, “low NMHC” event (DOY 88.5–88.6), a short, but sharp minimum in NMHC, O3 (Figure 2a), and SF6 (not shown) points to a maritime origin of this air mass, in contrast to the modestly enhanced levels of CO, CO2, and CH4. As only two samples are available and the trajectories changed during this time period the origin of these air masses could not be determined (trajectories g and h).
 For NHcT, “Bay of Bengal” event (DOY 88.9–89.3), tracers for combustion processes such as CO, δ18O(CO), C2H2, CO2, and CH4 were high (Table 3), in companion with enhanced levels of other NMHCs and O3. This again points to continental air masses imported into the Bay of Bengal by airflow F3.
 In summary, the five air mass types NHcT, NHcX, NHmT, SHmT, and SHmX crossed by the Ronald Brown show very different mean mixing ratios of the measured compounds (Table 2). The differences between these mean values are often higher than the variance within each type of air mass which confirms the suitable identification of the individual air masses. The mixing ratios of most longer-lived compounds (CH4, CO2, N2O, SF6, ethane, propane, acetylene) and the ratio C2H2/CO maximized in NHcX air, followed by NHcT, NHmT, SHmT, and SHmX air. For CH4, CO2, N2O, and SF6, the longest-lived and thus the most well-mixed trace gases examined, the interhemispheric gradient is much larger than the differences between the individual types of air masses within one hemisphere. The most pronounced pollution event (DOY 67.7–75.7, in sum 8 days) was attributed to long-range transport from the extratropics, which will be analyzed in more detail now.
4.2. Influence of Meridional Long-Range Transport on the Observed Trace Gas Variability
 It is analyzed now if these strong latitudinal gradients are, at least partially, responsible for the observed trace gas variability. Carbon monoxide 14 is an ideal tracer for the latitudinal origin of an air mass, as the nature of its main source and its sink lead to a clear latitudinal gradient in 14CO [Jöckel et al., 1999] with weak longitudinal variations. Carbon monoxide 14 is mainly produced by a well-defined diffuse source (cosmic radiation), its production rate varies only with latitude with a minimum in the tropics (changes in solar activity and solar proton events can be disregarded). The main sink of 14CO, the OH radical shows a meridional gradient with a maximum in the tropics. Typical NH background values for February-March are about 20 molecules cm−3 in the midlatitudes (47°N) [Gros et al., 2001] and about 9 molecules cm−3 in the tropics (13°N) [Mak and Southon, 1998].
 During INDOEX the NH 14CO concentrations (∼8 to 17 molecules cm−3, Figure 2b) are comparable to these values, although the latitudinal range is much smaller (∼3 to 18 °N), which already indicates that the observed large 14CO variation is related to the different latitudinal origins. This relationship is clearly illustrated in Figure 4 (except for samples 13–15 discussed later). Carbon monoxide 14 shows a compact correlation (r2 ≥ 0.99) with the latitudinal origin of the 5-day back trajectories in both hemispheres. Five-day back trajectories were used, being a compromise between the declining reliability of back trajectories with time [Stohl, 1998] and the typical transport timescale of Rossby waves that are mostly responsible for the advection of extratropical air masses to the Indian Ocean. Moreover, an excellent agreement is found for the gradients and the values with the estimated latitudinal 14CO gradient [Jöckel, 2000]. The foregoing reveals that the 14CO variability observed during INDOEX is almost exclusively caused by long-range transport, and that the available back trajectories are reliable. Furthermore, the high 14CO value of sample 7 impressively supports the extratropical origin of the air masses encountered around DOY 69.
 The 14CO excess detected in samples 13 to 15 (leg 3) shows the impact of (14C containing) biogenic CO (Figure 4), which is most probably related to biomass burning. This claim is supported by the relatively high δ18O(CO) values of those three samples (Figure 2b), as well as the enhanced C2H2 mixing ratios. Since 10 ppb CO of biogenic origin adds 0.38 14CO molecules cm−3 (STP air) [Brenninkmeijer, 1993], the 14CO excess of sample 15 would be equivalent to 85 ppb additional biogenic CO. This roughly agrees with the observed CO excess of sample 15 of ∼60 ppb when assuming the observed NHmT CO mixing ratio of (92.8 ± 6.9) ppb (Table 2) as background.
 The successful use of the back trajectories for 14CO is now applied on hydrocarbons. For this the mixing ratios of propane, ethane, and CH4 are displayed versus the latitudinal origin of the 5-day back trajectories (Figure 5). Two regimes become visible. First, low mixing ratios and no significant correlation with latitude south of ∼16 °N, in particular obvious for propane that is <25 ppt south of ∼16°N, representing background conditions. Second, progressively increasing mixing ratios with latitude north of ∼16 °N. This confirms that the observed trace gas variability primarily reflects the latitudinally varying origin of the sampled air masses and that 16°N is about the border between regions recently influenced by continental outflow and more pristine areas.
