Evidence of convective transport in tropical West Pacific region during SHIVA experiment

Air masses in the convective outflows of four large convective systems near Borneo Island in Malaysia were sampled in the height range 11–13 km within the frame of the SHIVA (Stratospheric Ozone: Halogen Impacts in a Varying Atmosphere) FP7 European project in November and December 2011. Correlated enhancements of CO, CH4 and the short‐lived halogen species (CH3I and CHBr3) were detected when the aircraft crossed the anvils of the four systems. These enhancements were interpreted as the fingerprint of vertical transport from the boundary layer by the convective updraft and then horizontal advection in the outflow. For the four observations, the fraction f of air from the boundary layer ranged between 15 and 67%, showing the variability in transport efficiency depending on the dynamics of the convective system.

Air masses in the convective outflows of four large convective systems near Borneo Island in Malaysia were sampled in the height range 11-13 km within the frame of the SHIVA (Stratospheric Ozone: Halogen Impacts in a Varying Atmosphere) FP7 European project in November and December 2011. Correlated enhancements of CO, CH 4 and the short-lived halogen species (CH 3 I and CHBr 3 ) were detected when the aircraft crossed the anvils of the four systems. These enhancements were interpreted as the fingerprint of vertical transport from the boundary layer by the convective updraft and then horizontal advection in the outflow. For the four observations, the fraction f of air from the boundary layer ranged between 15 and 67%, showing the variability in transport efficiency depending on the dynamics of the convective system.

K E Y W O R D S
aircraft, convective transport, in situ measurements, VSLS transport

| INTRODUCTION
The composition of the tropical upper troposphere (UT) is affected by the efficiency of the convective transport of chemical species (Fueglistaler et al., 2009). Tropical deep convection can efficiently transport surface emitted compounds from the lower troposphere into the tropical tropopause layer (TTL) altitude range (Marécal, Rivière, Held, Cautenet, & Freitas, 2006). Since in tropical regions large emissions of halogenated very short-lived species (VSLS) coincide with deep convection, one may expect rapid transport of VSLS into the TTL. Indeed, an efficient transport of chemical tracers from polluted air masses (Bechara, Borbon, Jambert, Colomb, & Perros, 2010) or biogenic sources from the oceans such as halogenated VSLS (CHBr 3 , CH 2 Br 2 , CH 3 I, etc.: Sala et al., 2014;Tegtmeier et al., 2013) was observed and modelled (Navarro et al., 2015;Werner et al., 2017).
In the last decades, several field campaigns (such as SHIVA-Stratospheric Ozone: Halogen Impacts in a Varying Atmosphere, Sala et al., 2014;Fuhlbrügge et al., 2016;TC4 [www.nasa.gov/mission_pages/TC4]; ATTREX, Jensen et al., 2015) and modelling studies (Hossaini et al., 2012) focused on determining the contribution of the VSLS to the bromine burden in the upper troposphere and lower stratosphere (UTLS). Observational campaigns can only address the transport into the UT on an event-like basis but they are of high interest because the convection parameterisation is a major source of uncertainty in chemistry transport models (Arteta, Marécal, & Rivière, 2009;Hoyle et al., 2011). Previous observational studies of convective outflows at midlatitudes (Bertram et al., 2007) and in the tropics (Ray et al., 2004) reported an effective transport of short-lived trace gases (Bechara et al., 2010;Cohan, Schultz, Jacob, Heikes, & Blake, 1999) by convection from the boundary layer (BL) into the UT. These studies showed that a fraction of 20-40% of BL air was present in individual convective plumes transported rapidly to the UT.
One of the objectives of the SHIVA FP7 European project was the investigation of the atmospheric transport of VSLS from the BL into the UTLS. The campaign took place in the tropical West Pacific during the boreal winter monsoon in November and December 2011 when strong convective transport is expected to occur (e.g., Aschmann, Sinnhuber, Atlas, & Schauffler, 2009;Levine, Braesicke, Harris, Savage, & Pyle, 2007;Liang et al., 2014). Here, we present airborne measurements for four mesoscale convective events indicating enhanced UT CO and CH 4 volume mixing ratios (vmr) that directly correlate with CHBr 3 and CH 3 I enhancements when the aircraft passed the anvil of convective cumulus clouds. The fraction of air originating from the BL is calculated by analysis of the CO enhancements.
The SHIVA campaign and instruments are presented in Section 2. In Section 3, we discuss the meteorological conditions of the flights and the detection of convective transport and its influence on CO and CH 4 concentrations. Implications for the vertical transport for VSLS are also addressed. Section 4 concludes the study.

