Tracer measurements in the tropical tropopause layer during the AMMA / SCOUT-O 3 aircraft campaign

Tracer measurements in the tropical tropopause layer during the AMMA/SCOUT-O3 aircraft campaign C. D. Homan, C. M. Volk, A. C. Kuhn, A. Werner, J. Baehr, S. Viciani, A. Ulanovski, and F. Ravegnani Institute for Atmospheric and Environmental Sciences, Goethe University Frankfurt, Germany Istituto Nazionale di Ottica Applicata, Florence, Italy Central Aerological Observatory, Dolgoprudny, Russia CNR Institute of Atmospheric Science and Climate, Bologna, Italy now at: Department of Physics, University of Wuppertal, Germany Received: 29 September 2009 – Accepted: 12 November 2009 – Published: 25 November 2009 Correspondence to: C. D. Homan (c.homan@iau.uni-frankfurt.de) Published by Copernicus Publications on behalf of the European Geosciences Union.

Abstract.We present airborne in situ measurements made during the AMMA (African Monsoon Multidisciplinary Analysis)/SCOUT-O3 campaign between 31 July and 17 August 2006 on board the M55 Geophysica aircraft, based in Ouagadougou, Burkina Faso.CO 2 and N 2 O were measured with the High Altitude Gas Analyzer (HAGAR), CO was measured with the Cryogenically Operated Laser Diode (COLD) instrument, and O 3 with the Fast Ozone ANalyzer (FOZAN).
We analyse the data obtained during five local flights to study the dominant transport processes controlling the tropical tropopause layer (TTL, here ∼350-375 K) and lower stratosphere above West-Africa: deep convection up to the level of main convective outflow, overshooting of deep convection, and horizontal inmixing across the subtropical tropopause.Besides, we examine the morphology of the stratospheric subtropical barrier.
Except for the flight of 13 August, distinct minima in CO 2 mixing ratios indicate convective outflow of boundary layer air in the TTL.The CO 2 profiles show that the level of main convective outflow was mostly located at potential temperatures between 350 and 360 K, and for 11 August reached up to 370 K.
While the CO 2 minima indicate quite significant convective influence, the O 3 profiles suggest that the observed convective signatures were mostly not fresh, but of older origin (several days or more).When compared with the mean O 3 Correspondence to: C. D. Homan (c.homan@iau.uni-frankfurt.de)profile measured during a previous campaign over Darwin in November 2005, the O 3 minimum at the main convective outflow level was less pronounced over Ouagadougou.Furthermore O 3 mixing ratios were much higher throughout the whole TTL and, unlike over Darwin, rarely showed low values observed in the regional boundary layer.
Signatures of irreversible mixing following overshooting of convective air were scarce in the tracer data.Some small signatures indicative of this process were found in CO 2 profiles between 390 and 410 K during the flights of 4 and 8 August, and in CO data at 410 K on 7 August.However, the absence of expected corresponding signatures in other tracer data makes this evidence inconclusive, and overall there is little indication from the observations that overshooting convection has a profound impact on gas-phase tracer TTL composition during AMMA.
We find the amount of photochemically aged air isentropically mixed into the TTL across the subtropical tropopause to be not significant.Using the N 2 O observations we estimate the fraction of aged extratropical stratospheric air in the TTL to be 0.0±0.1 up to 370 K during the local flights.Above the TTL this fraction increases to 0.3±0.1 at 390 K.
The subtropical barrier, as indicated by the slope of the correlation between N 2 O and O 3 between 415 and 490 K, does not appear as a sharp border between the tropics and extratropics, but rather as a gradual transition region between 10 • N and 25 • N where isentropic mixing between these two regions may occur.

Introduction
The tropical tropopause layer (TTL), the region in the tropics where air has characteristics of both the stratosphere and the troposphere, is a critical region of the atmosphere.It is the main region for air entering the stratosphere and therefore it sets the chemical boundary conditions for the stratosphere.The TTL is commonly defined as the layer extending from the level of main convective outflow at 10-14 km to the cold point tropopause at about 16-19 km altitude (Gettelman and Forster, 2002).Laterally, the TTL is bounded by the position of the subtropical jets.Processes active in the TTL include the global-scale circulations in the stratosphere (Brewer-Dobson) and troposphere (Hadley-Walker), largescale and small-scale atmospheric waves, convection organised on all time and space scales, mixing, radiative heating and cooling, chemistry and cloud microphysics.
