Co-emission of volcanic sulfur and halogens amplifies volcanic effective radiative forcing

Abstract. The evolution of volcanic sulfur and the resulting radiative forcing following explosive volcanic eruptions is well understood. Petrological
evidence suggests that significant amounts of halogens may be co-emitted alongside sulfur in some explosive volcanic eruptions, and satellite
evidence indicates that detectable amounts of these halogens may reach the stratosphere. In this study, we utilise an aerosol–chemistry–climate
model to simulate stratospheric volcanic eruption emission scenarios of two sizes, both with and without co-emission of volcanic halogens, in order
to understand how co-emitted halogens may alter the life cycle of volcanic sulfur, stratospheric chemistry, and the resulting radiative forcing. We
simulate a large (10 Tg of SO2) and very large (56 Tg of SO2) sulfur-only eruption scenario and a corresponding
large (10 Tg SO2, 1.5 Tg HCl, 0.0086 Tg HBr) and very large (56 Tg SO2, 15 Tg HCl,
0.086 Tg HBr) co-emission eruption scenario. The eruption scenarios simulated in this work are hypothetical, but they are comparable to
Volcanic Explosivity Index (VEI) 6 (e.g. 1991 Mt Pinatubo) and VEI 7 (e.g. 1257 Mt Samalas) eruptions, representing 1-in-50–100-year and 1-in-500–1000-year events, respectively, with plausible amounts of co-emitted halogens based on satellite observations and volcanic plume modelling. We show that co-emission of volcanic halogens and sulfur into the stratosphere increases the volcanic effective radiative forcing (ERF) by
24 % and 30 % in large and very large co-emission scenarios compared to sulfur-only emission. This is caused by an increase in both the
forcing from volcanic aerosol–radiation interactions (ERFari) and composition of the stratosphere (ERFclear,clean). Volcanic
halogens catalyse the destruction of stratospheric ozone, which results in significant stratospheric cooling, offsetting the aerosol heating
simulated in sulfur-only scenarios and resulting in net stratospheric cooling. The ozone-induced stratospheric cooling prevents aerosol self-lofting
and keeps the volcanic aerosol lower in the stratosphere with a shorter lifetime. This results in reduced growth by condensation and coagulation
and a smaller peak global-mean effective radius compared to sulfur-only simulations. The smaller effective radius found in both co-emission scenarios
is closer to the peak scattering efficiency radius of sulfate aerosol, and thus co-emission of halogens results in larger peak global-mean
ERFari (6 % and 8 %). Co-emission of volcanic halogens results in significant stratospheric ozone, methane, and water vapour
reductions, resulting in significant increases in peak global-mean ERFclear,clean (> 100 %), predominantly due to ozone loss. The
dramatic global-mean ozone depletion simulated in large (22 %) and very large (57 %) co-emission scenarios would result in very high levels
of UV exposure on the Earth's surface, with important implications for society and the biosphere. This work shows for the first time that co-emission of plausible amounts of volcanic halogens can amplify the volcanic ERF in simulations of
explosive eruptions. It highlights the need to include volcanic halogen emissions when simulating the climate impacts of past or future eruptions,
as well as the necessity to maintain space-borne observations of stratospheric compounds to better constrain the stratospheric injection estimates of
volcanic eruptions.


processes including the formation, growth, transport and loss of aerosol (Dhomse et al., 2014). GLOMAP-mode also calculates aerosol optical properties online which are used to calculate direct and indirect radiative effects 155 (Mulcahy et al., 2020).
In UKCA, stratospheric ozone concentrations are determined by sets of photochemical reactions as well as ozone destroying catalytic cycles involving chlorine, bromine, nitrogen, and hydrogen radical species (Archibald, 2020).
Photolysis reactions in UKCA utilise rates calculated from a combination of the FAST-JX scheme and look-up 160 tables (Telford et al., 2013). Ozone depleting radical species are produced by the photolysis of halogen containing compounds reacting on the surface of stratospheric aerosols, including hydrochloric acid (HCl), chlorine nitrate (ClONO2), hydrogen bromide (HBr), and bromine nitrate (BrONO2). Heterogeneous reactions in the presence of polar stratospheric clouds (PSCs) in the polar lower stratosphere or in the presence of sulfate aerosol following explosive volcanic eruptions are also important for stratospheric ozone concentrations. Eight additional 165 heterogeneous reactions involving chlorine and bromine species were added as described in Ming et al., (2020), with the main change being the explicit treatment of the reactions of four additional chemical species: Cl2, Br2, ClNO2, and BrNO2 which are photolysed to produce Cl and Br radicals.