 Moreover, it demonstrates that long-range advection from the extratropical NH strongly contributes to the pollution of the Indian Ocean (at least during INDOEX) and that a significant fraction of the “NH winter plume” is decomposed at lower latitudes, in agreement with model results [Gupta et al., 1998]. Nevertheless, this finding does not imply that the extratropical NH is the major source of pollution to the Indian Ocean (via F2). Severe pollution from local sources, especially from the Calcutta region, was intersected near the tip of India while F3 was active (DOY 58–59, 64–65) and (as already noted) F3 and F4 are usually more persistent. Additionally, these recent air masses are expected to be rich in NOx (in contrast to aged NHcX air) which is required to generate O3.
 Worth mentioning is that the latitudinal CH4 gradient of (1.7 ± 0.2) ppb / °N (r2 = 0.66, except samples south of the chemical ITCZ, Figure 5) agrees reasonably with ∼2.2 ppb/ °N and ∼1.9 ppb / °N extracted from several latitudinal distributions (annual means, 1984–1993) [Dlugokencky et al., 1994] and retrieved from NOAA/CMDL data for March, 1999.
4.3. Discussion of the Use of the C2H2/CO Ratio As a Measure for Atmospheric Processing
 Acetylene and CO have comparable atmospheric cycles. Both are largely combustion products and are removed by the reaction with OH. Owing to the difference in lifetime against OH (for C2H2 ∼1/3 of that of CO, Table 1) photochemical aging of an air mass after its emissions leads to decreasing C2H2/CO ratios. A competing process also leading to a decrease of the C2H2/CO ratio is mixing. Thus C2H2/CO contains information on the sum of mixing and photochemistry (“atmospheric processing”) [McKeen and Liu, 1993]. Note that local sources or mixing with different air masses corrupt the C2H2/CO ratio.
 On the basis of this tool, Smyth et al. [1996, 1999] concluded that “mixing” strongly dominated over “chemistry” during the Pacific Exploratory Missions (PEM) in the free atmosphere. Rapidly decreasing C2H2/CO ratios with a half-lifetime of typically 1–3 days were recorded, which primarily reflected the speed of mixing with background air.
 It is demonstrated now that the information derived from the C2H2/CO ratio differs strongly from the results by Smyth et al. [1996, 1999] when applied to the measurements over the Indian Ocean. As seen in Figure 2, maximum C2H2/CO ratios of up to 2.3 ppt/ppb were encountered in air masses that originated from the extratropical free troposphere (around DOY 69) and subsided to the marine boundary layer near the coastline. According to Smyth et al.  such C2H2/CO ratios should mark air mass ages of <2 days, which seems to be unrealistic as the back trajectories show that the “source area” is more than 5 days away. The following conditions encountered during INDOEX and the PEM missions are different.
In the extratropical NH winter/early spring, C2H2 strongly accumulates (see section 4.2), so that the C2H2/CO background ratio is much higher than in the free troposphere over the Pacific (<0.5 ppt/ppb). Therefore mixing with background air has a smaller impact on the decline of the C2H2/CO ratio compared to the free troposphere over the Pacific. This claim is supported by observations by Roths and Harris  and Boissard et al. , who found 100–130 ppb CO and 300–400 ppt C2H2, respectively, in the free troposphere at 30°–45°N in January (Tropospheric Ozone Experiment (TROPOZ) II), which corresponds to a C2H2/CO ratio of ∼3 ppt/ppb. These contrasting conditions are reflected in the INDOEX data (Figure 6) Indeed, the air masses with the highest C2H2/CO ratios originate from the extratropics (see NHcX, Table 2, Figure 2a) where the C2H2/CO background ratio is much higher than in the tropics.
Mixing with background air is reduced in the planetary boundary layer compared to the free troposphere, which slows down the decrease of the C2H2/CO ratio with time. Note that Smyth et al. [1996, 1999] have omitted the measurements made in the boundary layer from their analysis.
 Conclusively, the C2H2/CO ratio during INDOEX was less affected by mixing compared to the free troposphere over the Pacific (i.e., during the PEM missions), where mixing was found to be the dominant effect influencing the C2H2/CO ratio. Therefore the relationship between the C2H2/CO ratio and air mass age given by Smyth et al. [1996, 1999] which is based on the dominance of mixing is not applicable during INDOEX for air masses advected from latitudes north of ∼16°N.
 Note that in areas which were not recently affected by extratropical air masses (south of ∼16°N), enhanced C2H2/CO ratios of up to 1.5 (on DOY 87), the “high NMHC” event (Table 3, Figure 2a), were related to an aged biomass burning plume (section 4.1.2, second paragraph). However, compared to PEM-West B, the C2H2/CO ratios were much lower (Table 4), which demonstrates that the air masses intersected by the R/V Ronald Brown were less affected by fresh emissions from combustion processes, which is supported by other trace gas data (see section 5).
