The implementation of VSL halocarbons emitted from the ocean in CESM has been described in Ordóñez et al. (2012). The methodology follows a top-down approach based on chlorophyll-a monthly climatology from the SeaWIFS project for the 1998–2003 period (Melin, 2013), and the iterative adjustment of emission flux strength by comparing modelled distributions and vertical profiles of VSL halocarbons with a compilation of aircraft campaigns and surface observations (see Tables 1 and 2 in Ordóñez et al., 2012). Nine halogenated VSL species are included in the Ordóñez et al. (2012) emission inventory, including individual emissions of two bromocarbons (CHBr₃ and CH₂Br₂), three bromo-chlorocarbons (CH₂BrCl, CHBrCl₂ and CHBr₂Cl), two iodocarbons (CH₃I and CH₂I₂), one iodo-chloro (CH₂ICl) and another iodo-bromo (CH₂IBr) carbon, respectively. All of these species are well known to be emitted from biologically active oceans as side products of the metabolite of different micro- and macro-algae and phytoplankton colonies (Carpenter and Liss, 2000; Carpenter et al., 2003). In the Ordóñez et al. (2012) inventory, the oceanic emission shows a seasonal and latitudinal distribution, with most of the emission (∼70 %) occurring in the tropics (20° N–20° S), approximately 25 % arising from the north and south mid-latitudes (20–50° on both hemispheres) and the remaining 5 % released from ice-free oceanic regions from the high-latitudes and polar regions. The net oceanic emission flux of each halocarbon species (Espec) is given by Eq. (1); 1Espec=1.127×105×fspec×rcoast×chla, where fspec is a species-dependent scaling factor iteratively adjusted to improve model-observation agreement, rcoast is a constant enhancement factor applied only to coastal areas outside the tropics, and chl_a is the SeaWIFS monthly climatology chlorophyll-a distribution (Ordóñez et al., 2012). In comparison to other VSLS approaches considering only latitudinal bands (Warwick et al., 2006a, b; Butler et al., 2007; Jones et al., 2010), the Ordóñez et al. (2012) inventory introduces a tropical geographically-heterogeneous and seasonally dependent variability, which allows for an improved spatio-temporal representation of VSLS distribution in the global troposphere. Once the monthly-mean prescribed fields are read, hourly dependent profiles are applied to the flux strength of all halocarbons to represent the photosynthetic dependence on radiation intensity. Therefore, VSLS emissions follow a Gaussian diurnal profile with peak emissions at local solar noon (Ordóñez et al., 2012), with exception of CH₂I₂ that follows a top-hat shape with a uniform distribution during the day and no-null emissions at night (see Fig. S1 in the Supplement). Given that no global long-term trend has been observed or established (WMO, 2022; Tinel et al., 2023), the emission strength of oceanic halocarbons is assumed to follow a constant climatology.

Due to improvements in the representation of large-scale ascent and transport across the boundary layer in CESM2 (Simpson et al., 2020) as well as in the emission strength and speciation of air pollutants altering the oxidative capacity of the troposphere (Emmons et al., 2020) the original Ordóñez et al. (2012) inventory for chlorinated and brominates species was increased by a constant factor of 1.15 (i.e., 15 %) to reproduce the stratospheric halogen loading obtained in previous CESM1 studies (Fernandez et al., 2014, 2021). The rationale for increasing the offline emission fluxes is that the reduction of transport across the boundary layer and the increase near-surface OH abundance in CESM2-SLH results in more efficient conversion from SGs to PGs at lower model levels. Given the much shorter photochemical lifetime of iodine species, no scaling was applied to VSL iodocarbons.