| Measurement campaign
The SHIVA aircraft campaign took place in Malaysia between November 16 and December 11, 2011. Using the German Aerospace agency (DLR) Falcon-20 aircraft, 16 research flights were conducted from Miri (Malaysia) airport in northwestern Borneo. In the present study, the results of four flights performed on November 19 (F19NOV), December 9, 2011 (F09DEC) and two on December 11 (F11DECa and F11DECb) are described in Appendix S1 (Supporting Information). The RV Sonne cruise started on November 15 in Singapore, passed near the northern coast of Borneo and ended in Manila, Philippines on November 29.
Measurements on Sonne are used to estimate the variability of halocarbons concentration in the BL.

| Experimental method
The airborne CO and CH 4 measurements were performed with the SPIRIT instrument (Catoire et al., 2017), and CHBr 3 and CH 3 I with the GHOST instrument (Sala et al., 2014). Additionally, whole air samples taken in the RV Sonne were analysed for halocarbons, CO and CH 4 . More detail about instrumentation is provided in Appendix S1. Relative humidity from the Falcon-20 instrument and webcam imagery from mini-DOAS instrument (Großmann, 2014) are used to study the convective condition.  Figure 1(a) indicates the presence of a well-developed convective system around 6 N and 115.5 E during research flight F19NOV that reached a maximum height of 16 AE 0.5 km in altitude with an extended anvil on its west side reaching 14.5 AE 0.5 km. For F09DEC, a convective system with a smaller horizontal extent was detected at around 5.5 N and 118.5 E (Figure 1(b)). The convective part of the system reached~15.5 AE 1 km altitude and was embedded in stratiform clouds with maximum height 13.5 AE 1.5 km. In F11DECa (Figure 1(c)), a welldeveloped convective system was probed between 1 -2 N and 106 -107 E. The cloud top altitude for this system reached a maximum of~17 AE 0.5 km. The convection cell lasted throughout the day and was again probed during the back flight from Singapore to Miri in the afternoon for F11DECb (Figure 1(d)), though with a weakened strength.
3.2 | Impact of deep convection on trace gases 3.2.1 | CO and CH 4 Figures 2-4 show CO and CH 4 measured by the SPIRIT instrument. In all cases when the aircraft crossed convective outflows (period determined by webcam data, relative humidity and brightness temperature, see Figure 2), the mixing ratios of the measured tracers are increased. In the next section, such measurements are defined as [X] UTconv , and the lower tracer mixing ratios observed outside of the convective system are defined as [X] UT . CO and CH 4 are mainly emitted from anthropogenic sources in the BL. The sudden increases of [X] UTconv , larger than the UT mixing ratios ([X] UT ), i.e., between 15 and 60 pbbv for CO and between 20 and 50 ppbv for CH 4 , are thus indicative of transport of polluted air from the BL into the UT. Such enhancements of BL tracers due to convection and affecting the UT composition have also been previously reported by Bechara et al. (2010) and Borbon et al. (2012).

