Ozone is an important trace gas in the Earth's atmosphere. The stratospheric layer with its highest abundance, the ozone layer, absorbs harmful ultraviolet radiation threatening all life forms on the Earth's surface and acts as a potent greenhouse gas (GHG). In the troposphere, ozone is considered a harmful pollutant. Catalytic cycles involving bromine and mixed halogen reactions, namely with chlorine, efficiently deplete ozone e.g.,. The ozone depletion efficiency of bromine is strongly related to the available amount of activated chlorine in the atmosphere . Long-lived, anthropogenically emitted, halogenated source gases (SG), e.g., CH3Br and halons, have been restricted by the Montreal Protocol and its amendments. Their atmospheric concentrations have started to decline globally seeChap. 1. Still, they contribute about 75 % to the overall bromine loading in the stratosphere. The remainder is provided by organic SG of oceanic origin of which methyl bromide (CH3Br), bromoform (CHBr3), and dibromomethane (CH2Br2) are the most abundant. Minor brominated very short-lived substances (VSLS) include the mixed bromochlorocarbons CHCl2Br, CHClBr2, and CH2ClBr. The tropospheric lifetime of these gases lies between several days to weeks. They are produced by plankton and macroalgae and are predominantly produced in coastal waters . Through gas exchange governed by the concentration gradient between ocean water and atmosphere, solubility, and wind stress, VSLS are emitted into the atmosphere. Transport to the stratosphere, as shown by different model studies , occurs in tropical regions of deep convection, most importantly the western Pacific and Maritime Continent, in Southeast Asia, and over the Gulf of Mexico. Organic SG are transported through the tropical tropopause layer (TTL) together with their inorganic product gases (PG). PG are produced through photochemical decomposition of VSLS and provide reactive bromine (Bry, from Br, Br2, HBr, BrO, BrONO2, BrNO2, BrCl, and HOBr) to the stratosphere. This is schematically shown in Fig. . In recent years, several approaches have been taken to describe the stratospheric or regional abundance of bromine from VSLS. Top-down scenarios  match atmospheric observations by setting constant fluxes or boundary concentrations. Bottom-up scenarios e.g., developed emission climatologies by extrapolating measurements in the surface ocean and marine boundary layer and calculate emissions accordingly. As shown by , the bottom-up fluxes based on the oceanic water concentrations of are in good agreement with available atmospheric VSLS observations. Recently, have investigated the future evolution of the ocean–atmosphere fluxes of VSLS through the 21st century based on Coupled Model Intercomparison Project (CMIP) 5 model output and fixed atmospheric VSLS concentrations. They found fluxes of CH2Br2 and CHBr3 increasing by 6.4 % (23.3 %) and 9.0 % (29.4 %), respectively, dependent on the Representative Concentration Pathways (RCP) 2.6 (RCP8.0) scenario.

In this study, we will address the open questions on how these oceanic emissions of VSLS evolve in response to a changing climate and changing atmospheric concentrations (Sect. ), how transport and tropospheric chemistry influence the stratospheric bromine abundance in a changing climate (Sect. ), and how stratospheric ozone will be affected by the assumed changes in VSLS abundance (Sect. ). Details about the model and simulations used in this study will be given in Sect. .

All model experiments have been performed using the ECHAM/MESSy Atmospheric Chemistry (EMAC) model . Table  gives an overview over the key factors of the simulations.