For anthropogenic VSL sources of chlorocarbons, we used the Hossaini et al. (2019) and Claxton et al. (2020) emission inventories, which include time-dependent emission of two major contributors to the organic chlorine load (CH₂Cl₂ and C₂Cl₄.) as well as surface LBCs for two other compounds (CHCl₃ and C₂H₄Cl₂) during the recent past. Beside a fraction of CHCl₃ emissions arise from natural sources (with minor contributions from the other compounds), chlorinated VSLS are dominated by anthropogenic emissions that display a pronounced hemispheric asymmetry (WMO, 2022). Therefore, hereafter we assume all of these offline emissions to have only an anthropogenic origin. Consistent with the scaling factor applied to the Ordóñez et al. (2012) inventory, both VSL chlorocarbon surface emissions and LBCs were globally scaled by the same constant factor (1.15 or 15 % enhancement). It should be noted that additional anthropogenic halocarbon sources, as those arising from the waste treatment of power-plants as well as seaweed aquaculture are not considered (Carpenter et al., 2000; Leedham et al., 2013; Jia et al., 2023).

In addition to the VSL halocarbon emissions, abiotic sources of inorganic halogens are implemented in the model. Here it is worth noting that these ocean-related sources have different routes for iodine compared to bromine and chlorine, and therefore are described separately. Inorganic iodine (HOI and I₂) emissions from the ocean surface are computed online following the ozone-driven oxidation of aqueous iodide occurring at the seawater surface (Carpenter et al., 2013). Following Prados-Roman et al. (2015b), the parameterized expressions for HOI and I₂ fluxes from MacDonald et al. (2014) are computed online by Eqs. (2a) and (b), respectively, 2aFHOI=O3×(4.15×105×Iaq-w-20.6w-23600×Iaq-),2bFI2=O3×Iaq-1.3×(1.74×109-(6.54×108×ln⁡(w))), where [O₃] and w are the surface ozone abundance (ppbv) and wind speed (m s⁻¹), respectively, computed over the ocean at the lowest model level. [Iaq-] is the aqueous iodide seawater concentration computed by Eq. (2c), which in turn depends on the model SST, 2cIaq-=1.46×106×exp⁡-9134SST. Equations (2a, b) are only valid for w≥3 m s⁻¹. Below this threshold, anomalously high iodine fluxes are obtained. To avoid this overprediction, w=3 m s⁻¹ is assumed for all grid cells where w<3 m s⁻¹. Not imposing this wind speed mask result in an overestimation of iodine fluxes from the ocean during calm periods (Inamdar et al., 2020). The first order dependence on ozone abundance of FHOI and FI2 (Eqs. 2a and b) has important consequences for the long-term trend of inorganic iodine emissions during the historical period (see Sect. 4).

For bromine and chlorine, additional abiotic inorganic sources arise from the so-called SSA-dehalogenation reactions. In this case, the online computation of inorganic halogen emissions assumes an oxidized gas-phase species to be deposited on the sea-salt aerosol surface, followed by an heterogeneous reaction that captures a reduced halide reservoir within the substrate that is oxidized before being released to the gas-phase (Fernandez et al., 2014). This complex redox process is parameterized to proceed in a single-step heterogeneous reaction dependent on the collisional frequency of the gas-phase species and the substrate, as well as on the degassing efficiency of the halogenated product released to the atmosphere (Ordóñez et al., 2012). Given that CESM2 does not include an explicit treatment of aqueous phase chemistry and diffusion, we assume the rate-limiting step is the uptake of inorganic halogen species onto SSA and computes the first-order reaction rate based on the Free-Regime Approximation (FRA) approach (McFiggans et al., 2000) following 3FSSAXY=14×γox×100×8×R×Tπ×MWox×SADSSA×Specox×DF×maskSSA. Here, FSSAXY is the flux of inorganic halogens (in the form of X2 or XY, with X,Y = Cl, Br or I) in molec. cm⁻³ s⁻¹, γox represents the accommodation coefficient of the gas-phase halogenated oxidized reservoir colliding with the SSA, MW_ox is the molecular weight (g mol⁻¹) of the oxidized species used to compute the mean-root square molecular speed, which in turn depends on the modelled temperature at each gridpoint (T in K) and the universal gas constant (R). Furthermore, the surface area density of SSA (SAD_SSA in cm⁻² cm⁻³) is computed considering the three SSA bins represented in the model, while [Spec_ox] is the atmospheric molecular concentration (molec. cm⁻³) of the gas-phase oxidized halogen species (XONO₂, XNO₂ and/or HOX, with X = Cl, Br and I) colliding with SSA. Finally, DF is a seasonal-dependent normalized depletion factor that represents the efficiency of the recycling reaction for any given atmospheric condition (Yang et al., 2005), while mask_SSA is a logical mask that limits the occurrence of the SSA-dehalogenation processes below a pressure threshold (P<300 hPa). Further details on the SSA-dehalogenation implementation in CESM can be found elsewhere (Ordóñez et al., 2012; Fernandez et al., 2014, 2021).