Table 4. Mixing Ratios (Mean and 1 σ Variation) at Altitudes <2 km During PEM-West B (February-March 1994)
 Compared to the pre-INDOEX ship study by Rhoads et al.  (March-April 1995), more severe pollution was encountered during INDOEX in NH air masses, supported, among others, by the higher mixing ratios of CO and O3 [Stehr et al., 2002]. Tracers for combustion (see Table 1) peaked in the Arabian Sea (in agreement with Lal et al.  and Naja et al. ) and the Bay of Bengal, both areas not visited by Rhoads et al. . These regions are surrounded by densely populated areas and are affected by advection from the extratropics. Rhoads et al.  found two SH maritime regimes, based on a rise of CO and non-sea salt aerosol (but without a corresponding change in the back trajectories). The INDOEX trace gas data and the back trajectory calculations support the presence of two SH air mass types (SHmT and SHmX), see Table 2. The INDOEX data confirm the distinction in meteorological regimes made by Rhoads et al.  and W. P. Ball et al. (unpublished manuscript, 2002) and give further evidence for the importance of transport of anthropogenic emissions to the remote Indian Ocean.
 Numerous NHcT and most NHmT propane measurements are at the lower end of previous reports, which emphasizes the remote character already seen at the back trajectories. Only Rudolph and Johnen  and Atlas et al.  reported comparable low propane mixing ratios.
 During PEM-West B (February-March, 1994) where the Asian outflow to the Pacific Ocean was analyzed, distinctly higher concentrations of pollutants were encountered below 2 km altitude [Blake et al., 1997; Talbot et al., 1997]. All trace gases except CO2 were significantly higher (Table 4) in continentally influenced air masses (origin >20°N, classified by Talbot et al.  as continental north). Compared to the fresh continental outflow encountered during PEM below 5 km altitude, the air masses sampled by the Ronald Brown were aged, i.e., had no contact with landmasses for more than 4 days. Even less polluted air masses classified by Talbot et al.  as continental south (origin <20°N) showed higher levels of typical pollution tracers, e.g., CO and acetylene. Aged marine masses [Talbot et al., 1997] and NHmT air during INDOEX had a more comparable trace gas composition (Table 4) (except for propane), but the slightly higher acetylene and ethane mixing ratios document a less pristine character even for the aged maritime air.
 During the INDOEX R/V Ronald Brown cruise in the Indian Ocean (February-March 1999, track in Figure 1), pronounced variability in various medium- and long-lived trace gases, such as NMHC, CH4, CO, and CO2, was observed. This variability was partially related to nearby continental outflow from India and Southeast Asia via the airflows F3 and F4 (the four prevailing airflows are shown in Figure 1), but the strongest pollution event was caused by long-range transport from the extratropical NH (Middle East and probably even Europe) via airflow F2. In comparison to the PEM- West B expedition over the Pacific Ocean (February-March 1994), little fresh pollution was encountered by the R/V Ronald Brown; most air masses had no contact with landmasses for more than 4 days. Because of the atypical meteorological situation during INDOEX [Verver et al., 2001], it can be expected that the continental outflow from India and Southeast Asia via F3 and F4 is usually more persistent in the lower troposphere. Note that throughout the INDOEX campaign layers of fresh pollution were actually encountered, but mainly above the marine boundary layer (MBL), demonstrated by extensive airborne trace gas and aerosol measurements [Mayol-Bracero et al., 2002, Sheridan et al., 2002].
 The presence of long-range advection of pollutants in the Indian Ocean was excellently confirmed by the measured 14CO. Using 5-day back trajectories, a compact latitudinal gradient (r2 ≥ 0.99) of (0.29 ± 0.01) molecules cm−3/°N in the NH and (0.077 ± 0.004) molecules cm−3/°S in the SH was inferred, in excellent agreement with the 14CO climatology by Jöckel .
 For air masses that were imported from latitudes north of ∼16°N, the C2H2/CO ratio was found to be unsuitable as a measure for atmospheric processing as defined by Smyth et al., [1996, 1999] for the PEM missions. This difference is caused by the lower impact of mixing during INDOEX compared to PEM, where mixing was found to be the dominant effect influencing the C2H2/CO ratio. First, due to the strong accumulation of pollutants in the winter northern hemisphere and the corresponding elevated background C2H2/CO ratio of typically 3 ppt/ppb (much higher than the ∼0.5 ppt/ppb during PEM-West B), the impact of mixing on the C2H2/CO ratio is reduced. Second, mixing with background air in the marine boundary layer is small compared to the free troposphere. Therefore elevated C2H2/CO ratios during INDOEX mostly reflected import of air masses from the extratropics. However, south of ∼16°N enhanced C2H2/CO ratios, as observed in the Bay of Bengal, could be assigned to biomass burning, supported by other NMHC as well as the 14C/12C and 18O/16O isotope ratios of CO.
 The work was supported by the BMBF under the grant 01LA9831/2. We thank J. E. Johnson and J. W. Stehr for provision of the CO and O3 data, and the National Oceanic and Atmospheric Administration (NOAA) Climate Monitoring and Diagnostics Laboratory (CMDL) Carbon Cycle Group for reference data on CH4, CO, and CO2. We are grateful to R. Hofmann, W. Hanewacker, C. Koeppel, and R. Schmunck for their careful analyses. We especially thank the two anonymous reviewers for their valuable comments on the manuscript.