Satellite observations and 2-D or 3-D transport models provide a general picture of the TTL, but typically cannot well resolve spatial, in particular vertical, variations within the layer.Hence, highly resolved in situ measurements of trace gases and their use as tracers of atmospheric transport have proven extremely useful for studying the radiative, chemical, and dynamical properties of the TTL (e.g., Park et al., 2007a).Tracers are compounds whose lifetimes are longer than the timescales of the processes transporting them, such that their distributions are mainly determined by dynamical processes.In the stratosphere, the correlation between two long-lived tracers (tracer-to-tracer correlation) often turns out to be very compact (Plumb and Ko, 1992).The form and curvature of the correlation curve is dependent on the lifetimes of the tracers and on transport processes.Different atmospheric regions may exhibit different correlation curves between tracers.Variations from these correlations may be caused by horizontal or vertical mixing between the regions or by chemistry.Hence, tracer-to-tracer correlations are a particularly useful tool to study transport and mixing between the troposphere and the stratosphere (e.g., Hoor et al., 2002), or between the tropical and extratropical stratosphere (e.g., Volk et al., 1996).
Over the last decade various field campaigns have taken place to obtain in situ observations of the TTL and get more insight into the different processes taking place in this layer of the atmosphere.Several European campaigns were conducted with the Russian M55-Geophysica high-altitude aircraft.The APE-THESEO campaign took place in February and March 1999 over the western equatorial Indian Ocean (Stefanutti et al., 2004;MacKenzie et al., 2006).Cairo et al. (2008) show with observations made above a tropical cyclone during this campaign that cyclones may induce horizontal stirring of the lower stratosphere, possibly promoting irreversible entrainment of midlatitude stratospheric air into the tropical zone.The TROCCINOX campaign took place in February 2005 from Aracatuba, Brasil.Konopka et al. (2007) show with a comparison of the in situ measurements of ozone, water vapour, NO, NOy, CH 4 and CO with CLaMS model simulations that vertical mixing, mainly driven by the vertical shear in the tropical flanks of the subtropical jets and, to some extent, in the outflow regions of large-scale convection, offers an explanation for the upward transport of trace species from the main convective outflow at around 350 K up to the tropical tropopause around 380 K.The SCOUT-O3 campaign over Darwin, Australia in November 2005 was focused on studying the effect of deep convection on the composition of the TTL (Vaughan et al., 2008).Schiller et al. (2009) found highly localised layers of enhanced water vapour up to 420 K which could be traced to direct injection by overshooting turrets during this campaign.
In situ measurements of tracers in the TTL have also been conducted using the NASA ER-2 and WB-57F aircraft.Marcy et al. (2007) have presented measurements of HCl, O 3 , HNO 3 , H 2 O, CO, CO 2 and CH 3 Cl in the tropical upper troposphere and lower stratosphere (UT/LS) during the Pre-AVE campaign over Costa Rica in January 2004.They infer that a significant amount of stratospheric air and O 3 were present in the TTL, making it distinct from both the stratosphere and the remainder of the troposphere.Park et al. (2007a) have presented CO 2 measurements during the Pre-AVE, CR-AVE and TWP-ICE campaign in Costa Rica and Australia.They suggest that the TTL is composed of two layers, the lower TTL which is subject to significant inputs of convective outflow, and the upper TTL, where air ascends slowly and ages uniformly.They calculate a mean age of air entering the lower stratosphere of 26 days during NH winter.Tuck et al. (2003) have analyzed tracers and thermodynamical data from various ER-2 and WB57F aircraft campaigns and documented significant transport from the lower midlatitudes stratosphere toward the tropics, coming to the conclusion that the characteristics of the TTL are determined by a combination of subtropical jet stream dynamics and inner tropical ascent via deep convection.
Until now, in situ measurements of tracers throughout the TTL have not been reported during NH summer and none have been made above the African continent.However, this region may play an important role in troposphere-tostratosphere transport.Ricaud et al. (2007) present satellite data of N 2 O, CH 4 and CO and radar data in the tropical tropopause region during NH spring and suggest that rapid uplift over land convective regions, in particular over Africa, may be the dominant process of troposphere-to-stratosphere exchange.However, this view is in contrast to a number of other studies showing that the African region is not an important contributor to troposphere-to-stratosphere transport compared in particular to Southeast Asia and the Western Pacific (Fueglistaler et al., 2004;Berthet et al., 2007;Barret et al., 2008;Park et al., 2007b).
Beginning in 2001, the large project AMMA (African Monsoon Multidisciplinary Analyses) was set up to study the West-African monsoon and its influence on the physical, chemical and biological environment, regionally and globally.The overall objective of the project was to provide the underpinning science that relates variability of the West African Monsoon to issues of health, water resources, food security and demography for West-African nations and to define and implement relevant monitoring and prediction strategies (www.amma-international.org).Within the frame of AMMA and the project SCOUT-O3 (Stratospheric-Climate Links with Emphasis on the Upper Troposphere and Lower Stratosphere) an aircraft campaign took place at the peak of the summer monsoon period in 2006, probing the air from the boundary layer up to the stratosphere (Cairo et al., 2010).The campaign took place in July/August 2006 and was based in Ouagadougou, Burkina Faso.The main aim of this aircraft campaign was to quantify the contribution of different air mass origins on the chemical composition of the TTL during the summer monsoon over West Africa.At this time of year the TTL can be impacted by recent local or regional convective uplift, but also by uplift from the lower troposphere upwind (e.g. over Asia), intrusion of air from the midlatitude lower stratosphere or cross-hemispheric transport of air masses from the Southern Hemisphere (Law et al., 2010).A detailed description of the meteorological situation during the West-African monsoon in 2006 can be found in Janicot et al. (2008).