The FRA implementation in CESM2-SLH assumes that the bromide and chloride content of SSA is sufficiently large to act as an effectively infinite halide reservoir capable of sustaining the surface heterogeneous-redox reaction. This implies that the bromide and/or chloride content in the SSA bulk is always in excess compared to the abundance of the gas-phase halogen reservoir that deposits on the aerosol surface, which is valid for fresh SSA typically found close to the ocean surface. Therefore, the implementation of SSA-dehalogenation does not account for the slowing down of the halogen recycling when aged sea-salt aerosols become depleted in bromide and chloride (von Glasow et al., 2002). Table 1 summarizes all 9 independent reactions releasing either gas phase X2 or XY, where the non-stoichiometric coefficients between reactants and products for each gas-phase halogen species determines the net inorganic halogen flux. Given the large increase in SSA abundance and vertical distribution between CESM1 and CESM2 (see Fig. S2), particularly over the boundary layer where the latter show up to 10 times larger values than the former, the accommodation coefficients (γox) for halogenated reservoirs compiled in Table 4 of the Supplement of Ordóñez et al. (2012) was reduced by a factor of 3 for chlorine and iodine and by a factor of 8 for bromine (see Table 1). This scaling factors are within the reported uncertainty and variability of γox for the different halogen families and aerosol compositions (Burkholder et al., 2020) and allowed to reproduce previously estimated SSA-dehalogenation fluxes and burdens (see Sect. 4.2).

For chlorine, additional inorganic halogen sources arise from the heterogeneous recycling of nitrogen species (N₂O₅ and HNO₃) on halide-rich SSA (see Eq. 3 and Table 1). In the first case, the SSA uptake and recycling of N₂O₅ drives two key processes: (i) the well-documented hydrolysis to produce HNO₃ (het_ss_11, Lamarque et al., 2012); and (ii) the production of nitryl chloride (ClNO₂), which constitutes a net chlorine source (het_ss_12). The latter is of importance as ClNO₂ photolysis in polluted environments lead to ozone production (Knipping and Dabdub, 2003; Thornton et al., 2010). Within CESM2-SLH, the ClNO₂ yield is computed online based on a cubic expression dependent on the total sea-salt aerosol mass within each of the sea-salt aerosol bins (Li et al., 2022). In the second case (het_ss_9), the acid-displacement reaction results in the oxidation of the available chloride in the sea-salt bulk and depends on a fixed HNO₃ accommodation coefficient (Sander, 2015), following other models including chlorine chemistry (Hossaini et al., 2016). The implementation of both chlorine sources follows the FRA-approach described above for oxidized halogen reservoirs (Eq. 3), although in this case the uptake of oxidized nitrogen species drives the SSA-dehalogenation. We note that stoichiometric N₂O₅ recycling on halide-poor tropospheric aerosols is also considered in the model, although for these reactions the halogen uptake and recycling only constitute a partitioning shift between different halogen and nitrogen species (see Table 2).

