Delivery of halogenated very short-lived substances from the West Indian Ocean to the stratosphere during Asian summer monsoon

Abstract. Halogenated very short-lived substances (VSLSs) are naturally produced in the ocean and emitted to the atmosphere. When transported to the stratosphere, these compounds can have a significant influence on the ozone layer and climate. During a research cruise on RV Sonne in the subtropical and tropical west Indian Ocean in July and August 2014, we measured the VSLSs, methyl iodide (CH3I) and for the first time bromoform (CHBr3) and dibromomethane (CH2Br2), in surface seawater and the marine atmosphere to derive their emission strengths. Using the Lagrangian particle dispersion model FLEXPART with ERA-Interim meteorological fields, we calculated the direct contribution of observed VSLS emissions to the stratospheric halogen burden during the Asian summer monsoon. Furthermore, we compare the in situ calculations with the interannual variability of transport from a larger area of the west Indian Ocean surface to the stratosphere for July 2000–2015. We found that the west Indian Ocean is a strong source for CHBr3 (910 pmol m−2 h−1), very strong source for CH2Br2 (930 pmol m−2 h−1), and an average source for CH3I (460 pmol m−2 h−1). The atmospheric transport from the tropical west Indian Ocean surface to the stratosphere experiences two main pathways. On very short timescales, especially relevant for the shortest-lived compound CH3I (3.5 days lifetime), convection above the Indian Ocean lifts oceanic air masses and VSLSs towards the tropopause. On a longer timescale, the Asian summer monsoon circulation transports oceanic VSLSs towards India and the Bay of Bengal, where they are lifted with the monsoon convection and reach stratospheric levels in the southeastern part of the Asian monsoon anticyclone. This transport pathway is more important for the longer-lived brominated compounds (17 and 150 days lifetime for CHBr3 and CH2Br2). The entrainment of CHBr3 and CH3I from the west Indian Ocean to the stratosphere during the Asian summer monsoon is lower than from previous cruises in the tropical west Pacific Ocean during boreal autumn and early winter but higher than from the tropical Atlantic during boreal summer. In contrast, the projected CH2Br2 entrainment was very high because of the high emissions during the west Indian Ocean cruise. The 16-year July time series shows highest interannual variability for the shortest-lived CH3I and lowest for the longest-lived CH2Br2. During this time period, a small increase in VSLS entrainment from the west Indian Ocean through the Asian monsoon to the stratosphere is found. Overall, this study confirms that the subtropical and tropical west Indian Ocean is an important source region of halogenated VSLSs, especially CH2Br2, to the troposphere and stratosphere during the Asian summer monsoon.