Volcanic effective radiative forcings (hereafter ERF) are calculated as differences (Δ) in the net top of atmosphere 170 (TOA) radiative fluxes (F) between perturbed and control simulations, as follows:

= ∆
Volcanic ERF is decomposed as described in Schmidt et al., (2018)  (2013) to obtain Fclean and Fclear,clean. Where Fclean denotes a radiation flux diagnostic calculation without aerosolradiation interactions but including aerosol-cloud interactions through microphysics. Fclear,clean denotes a radiation flux diagnostic calculation that ignores both aerosol and cloud-radiation interactions. Thus, F -Fclean, determines the impact of all aerosols and Δ(F -Fclean) is an estimate of the forcing from volcanic aerosol-radiation interactions (ERFari). The second term Δ(Fclean -Fclean,clear) represents the difference in the clean-sky cloud radiative forcing, 185 and is an estimate of the aerosol-cloud interactions (ERFaci) due to volcanic emissions. The third term, ERFclear,clean accounts for changes not directly due to aerosol or cloud interactions, largely the result of changes in surface albedo and atmospheric composition. In this study, we fix surface temperature and sea-ice fields meaning that surface albedo is expected to be unchanged and any Fclear,clean changes are the result of atmospheric compositional changes. 190 https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License.

Experimental Design
We utilise atmosphere-only, time-slice experiments whereby the initial sea surface temperature, sea ice fraction and forcing agents are prescribed climatologies with values taken from fully coupled UKESM1.0 historical runs produced for CMIP6 (Eyring et al. 2016) and averaged over the years 1990 -2000. The 1990s, and thus these timeslices, were characterised by high background halogen levels due to anthropogenic emissions of CFCs 195 throughout the preceding decade. A control simulation was run with a 15 year spin up followed by a further 20 years. The effect of explosive volcanic eruptions was investigated by running a series of 10 year volcanic perturbations spun off from 6 different years in the control run to represent the variability in QBO states. Changes are plotted as the difference between the average of the 6 ensembles and the climatology derived from the 20 year control run, cumulative forcings are calculated as the sum of the forcing over the full 10 year simulation duration. 200 The volcanic emissions are prescribed by direct injection of SO2, HCl and HBr into the stratosphere with a Gaussian plume vertical distribution centred on 21 km, lasting for 24 hours on July 1st. The gases were injected in the tropics (5 o S latitude and 0 o longitude) to represent a typical tropical explosive eruption (Newhall et al., 2017).

205
Since historical stratospheric volcanic SO2 fluxes are variable and the volcanic flux of HCl and HBr into the stratosphere remains uncertain, we developed a simulation matrix that spans a range of possible explosive volcanic emissions. The four sets of experiments have one high SO2 (56 Tg), and one low SO2 (10 Tg) emission scenario both with (HAL56 and HAL10) and without halogens (SULF56 and SULF10), as shown in Table 1. These eruption sizes (56 and 10 Tg SO2) are similar in size to a VEI 7 (e.g. 1257 Mt. Samalas) and VEI 6 (e.g. 1991 Mt. 210 Pinatubo) eruption, representing 1 in 500 -1000 year and 1 in 50-100 year events respectively. HAL56 utilises the 1257 Mt. Samalas HCl and HBr emission estimates from Vidal et al. (2014) and assumes a conservative ~5% stratospheric halogen injection efficiency, less than the 10-20% predicted by plume modelling in Textor et al (2013) and closer to the observed efficiency following El Chichon (>2.5%) and in the ice core record of Mt.