Finally, we highlight the importance of adjusting the strong model dependence of online SLH sources and burdens depending on the sea-salt abundances and distributions represented in different model configurations (see Fig. S2). Note that the accommodation coefficients (γ) and depletion factors (DF) altering the reactive efficiency for the different species reported in the literature (Sander, 2015; Burkholder et al., 2020) possess a wide range of values and large uncertainties that depend on variables that are unconstrained in the simplified parameterization implemented in CESM2-SLH (e.g., pH dependence, halide content, etc.). Consequently, variations in SSA fields, pollutant emissions, and atmospheric composition (e.g., HO_x/NO_x partitioning), combined with non-linear heterogeneous recycling (Table 1), may cause regional over-/under-estimations of halogen content in localized regions in comparison with previous CESM1 studies (see Sect. 5).

A comprehensive list of all photochemical halogen reactions included in the SLH scheme is shown in Table S1, based on the original publication compiled in Ordóñez et al. (2012) and Saiz-Lopez et al. (2014), with further updates from Badia et al. (2021) and Saiz-Lopez et al. (2023). This includes 11, 10 and 17 new photolysis reactions for chlorine, bromine and iodine, respectively, which adds up to previously defined photolysis reactions of long-lived halocarbons as well as odd-chlorine and odd-bromine reactions that are required for properly representing stratospheric ozone depletion in CESM2 (Gettelman et al., 2019b). The absorption cross sections and quantum yields are taken from the latest JPL 19-5 (Jet Propulsion Laboratory; Burkholder et al., 2020) and IUPAC (International Union of Pure and Applied Chemistry; Atkinson et al., 2007, 2008) handbooks, and are included within the xs_long and xs_short file typically used in CAM-Chem and WACCM. Wavelength integrated J-values are computed as a function of temperature and height considering the model actinic flux and a Look-Up-Table (LUT) approach (Kinnison et al., 2007; Lamarque et al., 2012; Emmons et al., 2020).

Note that the photodegradation of anthropogenic VSLS (CHCl₃, CH₂Cl₂ and C₂Cl₄) results in the formation of phosgene (COCl₂), which constitutes a halogen intermediate that accumulates in the upper troposphere until it photo-decomposes in the lower stratosphere (see Fig. S3) (Hossaini et al., 2016). Therefore, phosgene does not participate in direct ozone destruction in the lower-stratosphere until it releases the Cl atoms through photolysis. However, and given that the only source of phosgene in the model is the degradation of chlorinated VSLS, we consider the contribution of COCl₂ within the PGs chlorine fraction, in line with the latest WMO (2022) report. In addition, we follow the JPL 19-5 recommendation (Burkholder et al., 2020) of neglecting the C₂Cl₄ absorption cross-section beyond λ>270 nm (Keller-Rudek et al., 2013), as not doing so resulted in a complete photodecomposition in the lower troposphere which is not in line with observations (Roozitalab et al., 2024). For iodine, Table S1 considers the photodissociation of higher order oxides (I_xO_y, with x=2 and y=2,3,4). The absorption cross sections of these species were derived from solution spectrum measured at the University of Leeds (Lewis et al., 2020), which were subsequently scaled to estimate absolute absorption cross-sections (Gómez Martín et al., 2005). Therefore, and supported by observationally-constrained modelling studies demonstrating that considering I_xO_y photolysis was required to reproduce gas-phase (Saiz-Lopez et al., 2014, 2015) and particulate iodine (Koenig et al., 2020) vertical profiles, the released CESM2-SLH chemical mechanism is based on the JIxOy scheme for iodine proposed in Saiz-Lopez et al. (2014).