Introduction
Natural halogenated volatile organic compounds in the ocean originate from chemical and biological sources like phytoplankton and macro algae (Carpenter et al., 1999;Quack and Wallace, 2003;Moore and Zafiriou, 1994;Hughes et al., 2011). When emitted to the atmosphere, the halogenated very short-lived substances (VSLS) have atmospheric lifetimes of less than half a 60 year (Law et al., 2006). Current estimates of tropical tropospheric lifetimes are 3.5, 17, and 150 days for methyl iodide (CH 3 I), bromoform (CHBr 3 ), and dibromomethane (CH 2 Br 2 ), respectively (Carpenter et al., 2014). VSLS can be transported to the stratosphere by tropical deep convection, where they contribute to the halogen burden, take part in ozone depletion and thus impact the climate (Solomon et al., 1994;Dvortsov et al., 1999;Hossaini et al., 2015). 65 CHBr 3 is an important biogenic VSLS due to its large oceanic emissions and because it carries three bromine atoms per molecule into the atmosphere (Quack and Wallace, 2003;Hossaini et al., 2012). CH 2 Br 2 has a longer lifetime than CHBr 3 , and thus a higher potential for stratospheric entrainment. CH 3 I is an important carrier of organic iodine from the ocean to the atmosphere and the most abundant organic iodine compound in the atmosphere (Manley et al.,70 1992; Moore and Groszko, 1999;Yokouchi et al., 2008). Despite its very short atmospheric lifetime, it can deliver iodine to the stratosphere in tropical regions (Solomon et al., 1994;Tegtmeier et al., 2013). Ship-based observations showed that bromocarbon emissions near coasts and in oceanic upwelling regions are generally higher than in the open ocean, because of macro algal growth near coasts (Carpenter et al., 1999) and enhanced primary production in 75 upwelling regions (Quack et al., 2007), while coastal anthropogenic sources also need to be considered (Quack and Wallace, 2003;Fuhlbrügge et al., 2016b). Measurements of VSLS in the global oceans are sparse and the data shows a large variability. Thus, attempts at creating observation based global emission estimates and climatologies (bottom-up approach) (Quack and Wallace, 2003;Butler et al., 2007;Palmer and Reason, 2009;Ziska et al., 2013), modeling the 80 global distribution of halogenated VSLS emissions from atmospheric abundances (the top-down approach) Liang et al., 2010;Ordóñez et al., 2012), as well as biogeochemical modeling of oceanic concentrations (Hense and Quack, 2009;Stemmler et al., 2014) are subject to large uncertainties (Carpenter et al., 2014). Global modeled top-down estimates Liang et al., 2010;Ordóñez et al., 2012) yield higher emissions 85 Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-8, 2017 Manuscript under review for journal Atmos. Chem. Phys. Published: 11 January 2017 c Author(s) 2017. CC-BY 3.0 License. than bottom-up estimates (Ziska et al., 2013;Stemmler et al., 2014Stemmler et al., , 2015, which may indicate the importance of localized emission hot spots underrepresented in current bottom-up estimates. The amount of oceanic bromine from VSLS entrained into the stratosphere is estimated to be 2-8 ppt, which is 10-40 % of the currently observed stratospheric bromine loading (Dorf et al., 2006;Carpenter et al., 2014). This wide range results mainly from uncertainties in tropospheric 90 degradation and removal, transport processes, and especially from the spatial and temporal emission variability of halogenated VSLS (Carpenter et al., 2014;Hossaini et al., 2016).
Analyzing the time period 1993-2012, Hossaini et al. (2016) found no clear long-term transportdriven trend in the stratospheric injection of oceanic bromine sources during a multi-model intercomparison. 95 Transport processes strongly impact stratospheric injections of VSLS, because their lifetimes are comparable to tropospheric transport timescales from the ocean to the stratosphere.
The main entrance region of tropospheric air into the stratosphere is above the tropical West Pacific. Another active region lies above the Asian monsoon region during boreal summer (Newell and Gould-Stewart, 1981), when the Asian monsoon circulation provides an efficient 100 transport pathway from the atmospheric boundary layer to the lower stratosphere (Park et al., 2009;Randel et al., 2010). Above India and the Bay of Bengal, convection lifts boundary layer air rapidly into the upper troposphere (Park et al., 2009;Lawrence and Lelieveld, 2010). As a response to the persistent deep convection, an anticyclone forms in the upper troposphere and lower stratosphere above Central, South and East Asia (Hoskins and Rodwell, 1995). This so-105 called Asian monsoon anticyclone confines the air masses that have been lifted to this level within the anticyclonic circulation (Park et al., 2007;Randel et al., 2010). For the period 1951-2015, a decreasing trend in rainfall and thus convection has been reported over northeastern India caused by a weakening northward moisture transport over the Bay of Bengal (Latif et al., 2016).
Chemical transport studies in the Asian monsoon region have mostly focused on water 110 vapor entrainment to the stratosphere (Gettelman et al., 2004;James et al., 2008), or the transport of anthropogenic pollution (Park et al., 2009). The chemical composition and source regions for air masses in the Asian monsoon anticyclone have been the topic of more recent studies (Bergman et al., 2013;Vogel et al., 2015;Yan and Bian, 2015). Chen et al. (2012) investigated air mass boundary layer sources and stratospheric entrainment regions based on a climatological 115 domain-filling Lagrangian study in the Asian summer monsoon area. The West Pacific Ocean Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-8, 2017 Manuscript under review for journal Atmos. Chem. Phys. Published: 11 January 2017 c Author(s) 2017. CC-BY 3.0 License. and the Bay of Bengal are found to be important source regions, while maximum stratospheric entrainment occurred above the tropical West Indian Ocean.
The Asian monsoon circulation could be an important pathway for the stratospheric entrainment of oceanic VSLS (Hossaini et al., 2016), because the steady southwest monsoon 120 winds in the lower troposphere during boreal summer deliver oceanic air masses from the tropical Indian Ocean towards India and the Bay of Bengal (Lawrence and Lelieveld, 2010), where they are lifted by the monsoon convection and the Asian monsoon anticyclone. However, little is known about the emission strength of VSLS from the Indian Ocean and their transport pathways.
A few measurements in the Bay of Bengal (Yamamoto et al., 2001) and Arabian Sea (Roy et al.,125 2011) as well as global source estimates suggest that the Indian Ocean might be a considerable source (Liang et al., 2010;Ziska et al., 2013). No bromocarbon data is available for the equatorial and southern Indian Ocean, yet, but CH 3 I, which has been measured around the Mascarene Plateau, showed high oceanic concentrations (Smythe-Wright et al., 2005). Liang et al. (2014) use a Chemistry Climate Model for the years 1960 to 2010 and project that the tropical Indian