Mazama (8%), as well as the fraction supported by Wade et al. (2020). This results in a HCl:SO2 molar ratio of 215

Sulfur Microphysics and ERFari
Atmospheric burdens of volcanic sulfur species are summarized in Figure 1. As found by Lurton et al., (2017), volcanic halogens deplete OH via equation 1 which limits the availability of OH for SO2 oxidation, leading to slower destruction of volcanic SO2 and an increase in SO2 e-folding time of 21% and 40% in HAL10 and HAL56 compared to SULF10 and SULF56 respectively. As the rate of formation of sulfuric acid is decreased, we simulate a corresponding delay in the formation in sulfate aerosol and a reduction in the peak sulfate aerosol burden by 8% in both HAL10 and HAL56. 235 https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. Despite the slower rate of SO2 oxidation, the co-emission of halogens reduces the lifetime of the sulfur burden to 17.3 and 11.7 months in HAL10 and HAL56, compared with 21.2 and 13.6 months in SULF10 and SULF56, a decrease of 18% and 14% respectively. This indicates that co-emission of halogens alters the rate at which sulfur 245 is removed from the atmosphere. Significant differences in temperature change are simulated between the sulfuronly and halogen simulations. In sulfur-only simulations, strong positive temperature anomalies of ~3 K due to sulfate aerosol absorption of infra-red radiation are simulated across the tropical stratosphere ( Figure 2). This aerosol heating lofts volcanic aerosol to altitudes higher than the initial injection height in the model. By contrast, co-emission of volcanic halogens results in significant stratospheric ozone depletion of 22-57% (see section 3.2) 250 https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. and in turn this results in large negative temperature anomalies over most of the lower and middle stratosphere ~ -3 K (Figure 2 ). Ozone generates heat in the stratosphere by absorbing both incoming SW radiation from the sun and by absorbing upwelling LW radiation from the troposphere. Thus, decreasing stratospheric ozone results in lower stratospheric temperatures; the effect of which is larger in magnitude than the aerosol heating and prevents volcanic sulfate aerosol being self-lofted. The volcanic sulfate aerosol thus remains at significantly lower altitudes 255 in HAL10 and HAL56 (~21-22km) compared with SULF10 and SULF56 (~24-25km) ( Figure 2). Lower altitude aerosol remains in a faster branch of the Brewer-Dobson Circulation ( Figure S1) which results in faster transport to high-latitudes and removal from the stratosphere ( Figure 1d). https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. The radiative impact of sulfate aerosols depends on the particle size (Timmreck et al., 2010). Using Mie scattering theory, Lacis (2015) found that the cross section per unit mass is largest for sulfate aerosol with effective radius of ~0.25 μm. The smaller Reff in HAL10 and HAL56 is closer to 0.25μm and results in more efficient scattering of radiation per unit mass (Timmreck, 2012). Therefore, we simulate 11% and 22% higher peak global-mean 290 stratospheric aerosol optical depth (SAOD) anomalies in HAL10 and HAL56 than their equivalent SULF https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. simulations (Figure 4), despite having a 14% and 9% smaller peak aerosol burden. Correspondingly, we simulate an 8% and 6% increase in the peak global-mean ERFari in HAL10 and HAL56 compared to SULF10 and SULF56 ( Figure 4). The SAOD and ERFari anomalies are a balance between the offsetting effects of smaller aerosol and shorter lifetime which result in a net-zero impact on cumulative ERFari despite a significant increase in the peak 295 global-mean ERFari ( Figure S2a,b).

Composition Changes and Resulting ERFclear,clean
Co-emission of volcanic sulfur and halogens causes significant perturbations to the chemistry of the stratosphere beyond the depletion of OH in HAL10 and HAL56 mentioned in section 3.1 Stratospheric methane, stratospheric 305 water vapour (SVW) and, in particular, stratospheric ozone are all impacted.