In this paper we will present in situ tracer data of CO 2 , N 2 O, O 3 and CO that were obtained on board the M55 Geophysica aircraft during the AMMA/SCOUT-O3 project and analyse them regarding the principal transport processes that control the chemical composition of the TTL and tropical lower stratosphere above West-Africa.Vertical profiles and correlations between the various species, serving as stratospheric tracers, as boundary layer tracers, or ageof-air tracers will be used to contrast observations of the background TTL with convectively influenced air, to diagnose irreversible mixing of convectively overshooting air with the background TTL, to assess isentropic mixing across the subtropical tropopause and to study the morphology of the stratospheric subtropical transport barrier.

In situ measurements
Tracer measurements were made between 31 July and 17 August 2005 on board the M55 Geophysica aircraft from (and during transfer to/from) Ouagadougou, Burkina Faso (12 • N, 1 • W).A total of nine flights were performed with measurements from the free troposphere up to the tropical lower stratosphere around 20 kilometer altitude.Four transfer flights were made between Verona (Italy), Marrakech (Morocco) and Ouagadougou (Burkina Faso).Five local flights between 4 • N and 17 • N latitude and 3 • W and 3 • E longitude were made to study mesoscale convective systems (MCS) and long range transport, and to validate the CALIPSO satellite.A brief overview of the flights is given in Table 1.A more extended overview of the flights and the Geophysica campaign can be found in Cairo et al. (2010).
A suite of long-lived tracers (CO 2 , N 2 O, CFC−12, CFC−11, H−1211, SF 6 , CH 4 , H 2 ) was measured by the University of Frankfurt's High Altitude Gas Analyzer (HA-GAR) on all flights except on 7 August when a software failure occurred.HAGAR is a two-channel in situ gas chromatograph (GC) that is combined with a CO 2 sensor (LI-COR 6251).The two GC-channels with electron capture detectors (ECD) measure the long-lived tracers N 2 O, CFC−12, CFC−11,H−1211, CH 4 , SF 6 and H 2 with a time resolution of 90 s.The CO 2 sensor achieves a time resolution of 5 s using non-dispersive infrared absorption (NDIR).A detailed description of the instrument can be found in Riediger (2000) and Strunk (1999).
Frequent in flight calibrations within the instrument were performed in order to meet the specifications for accuracy and precision.All measurements are directly traceable to the current WMO scales based on intercalibrations against standards of NOAA/GMD or (for CO 2 ) the University of Heidelberg (I.Levin).For CO 2 the mean precision during all flights was 0.3 ppm.The precisions for the measurement of the other tracers are in the order of 0.5-3%; for the N 2 O measurements used in this study it is better than 1.5%.Accuracies of the HAGAR measurements are limited by the precision and are thus only slightly larger than these precision values.
Additionally, CO was measured with a precision of 1% and accuracy of 6-9% by the Italian institute INOA (Istituto Nazionale di Ottica Applica) with the tunable diode laser COLD (Cryogenically Operated Laser Diode) as described in detail by Viciani et al. (2008).Ozone was measured with the Fast Ozone ANalyzer (FOZAN) of the Central Aerological Observatory (CAO, Russia) with an accuracy of 10% (Ulanovsky, 2001).Mixing ratios for all gas measurements are expressed in dry mole fractions.Flight parameters as altitude, pressure and temperature were recorded by the aircraft.
The results of the analysis of the different transport processes occurring in the summer TTL over West-Africa will be discussed in the following sections.Section 3.1 describes both main convective outflow as well as overshooting convection, Sect.3.2 describes isentropic stratospheric inmixing across the subtropical tropopause and Sect.3.3 describes isentropic mixing across the subtropical barrier in the lower stratosphere.Each section will have a short introduction about the concepts of the transport process described.

Convection
Convection plays an important role in determining the thermodynamic and chemical properties of the TTL.It can provide a fast pathway for halogenated very short lived species and other boundary layer trace gases to reach the TTL and subsequently the stratosphere, where they could contribute to the depletion of ozone.The amount of very short lived species that can reach the stratosphere thus depends crucially on the convective mass flux into the TTL as well as the maximum altitude level that is reached by deep convection.The impact of convection on the TTL may also significantly depend on the fraction and composition of mid-tropospheric air entrained into convection (Fridlind et al., 2004).