Future changes in fluxes of brominated VSLS from the ocean are studied with a free-running long-term simulation (SC_free, 1979–2100) using a simplified chemistry (Sect. ), augmented by a similar simulation, but nudged towards the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim reanalysis over the period 1979–2012. Therein, VSLS emission fluxes are computed online from prescribed seawater concentrations. The simplified chemistry simulations use EMAC version 2.50 with submodels airsea (with the water-side transfer velocity parameterization kw=0.222u2+0.333u with respect to wind speed according to ), cloud, cloudopt, convect (with operational ECMWF convection scheme), cvtrans, ddep , ptrac , rad, scav , surface, and tnudge . The setup is as in . Chemical reactions are not computed interactively, e.g., via the EMAC submodel mecca . The VSLS lifetime due to reaction with OH has been fixed to monthly mean values from the National Centre for Meteorological Research (CNRM)  model calculations, while photolysis rates are computed within the EMAC submodel jval . Only in these simulations with simplified chemistry, OH concentrations have been set to zero in the lower troposphere (700–1000 hPa) to reduce the variability of ground level volume mixing ratio (VMR) of VSLS. The chemical lifetime of VSLS in the lower troposphere is therefore overestimated. Due to the longer lifetime, VSLS are more abundant in the lower troposphere leading to a flux suppression. Water concentrations of CH2Br2 and CHBr3 have been held constant using the climatology of . For mixed bromochlorocarbons (CHBrCl2, CHBr2Cl, CH2BrCl), water concentrations have been estimated by scaling CHBr3 concentrations to obtain a better agreement between model simulation and tropical mean profile and surface observations of these VSLS. Based on the lifetime estimate, VSLS are decomposed and converted to Bry. The partitioning of Bry into Br, Br2, HBr, BrO, BrONO2, BrNO2, BrCl, and HOBr in these simplified chemistry simulations has been computed offline from a full-chemistry EMAC simulation of 1-year duration with 6-hourly output. Scavenging is applied to Br, Br2, HBr, BrNO2, BrONO2, BrCl, and HOBr. Concentrations of CO2, CH4, CFC, and N2O in SC_free are taken from a CNRM CM5 model  simulation with RCP6.0 scenario .

Data of a full-chemistry long-term simulation RC2-base-05, over a timespan of 150 years (1950–2100) and performed as part of a Chemistry-Climate Model Initiative (CCMI) recommended set of simulations by the Earth System Chemistry integrated Modelling (ESCiMo) Consortium will be used for studying changes in transport and photochemical transformation of bromine species (Sect. ). In this simulation, VSLS fluxes have been held constant following scenario five of .

An intermediate-term experiment, consisting of a set of two simulations and spanning the years 2075–2100, has been performed for assessing implications on ozone depletion in a future climate with significantly lower chlorine loading in the atmosphere (Sect. ). The simulations named RT1a and RT1b both include online computation of aerosol formation. Fluxes of CH2Br2 and CHBr3 are computed online from seawater concentrations of using the EMAC submodel airsea as in RC2-base-05 with the kw parameterization according to , which is strictly quadratic with respect to wind speed (kw=0.31u2). For assessing the impact of VSLS on ozone, all VSLS emissions have been switched off in RT1b. The impact of various kw parameterizations on VSLS emission has been previously studied. The differences on the global level are <15 % (see Table 4 in , when comparing , and ). For wind speeds exceeding 10 ms-1 the kw parameterization diverges slightly stronger towards higher transfer velocities compared to the parameterization (cf. Fig. 1 in , and Fig. 2 in ). Regarding integrated global emissions of VSLS, both parameterizations result in similar fluxes given that the mean global wind speed lies in a range where these parameterizations do not differ drastically. However, the parameterization reacts more sensitively to changes in wind speed, which introduces a further uncertainty when assessing changes over time in a changing climate. The full-chemistry experiments use EMAC version 2.51 (RC2-base-05) and 2.52 (RT1a/b), respectively. The dynamics have not been specified except for a weak nudging of the equatorial wind quasi-biennial oscillation (QBO). RC2-base-05 combines hindcast with future projections. The setup of RT1a and RT1b is almost identical to RC2-base-05, and so we refer to the corresponding paper by the ESCiMo Consortium  for general information. The major difference lies in the aforementioned treatment of VSLS emission, which is handled analogous to SC_free except for mixed bromochlorocarbons emissions taken from . Since heterogeneous reaction and chlorine activation are important for the depletion process of ozone, tropospheric and stratospheric aerosol formation is computed online using the submodel GMXe  of EMAC. The setup has been adapted from RC1-aero-07  with modifications as described by . Radiation coupling had been activated in GMXe, but cloud coupling had not been activated. In this regard, an additional oceanic sulfur source, carbonyl sulfide (COS), which is a major source of stratospheric sulfur, has been included in addition to dimethyl sulfide (DMS). Whereas the emission of the latter is computed from prescribed ocean concentrations, constant fluxes of COS have been adopted from . Additional reaction pathways of sulfur have been enabled accordingly. RT1a and RT1b have been initialized with available monthly mean values from RC2-base-05. COS has been initialized from a simulation whose results have been published recently , including an artificially increased oceanic source to close the atmospheric budget.