In sulfur-only simulations, we simulate a modest reduction in global-mean ozone column, -9 DU (-3.9%) in SULF10 and -15 DU (-6.6%) in SULF56 (Figure 5a,c). This ozone depletion is catalysed by halogen radicals activated from background halogens on the surface of volcanic aerosol. In simulations with co-emitted halogens 310 we simulate more dramatic ozone depletions; HAL10 resulted in a peak global-mean ozone reduction of 65 DU (-22%) 1-2 years after the eruption followed by a gradual recovery over the next 3-4 years ( Figure 5d). HAL56 https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. resulted in a peak global-mean ozone reduction of 175 DU (-57%) 1-2 years after the eruption followed by a gradual recovery the remainder of the 10 year simulation, with an average reduction of 82 DU (-27%) over the 10 year simulation (Figure 5b). 315 Volcanic halogen catalysed ozone depletion was simulated across all latitudes, but the largest magnitude changes in HAL10 (-40%) and HAL56 (-80%) were within the aerosol cloud and the polar regions, where the co-emitted halogens were activated on aerosol surfaces and PSCs respectively ( Figure 5). In both HAL10 and HAL56 tropical ozone was found to recover first with significant depletions recurring during the winter in the polar regions for 320 the remainder of the simulation. Ozone depletion shows a similar bimodal altitude distribution in the stratosphere similar to that found in Brenna et al., (2020), with 3 year mean depletion maxima (-1 ppmv and -3.5 ppmv in HAL10 and HAL 56) in the lower (20 km) and upper (40km) stratosphere ( Figure 6). Following sulfur-only eruptions we simulate small enhancements in stratospheric water vapour (SWV) and methane (Figure 8). SULF10 results in a peak increase in SWV of 0.4 ppmv (+7%) and a 10 ppbv (0.8%) increase in stratospheric methane 3-4 years after the eruption. SULF56 results in a peak increase in SWV of 1.1 ppmv 335 (+17%) and a 30 ppbv (2.5%) increase in stratospheric methane 3-4 years after the eruption, perturbations recover gradually over the remainder of the simulation. Unlike in sulfur-only eruptions, following eruptions with coemitted halogens we simulate a reduction in SWV and methane (Figure 8). HAL10 and HAL56 result in a peak SWV reduction of 1.0 ppmv (-16%) and 2.3 ppmv (-36%) respectively 3-4 years after the eruption followed by a gradual recovery. In HAL10 SWV perturbation levels return to the background levels over 6-7 years whereas in 340 HAL56 the perturbation does not fully recover within the 10 year duration of the simulation. HAL10 and HAL56 result in a peak mean stratospheric methane reduction of 37 ppbv (-3%) and 214 ppbv (-18%) respectively 2 years after the eruption. In HAL10 the stratospheric methane perturbation returns to the background levels over 3-4 years whereas in HAL56 the perturbation remains below zero for between 7-8 years. Stratospheric SWV and methane levels are linked. SWV has two main sources: transport from the troposphere and, chemical production from methane (Loffler et al., 2016). Whereas, as well as being oxidised by OH to form SWV, stratospheric methane is also destroyed by reaction with halogens via equation 2 and is sourced mainly 355 from transport from the troposphere. .
Sulfur-only eruptions cause an increase in the levels of stratospheric methane, in agreement with Loffler et al. 360 (2015), who showed stratospheric methane mixing ratios increased by ~5% following simulations of El Chichon and 15-20% following the larger Mt Pinatubo. They showed that major volcanic eruptions do not increase the upward transport of methane from the troposphere to the stratosphere, but rather, the increased stratospheric methane levels are due to lengthening of stratospheric methane lifetime. When we co-emit halogens, enhanced destruction of methane by chlorine via Eq 2 results in the significant decrease in the HAL10 and HAL56 365 stratospheric methane levels.
Using the whole atmosphere ozone radiative kernel from Rap et al., 2015, we are able to show that the stratospheric ozone change is the dominant driver of the ERFclear,clean accounting for ~75% of the ERFclear,clean (Figure 9 a,b). The 405 remainder is likely predominantly due to SWV changes with a small contribution from stratospheric methane changes. The latitudinal pattern of ozone radiative forcing reflects the locations of the ozone change, with largest forcings at the poles.
In both HAL10 and HAL56, ~25% of the additional peak global-mean volcanic ERF simulated compared to SULF10 and SULF56 respectively comes from the changes to the ERFari, with the remainder coming from changes to ERFclear,clean.

Figure 10
Evolution of the global-mean top of atmosphere total volcanic forcing ERF forcing (volcanic ERF) in SULF56 and HAL56 (a), SULF10 and HAL10 (b). Volcanic ERF is the sum of ERFari , ERFacii and ERFclear,clean.
