The contribution of natural and anthropogenic very short-lived species to stratospheric bromine

R. Hossaini, M. P. Chipperfield, W. Feng, T. J. Breider, E. Atlas, S. A. Montzka, B. R. Miller, F. Moore, and J. Elkins Institute for Climate and Atmospheric Science, School of Earth and Environment, University of Leeds, UK School of Engineering and Applied Sciences, Harvard University, Cambridge, USA Rosenstiel School of Marine and Atmospheric Science, University of Miami, USA National Oceanic and Atmospheric Administration, USA


Introduction
The bromine (Br) budget of the stratosphere is largely controlled by anthropogenic emissions of long-lived source gases (SGs) such as halons (e.g.CBrF 3 ) and methyl bromide (CH 3 Br).These SGs are sufficiently long-lived that tropospheric gradients are small.In recent years, it has been suggested that very short-lived substances (VSLS) of predominately natural origin may provide an additional source of stratospheric inorganic bromine (Br VSLS y ) (e.g., Sturges et al., 2000;Dorf et al., 2008).The magnitude of this supply is thought to be 6(3-8) ppt, representing ∼25 % of total stratospheric bromine (Montzka and Reimann et al., 2011).
Natural emissions of VSLS are predominately from marine macro-algae/phytoplankton and show large variability in space and time (e.g., Carpenter et al., 1999).From the tropical marine boundary layer (MBL), troposphere-stratosphere transport can be rapid (hours) in deep convective updrafts.The Western Pacific is a strong convective source region for air entering the tropical tropopause layer (TTL) (e.g., Aschmann et al., 2009).Rapid convective transport allows short-lived SGs to escape oxidation in the troposphere and detrain in the lowermost stratosphere (LMS).Once photolysed, reactive inorganic products (e.g., BrO) take part in catalytic ozone (O 3 ) depleting cycles (e.g., BrO + ClO) (e.g., Sinnhuber et al., 2002).The most important natural VSLS are bromoform (CHBr 3 ) and dibromomethane (CH 2 Br 2 ) due to their relatively large emissions and for CH 2 Br 2 in particular its long local lifetime (e.g., Hossaini et al., 2010).To date, most model work has focused on quantifying the stratospheric Br y supply from these VSLS (e.g., Hossaini et al., 2010;Liang et al., 2010).These studies estimate a supply of ∼2.4 and 5 ppt, respectively.The larger latter value is likely due to the explicit modelling of wet deposition (i.e.removal in precipitation) of soluble Br y .The assumed uniform 10-day lifetime of Br y , employed in Hossaini et al. (2010), is likely an underestimate (in some regions) and thus results in too much scavenging of soluble products (e.g., Aschmann et al., 2011).Other model studies have considered dibromochloromethane (CHBr 2 Cl), bromodichloromethane (CHBrCl 2 ) and bromochloromethane (CH 2 BrCl) in addition (e.g., Warwick et al., 2006;Schofield et al., 2011).Warwick et al. (2006) reported a total Br VSLS y contribution of 6-7 ppt from these species.Variation in the calculated contribution of VSLS is expected due to different treatment of emissions and transport processes (Hoyle et al., 2011;Feng et al., 2011).
Some VSLS have a predominately anthropogenic source; e.g.n-propyl bromide (n-C 3 H 7 Br, nPB), i-propyl bromide (i-C 3 H 7 Br, iPB), ethyl bromide (C 2 H 5 Br) and ethylene dibromide (CH 2 BrCH 2 Br, EDB).nPB is currently used as a replacement compound for some chlorofluorocarbons (CFCs, e.g., CFC-113) and for methyl chloroform (CH 3 CCl 3 ).It is used for industrial metal degreasing and electronic part cleaning (Montzka and Reimann et al., 2011).C 2 H 5 Br is used in chemical synthesis, as a refrigerant and as a flame retardant.Observations show surface mixing ratios of <0.3 ppt (Low et al., 2003) and 0.09-0.49ppt (Carpenter et al., 1999).EDB is used as a fumigant and has been observed at ∼5 ppt and 1 ppt in urban areas (Pratt et al., 2000) and the South Pole (Khalil and Rasmussen, 1985), respectively.
In general these minor VSLS are poorly studied and observations are sparse.Laube et al. (2008) have suggested unidentified peaks in their chromatogram from air sampled at 15.2 km point towards the presence of C 2 H 5 Br, nPB and EDB.These balloon-borne measurements were made in 2005 in the tropics (Teresina, Brazil).These species are estimated to currently contribute <0.2 ppt of organic bromine in the TTL region (Montzka and Reimann et al., 2011).Quantifying the present-day contribution of these unregulated anthropogenic VSLS on stratospheric Br y is important.Section 2 describes the model, improvements on our previous model set up (Hossaini et al., 2010) and experiments performed.Results are presented in Sect. 3 along with a description of the NSF HIPPO-1 data.Conclusions and recommendations for future work are discussed in Sect. 4.