The model's spatial resolution is T42L39MA for the simplified chemistry experiments, T42L47MA for RC2-base-05, and T42L90MA for RT1a/RT1b, respectively, corresponding to a 2.8∘×2.8∘ grid, with a top level at 0.01 hPa, and 39, 47, or 90 vertical hybrid-pressure levels. The mean tropical troposphere (below 100 hPa) is discretized into 16, 26, or 27 levels, and the mean tropical stratosphere between 100 and 1 hPa consists of 15, 15, or 48 levels, respectively. Emissions of GHG follow the RCP6.0 scenario and sea surface temperature (SST) and sea ice cover (SIC) are prescribed from Hadley Centre Global Environment Model version 2 (HadGEM2) forced with the RCP6.0 scenario for all simulations accordingly.

In this section, we investigate how a changing climate may influence emission fluxes of VSLS from the ocean. We will assess the impact of changing physical factors (e.g., SST, SIC, and wind speed) on ocean–atmosphere gas exchange driven by the RCP6.0 scenario. Here we assume constant oceanic concentrations of VSLS over the course of the century (following ). This specific assumption might not hold since the effects of climate change, e.g., increase of ocean temperature, acidification, change of salinity, and nutrient input, on marine organisms and thus the production of CH2Br2 and CHBr3 is not yet fully understood. Recent combined marine ecosystem model studies imply a global decrease of net primary production by plankton over the course of the 21st century . However, the impact on bromocarbon concentration, predominantly produced by macroalgae in coastal regions, remains unclear.

As implemented in the EMAC submodel airsea , the flux of a gas dissolved in ocean water to the atmosphere is governed by its concentration gradient Δc and transfer velocity k: Φ=k⋅Δc=k⋅cw-H⋅cair, with k=(1/kw+R⋅H⋅Tair/kair)-1, wherein R is the universal gas constant and H is the Henry coefficient for a specific gas. The transfer velocity depends largely on air temperature Tair and surface wind speed, which is taken into account by distinguishing between water- and air-side transport velocities (kw, kair). kw is a polynomial function of wind speed depending on the chosen parameterization as mentioned in Sect. . The corresponding water and atmospheric concentrations are named cw and cair.

In Fig. , the difference of VSLS fluxes with respect to the start of the simulation in 1979 is shown for the free-running and nudged simplified chemistry simulation. For both CH2Br2 and CHBr3, all zonal bands display linearly rising fluxes. The strongest increase with respect to 1979 values is found in the tropical zone (20∘ N–20∘ S) with roughly 2.5 and 13 Ggyr-1 for CHBr3 and CH2Br2 respectively. Relative to the absolute value of the zonally averaged fluxes, this yields an increase of about 10 % over the course of the century (Table ). The increase is slightly stronger in the southern tropics. The strongest relative increase in flux is found in the northern hemispheric polar region (90–50∘ N), with 25 % and roughly 55 % for CHBr3 and CH2Br2, respectively.