Comparing the perturbations in HAL56 to HAL10, we find that increasing the volcanic halogen flux by 10 times 430 only results in a ~2.5 times larger global ozone response and, as ERFclear,clean is dominated by changes in stratospheric ozone, only a ~2 times larger ERFclear,clean. This suggests that there is a saturation in the ozone depleting potential of co-emitted volcanic halogens. Plotting the column ozone percentage change against the injected Equivalent Effective Stratospheric Chlorine (EESC) from this study and a number of previous studies, we find an exponential decay curve describes this relationship: as the EESC increases the efficiency of volcanic 435 halogen ozone depletion decreases ( Figure 11). This relationship suggests that column ozone is most sensitive to volcanic halogens when the additional EESC is < 20 Tg, and that increasing the volcanic EESC flux beyond 60 Tg has little impact on column ozone change. This analysis spans simulations with very different background EESC and column ozone values. Wade et al. (2020), Brenna et al. (2019), and Brenna et al. (2020 simulations are all in a pre-industrial atmosphere background states with low background chlorine levels, whereas, the 440 background chlorine levels in HAL10 and HAL56 are significantly higher and with lower initial ozone columns. This relationship suggests that the peak global-mean ozone loss (%) is dependent more on the volcanically injected EESC than the background chlorine and initial ozone columns. In other words, this relationship is timeindependent and this exponential decay curve can be used to estimate the peak global-mean ozone loss for an eruption in any climate state, including future eruptions where the background EESC will have decayed back to 445 pre-1980s levels. This will be especially useful for rapid estimates of ozone change as new or better constrained volcanic halogen data becomes available. https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. The implications of ozone depletion in HAL10 and HAL56 go further than enhancing the ERFclear,clean. High 455 anthropogenic fluxes of halocarbons into the atmosphere during the 1980s caused background chlorine levels to be elevated during the 1990s and an ozone hole is simulated to develop in the polar region every SH winter ( Figure   S4) of our control simulation. Using the definition for ozone hole conditions as <220 DU, we simulate enhanced ozone hole conditions following both HAL10 and HAL56 eruptions ( Figure 12). In HAL10, ozone hole conditions are simulated in the tropics for one year after the eruption, and a deepening of ozone hole conditions is seen in 460 northern hemisphere polar regions for two and in the southern hemisphere polar regions for four winters. In HAL56, we simulate ozone hole conditions globally for 5 years, continuing for a further three winters in the NH polar regions and six winters in the SH polar regions. https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. This shows that in the HAL56 scenario, on average 'Very High' or 'Extreme' UV levels would be expected all day for much of the globe in the three to four summers after the eruption, with noon values being even higher. 485 Living under such a high UV exposure would cause immediate immunosuppression, epidemic outbreaks, increases in the occurrences of eye damage and, in the longer term, skin cancers among the population living between the equator and the mid-latitudes, which equates to >95% of the global population. It is worth noting, that the assessment of surface UV changes is made more challenging by the presence of volcanic aerosols, which also scatter UV radiation. However, damaging UVB and UVC radiation will not be scattered effectively by larger 490 aerosol size distributions and volcanic aerosol levels reduce rapidly after peaking in the first post eruption year.
Whilst we have been able to calculate the composition and climate impacts of the co-emission of halogens and Although this work has focused on simulations of explosive volcanic eruptions in a background climate representative of the 1990s, Figure 11 demonstrates the simulated ozone depletion predominantly depends on the volcanic halogen injection size and not the background atmospheric state. Using the relationship outlined in Figure  505 11, we can estimate the peak global-mean ozone percentage loss for any size of volcanic halogen injection, past or present. We plan to explore this more and understand the impacts that plausible future background atmospheric states (such as different greenhouse gas concentrations, background halogen levels and stratospheric temperatures) may have on the simulated ozone response and volcanic ERF due to co-emitted sulfur and halogen volcanic emissions. 510 In addition to the co-emission of volcanic halogens, there is also scope to model the co-emission of volcanic water vapour and ash directly into the stratosphere. LeGrande et al., 2016, provided a mechanism explaining how SWV originating from volcanic eruptions may alter the chemistry of the stratosphere and the nucleation rate of sulfate aerosol, and suggested that this may severely alter the climate impacts. In addition, SVW proved to be an 515 amplifying feedback in simulations in this work and it would be interesting to see how co-emission of water vapour, halogens and sulfur would further alter the volcanic forcing in simulations of explosive volcanic eruptions.