Model description
TOMCAT/SLIMCAT is a three-dimensional (3-D) chemical transport model (CTM) described in detail by Chipperfield (2006).For this study we ran the CTM with 38 hybrid σ -p vertical levels extending from the surface to ∼35 km and with a resolution of 5.6 • ×5.6 • .The offline model is forced by analysed winds taken from the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim reanalysis.The CTM includes an online parameterisation of cumulus convection based on the Tiedtke (1989) mass flux scheme (Stockwell and Chipperfield, 1998).This CTM set up is described in Hossaini et al. (2010) and has been shown to simulate the transport/chemistry of short-lived tracers reasonably well.
We have made two improvements compared to Hossaini et al. (2010).First, as model convection is an uncertainty, we also now include an additional approach to the Tiedtke (1989) mass flux scheme.Here the CTM can be forced with ERA-Interim archived convective updraft/downdraft mass fluxes and detrainment rates (Feng et al., 2011).Second, we model the wet deposition of soluble Br y in both large-scale dynamic and also convective precipitation (Giannakopoulos et al., 1999).The fraction of Br y available for wet deposition is altitude-dependent.Here we partition Br y among nonsoluble products (e.g.BrO) and soluble HBr using a HBr:Br y ratio from a previous full chemistry TOMCAT integration (Breider et al., 2010).
In this study we include 9 bromine-containing VSLS SG tracers (see Table 1) and the following long-lived species: methyl bromide (CH 3 Br), Halon-1211 (H1211, CF 2 BrCl), Halon-1301 (H1301, CBrF 3 ) and Halon-2402 (H2402, C 2 Br 2 F 4 ).The CTM here also includes methyl iodide (CH 3 I) and an idealised tracer with a 6 h lifetime (i6hr).These shorter lived species are useful for assessing troposphere-stratosphere transport in modelled convection.The model surface mixing ratio of most SGs are constrained by HIPPO-1 observations in the tropical regions.For CHBrCl 2 and CH 2 BrCl the model is constrained by the estimated MBL mixing ratios, based on a compilation of observations, provided in Law and Sturges et al. (2007).Observations of EDB, nPB and iPB are sparse.For these gases, the model surface mixing ratio cannot be constrained by observations and thus have an arbitrary value at the surface.Therefore, these gases are not considered in the calculated Br VSLS y .c Estimated tropical MBL mixing ratio (Law and Sturges et al., 2007).

Atmos
Source data both natural (N) and/or anthropogenic (A) taken from Law and Sturges et al. (2007).
There is significant uncertainty in the surface emissions of natural and anthropogenic VSLS.There is likely to be considerable spatial variation in the location of emission (e.g., Law and Sturges et al., 2007).Warwick et al. (2006) found significant variation in modelled CHBr 3 profiles depending on source region.While use of spatially varying emissions for CHBr 3 and CH 2 Br 2 is possible (e.g., Liang et al., 2010), emissions of minor VSLS (e.g., CHBr 2 Cl) are poorly quantified.For simplicity, and to overcome some of this uncertainty, in the model runs presented here we do not specify emissions but instead we overwrite the surface volume mixing ratio (vmr) of the halogenated source gases.The surface vmr is constant with latitude and longitude within the tropics and is scaled so that the model gives good agreement with observation in the free troposphere, where available.This approximation is better for longer-lived species which will be mixed by transport processes and are the species which are potentially more important for transport to the stratosphere.For VSLS, our approach will not capture potential hot spot regions where surface fluxes are relatively large.
Loss of SG occurs via oxidation by OH and photolysis.Rate constants and absorption cross sections are taken from Sander et al. (2006) where available.SG degradation is assumed to release Br y immediately (i.e.CHBr 3 + OH → 3×Br y ).This assumption, which ignores any organic product gases (e.g.CBr 2 O), has been shown to be good for CHBr 3 and CH 2 Br 2 (Hossaini et al., 2010).Henry's law constants needed for calculating the wet deposition of soluble Br y products are taken from Law and Sturges et al. (2007).
We have performed 2 model simulations; run S diagnosed where we diagnose convective transport and run S archived where convective transport is forced with archived mass fluxes.Both simulations were initialised on 1 January 2005 and run for 3 years.