Regarding the changing physical factors, the HadGEM2 prescribed SSTs are increasing almost linearly over the course of the century (Fig. a). Under the RCP6.0 scenario, this increase in SST ranges between 1 and 3.5 ∘C. The weak rise in Antarctic SST is accompanied by a weakly increasing Antarctic flux of VSLS. The corresponding HadGEM2 prognosticated retreat of Arctic sea ice is shown in Fig. b. Sea ice is not regarded as a source of VSLS in our study and therefore only acts as a lid blocking the ocean-to-atmosphere flux. Since the water concentrations from used in our simulations do not take SIC into account, water concentrations have been extrapolated for regions typically covered by ice at present. In the airsea submodel, if SIC (fraction of grid box) is larger than 0.5, the transfer velocity (kw) is equal to zero, in other cases kw is scaled depending on the fraction of SIC. Hence, a polar sea which is to a large extent free of sea ice has increased fluxes of VSLS in our future simulations. However, there are large uncertainties regarding the VSLS water concentrations in the future polar sea, for the polar ecosystem as a whole is undergoing a drastic change. In accordance to the general increase in flux, the Arctic August–September maximum of flux is expected to be more pronounced. In both hemispheres, seasonal cycles in zonally averaged VSLS fluxes peak in the summer months and show minima in late winter. There is a slightly stronger increase of fluxes in the future during the time periods of maxima but no change in phase. Negative emissions representing a net sink of atmospheric CH2Br2 are found during winter at high latitudes in the Northern Hemisphere. In the northern tropics, CHBr3 shows a distinct maximum in northern hemispheric summer, while the southern tropics do not display any seasonal cycle. Despite increased ocean–atmosphere fluxes in the future, only taking the changes of physical factors into account, seasonal cycles remain largely the same. In our simulations, zonally averaged absolute wind speed at 10 m is only slightly changing over the course of the 21st century and with varying sign (-4–2 %). Thus, it is indicated by our simulations but not explicitly shown that the important factor regarding an increase of ocean–atmosphere flux of VSLS is the change in SSTs.

In Fig. , resulting VMR profiles of organic (Brorg) and inorganic bromine (Bry) from VSLS are shown. VSLS data have been averaged over a time period 1990–2000 and 2090–2100. The VMR profile of Brorg displays a steep decline in the lower troposphere (400–1000 hPa) by more than 50 % of ground level VMR and stays almost constant before entering the stratosphere, where VSLS are quickly dissociated. Comparing present-day and future values (keeping in mind that OH concentrations are nudged towards monthly mean values of OH and photolysis rates are fixed in SC_free), Brorg is found to have increased throughout the atmosphere by about 0.1–0.4 ppt. Surface values of VSLS increase by 0.47 ppt (9 %), while in the lower stratosphere the increase amounts to 0.3 ppt (8 %). VMR of Bry is increased from the lower stratosphere upwards by roughly 0.4 ppt (10 %). These changes in the vertical profiles can be attributed to enhanced emissions in a future climate which, as shown, are of the order of 10 % in the tropics.

As can be inferred from Eq. (), ocean–atmosphere fluxes are sensitive to the abundance of VSLS in the atmosphere and differing wind speed parameterization. An increased chemical dissociation of VSLS in the lowermost troposphere (e.g., due to a probable future increase in OH) would reduce the atmospheric concentration and therefore increase the flux from the ocean to the atmosphere without necessarily increasing the actual amount of bromine which is transported to the stratosphere. The total amount of bromine from VSLS transported through the UTLS strongly depends on the washout of inorganic PG (BryVSLS) and hence on the partitioning and heterogeneous reactions converting BryVSLS between soluble, e.g., HBr and HOBr, and insoluble, e.g., BrO, species e.g.,. Since OH concentrations in the lower troposphere have been set to zero in SC_free, the atmospheric lifetime and the resulting abundance of VSLS in the lower troposphere is enhanced. Therefore, the total ocean–atmosphere flux is suppressed. In this regard, fluxes from RT1a at the end of the 21st century have been compared to SC_free within the same time period. Much stronger fluxes (1.3–1.5 times) have been found in RT1a in comparison to SC_free. Particularly, no net sink for CH2Br2 occurs at high latitudes in RT1a. This partly explains the smaller increase in comparison to results recently published by . diagnosed the flux from parameters such as SST and wind speed for a fixed VSLS concentration gradient and for different CMIP5 model simulations. They found an increase in flux of CHBr3/CH2Br2 of 29.4/23.3 % for the RCP8.0 scenario and 9.0/6.4 % for RCP2.6, respectively. In addition to the smaller absolute fluxes due to the artificial suppression caused by setting OH to zero in the lower troposphere, we expect a smaller increase in flux from a theoretical point, taking Eq. () in to account, since we allow atmospheric concentrations to respond to changing flux. Underlying changes in photochemical dissociation and tracer transport due to a changing climate have not been disentangled at this point and will be studied in detail in the following section.