There are two mechanisms in which convection contributes to vertical transport through the TTL.The first is direct convective uplift to the level of neutral buoyancy, which is located at a potential temperature of at most 365 K (Folkins et al., 2000).At this level the potential temperature becomes equal to the highest equivalent potential temperatures realized in the marine boundary layer.It is therefore near the maximum altitude an air parcel from the boundary layer can reach by undiluted, non-overshooting ascent (Reid and Gage, 1981).This convection up to the level of neutral buoyancy can be followed by slow diabatic ascent up to and across the tropopause due to radiative heating and the dynamical forcing by the Brewer Dobson circulation (Folkins et al., 1999).The typical time scales for this diabatic transport, as derived from numerical models and tracer measurements, range between 2 and 3 months for the upward transport from a potential temperature of 350 K up to 390 and 420 K, respectively (e.g., Andrews et al., 1999).
The second mechanism, by which convection can influence the TTL to much higher levels than its neutral buoyancy level, is irreversible mixing of air following dynamic overshooting (Danielsen, 1982(Danielsen, , 1993;;Sherwood, 2000).When an air parcel overshoots its level of neutral buoyancy it will be colder than the surrounding air and start descending back to its equilibrium level.However, when it entrains and mixes with warmer surrounding air it will come to rest at a warmer, higher equilibrium level, i.e. its potential temperature increases (Danielsen, 1982).This overshooting of convection occurs predominantly above continental areas (Liu and Zipser, 2005;Zipser et al., 2006) and in large mesoscale convective systems (Rossow and Pearl, 2007), which are generated regularly over West-Africa in the monsoon period, thereby providing a fast pathway for boundary layer air into the upper TTL or even the lower stratosphere.During AMMA, in fact, moist layers were observed above the tropopause over Niger up to 19 km and have been attributed to geyser-like injection of ice particles over overshooting turrets (Khaykin et al., 2009).Although impact of overshooting convection on the upper TTL and lower stratosphere is generally accepted and can be numerically simulated (e.g., Arteta et al., 2009;Liu et al., 2010), its importance at global scale is less clear.
In this section the two mechanisms for convective influence will be studied with help of the tracer data measured during AMMA/SCOUT-O3.In Sect.3.1.1we will examine the first mechanism and assess the level of main convective outflow and the influence of local and aged convection on the CO 2 profiles.In Sect.3.1.2we will examine the data for signatures of overshooting air.

Main convective outflow
CO 2 can be used as a tracer for continental convection because its mixing ratio is reduced in the boundary layer due to uptake by vegetation during daytime.Deep convection during the African monsoon season peaks in the evening (Sultan et al., 2007) when boundary layer CO 2 is expected to be around its minimum.Convective transport of boundary layer air into the TTL during the monsoon season therefore results in a layer of low CO 2 mixing ratios around the level of main convective outflow.Above this level, diabatic ascent is expected to dominate convection, vertical mixing and mixing of older air from mid-latitudes into the tropics.Hence, a coherent "tape recorder" signal due to the monotonic aging of the slowly ascending air and the progressing CO 2 seasonal cycle in the tropospheric boundary layer can be observed (Boering et al., 1996;Park et al., 2007a).
In this section we will discuss the strength and height of the main convective influence with help of the CO 2 profiles and compare with results of an analysis by Law et al. (2010).Law et al. (2010) calculate the fraction of measured air sampled between 350 and 365 K that has potentially been influenced by recent convection with help of ECMWF backtrajectories and infrared satellite images identifying convective clouds.They label an airmass as possibly recently convectively influenced when its backward trajectory crossed a region whose cloud top radiance was below 200 K within a longitudinal band from 30 • W to 40 • E, corresponding to an age of at most three to four days.
Figure 1 shows the vertical CO 2 profiles for the four local flights on 4, 8, 11 and 13 August, respectively.Mixing ratios as low as 372 ppm are observed in the daytime boundary layer over Ouagadougou; even lower values could be present in rural areas.The influence of vegetative uptake of CO 2 is getting smaller in the free troposphere, resulting in higher mixing ratios of around 380 ppm.Convection during afternoon and evening transports the CO 2 -depleted boundary layer air up to about 13-14 km, or a potential temperature level of 350 K (Fig. 2), where again a distinct layer with low mixing ratios is observed.Above this level the typical coherent "tape recorder" signal is observed.
The flight of 11 August shows the highest and strongest outflow signature, with mixing ratios as low as 374 ppm found up to 370 K.This flight aimed specifically at sampling air influenced by a large and intense MCS that had crossed westward across Burkina Faso during the previous days (Cairo et al., 2010).Law et al. (2010) also find that of all flights performed during the campaign, the TTL sampled during the flight on 11 August was most influenced by recent convection.They calculate that for this flight ~55% of the measured air between 350 and 365 K was possibly influenced by recent convective activity.Our CO 2 observations suggest that convective outflow of this intense MCS in fact reached unusually high potential temperatures of 370 K.
For the flight of 8 August, Law et al. ( 2010) also indicate possible convective influence for ~40% of the air sampled between 350 and 365 K.This agrees well with the CO 2 profile, which shows a level of main convective outflow at 350 K, albeit with a CO 2 minimum that is less pronounced than on 11 August.However, above this level the "tape recorder" signal is less compact than during the other flights, which might be an indication of overshooting convection, which will be discussed in the next section.