130
Ocean delivers more bromine to the stratosphere than the tropical Pacific because of its higher atmospheric surface concentrations based on the global top-down emission estimate by Liang et al. (2010).
In this study, we show surface ocean concentrations and atmospheric mixing ratios of the halogenated VSLS CH 3 I, and for the first time for CHBr 3 and CH 2 Br 2 , in the subtropical and 135 tropical West Indian Ocean during Asian summer monsoon. We use the Lagrangian transport model Flexpart to investigate the atmospheric transport pathways of observation-based oceanic VSLS emissions to the stratosphere.
Our questions for this study are: Is the tropical Indian Ocean a source for atmospheric VSLS? What is the transport pathway from the West Indian Ocean to the stratosphere during the 140 Asian summer monsoon? How much VSLS are delivered from the West Indian Ocean to the stratosphere during Asian summer monsoon? How large is the interannual variability of this VSLS entrainment?
In Sect. 2, we describe the cruise data and the transport model simulations. In Sect. 3, the results from the cruise measurements and trajectory calculations are shown and discussed. Then 145 the spatial and interannual variability of transport is presented in Sect. 4. In Sect. 5, we address uncertainties before summarizing the results and concluding in Sect 6. Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-8, 2017 Manuscript under review for journal Atmos. the tropical open West Indian Ocean, and were designed to cover biologically productive and non-productive regions (Fig. 1). In the following, we will refer to the combined cruises as the "OASIS cruise".
We collected meteorological data from ship based sensors including surface air temperature (SAT), relative humidity, air pressure, wind speed and direction taken every second 165 at about 25 m height on RV Sonne. Sea surface temperature (SST) and salinity (SSS) were measured in the ship's hydrographic shaft at 5 m depth. We averaged all parameters to 10 minute intervals for our investigations.
During the cruise, we launched 95 radiosondes and thus obtained high resolution atmospheric profiles of temperature, wind, and humidity. During the first half of the cruise, to improve meteorological reanalyses (e.g. European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis Interim (ERA-Interim)) and operational forecast models (e.g. opECMWF (operational ECMWF)) in the subtropical and tropical West Indian Ocean. Trace gas emissions are generally well mixed within the marine atmospheric boundary 180 layer (MABL) on timescales of an hour or less by convection and turbulence (Stull, 1988). We determined the stable layer that defines the top of the MABL with the practical approach described in Seibert et al. (2000). From the radiosonde ascent we computed the vertical gradient of virtual potential temperature, which indicates the stable layer at the top of the MABL with positive values. A detailed description of our method can be found in Fuhlbrügge et al. (2013). 185 We collected a total of 213 air samples with a 3-hourly resolution at about 20 m height above sea level. These samples were pressurized to 2 atm in pre-cleaned stainless steel canisters with a metal bellows pump, and they were analyzed within 6 months after the cruise. Details about the analysis, the instrumental precision, the preparation of the samples, and the use of We collected 154 water samples every three hours from the hydrographic shaft of RV Sonne at a depth of 5 m. The samples were then analyzed for halogenated compounds using a purge and trap system onboard, attached to a gas chromatograph with electron capture detector.
Analytical reproducibility of 10 % was determined from measuring duplicate water samples.