In addition to the possible increase in oceanic VSLS emissions due to climate change, discussed in the previous section, atmospheric transport and chemical transformation processes are also sensitive to climate change and may contribute to a change in the future stratospheric bromine loading from VSLS. These aspects will be studied in this section, based on the RC2-base-05 ESCiMo simulation, spanning 150 years from 1950 to 2100, assuming constant VSLS fluxes. Hence the fluxes of VSLS do not response to changes in the ground level abundances of VSLS.

In Fig. , profiles of brominated substances are shown for the tropics. The profiles are weighted by the amount of bromine atoms per molecule. The whole 150-year data set has been smoothed using a moving average with a box window size of 11 years to account for, e.g., seasonal variations and the solar cycle. From these smoothed data, three reference years have been chosen for the analysis: 1980, 2016, and 2100. Therefore, 2100 is referring to June of the last valid year of the smoothed data (2094). To guide the eye, the corresponding mean tropical tropopause heights from the model output are shown together with the profiles. There is an upward shift of the tropopause height of about 8 hPa between present day and future. An upward shift of VSLS VMR profiles in 2100 in comparison to past/present-day profiles is also visible. In the RCP6.0 scenario, ground level VMR of CH3Br and VSLS are constant from 2016 onward. In case of CH3Br, this roughly amounts to 1980 values. For all years, we find a fast decrease of VSLS of 5 ppt with a standard deviation of 0.25 ppt (or about 50 % compared to ground level VMR) between the surface and the mid-troposphere at about 500 hPa. The comparison of the difference of profiles between future and past/present (Fig. a, lower panel) reveals decreasing bromine values from VSLS by about 0.1–0.8 ppt throughout the troposphere, while there is an increase of the same order of magnitude in the lower stratosphere. Similar results have been published for RCP4.5 and RCP8.5 scenarios, attributing these to changes in the tropospheric circulation and to the primary oxidant OH . The amount of inorganic PG from VSLS (BryVSLS) in the UTLS is decreasing by the same order of magnitude due to the enhanced upwelling in the tropics. As air in the UTLS becomes younger in a future climate, less SG (BrorgVSLS) will be dissociated into PG (BryVSLS) compared to present day. For 2016, this decline is compatible with a decreasing amount of VSLS in the troposphere. A slight excess of BryVSLS in the stratosphere is found for 1980 in comparison to 2016 and 2100. This excess and the strong variability (denoted by the shown standard deviation) can be attributed to the hindcast period (1950–2005) of the simulation including volcanic eruptions. Large volcanic eruptions can influence the transport of bromine from VSLS into the stratosphere which may be related to a similar effect as seen in stratospheric water vapor . Since volcanic activity has not been included in the future scenario, there is no such impact on BryVSLS from 2005 onward.

The largest change between present and future stems from the estimated decrease of long-lived SG, in particular halons and CH3Br. At present, halons contribute about 6–7 ppt to the total bromine loading of the lower stratosphere (∼23 ppt), which is about the same amount as VSLS and CH3Br in RC2-base-05, whereas by the end of the century their contribution is reduced significantly to 1–2 ppt of total bromine (∼17 ppt). This decline in long-lived, anthropogenically emitted SG is altering the amount of bromine released in the stratosphere on longer timescales. VSLS are already reduced due to photochemical dissociation when entering the TTL, while halons are dissociating more slowly, providing a long lasting source of bromine to the stratosphere (Fig. b, lower panel). It is important to note that although there is an increase of BrorgVSLS of 0.5 ppt in the stratosphere assuming constant ocean–atmosphere fluxes, the overall amount of bromine in the stratosphere due to VSLS (BrtotVSLS) might be decreasing in the future. This depends on whether PG (BryVSLS) are transported alongside the VSLS into the UTLS or removed through washout in the troposphere. The model representations of underlying processes, e.g., conversion between soluble and insoluble inorganic bromine species through heterogeneous chemical reactions, are still uncertain.