For the flight of 4 August, the CO 2 profile shows again a recognisable level of main convective outflow at 355 K, again less distinct as for the flight of 11 August.Law et al. (2010) infer here a possible influence of recent convection for only ∼10% of the air between 350 and 365 K during that day.
Least influence of convection is evident in the profile of 13 August, where an outflow level can hardly be discerned, with mixing ratios at 13 km as high as in the free troposphere.This again agrees well with the analysis of Law et al. (2010), who also find least possible influence of recent convection (∼5%) during this flight, and in fact not even convective uplift of the trajectories within the previous 10 days.Overall, the observed features in the CO 2 profiles and their relative strengths are thus qualitatively in line with the fractions of sampled air having recently (within about 4 days) passed over convective systems, as calculated by Law et al. (2010).An exception is the flight of 4 August where the observed pronounced CO 2 reductions by 2-3 ppm in the convective outflow region cannot easily be explained by the small recent convective input of 10%, unless regional boundary layer mixing ratios are much lower on the average than those actually observed over Ouagadougou.Another, more likely, explanation is that the CO 2 minima are not only due to recent convection, but to a larger part caused by older convection having occurred more than four days earlier.Since CO 2 is chemically conserved in the TTL, it is not possible to distinguish between the effects of recent and older convection with help of the CO 2 profiles.However O 3 profiles can give an indication whether the convection was of recent or of older origin.
In the tropical boundary layer typical ozone mixing ratios range from 15 to 40 ppb (Folkins and Martin, 2005).Deep convection will transport this O 3 poor air together with ozone precursors from the tropical boundary layer into the TTL, where, as long as no new convective input takes place, the mixing ratio will steadily increase by photochemical production until it reaches its steady state.Therefore, O 3 mixing ratios are an indication for the convective replacement timescale over a region.However, not only convection, photochemical production, and the time since the air has last experienced convective flushing determine the O 3 mixing ratios in the TTL.The O 3 budget in the TTL can also be significantly affected by isentropic stratospheric inmixing bringing in extratropical stratospheric air with high O 3 mixing ratios.Vertical mixing, e.g.following overshooting of air, may also mix down higher ozone concentrations from higher altitudes.We show in Sect.3.2 that horizontal stratospheric inmixing, however, is not significant during the four flights considered here.Mixing following overshooting will be discussed in the next section (Sect.3.1.2).
Figure 3 shows the ozone profiles during the AMMA campaign.The average profile of all five local flights is indicated as the black line.For comparison, the dark green line represents the average ozone profiles during the SCOUT-O3 campaign, which took place in November and December 2005 above northern Australia.This latter profile shows the typical S-shape, with low O 3 mixing ratios of 30 ppb over the marine boundary layer, increasing concentrations in the free troposphere, again low O 3 mixing ratios at the level of main convective outflow, and above that increasing mixing ratios due to the photochemical production of O 3 in the lower stratosphere.
The AMMA-SCOUT-O3 profiles also show this S-shape profile, although less pronounced.Again, concentrations are on average about 30 ppb in the boundary layer; this mean value above Ouagadougou agrees also well with that observed in the boundary layer over Niamey, Niger, by regularly launched ozone sondes (Cairo et al., 2010) and is thus thought to be fairly representative for a wider region.In the free troposphere O 3 increases with height up to around 60 ppb at a potential temperature level of 340 K. O 3 mixing ratios at the level of main convective outflow, however, are much higher on average than those observed during the SCOUT-O3 campaign or in the boundary layer during AMMA-SCOUT-O3.For most of the flights no boundary layer values are observed at the main level of convective outflow (~355 K), thereby suggesting that there is no major flushing of the TTL by recent convection of O 3 -poor bound-ary layer air.The absence of this flushing results in increased O 3 mixing ratios due to production by NOx from lightning (LiNOx) (e.g., Thompson et al., 2000;Sauvage et al., 2007;Barret et al., 2010) and other ozone precursors.Law et al. (2010) indicate that a large part of the air in the TTL originated from Asia about a week earlier, where uplift of these ozone precursors could have taken place.An exception appears to be the flight on 4 August when values as low as 30 ppb observed at a level of 355 K indicate more recent convective flushing.
Above 370 K the O 3 mixing ratios are up to twice as high as during the SCOUT-O3 campaign.Randel et al. (2007) show that in a narrow vertical layer between ∼16 and 19 km (∼375-450 K) approximately a factor 2 change in ozone between the minimum (during NH winter) and maximum (during NH summer) takes place due to variations in vertical transport associated with mean upwelling in the lower stratosphere (the Brewer Dobson circulation).Thus, the higher mixing ratios observed during AMMA in the lower stratosphere are probably caused by the seasonal difference in upwelling due to the Brewer Dobson circulation.