Zhu et al., 2020, showed the importance of including volcanic ash injections in climate simulations. When heterogeneous chemistry on ash particles was included they found that 43% more volcanic sulfur was removed https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License. from the stratosphere in the first 2 months. Volcanic ash is also likely to alter the lifetime, activation and impact 520 of co-emitted volcanic halogens in climate simulations.

Conclusions
In this study we utilised UKESM-AMIP simulations of volcanic eruptions to investigate how the co-emission of volcanic halogens and sulfur alters the effective radiative forcing (ERF) of explosive volcanic eruptions under atmospheric conditions representative for the mid-1990s. As the volcanic flux of HCl and HBr into the 525 stratosphere remains uncertain, a range of plausible explosive volcanic emissions scenarios based on petrological degassing estimates, satellite observations and volcanic plume modelling were simulated. The four sets of experiments included one high SO2 (56 Tg), and one low SO2 (10 Tg) emission scenario, both with (HAL56 and HAL10) and without halogens (SULF56 and SULF10), each with an ensemble size of 6 sampling different QBO states. These eruption sizes (56 and 10 Tg SO2) are similar in size to a VEI 7 (e.g. 1257 Mt. Samalas) and VEI 6 530 (e.g. 1991 Mt. Pinatubo) eruption, representing 1 in 500 -1000 year and 1 in 50-100 year events respectively.
HAL56 utilises the 1257 Mt. Samalas HCl and HBr emission estimates from Vidal et al. (2014) and assumes a conservative ~5% stratospheric halogen injection efficiency. HAL10 has a SO2 injection similar to 1991 Pinatubo and a 10 times smaller injection of HCl and HBr than HAL56.

535
We have shown that the co-emission of halogens and sulfur in simulations of explosive volcanic eruptions increases the peak and cumulative volcanic ERF significantly. This is due to a combination of increased forcing from i) volcanic aerosol-radiation interactions (ERFari) and ii) composition of the stratosphere (ERFclear,clean).
Co-emitting halogens results in a larger global-mean ERFari in both HAL10 (+8%) and HAL56 (+6%). Ozone 540 depletion catalysed by volcanic halogens leads to stratospheric cooling (HAL10 ⋍ -2 K, HAL56 ⋍ -3.5 K) which more than offsets the volcanic aerosol heating (SULF10 ⋍ 1.5 K, SULF56 ⋍ 3.5 K). The ozone induced stratospheric cooling prevents aerosol self-lofting and keeps the volcanic aerosol lower in the stratosphere with a shorter lifetime, resulting in reduced growth via condensation and coagulation and smaller peak global-mean effective radius compared to sulfur-only simulations. The peak global-mean effective radii of the HAL10 and 545 HAL56 sulfate aerosols are found to be 15% and 10% smaller than SULF10 and SULF56 sulfate aerosol, closer to the most efficient radii for radiation scattering per unit mass, ~0.25 μm. Subsequently, we find HAL10 and HAL56 have higher peak global-mean SAOD anomalies (+11%, +22%) and ERFari (+8% + 6%).
Co-emission of halogens also results in significant perturbations to the stratospheric chemistry and compositional-550 driven forcing. Stratospheric methane was found to decrease by 3% and 18% and stratospheric water vapour (SWV) was found to reduce by 16% and 36% in HAL10 and HAL56 respectively. The methane reductions were driven by the enhanced destruction flux by volcanic Cl radicals and the SWV changes were attributed to the same stratospheric temperature reductions mentioned previously. Cooling in the tropical tropopause vicinity increased the efficiency of the tropical cold trap dehydration effect, reducing the flux of water vapour being transported 555 from the troposphere. The most dramatic change in chemistry was found to be in stratospheric ozone. Significant ozone depletions were simulated globally in both HAL10 (22%) and HAL56 (57%) with prolonged depletion in both NH and SH winter polar regions. In HAL10, ozone hole conditions (<220 DU) were simulated globally for https://doi.org/10.5194/acp-2020-1110 Preprint. Discussion started: 6 November 2020 c Author(s) 2020. CC BY 4.0 License.