In the following, we will derive a semi-analytic model to separate various aspects affecting the future VSLS distribution in the atmosphere (Sect. ). Since the atmospheric window for air entering the stratosphere is located in the tropics, we will focus on averaged tropical atmospheric quantities. Subsequently, the transition between troposphere and stratosphere caused by a rising tropopause is influencing the interpretation of VMR profile differences between present and future. This will be discussed in Sect. .

In this section, the influence of brominated very short-lived source gases on ozone depletion will be discussed. Based on RT1a and RT1b, we assess the impact of VSLS on a zonally averaged ozone distribution at the end of the 21st century. For a thorough discussion of future trends, the 25-year data set is too short. However, from our long-term simulations (SC_free, SC_nudged, RC2-base-05), long-term influence of emission perturbations on ozone is not assessable. As described in Sect. , SC_free does not include interactive ozone chemistry, whereas RC2-base-05 incorporates prescribed fluxes based on scenario five by . Furthermore, results from RC2-base-05 cannot be compared to RT1a and RT1b directly because of significant differences in ozone distribution and amount between the differing vertical resolutions (L47MA and L90MA) of the model. This issue has been already reported by .

Zonally averaged data of total column ozone have been smoothed using a moving average algorithm with box window size of 11 years (Fig. ). In general, ozone trends at the end of the century are roughly the same for both resolutions. The actual amount of ozone differs, with L90 showing more ozone except for the northern hemispheric polar region and midlatitudes. In the Arctic, RT1a and RT1b even indicate, in contrast to RC2-base-05, slightly decreasing total column ozone. In the case of RT1a and RT1b, this might be partially caused by interactive aerosol and accordingly added oceanic COS source. This bias in total column ozone between the vertical resolutions is larger than the difference between RT1a and RT1b.

For estimating the effect of brominated VSLS on ozone depletion, the difference in zonally averaged ozone of RT1a and RT1b has been computed. A period of 20 years (2080–2100) has been used accounting for an estimated model spinup of 5 years in the beginning. In Fig. , the relative difference ((RT1a-RT1b)/RT1a×100) is shown as contour plot. Dashed lines indicate a decrease of ozone in the simulation with VSLS turned on (RT1a) compared to the one with VSLS turned off (RT1b), while solid lines indicate an increase. Significance has been estimated as divergence from zero in units of standard error of mean. It is indicated by shades of blue. VSLS cause a tropospheric ozone reduction on the order of 1–2 %, mainly at high latitudes. The UTLS region in the tropics is most affected; there VSLS cause a decrease of ozone of about 3 %. The decrease of ozone in the high latitude troposphere and tropical UTLS is rendered significant. Increasing amounts of ozone (∼1 %) are found in the Antarctic middle and upper stratosphere, but these are mainly not significant. This increase in ozone abundance may be due to dynamical feedback .

While VSLS have a large impact on Antarctic ozone depletion during the ozone hole period, i.e., during times with high-stratospheric chlorine loading from about the late 1970s to the second half of the 21st century (e.g., ), we find that by the end of the 21st century VSLS have less impact on total Antarctic stratospheric ozone depletion under low chlorine loading, although their importance relative to the total stratospheric halogen load is increasing (about 40 % in accordance to ). Assuming an adherence to the Montreal Protocol, stratospheric volume mixing ratios of Cly will decrease exponentially in the course of the 21st century from its peak values in 2000. From the Chap. 2, Figs. 2–21, a decline of Cly loading at 1 hPa of about 2 ppbv between 2000 and 2080 can be deduced. In accordance, we find a reduction of zonally averaged stratospheric Cly at 1 hPa of 2.1 ppb by the end of the 21st century compared to year 2000 in RC2-base-05. Since RT1a and RT1b have identical chlorine load and do not include present day, we cannot assess the chlorine moderation effect from these two simulations. have shown a reduction of ozone due to VSLS in the TTL region on the order of 6 % during the period 1970–1982 with significantly less stratospheric chlorine compared to the later period 1983–2005 (7 %), while SupplementFig. S3 have shown a change of total ozone column due to VSLS (on/off scenario) of the order of 1–3 % in pre-industrial conditions. Given that the VSLS emission scenario five of  used in RC2-base-05 has about twice the amount of VSLS compared to RT1a and chlorine abundance in the stratosphere will drop to 1970 values by the end of the 21st century seeChap. 12, our results are in good agreement with these previous studies. In more detail, investigated the combined influence of brominated VSLS and chlorine on ozone. For two stratospheric Cly loadings (3, 0.8 ppb), which correspond to 2000 and 2100, they have varied the amount of VSLS. have shown the more chlorine in the stratosphere the stronger ozone is affected by an increase of bromine VMR from VSLS. In concert with our results, although by doubling the initial amount of VSLS on a varying bromine background from anthropogenic sources, they have found a significant decrease of ozone in the tropical UTLS and polar region on the order of 2–4 % and slight, insignificant increases in the Antarctic mid-stratosphere Fig. 1e.