In summary, the CO 2 profiles show that the region around West-Africa is highly influenced by convection up to 355 K, during 11 August even up to 370 K. Clearly pronounced levels of main convective outflow are observed during three of the four flights.However, O 3 profiles strongly suggest that this convectively influenced air is mostly of older origin, and has been transported for at least several days before being measured.Only the observations of 4 August exhibit some signatures of recent convection in the O 3 profile.Overall, these results are in good agreement with the findings of Law et al. (2010).

Overshooting of convection
Now we examine potential signatures of overshooting convection in the tracer data by observations of CO, CO 2 and O 3 .
In Fig. 4 the CO 2 profiles during AMMA/SCOUT-O3 are plotted in one figure, with their average represented by the black line.On 8 August there are large deviations from this average between 390 and 420 K.Both enhanced and reduced values of CO 2 can be found at this level.On 4 August there are also some reduced values between 400 and 420 K.
The enhanced values on 8 August cannot be explained by transport processes; given they occur shortly after a calibration phase of the instrument we cannot completely rule out an unusual instrumental instability.The low values could be an indication of irreversible mixing of overshooting air.If irreversible mixing takes place during or following convective overshoot, air parcels originating from the boundary layer with a potential temperature of 350-360 K (the level of neutral buoyancy) will mix with air masses with a higher potential temperature along a mixing line.In practice, mixing may proceed along a multitude of such mixing lines resulting in a mixing band in a profile plot (provided sufficient sampling).In Fig. 4, two possible idealized mixing lines are displayed.They indicate that, in order to explain the signatures, mixing would have occurred between air parcels with potential temperatures of approximately 355 K and 400-420 K, that is overshooting convection would have reached up to 17-18 km.
Overshooting convection can also be diagnosed in tracertracer correlations.Without mixing, the correlation between a tropospheric (e.g.CO, CO 2 ) and a stratospheric tracer (e.g.O 3 ) will form an L-shape: in the troposphere an almost constant mixing ratio of the stratospheric tracer and much variation in the mixing ratio of the tropospheric tracer, and the opposite in the upper TTL and stratosphere.Thus, the correlation plot exhibits two distinct branches for the stratosphere and the troposphere, which meet in the lower TTL.Mixing following overshooting would again be seen along mixing lines between these two branches, connecting boundary layer observations with values observed at the highest level of overshoot.
Figure 5 shows the correlation plot between O 3 and CO 2 for the local flights, coloured according to potential temperatures and with the average correlation in black.The figure shows a flipped L-shape with variable CO 2 mixing ratios and constant O 3 mixing ratios in the troposphere up to about 360 K, and above that increasing O 3 mixing ratios and slowly decreasing CO 2 mixing ratios.Again, the correlation plot shows a deviation from the average correlation curve be- tween potential temperatures of about 360 and 420 K.They occur along mixing lines corresponding to those displayed in Fig. 4.However, similar signatures are not observed during that part of the flight in other species like H 2 O, NO y and particles (not shown).If overshooting of convection had occurred one would expect clear signals of enhanced H 2 O and particles as moister air with more particles from the boundary layer would be entrained.Voigt et al. (2008) indicate layers with enhanced particles and NO y , but these are found at other locations during the flight.The latter study does, however, show with backtrajectories that the air measured during this flight was located above a mesoscale convective system for at least 1.5 days prior to the measurements, with cloud top levels up to at least 120 hPa (16 km altitude).
As during the flight of 7 August CO 2 was not measured due to failure of the HAGAR instrument we examine the CO data measured by the COLD instrument.Figure 6 shows the CO profiles for both 7 and 8 August (the only flights during which COLD measured).Around 410 K, very high CO mixing ratios are observed, which could again be an indication of vertical mixing following overshooting.This feature is also present in the correlation plot between CO and O 3 (not shown here).However, as the two possible mixing lines in Fig. 6 indicate, this would imply mixing of air from the main convective outflow level (355 K) with air parcels at potential temperatures of at least 430 K, i.e. overshoot to extremely high levels.Again, other tracers do not show similar signatures.Although enhancements are observed in mixing ratios of NO y and CCN during that day, these only indicate an influence up to 14-15 km, and during a later section of the flight (not shown).Water vapour mixing ratios do not show any signatures indicative of overshooting convection (not shown).Fig. 6.CO profiles for 7 and 8 August.The black lines represent the two most extreme possible mixing lines that could explain the high CO values at 410 K.One mixing line would mean mixing of air parcels with a potential temperature of 355 K and 430 K, the other line mixing of air parcels with potential temperatures of 355 K and 470 K. Any mixing line in between these two extremes is possible to explain the enhanced CO values at 410 K.
Overall, we conclude that signatures potentially indicating impact of overshooting convection are observed in the lower stratosphere in the CO and CO 2 mixing ratios during the AMMA campaign.However, these signatures are not corroborated by measurements from other instruments on board the airplane.Thus, evidence for the impact of overshooting is inconclusive.In particular, there is no clear indication that overshooting of convection plays a major role in troposphereto-stratosphere transport during the time of the campaign.