Modelled tracer profiles
The modelled source gas profiles of CH 3 Br, H1211, H1301, H2402, CHBr 3 , CH 2 Br 2 , CHBr 2 Cl, C 2 H 5 Br and CH 3 I are shown in Fig. 1.We compare with aircraft observations in the tropics (± 20 • ) from the recent NSF HIPPO-1 campaign (January 2009, http://hippo.ornl.gov/).The HIPPO program is an ongoing series of global aircraft measurements spanning global latitudes from the near North Pole to coastal Antarctica.The NSF Gulfstream V aircraft is used to sample the atmosphere from the surface to ∼14 km.Numerous trace gases have been measured including the greenhouse gases CO 2 , CH 4 , N 2 O and also CO, SF 6 , CFCs, VSLS etc. (Wofsy et al., 2011).Whole air samples, collected in stainless steel and glass flasks, from 5 flights were analysed using gas chromatography/mass spectrometry (GC-MS) in two different laboratories; the University of Miami and NOAA/ESRL (Schauffler et al., 1999).Mixing ratios reported by the Miami laboratory were multiplied by factors of 0.90, 0.94, 0.98, 0.94, 0.81 and 1.01 for CH 3 Br, H1301, H1211, H2402, CH 2 Br 2 and CHBr 3 , respectively, to be consistent with the NOAA/ESRL laboratory calibration scale.
We also consider January 2009 monthly mean surface observations from an ongoing NOAA/ESRL program at three sites in the Pacific Basin; American Samoa (SMO, 14.3 • S, 170.6 • W), Mauna Loa (MLO, 19.5 • N, 155.6 • W) and Cape Kumukahi (KUM, 19.5 • N, 154.8 • W).Whole air samples were collected weekly into paired steel or glass flasks and have been analysed (GC-MS) in the same laboratory as the HIPPO measurements.Aircraft observations up to ∼6.5 km a.s.l.over Raratonga (21.5 • N, 160.0 • W) are also considered.These additional observations are consistent in space/time with the observations from HIPPO.Further description of the NOAA/ESRL surface network can be obtained in Montzka et al. (2003).For this comparison, the model is averaged over the flight days and also in space.The TTL base is defined as the level of main convective outflow (∼12 km) and the top as the cold-point tropopause (CPT, ∼17 km).The upper TTL is defined as beyond the level of zero radiative heating, where slow radiatively-driven ascent dominates over convective mixing (Law and Sturges et al., 2007).
Both runs S diagnosed and S archived show good agreement with observations.Model profiles of CHBr 3 and CH 2 Br 2 give good agreement when constrained by 1-1.2 ppt at the surface.This is consistent with previous aircraft comparisons (Hossaini et al., 2010).The profiles indicate ∼0.5 and 0.8 ppt of CHBr 3 and CH 2 Br 2 in the lower TTL, respectively.This is consistent with the lower TTL (12-14 km) mean mixing ratio and range of 0.61(0.3-1.1)ppt and 0.92(0.77-1.15)ppt reported by Montzka and Reimann et al. (2011) and based on a compilation of previous observations.The HIPPO-1 observations show CHBr 2 Cl mixing ratios of 0.1-0.2ppt in the lower TTL.This is again consistent with previously observed mixing ratios of 0.1(0.06-0.15)ppt (Montzka and Reimann et al., 2011).The anthropogenic short-lived substance, C 2 H 5 Br, was also detected in the TTL (∼0.4 ppt).nPB was below the detection limit of ∼0.5 ppt.
Overall, run S archived shows more tracer transported to the upper TTL.This can be expected given the archived mass fluxes show strong convection in the tropics (e.g., Hoyle et al., 2011;Feng et al., 2011).However, the averaged profiles in Fig. 1 show only small differences between runs S diagnosed and S archived , particularly for the relatively longerlived gases for which vertical transport in convective cells is less important.The differences between the two model runs is discussed further in Sect.3.2.