A complete description of all bimolecular Arrhenius type reactions of halogen species are shown in Table S2, where for the sake of simplicity we have ordered reactions in the following groups: Odd-Oxygen, Odd-Chlorine, Odd-Bromine, Odd-Iodine, VSLS degradation and reactions with sulphur and carbon compounds. All these developments have been initially described in Ordóñez et al. (2012), with additional updates for iodine described in Saiz-Lopez et al. (2014). In addition, all termolecular reactions considered in the default SLH chemical mechanism are shown in Table S3. Note that for the final implementation of SLH chemistry in CESM2, all expressions, reaction-rate coefficients and temperature dependence factors have been updated to the reported values in the last JPL 19-5 handbook (Burkholder et al., 2020).

Oxidation of DMS to produce sulphur dioxide (SO₂) by BrO, Cl and IO are also included in the updated SLH chemical scheme. These improvements in sulphur chemistry representation in CESM2 with respect to CESM1, including tropospheric sulphate formation and washout, are a major source of uncertainty for radiative forcing estimations (Ge et al., 2022) and are currently under development. Therefore, the current CESM2-SLH mechanism has been developed starting from the benchmark sulphur scheme implemented in TS1.2 (Emmons et al., 2020). Similarly, the Modal Aerosol Model with 4 modes (MAM4) scheme is the default aerosol option in CESM2, which includes updates in the formation and growth of sulphate, black carbon and organic matter, secondary organic aerosols, sea salt, and dust (Liu et al., 2016). Although no direct SLH changes to the MAM4 scheme were implemented, we note that indirect perturbations of reactive halogens on OH abundance can influence the global aerosol burden (Saiz-Lopez et al., 2023). In the following section, we describe the stoichiometric heterogeneous processes involving inorganic halogens species.

Heterogeneous SLH reactions occurring on different types of atmospheric substrates have also been implemented (see Table 2). It is worth noting that unlike the SSA-dehalogenation source described in Sect. 2.1.3, all of these reactions are stoichiometric and therefore do not represent a net halogen source from the aerosol to the gas-phase, but instead, result in a change in individual species partitioning between gaseous reactants and products. Indeed, in all cases, the substrate surface (e.g., ice-crystals, cloud droplets and/or other anthropogenic aerosols typically considered in CAM6 like black-carbon, organic carbon, sulphate, nitrate, etc.) acts like a catalyst and does not contribute with any halogen content. For heterogeneous reactions occurring over ice-crystals, we use the FRA approach to compute the reaction rate constant (rate_ice) considering the total number of gas-phase species that suffer uptake on the aerosol surface, resulting in the following expressions for uni-molecular 4arateiceuni=14×γX×100×8×R×Tπ×MWX×SADice×maskicestrat, and bimolecular 4brateicebi=14×γX×100×8×R×Tπ×MWX×SADice×maskicestrat×1[Y], reactions (see Table 2). Here, γx and MW_X are, respectively, the accommodation coefficient and the molecular weight of the gas-phase halogenated reservoir species (X) that is initially taken up by the ice-crystal, SAD_ice represents the surface area density of ice crystals and is computed online based on the ice-water content in clouds (CLDICE) from the CAM6 model, and maskicestrat is a logical mask imposed to limit the computation below the model tropopause and avoid double-counting recycling reactions historically implemented to occur in the stratosphere (Kinnison et al., 2007; Fernandez et al., 2014). For bimolecular reactions, note that the reaction rate is normalized by the atmospheric concentration of the most abundant halogenated species (either X or Y) involved in the reaction. The original implementation of these reactions has been described in detail in Fernandez et al. (2014) for bromine and chlorine and in Saiz-Lopez et al. (2015) for the case of iodine.