195
Calibration was performed with a liquid mixed-compound standard prepared in methanol. Details of the procedure can be found in Hepach et al. (2016).
The sea-air flux (F) of the VSLS was calculated from the transfer coefficient ( ) and the concentration gradient ( ) according to Eq. (1). The gradient is between the water concentration ( ) and theoretical equilibrium water concentration ( ), which is derived from the 200 atmospheric concentration ( . We use Henry's law constants (H) of Moore and coworkers (Moore et al., 1995a;Moore et al., 1995b). (1) Compound-specific transfer coefficients were determined using the air-sea gas exchange . This wind speed is derived from a logarithmic wind profile using the von Kármán constant ( ), the neutral drag coefficient (C d ) from Garratt (1977), and the 10 min average of the wind speed (u(z)) measured at z = 25 m during the cruise (Eq. (3)):

Trajectory calculations
For our trajectory calculations, we use the Lagrangian particle dispersion model Flexpart of the Norwegian Institute for Air Research in the Department of Atmospheric and Climate Research (Stohl et al., 2005), which has been evaluated in previous studies (Stohl et al., 1998;Stohl and 215 Trickl, 1999). The model includes moist convection and turbulence parameterizations in the atmospheric boundary layer and free troposphere (Stohl and Thomson, 1999;Forster et al., 2007).
In this study, we employ the most recently released version 9.2 of Flexpart. We use the ECMWF reanalysis product ERA-Interim (Dee et al., 2011) with a horizontal resolution of 1° x 1° and 60 vertical model levels as meteorological input fields, providing air temperature, winds, boundary 220 layer height, specific humidity, as well as convective and large scale precipitation with a 6-hourly temporal resolution. The vertical winds in hybrid coordinates were calculated mass-consistently from spectral data by the pre-processor (Stohl et al., 2005). We record the transport model output every 6 hours.
We ran the Flexpart model with three different setups, which are described in Table 1.
We calculate OASIS backward trajectories from the 12 UTC locations of RV Sonne during the cruise. These trajectories are later used to determine the source regions of air masses investigated along the cruise track.

230
With the OASIS setup, we study the transport of oceanic CHBr 3 , CH 2 Br 2 , and CH 3 I emissions from the measurement locations into the stratosphere similar as was carried out in the corresponding study by Tegtmeier et al. (2012). At every position along the cruise track at which emissions were calculated (Section 2.1), we release a mass of the compound equal to a release from 0.0002° x 0.0002° in one hour. The mass is evenly distributed among 10,000 trajectories. Tongue .
The transport calculations based on the measured emissions from OASIS give insight into the contribution of oceanic emissions to the stratosphere during Asian summer monsoon.

VSLS observations and oceanic emissions
CHBr 3 , CH 2 Br 2 , and CH 3 I surface ocean concentrations, atmospheric mixing ratios, and emissions for the OASIS cruise are plotted as time series in Fig. 2c-e, and are summarized in 295   Table 2. We calculated oceanic emissions from the synchronized measurements of surface water concentration and atmospheric mixing ratio as described in Section 2.1 ( Fig. 2e and Fig 1). High During the second part of the cruise, the atmospheric mixing ratios of CHBr 3 and CH 2 Br 2 increased from south to north and in the direction of the wind maximizing close to the equator 345 (Fig. 2d). The emissions were high between Mauritius and the equator (Fig. 2e). This suggests that the air around the equator was enriched by the advection of the oceanic emissions with the trade winds from south to north. We assume that the bromocarbons accumulate because of the steady wind directions and the suppression of mixing into the free troposphere by the top of the MABL and the trade inversion layer (Fig. S2, July 27-Aug. 2) acting as strong transport barriers 350 for VSLS as was observed for the Peruvian upwelling (Fuhlbrügge et al., 2016a).