Archived versus diagnosed convection
The implementation of 6-hourly ECMWF archived convective mass fluxes in the TOMCAT CTM is described in Feng et al. (2011).Here we use observations of short-lived halogen species in the mid-upper troposphere to examine the performance of these fluxes compared to fluxes diagnosed within the model.The archived convective updraft mass flux shows non-zero values at 100 hPa (∼16.5 km).The diagnosed convection typically does not extend above 200 hPa (∼12.5 km) and this does not capture deep convective events (Feng et al., 2011).To investigate the differences between runs S diagnosed and S archived further, we also consider CH 3 I; a trace gas with a local lifetime of ∼7 days and thus shorter lived than any of the bromine-containing VSLS previously discussed (Law and Sturges et al., 2007).Figure 2 shows modelled profiles of CH 3 I in the tropics from both runs S diagnosed and S archived versus observations made on board the DC8 aircraft during the 2007 NASA TC4 campaign (see Hossaini et al. (2010) for campaign description).
In Fig. 2, the difference between the two convective set ups is more pronounced than for the longer-lived brominecontaining VSLS.With an assumed uniform surface mixing ratio of 1 ppt in the tropics, both runs S diagnosed and S archived give reasonable agreement with the observations.A surface mixing ratio of 1 ppt is within the estimated range of 0.3-1.9ppt for the marine boundary layer reported by Montzka and Reimann et al. (2011).Run S archived shows significantly more CH 3 I in the TTL.This appears to provide a better fit to observations, particularly for flights on the 5/8 August 2007 (see Fig. 2).
The zonally averaged mixing ratio of the idealised tracer with a 6 h lifetime (i6hr), CH 3 I, CHBr 3 and CH 2 Br 2 at the approximate base of the TTL (∼200 mb) and the tropical cold point tropopause (CPT, ∼100 mb) is shown in Fig. 3. i6hr was initialised at 1 ppt at the surface at the start of the simulation.Run S archived transports more tracer to higher altitudes in all seasons with the largest difference between runs S diagnosed and S archived being for the relatively shortest lived species (e.g.i6hr, CH 3 I).For CH 3 I, CPT mixing ratios within the estimated range of 0.02-0.18ppt reported by Law and Sturges et al. (2007) for the tropical upper troposphere.The CPT CH 3 I mixing ratios from S diagnosed are up to ∼6× lower.For CHBr 3 and CH 2 Br 2 , both the TTL and CPT mixing ratios are less sensitive to the choice of convection scheme.
Figure 4 shows the annual mean mixing ratio of i6hr at the CPT for runs S diagnosed and S archived .Clearly, stronger convection with run S archived leads to a larger modelled mixing ratio at this level.Here, the strongest convection occurs over the Western Pacific, indicating a strong source region for the troposphere-stratosphere of VSLS -consistent with Aschmann et al. (2009).For studying the tropospherestratosphere transport of relatively shorter lived gases (including species whose lifetime is equal to or less than that of CH 3 I), the choice of convection scheme is clearly important.

Source gas local lifetimes
We have calculated the local lifetimes (τ local ) of source gases in the model using: Here τ OH and τ hν are the lifetimes with respect to OH oxidation and photolysis, respectively.The annual tropical mean lifetimes of C 2 H 5 Br, EDB, nPB and iPB are shown in Fig. 5. Note, the equivalent lifetime plots of CHBr 3 and CH 2 Br 2 can be found in Hossaini et al. (2010).
For these species OH oxidation is the dominant sink in the troposphere.For C 2 H 5 Br, the simulated local lifetime is ∼18 days at the surface and up to ∼183 days in the TTL.Similarly for EDB, the local lifetime is ∼30 and up to ∼603 days at the surface and in the TTL, respectively.The long local lifetime in the cold TTL is due to the strong temperature dependence of the OH sink reaction.Once in the TTL, C 2 H 5 Br and EDB are thus potentially important carriers of bromine to the stratosphere.
It is clear that local lifetimes of VSLS can vary substantially with altitude.The simulated local lifetime of nPB is ∼6 days at the surface and up to 38 days in the TTL.Similarly for iPB, we find local lifetimes of 8 (surface) and 49 days (TTL).For these species with relatively shorter lifetimes at the surface, rapid convective transport to the TTL is likely required for their source gas contribution to be significant.The tropospheric mean local lifetimes of VSLS have also been calculated.These values are shown in Table 2.
Our simulated local lifetimes in the TTL vary from those quoted in the recent World Meteorological Organisation (WMO) Assessment Report (Montzka and Reimann et al., 2010).The reason for this is that the WMO values are based on OH loss rates calculated using a uniform tropospheric OH concentration (1×10 6 molecules cm −3 ) and for a temperature of 275 K, i.e. representative of the surface.The simulated local lifetimes here are derived using a spatially/temporally varying OH field (and photolysis) and can be quoted as a function of temperature/altitude.A direct comparison of the two is therefore not appropriate, but if a gas is sufficiently long-lived to be transported away from the surface then the WMO lifetimes may not be applicable.