The heterogeneous recycling reaction of N₂O₅ shown in Table 2 (usr_N2O5_aer2) differs from the net source of chlorine occurring on sea-salt aerosols (het_ss_12 in Table 1). Indeed, tagged reaction usr_N2O5_aer2 occur on other tropospheric aerosols (e.g., black-carbon, organic-carbon, sulphate and/or nitrate) over continental domains, considering accommodation coefficients and production yields from McDuffie et al. (2018, 2019). The resulting ClNO₂ can further photolyze to release NO₂ and consequently result in additional ozone production within polluted environments (i.e., high-NO_x regimes). Given that main aerosol components within MAM4 do not represent a halide reservoir, reaction usr_N2O5_aer2 assumes that the gas phase HCl is initially condensed at the substrate surface (Osthoff et al., 2008), which further react to produce ClNO₂ that is released back to the gas phase (reactive uptake). This stoichiometric recycling of nitrogen oxides is based on Li et al. (2022) following the original implementation of chlorine activation on non-SSA substrates described in Hossaini et al. (2016).

Implementation of stratospheric heterogeneous reactions of inorganic halogen reservoirs arising from the photodecomposition of long-lived halogens in WACCM are important for representing the ozone layer (Marsh et al., 2013). This includes temperature and water-vapour dependent re-activation reactions of chlorinated and brominated reservoirs occurring on stratospheric sulphate (SAD_sulfc), ice-crystals (SAD_ice) and nitric acid trihydrate (SAD_NAT) surfaces (Kinnison et al., 2007), including recent updates of individual stratospheric accommodation coefficients (γstrat; Solomon et al., 2015). Based on Cuevas et al. (2022), in CESM2-SLH we extended WACCM developments to both low-top (32L) and high-top (70L) vertical configurations (see Sect. 3.1) by mapping equivalent heterogeneous reactions in the stratosphere for the minor halogenated reservoirs (mainly HOCl and HOBr), as well as the previously neglected iodine species (see Table 3). Here, we assumed identical accommodation coefficients and reaction yields for iodine as those used for the equivalent chlorinated and brominated reservoirs. This assumption represents a lower-limit of iodine impacts, as previous studies suggest that the reactive efficiencies of iodine species are larger than those for bromine and chlorine (Solomon et al., 1994; Koenig et al., 2020). Based on the original implementation in CESM1 (Fernandez et al., 2017; Saiz-Lopez et al., 2023), for low-top configurations (see Table 6) we imposed prescribed stratospheric sulphate aerosols above the tropopause (Mills et al., 2016), which results in smaller SAD_sulfc fields compared to those computed with MAM4 (see Fig. S4). Note that equivalent logical conditions based on maskicestrat have also been applied to all equivalent tropospheric reactions occurring over tropospheric ice-crystals to avoid double counting.

Several studies highlight the importance of properly distinguishing between the contribution of Source Gas Injection (SGI) and Product Gas Injection (PGI) to the stratosphere when representing the contribution of SLH to stratospheric ozone depletion (Fernandez et al., 2014, 2021; Salawitch et al., 2005; WMO, 2018, 2022). This is mainly because the ozone destruction efficiency of SLH in the lowermost stratosphere depends on the net fraction of the emitted VSLS that has already been converted to the reactive inorganic form (Fernandez et al., 2021). Figure 4 shows the tropical mean (20° N–20° S) vertical profiles of all VSL source gases (SG_X) along with the associated increment of inorganic product gases (PG_X). Since the simulations neglecting SLH have non zero Cl_y and Br_y abundance, the PG_Cl and PG_Br vertical profiles were computed as ΔCl_y and ΔBr_y, respectively (i.e., the difference between the CC-SLH and CC-NOH experiments). Therefore, PG_Cl and PG_Br exclusively account for the additional contribution of SLH to the total inorganic halogen loading at any given height, without considering the small (but not negligible) contribution from long-lived chlorine and bromine photodecomposition close to the tropopause (see Fig. 10a and c). In addition, Fig. 4 shows the sum of all organic and inorganic species for each halogen family (SG_X + PG_X).