Stratospheric isentropic inmixing into the TTL
Horizontally, the tropical tropopause layer is confined by the subtropical jets.The strength and position of these jets vary by season, with a strong jet close to the equator in the winter hemisphere and a weak, poleward shifted jet in the summer hemisphere.Haynes and Shuckburgh (2000) show that a strong subtropical jet forms an effective transport barrier for the meridional, isentropic transport between the lower part of the TTL and the extratropical lower stratosphere.Isentropic, quasi-horizontal transport from the extra-tropical stratosphere may play a significant role in determining the chemical (trace species) and radiative character of the TTL (Gettelman and Forster, 2002).Analysis of previous aircraft measurements suggests that there may be significant quasiisentropic transport from the lower mid-latitude stratosphere toward the tropics (Tuck et al., 2003;Marcy et al., 2007).
In order to quantify the amount of extratropical stratospheric air entering the TTL we examine the N 2 O data.N 2 O has its source located at the surface and is well mixed In order to estimate the fraction of air transported from the extratropical lower stratosphere the N 2 O mixing ratios of the extratropical stratosphere are compared with the values in the TTL.The fraction of aged extratropical air (χ) can be expressed as: cold point tropopause during the campaign.The fraction of aged extratropical air in the TTL is around zero in the lower part up to 370 K and is only slightly increasing towards the top of the TTL (here defined as the cold point tropopause).
Note that this result refers to in-mixing of photochemically aged air only and thus does not necessarily rule out in-mixing of extra-tropical stratospheric air altogether.In fact active isentropic exchange across the weak summer subtropical jet with a net export of air masses from the TTL to the extratropical stratosphere is known to flush the lowermost stratosphere with young air over the course of the summer.Any re-entrainment of these young air masses into the TTL would not be detected by the above diagnostic (but would also be of little relevance for the chemical composition of the TTL).

Isentropic mixing across the subtropical barrier
Except for its lowest part where isentropic mixing with the extratropics is still relatively efficient, the tropical stratosphere is more or less isolated from the extratropical part of the stratosphere (Volk et al., 1996;Minschwaner et al., 1996).The subtropical barrier constitutes a region of strong horizontal shear and maximum PV gradient along isentropes, thereby prohibiting fast isentropic mixing between the tropical stratosphere and mid-latitude stratosphere, especially in the winter hemisphere.
Due to the isolation of the tropical stratosphere, tracer pairs with differing source/sink structures (e.g.O 3 -N 2 O, O 3 -NO y ) show different correlation slopes in the tropical and extratropical stratosphere (Volk et al., 1996;Fahey et al., 1996).At mid-latitudes, where quasi-horizontal mixing is faster than both vertical transport and chemical time scales, a compact correlation evolves whose slope at any point is determined by the ratio of globally integrated sources and sinks of the two species above the given altitude (Plumb and Ko, 1992).Further, since the sources and sinks for O 3 and N 2 O are insignificant in the lower stratosphere, this ratio remains approximately constant there, and thus the lower stratospheric correlation is fairly linear.In the tropics, on the other hand, quasi-horizontal mixing is slower, and the correlation is to a large part determined by vertical ascent and local chemistry (Volk et al., 1996); in situ production of O 3 in the lower stratosphere thus results in a correlation slope different from the mid-latitudes.The correlation slope is therefore a reliable indicator for the origin of an air mass.A change in the slope of these correlation branches marks the position of the subtropical barrier.Mixing events between the two regions manifest themselves as lines or bands connecting the characteristic tropical and extratropical correlation branches.
Figure 9 shows the correlation between O 3 and N 2 O for the local and the transfer flights during AMMA/SCOUT-O3.The colours indicate the latitude of the measurements.Different correlation slopes are observed at different latitudes.
The slopes of these correlations are visualised in Fig. 10 by plotting the ratio of the differences between the measured O 3 and N 2 O values from tropospheric reference values, i.e. plotting the correlation slope from the tropospheric origin of the correlation (chosen as 320 ppb N 2 O and 50 ppb O 3 ).The figure shows only data above a potential temperature of 415 K, above which a significant separation between the tropical and midlatitude correlations can be discerned.The lowest slope values of 0.02 observed southward of 15 • N indicate air within the isolated tropical region (the "tropical pipe"), whereas the midlatitudes exhibit slope values exceeding 0.04.Values in between these extremes are found in a band between 10 • N and 30 • N, suggesting a wider transition zone rather than a sharp subtropical barrier.In order to examine whether the subtropical barrier might nevertheless be  sharp, but undulate within that latitude band due to wave motions, we also plot the slope values against equivalent latitude in Fig. 11.Reversible wave motions should be mostly eliminated in this projection as they occur at roughly conserved potential vorticity, and thus equivalent latitude.Nevertheless, the transition between inner tropical and midlatitude slope values still spans a band between 10 • N and 25  (1996) and in fact show fairly sharp transitions between low tropical values and extratropical values in March (around 10 • N) and in October (around 15 • N).In contrast, the transition during AMMA in August is more gradual and ranges between 10 • N and 25 • N. We note here that the ER-2 data from ASHOE/MAESA in the region in question are on average sampled at a higher altitude than the M55 Geophysica data from AMMA, the former being mostly collected during transfer flights at maximum altitude (20-21 km).As the subtropical barrier increases in strength above 20 km, the ASHOE/MAESA data are thus expected to be more suitable to indicate the location and width of the barrier, whereas the Geophysica during AMMA may have sampled at an altitude range (below 20 km) at which the barrier is weaker and thus more permeable to mixing.On the other hand, the subtropical barrier is also expected to be weakest during summer on the summer hemisphere due to the absence of strong wave activity in the surf zone, which tends to sharpen the barrier.Thus the observed differences between ASHOE/MAESA and AMMA in the width of the transition between tropical and extratropical tracer values may well be explained by differences in both sampling altitude and season.