Comparison of OASIS VSLS emissions with other oceanic regions
Average emissions of the three VSLS from OASIS and other tropical cruises and modeling studies are summarized in Table 3 (Hepach et al., 2014), and 1.5 times higher than during MSM18/3 in the Atlantic equatorial upwelling . Only the SHIVA campaign decayed before reaching the stratosphere. In the following, we will use the expressions VSLS transit time, which is the transit time including loss processes of the VSLS in the atmosphere during the transport, and transit half-life, which is the time after which half of the total entrained compound has reached 17 km. We also calculated the relative emission and entrainment by regime. Table 4 displays the absolute and relative emissions and entrainment, the transport 415 efficiency, and the transit half-life for the whole cruise and the four regimes.
The mean sea surface release of CHBr 3 in Flexpart is 0.43 µmol (on 0.0002° x 0.0002° hr -1 ) during the cruise and the mean entrainment to the stratosphere is 5.5 nmol resulting in a mean transport efficiency of 1.3 %. CH 2 Br 2 has a higher transport efficiency of 6.4 % with mean emissions of 0.43 µmol (on 0.0002° x 0.0002° hr -1 ), and very high stratospheric The transport efficiency for all three compounds is highest in the Local Convection regime (CHBr 3 ~3 %, CH 2 Br 2 ~9 %, and CH 3 I ~1 %), because this regime has the shortest transit half-life for all three VSLS. For CH 3 I, the compound with the shortest lifetime, the fast transport plays the largest role, and thus this regime is by far the most efficient.  Although these emissions occur in the subtropics, they reach 17 km mainly in the tropics (Fig.   S3). The transport efficiency of 4 % still allows a large amount of 345 nmol CH 2 Br 2 to enter the stratosphere from the maximum emissions at 23 UTC on July 12, 2014 (Fig. 4). CH 3 I absolute entrainment (2.8 nmol, 79 %) is highest in the Local Convection regime, because of both highest emissions and highest transport efficiency (Table 4). for CHBr 3 is shown in higher transport efficiencies of 7.9 %, which lead to an entrainment of 48.4 nmol CHBr 3 (Table 5), because of the strong convective activity in that region during the time (Fuhlbrügge et 460 al., 2016b). The MSM18/3 cruise in the equatorial Atlantic  has the smallest emissions, entrainment, and a transport efficiency of 0.8 % (Table 5). Overall, the comparison indicates that more CHBr 3 was entrained to the stratosphere from the tropical West Pacific than from the tropical West Indian Ocean during Asian summer monsoon, using available in situ emissions and 6-hourly meteorological fields. This is in contrast to the study by Liang et al. 465 (2014), who projected with a chemistry climate model climatology that emissions from the tropical Indian Ocean deliver more brominated VSLS into the stratosphere than tropical West Pacific emissions.
CH 2 Br 2 entrainment to the stratosphere for the TransBrom ship campaign was ~8 nmol with 470 transport efficiencies of 15 % (Tegtmeier et al., 2012). This is much higher than the Indian Ocean transport efficiency of 6.4 %, but the absolute entrainment of 23.6 nmol CH 2 Br 2 we calculated for the OASIS cruise (  Table 4), applying a uniform lifetime profile of 3.5 days, is lower than in the West Pacific, but higher than in the Atlantic.

480
Uncertainties of VSLS emissions and modeling their transport to the stratosphere will be further discussed in Sect. 5.