Stratospheric injection is computed at approximately 85 hPa (17 km) within the tropics, just above the mean model tropopause. At this level, the simulated contribution of SGI and PGI to the total bromine injection reaches 1.8–2.2 pptv + 2.5–3.0 = 4.3–5.2 pptv during present-day (Fig. 4b). Notably, all experiments simulate SGs vertical profiles that fall within the accepted range reported in the last Ozone Assessment Report (WMO, 2022) throughout the tropical troposphere, regardless of the compset and resolution considered. Indeed, our simulations show a continuous conversion of SGs to PGs just below the tropopause that results in a larger fraction of inorganic bromine injection (PGI) relative to the organic halocarbon fraction (SGI). Once all organic VSL_Br has been converted into reactive Br_y, the total contribution of SLH to the stratospheric bromine loading remains nearly constant at ∼5 pptv roughly from the tropopause to the model top (grey line in Fig. 4b). Here we note that the dominant contribution of VSLS to the total SGI arises from the major CH₂Br₂ and CHBr₃ emissions, with only minor contributions from the remaining VSL bromocarbons (Fernandez et al., 2014). In contrast, CH₂IBr emissions, which constitute the largest source of VSL bromine mass to the atmosphere (Ordóñez et al., 2012), barely contributes with a negligible fraction to the stratospheric SGI due to its particularly short lifetime. Unlike bromine, the larger lifetime of chlorinated VSLS imply that the total chlorine injection (114–126 pptv) is dominated by SGI (91–97 pptv), with a smaller contribution from PGI (17–35 pptv), in agreement with reported ranges (WMO, 2022). For the case of iodine, much faster photochemistry compared to chlorine and bromine implies that almost the complete injection occurs through PGI = I_y (Fig. 4c), with a total injection of approximately 0.65–0.72 pptv, in agreement with observations and previous estimations using CESM1 (Saiz-Lopez et al., 2015; Koenig et al., 2020).

Figure 5 shows the temporal evolution of SGI_X and PGI_X occurring at the model tropical tropopause for chlorine, bromine and iodine for the CC-SLH [2×2-ndg] and CC-SLH [2×2-hst] experiments, where the modelled values are compared with the latest assessed values of SLH injection to the stratosphere (i.e., see Tables 1–6 in WMO, 2022). The modelled contribution of SLH to PGI for all three halogens does not show a significant trend and lies within the WMO estimates across the simulated period. For the particular case of bromine and chlorine, CC-SLH [2×2-ndg] shows slightly larger values than the CC-SLH [2×2-hst], coherent with the different representation of tropical convective transport and Brewer-Dobson circulation between nudged and free-running experiments. Modelled SGI_Cl shows a growing trend across 2000–2020 in agreement with previous reports (Hossaini et al., 2019) approaching the assessed 2019 atmospheric abundance by the end of the simulated period (WMO, 2022). Here we recall that in CESM2-SLH we applied the same 15 % enhancement to natural oceanic bromine sources as well as to anthropogenic chlorine emissions and LBCs. In quantitative terms, SLH contributes to a significant enhancement on present-day stratospheric bromine loading of approximately 25 %–26 % compared to other long-lived halogen (LLH) sources, while for chlorine, the contribution of SLH represents only an increase of 3.7 %–3.8 %, in agreement with the latest estimation from WMO (2022). The iodine SGI contribution is almost zero and all the iodine injection to the stratosphere occurs as PGI (Fig. 5c).