| Fraction of BL air detected in the UT
CO has proven to be a particularly good tracer to study convection due to its source at the surface and tropospheric lifetime of 1-3 months (Dessler, 2002). Following Bertram et al. (2007), the measured tracer's mixing ratio [X] is used to quantify the air fraction f originating from the BL and transported by convection, using the following equation: where [X] BL represents the vmr of the tracer in the BL. For the air masses affected by convection, [X] UTconv is determined from the calculated means for CH 4 and CO. For F19NOV, [X] BL is determined from the air directly probed below the convective system during the take-off and landing since the convective system was located near Miri. For F09DEC, the surface air was directly sampled during a dive under the convective system down to 1 km altitude and for F11DECa and F11DECb, the BL measurements from RV Sonne are used. All relevant parameters are summarised in Table 1. In all, 18-50% of air present in the outflow of convective systems was recently transported from the BL, based on measured CO and CH 4 .

| Impact of deep convection on upper tropospheric CH 3 I and CHBr 3
For the four studied flights, Figures 2-4 (upper panel) show GHOST-MS measurements of CHBr 3 and CH 3 I for air affected by deep convection. As for CH 4 and CO, flight-dependent enhancements ranging from 0.3 to 0.5 pptv for CH 3 I and 0.6-1.0 pptv for CHBr 3 are observed in the air of convective outflow. Three areas are defined in order to calculate the BL mean concentration depending on the location of the flight by using a combination of GHOST and RV Sonne measurements. For F19NOV, [X] BL only takes into account measurements in the region northeast of Miri, for F09DEC the region on the eastern side of Borneo and for F11DEC the region east of Singapore. Averaging over the designated areas separately removes the    Table 1 summarises measured and averaged mixing ratios of all gases and fractions f for CHBr 3 and CH 3 I. Note, the calculated fractions f depend on the actual source strength at the marine boundary surface, which for CHBr 3 are known to strongly vary in space and time. Furthermore, since the time resolution of the CHBr 3 and CH 3 I measurements is longer than for CO and CH 4 measurements, the different averaging time may also affect the inferred fractions f by probably biasing them low relative to f calculated from other trace gases.

| Comparison with previous studies
Considering all species from all flights, a mean fraction of 29 AE 25% is obtained (mean of the fractions f with standard deviations σ < 0.4). Table 1 2006) and Bechara et al. (2010) given the range uncertainties. Like in our study, these authors used CO and CH 4 measurements among other tracers to calculate the fractions f. The fractions of Cohan et al. (1999) and Barth et al. (2016) using VSLS CH 3 I, CH 3 O 2 H, CHBr 3 and VOCs are in agreement with our results (15-67%) derived from CHBr 3 and CH 3 I. Also, from the previous studies related to SHIVA project, Großmann (2014) inferred a fraction of 19% for the short-lived species HCHO measured during F19NOV using a mini-DOAS instrument  and Fuhlbrügge et al. (2016) calculated similar contributions of marine BL air to the free troposphere (30-50%) up to 13 km height for the whole SHIVA-campaign with a trajectory model, again in agreement with the results of the present study.

| CONCLUSIONS
Within the frame of the SHIVA project, air of the anvil from mesoscale large convective systems was sampled at altitudes around 11-13 km near  and Singapore (1 N-106 E) on November 19, December 9 and 11, 2011, respectively. Correlated measurements of CO, CH 4 , CHBr 3 and CH 3 I were interpreted with respect to the strength of air mass transported from the BL to the UT by convective systems. The fraction f of BL air contained in the fresh convective outflow was calculated to range between 18 and 50% based on measured CO and CH 4 . Correlative measurements of CHBr 3 and CH 3 I indicated a fraction between 15 and 67%. The inferred range of f indicates the variability in mixing due to air mass entrainment into the convective system, but also points to limitations in the method due to its dependence on the variability of the tracer's source strength and lifetime of the species.
To go a step further, modelling or measurements from higher flying platforms, such as recently performed from the Global Hawk in the NASA ATTREX project over the Pacific, may provide estimates of the transport of halogenated VSLS due to deep convection reaching the TTL (e.g., Werner et al., 2017).