Conclusions
We have presented in situ tracer data from the AMMA-SCOUT-O3 campaign in July/August 2006 over West-Africa.Data were obtained on board the M55 Geophysica with the High Altitude Gas ANalyzer (HAGAR), measuring CO 2 , N 2 O, CFC−11, CFC−12, H−1211, CH 4 , SF 6 and H 2 , as well as from the FOZAN and COLD instruments, measuring O 3 and CO, respectively.We have used the data to examine the dominant transport processes in the TTL: deep convection up to the level of main convective outflow, irreversible mixing after overshooting of deep convection, isentropic inmixing into the TTL across the subtropical tropopause, and horizontal transport in the region of the stratospheric subtropical barrier.The CO 2 profiles (with the exception of 13 August) show distinct minima in the TTL, reflecting the outflow of boundary layer air depleted in CO 2 .These reductions in CO 2 suggest that i) convective influence in the TTL is quite significant in the sampled air masses and that ii) the main convective outflow was usually located at potential temperature levels up to ∼350-360 K (13-14 km), and for the flight of 11 August even reached up to 370 K (∼15 km).The O 3 profiles, on the other hand, indicate that the larger part of the convective fraction in the sampled air masses must be of rather older origin (several days or more), as regional boundary layer mixing ratios of 30 ppb are only occasionally found in the TTL (only during the flight of 4 August), while average O 3 mixing ratios in the TTL are observed to be 50 ppb or higher.These results agree with those of Law et al. (2010) who find that a large part of the measured air was already residing in the tropical upper troposphere and lower stratosphere when arriving over West Africa and was uplifted over Asia within ten days prior to the measurements.According to their study, regional uplift over Africa above 350 K was only important on 11 August, and potentially on 8 August.Overshooting convection does not appear to have a large impact on vertical profiles of gas-phase tracers, which exhibit quite coherent shapes above the maximum level of convective outflow.Only a few small signatures potentially indicative of mixing following overshooting convection were observed in the CO 2 data during the flights of 4 and 8 August; another potential signature was observed in CO on 7 August.However, similar signatures were not simultaneously observed in other tracers.The absence of frequent high reaching local convection and the lack of major overshooting events impacting the gas-phase composition of the TTL inferred in this study is in accord with the notion that the African region is not a major player in troposphereto-stratosphere transport (Berthet et al., 2007;Barret et al., 2008), at least not in NH summer.Stratospheric inmixing of photochemically aged air from the extratropical stratosphere appears to be minimal up to the mean local tropopause at 376 K.The fractions of aged extratropical air in the TTL, as estimated from N 2 O profiles is 0.0±0.1 up to 370 K and is increasing above this level up to about 0.3±0.1 at 390 K.
The subtropical barrier does not manifest itself as a sharp boundary in tracer distributions, but rather as a gradual transition region between around 10 • N and 25 • N where tracer mixing ratios change from characteristic tropical to extratropical values.The subtropical barrier thus appears to be rather permeable to horizontal mixing in the summer subtropics below 20 km altitude.

Fig. 3 .
Fig. 3. Ozone profiles for the five local flights.The average profiles is indicated with the black line.The green line represents the average ozone profile for the SCOUT-O3 campaign over Australia in November/December 2005.

Fig. 4 .
Fig. 4. CO 2 profiles for the local flights during AMMA/SCOUT-O3.Average profile in black.Two examples of possible mixing lines are presented by the grey lines.

Fig. 5 .
Fig. 5. Correlation between O 3 and CO 2 .The data are coloured to potential temperature.The black line represents the average CO 2 profile for all local flights.The grey lines are the possible mixing lines that are also displayed in Fig. 4.

Fig. 8 .
Fig. 8. Fraction of aged extratropical air.The blue line represents the average height of the tropopause.
O vs. O 3 ) found earlier during the ASHOE/MAESA campaign in March and October/November 1994.The slope values from AMMA data are coloured and plotted against latitude.The ASHOE/MAESA slopes (in grey) can be viewed in analogy to the NO y /O 3 ratios published inFahey et al.