4
General transport from West Indian Ocean to the stratosphere 4.1 Spatial variability of stratospheric entrainment 485 We calculate the entrainment at 17 km for CHBr 3 , CH 2 Br 2 , and CH 3 I tracers by weighting the trajectories from the West Indian Ocean release region for July 2000-2015 with the transit-timedependent atmospheric decay plotted in Fig. 5. A summary of transport efficiency, transit halflife, and entrainment correlations for all three VSLS can be found in Table 6. Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-8, 2017 Manuscript under review for journal Atmos. Chem. Phys. Published: 11 January 2017 c Author(s) 2017. CC-BY 3.0 License.
The distribution of VSLS transit times shows that the shorter the lifetime of a compound, 490 the more important is the transport on short timescales (Fig. 5). For CHBr 3 , CH 2 Br 2 , and CH 3 I tracers the transit half-lifes are 8.5, 27.2, and 1.9 days, respectively (Table 6). For the two bromocarbons, the transit time distribution shows two maxima, one for the 0-2 days bin, and the second between 4-10 days for CHBr 3 and 6-12 days for CH 2 Br 2 . CH 3 I tracer entrainment occurs mainly on timescales up to 2 days (Fig. 5). tropical West Indian Ocean, and belongs to the Local Convection regime. We define these two regions to enclose the core entrainment and to be evenly sized (colored boxes in Fig. 6).
The larger West Indian Ocean release area and longer time series analysis (Table 6) confirms the results of our OASIS analysis (Table 4). The longer-lived VSLS tracers (CHBr 3 and CH 2 Br 2 ) are mainly entrained through the Monsoon Circulation regime, while the Local 505 Convection regime is more important for the shortest-lived tracer (CH 3 I). Chen et al. (2012) also identified these two stratospheric entrainment regions, analyzing the air transport from the atmospheric boundary layer to the tropopause layer in the Asian Summer monsoon region for a 9 year climatology. Additionally, they registered entrainment over the West Pacific Ocean, but the Local Convection entrainment above the central Indian Ocean 510 was by far the strongest. Similar to our VSLS transit times, the study of Chen et al. (2012) found very short transport timescales of 0-1 days in the equatorial West Indian Ocean, while transit times above the Bay of Bengal and northern India were between 3 and 9 days.

515
The time series of stratospheric entrainment from the West Indian Ocean to the stratosphere shows interannual variability for all three VSLS tracers (Fig. 7). Overall, July 2014 revealed high entrainment for CHBr 3 and CH 2 Br 2 tracers and low entrainment for CH 3 I tracer. The coefficient of variation (CV) for total entrainment is 0.13, 0.09, and 0.21 for CHBr 3 , CH 2 Br 2 and CH 3 I, respectively. Thus, the shortest-lived compound CH 3 I has the strongest interannual variation, the 520 longest-lived CH 2 Br 2 the weakest variation. In order to analyze which transport regime has a stronger influence on the total entrainment variability, we correlated the interannual entrainment time series of total entrainment with the entrainment in the Monsoon Circulation and Local Convection regimes (Table 6) correlation for all three compounds, suggesting that more entrainment in one regime is related to less entrainment in the other (Fig. 7).
The interannual time series of total VSLS tracer entrainment displays a small increase over time. This increase is independent of the chosen entrainment height (between 13 km and 19 km, and weaker for the other two compounds. It arises mainly from an increase of entrainment in the Monsoon Circulation regime (Fig. 7). Analyzing changes of rainfall revealed an increase in precipitation over northeastern India for the time interval of our transport study (Latif et al., 2016;Preethi et al., 2016). This indicates an increase of convection in our Monsoon Circulation regime over the years from 2000 to 2015, which can explain the increase in stratospheric 540 entrainment. However, for the long time period from the 1950s to the 2010s the same authors found a decrease of precipitation over the above mentioned area, potentially impacting the VSLS entrainment to the stratosphere.
In a follow-up study we will investigate the influence of the seasonal cycle of the Asian Monsoon circulation and interannual influences through atmospheric circulation patterns on the 545 West Indian Ocean VSLS entrainment to the stratosphere in more detail.