We have investigated long-term changes in emission and transport of brominated VSLS under a changing climate (RCP6.0). Under the implicit assumption of constant concentrations of VSLS in the ocean waters, over a timespan of 120 years, we have found an enhancement of zonally averaged fluxes of CH2Br2 and CHBr3 on the order of 10 % between present day and the end of the 21st century. A strong increase of flux (up to 55 % in CH2Br2 and 25 % in CHBr3) has been found in the northern hemispheric polar region. There, the retreat of sea ice is playing a key role. Exposing almost the entire polar ocean in August–September by the end of the 21st century, sea ice does not longer act as a lid to ocean–atmosphere fluxes of VSLS. Sea ice itself has not been considered a source of VSLS in our simulations. Subsequently, an increase of organic bromine in the UTLS is found of the same order of magnitude (8–10 %).

Ocean–atmosphere fluxes are sensitive to the abundance of VSLS in the atmosphere as well as on wind speed. An increased dissociation of VSLS in the lowermost troposphere, e.g., due to increasing OH concentrations in the RCP6.0 scenario, reduces the atmospheric concentration and therefore increases the flux from the ocean to the atmosphere without necessarily increasing the actual amount which is transported to the stratosphere. The total amount of bromine from VSLS transported through the UTLS strongly depends on the washout of inorganic PG (BryVSLS) and hence on the partitioning and heterogeneous reactions converting BryVSLS between soluble (e.g., HBr, HOBr) and insoluble (e.g., BrO) species. But these mechanisms have not been subject to our study.

For prescribed, constant VSLS fluxes, an increase of lower-stratospheric VSLS of about 5–10 % is due to enhanced vertical transport in the tropics. This increase is counteracted by a corresponding decrease in inorganic bromine. Everything else unchanged, an increase in tropical upwelling will not change the total amount of bromine in the future stratosphere. Additionally, due to enhanced future OH concentrations in the RCP6.0 emission scenario, the tropospheric lifetime of VSLS is reduced, which leads to a decrease of total bromine from VSLS. Furthermore, we have diagnosed a decrease of BrtotVSLS by 0.5–2 ppt for a fixed pressure level with respect to the mean tropical tropopause ΔTPP≈20 hPa, if the upward shift of the mean tropical tropopause of 0.81 hPadecade-1 is taken into consideration.

The impact of enhanced fluxes of brominated VSLS on future ozone abundance has been evaluated by comparing two experiments of which one has no VSLS emission and the other interactively computed fluxes from constant ocean concentrations of VSLS. We have found a significant reduction of ozone in the tropical UTLS of about 3 %. In the troposphere the largest significant decrease of ozone amounts to 1–2 %. Thus, bromine from VSLS may not act as a major source to future stratospheric ozone depletion. While interactive emissions from constant ocean concentrations have been taken into consideration, the actual climate change inflicted change in the production of VSLS by macroalgae in the ocean remains an open question. Whether the found increase of ocean–atmosphere fluxes of VSLS and a future decrease of VSLS in the troposphere will cancel out or overcompensate would need further simulation studies.