Modelled total bromine
The modelled source gas injection (SGI), product gas injection (PGI) and total bromine (SGI + PGI) for VSLS, CH 3 Br and the halons is shown in Table 2. SGI is the organic bromine resulting from direct transport of a source gas to the lower stratosphere (i.e. for CH 2 Br 2 , 2 × the SG mixing ratio at the tropopause).Similarly, PGI is the contribution from the inorganic products, formed following SG degradation.
The total bromine contribution from major VSLS CHBr 3 and CH 2 Br 2 is ∼2.09 and 1.69 ppt (tropical mean), respectively from simulation S diagnosed .The contribution is   precipitation.This increases the likelihood of wet deposition of soluble Br y in certain regions.We would also expect variation depending on the parameterisation of cumulus convection used in the model.Here the Tiedtke (1989) scheme is used but different parameterisations that provide more (less) frequent/intense convective upwelling would likely increase (decrease) the modelled SGI.As the removal of soluble products are coupled to the convective precipitation this could also lead to a decrease (increase) in the delivery of Br y from VSLS reaching the stratosphere via the PGI route.Physical atmospheric removal of Br y in the TTL is also possible from adsorption on to ice.However, according to recent model work, this process (not considered here) represents only a small sink (Aschmann et al., 2011).
Our modelled Br VSLS y from CHBr 3 and CH 2 Br 2 alone (∼4 ppt) is slightly lower than the 5 ppt reported by Liang et al. (2010) but larger than our previously reported 2.4 ppt (Hossaini et al., 2010).This is due to the explicit modelling of wet deposition, resulting in a larger contribution from the PGI route in our present work.For CHBr 3 , we find here 0.97 ppt of Br y reaches the tropical lower stratosphere via the PGI route.This is ∼3× larger than the 0.3 ppt reported in Hossaini et al. (2010).
The minor VSLS individually provide a modest amount of total bromine to the lower stratosphere.The largest contributor, CH 2 BrCl, provides about one fifth that of CHBr 3 .However, their accumulated total (∼1.2 ppt from S archived ) is significant and is likely a lower limit due to nPB, iPB and EDB not considered in the calculation.The contribution of all VSLS considered here and also the long-lived species to total stratospheric bromine is shown in Fig. 6.The model (S archived ) shows ∼21 ppt of total bromine in the stratosphere.We find a total Br VSLS y supply of ∼4.9-5.2 ppt for runs S diagnosed and S archived , respectively.This range is consistent with the inferred Br VSLS y supply of 5.2(± 2.5) ppt by Dorf et al. (2008).This study used ballon-borned DOAS (Differential Optical Absorption Spectroscopy) measurements of BrO in the stratosphere.Combined with modelling work the authors were able to infer the VSLS Br y contribution (from the BrO/Br y ratio).Our Br VSLS y estimate of ∼4.9-5.2 ppt is also in reasonable agreement with estimates based on ground-based BrO observations; for example, 6 (3-9) ppt (Schofield et al., 2006).However, it is significantly lower than the estimate of 8.4 (6.4-10.4)ppt of Sioris et al. (2006) from SCIAMACHY satellite BrO profiles.
Of the VSLS with a solely anthropogenic source, we have only considered C 2 H 5 Br in our calculated value of Br VSLS y .Our result may therefore represent a lower limit due to EDB, nPB and iPB not being included in the calculation.Montzka and Reimann et al. (2011) estimate that minor anthropogenic VSLS including C 2 H 5 Br, EDB and nPB species contribute <0.2 ppt via the SGI pathway.Our model shows C 2 H 5 Br and EDB are sufficiently long-lived to be potentially important carriers of bromine to the stratosphere.Simulations show, regardless of the surface vmr, ∼70-73 % and 80-85 % of total surface bromine from C 2 H 5 Br and EDB in the tropics can reach the lower stratosphere.
We have also quantified tropospheric loss of the major long-lived SGs.Tropospheric loss of the halons in the model was found to be negligible.For CH 3 Br we find ∼0.7 ppt of source gas is removed in the troposphere from S diagnosed ; 9 % of the assumed 7.5 ppt surface boundary condition.This is consistent with the estimated CH 3 Br gradients of 0-15 % reported by Schauffler et al. (1999).The PGI route still allows ∼0.33 ppt of the resulting Br y to enter the stratosphere.Therefore, ∼0.4 ppt of bromine is physically removed from the atmosphere (∼5 % of the surface boundary condition).This result implies that estimates of Br VSLS y that ignore tropospheric gradients of CH 3 Br may have a low bias of 0.4 ppt or so (though this may be compensated for by omission of minor VSLS).This result is likely sensitive to the model OH field and also treatment of wet deposition; both of which carry some uncertainty.For OH, our model reads monthly averaged 24-h mean values from a previous full chemistry integration.Tropospheric variability in OH is large and observations are sparse.However, comparison of OH profiles from the 1999 NASA PEM TROPICS-B campaign with modelled profiles (not shown), show reasonable agreement.The model OH field in this case was found to be within the min-max variability of the available observations from the surface to 200 mb.However, we acknowledge that model [OH] is still a potential source of error.