Uncertainties in the analysis
This study confirms that the subtropical and tropical West Indian Ocean is a source region of oceanic halogenated VSLS to the stratosphere during the Asian summer monsoon. The amount of 550 VSLS entrained depends on the emission strength, the lifetime of the compound, and the transport of trajectories in the regime, which have been quantified in this study. Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-8, 2017 Manuscript under review for journal Atmos. Chem. Phys. Published: 11 January 2017 c Author(s) 2017. CC-BY 3.0 License.
However, uncertainties of this study are present in various aspects of the analysis. The uncertainties result from the calculation of VSLS emissions, the Flexpart transport using ERA-Interim reanalysis fields, and the definition of entrainment to the stratosphere.

555
The calculation of VSLS emissions from the concentration gradient between the ocean at 5 m depth and the atmosphere at 20 m height is subject to measurement uncertainties and a possible different concentration gradient directly at the air-sea interface. Also the applied windspeed-based parameterization for air-sea flux, which represents a reasonable mean of the published parameterizations, is uncertain by more than a factor of two (Lennartz et al., 2015).

560
Both factors may lead to a systematic flux under-or overestimation in our study.
A vital part of this study is the meteorological reanalysis data ERA-Interim and the Flexpart model for determining the VSLS transport. With delivery of our radiosonde launches to the GTS we have improved the data coverage over the Indian Ocean for the time of our study, and thus the quality of meteorological reanalysis. Indeed, horizontal wind speed and direction 565 from ship sensors and sondes agree well with the ERA-Interim data (Fig. S1). As the scale of tropical convection is below the state of the art grid-scale of global atmospheric models it is not sufficiently resolved and must be parameterized. The Lagrangian model Flexpart uses a convection scheme, described and evaluated by Forster et al. (2007), to account for vertical transport. Using Flexpart trajectories with ERA-Interim reanalysis, Fuhlbrügge et al. (2016b) 570 were able to simulate VSLS mixing ratios from the surface to the free troposphere up to 11 km above the tropical West Pacific in very good agreement with corresponding aircraft measurements applying a simple source-loss approach. Tegtmeier et al. (2013) showed that Tropopause (CPT) is commonly used as boundary between the troposphere and the stratosphere (Carpenter et al., 2014). The average measured CPT height during OASIS was 17 km (Fig. S2), but it can be up to 17.6 km high within the Asian monsoon anticyclone during boreal summer season (Munchak and Pan, 2014). To test the sensitivity of our results with regard to the Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-8, 2017 Manuscript under review for journal Atmos. Chem. Phys. Asian monsoon anticyclone, being entrained to stratospheric levels in its southeastern part. The transport to the stratosphere in this regime is effective for CHBr 3 and CH 2 Br 2 (2 % and 8 % transport efficiency, respectively), but less effective for CH 3 I (0.3%) as its lifetime is shorter than the transport timescale. Absolute CHBr 3 entrainment from the OASIS cruise was strongest in the

Data availability
The underlying data will be available at the open-access library Pangaea (http://www.pangaea.de).

Author contribution
A. Fiehn, K. Krüger and B. Quack designed the experiments and A. Fiehn carried them out. All coauthors were involved in the VSLS measurements and analyses taken during the OASIS cruise.

665
A. Fiehn, K. Krüger and B. Quack prepared the manuscript with contributions from all co-authors.

Competing interests
The authors declare that they have no conflict of interest.        . 6).
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-8, 2017 Manuscript under review for journal Atmos. Chem. Phys. Published: 11 January 2017 c Author(s) 2017. CC-BY 3.0 License. Figure 8: Schematic illustration of emission, transport pathways and timescales, and entrainment of CH 3 I, CHBr 3 , and CH 2 Br 2 tracer from the tropical West Indian Ocean to the stratosphere during the Asian summer monsoon.