Conclusions
We have used a 3-D offline CTM and improved our previous estimate of the contribution of VSLS on the stratospheric bromine budget.We have considered 9 VSLS, methyl bromide and also four halons source gases.Modelled profiles of these gases give good agreement with tropical observations from the recent 2009 HIPPO-1 aircraft campaign.Both model and observations reiterate that a significant amount of total stratospheric bromine may arise from naturally-emitted VSLS.The observations did not detect significant levels of anthropogenic-VSLS.However, ethyl bromide (C 2 H 5 Br) was detected at ∼0.4 ppt in the TTL.Anthropogenic VSLS likely contribute only a small amount of bromine to the present day stratosphere.However, observations of these species are limited and more data is needed to support this conclusion.
Both ethyl bromide and ethylene dibromide (EDB) are long-lived in the tropical tropopause layer and thus are potentially efficient carriers of bromine to the stratosphere.The simulated local atmospheric lifetime of these anthropogenic gases are up to ∼183 and 603 days in the TTL, respectively.Therefore locally, these gases may breach the WMO working defintion of a very short-lived substance (i.e.lifetimes less than 6 months).Other anthropogenic gases considered, npropyl bromide and i-propyl bromide are shorter-lived.The maximum simulated local lifetimes in the TTL are ∼39 and 49 days.
Our model shows the contribution of the VSLS CHBr 3 , CH 2 Br 2 , CHBr 2 Cl, CHBrCl 2 , CH 2 BrCl and C 2 H 5 Br to stratospheric Br y is 4.9-5.2ppt for the present day, consistent with balloon-borne estimates.This may represent a lower limit as three anthropogenic gases, EDB, n-propyl bromide and iso-propyl bromide are not included in the calculation.Approximately 76 % of this supply comes from CHBr 3 and CH 2 Br 2 .To date, most model work has focused on these species.Our study shows minor VSLS (e.g.CHBr 2 Cl, CH 2 BrCl) individually contribute modest amounts.However, their overall total is significant and comparable to that of one of the major VSLS (e.g.CH 2 Br 2 ).Omission of these minor species in model simulations may result in an underestimate of stratospheric Br y by ∼1.2 ppt.More observations of these gases, particularly in the upper TTL, are required to further constrain estimates of their contribution to stratospheric Br y .Ongoing work to identify the largest geographical source regions of VSLS and to determine sea-air fluxes will ultimately lead to an improved treatment of emissions in global models.

Fig. 2 .Fig. 3 .Fig. 4 .
Fig. 2. Comparison of modelled CH 3 I profiles with observations made on board 4 DC8 flights during the NASA TC4 campaign (July-August 2007, http://www.espo.nasa.gov/tc4/).The CTM profiles are from both runs S diagnosed and S archived .The approximate base of the TTL is indicated with a dashed line.A

Fig. 5 .
Fig. 5. Modelled mean local tropical lifetime (days) with respect to OH oxidation, photolysis and total for (A) C 2 H 5 Br, (B) CH 2 BrCH 2 Br, (C) n-C 3 H 7 Br and (D) i-C 3 H 7 Br.Horizontal lines denote the approximate TTL.

Fig. 6 .
Fig. 6.Modelled tropical mean total bromine (ppt) for 2007 and from simulation S archived .Dashed horizontal lines denote the approximate base of the TTL and the cold-point tropopause (CPT).

Table 2 .
2007 tropical mean source gas injection (SGI), product gas injection (PGI) and total bromine (SGI + PGI) loading from VSLS, CH 3 Br and the halons in the lower stratosphere.Results are from simulations S diagnosed and S archived .