Additional model validation was performed extracting hourly surface reactive halogens mixing ratios for the CC-SLH [2×2-ndg] experiment at the same location and time as previously published observations obtained in multiple measurement campaigns. Three ship-based campaigns measured IO concentrations: TransBrom, conducted in October 2009 from Japan to Australia (Großmann et al., 2013), Poseidon, which surveyed the coasts of Northwest Africa between the Canary Islands and Cape Verde in June 2010 (Bange, 2011) and Malaspina, conducted between January and July 2011, which circumnavigated the globe from Brazil to Spain via Australia and the Panama Canal (Prados-Roman et al., 2015a). Additionally, we compared measurements from a coastal station in Cape Verde between April and June 2007 (Read et al., 2008; Mahajan et al., 2010). The ship routes and the location of Cape Verde are shown in the central panel of Fig. 6. The side panels compare daily mean modelled (red) and observed (blue) values, where the shading and error bars represents the modelled spread and the reported observational uncertainty, respectively. A general good agreement is achieved, especially for IO where observed values ranging between 0.5 and 2 pptv with a sharp variability that can exceed 1 pptv between consecutive days is captured in magnitude by the model, although not always for the exact time and location. The maximum values predicted in Malaspina during late May and early June (Fig. 6d) are attributed to highly localized IO bursts due to the large partitioning shift between I_y species driven by SSA recycling (Prados-Roman et al., 2015a). Similarly, near-zero BrO values are modelled in Cape Verde (Fig. 6a) due to the negligible contribution of SSA-debromification predicted by the model during early April. However, during late May the model is capable of reproducing the magnitude of sporadic BrO peaks due to the improved representation of SSA abundance over the east-Atlantic. Despite the remaining model-observation discrepancies, we highlight that CESM2-SLH results in a general underestimation of observed reactive halogen abundance, which implies that the predicted influence of SLH on atmospheric composition is, in any case, a lower limit.

Figure 7 shows the zonal average latitudinal-height distribution of total organic long-lived (Fig. 7a and b) and short-lived (Fig. 7c–e) halogens for the CC-SLH [2×2-ndg] experiment during present-day. While long-lived halogenated ODSs remain unreactive throughout the troposphere and are converted to inorganic Cl_y and Br_y only after crossing the tropopause, VSL halocarbons photolyze at much lower heights in the atmosphere. Consequently, iodine VSL is nearly completely decomposed in the lower troposphere (Fig. 7e), whereas only a small fraction of the emitted VSL bromine and chlorine SGs is transported unaltered to the stratosphere (Fig. 7c and d). In contrast to the naturally emitted bromine and iodine VSLS, the modelled VSL chlorine abundances show a remarkable inter-hemispheric difference with larger values over the NH, due to its dominant anthropogenic sources (Fig. 7c).

In addition, Fig. 8 shows the corresponding zonal average distribution of inorganic halogens (Cl_y, Br_y and I_y) from the model surface up to the middle stratosphere for CC-SLH [2×2-ndg] (Fig. 8a–c) and CC-NOH [2×2-ndg] (Fig. 8d and e), as well as the difference between both experiments (Fig. 8f and g). Note that CC-SLH [2×2-ndg] does not consider any long-lived iodine source (Fig. 7) while CC-NOH [2×2-ndg] does not include iodine chemistry at all (Fig. 8). Notably, not only the total Xy abundance at the model top, but also the conversion from organic SGs to inorganic PGs occurs at lower heights for CC-SLH [2×2-ndg] (Fig. 8a and b) than for CC-NOH [2×2-ndg] (Fig. 8d and e), following the different ODS and VSLS photodecomposition shown in Fig. 7. Given the typically longer photochemical lifetimes of long-lived chlorinated compounds compared to brominated halons (see Table A1 in the Annex of WMO, 2022), the latter presents a steeper vertical gradient in the lower stratosphere and an almost complete conversion to inorganic bromine in the upper stratosphere (Fig. 9b).

The change in inorganic halogen abundance Xy due to SLH (Fig. 8f and g) highlights the different behaviour for each individual halogen family in different regions of the atmosphere. For example, within the troposphere, both Cl_y and Br_y show a pronounced enhancement in the MBL and close to the surface mostly in the NH mid-latitude regions that is rapidly reduced in the free-troposphere, which is driven by the efficient SSA-dehalogenation source and acid displacement that presents a sharp vertical profile (see green lines in Fig. 10a and c). For the case of iodine, the largest abundances at the surface are driven by the strong HOI/I₂ emission occurring in the tropical regions (Fig. 1f). In this region, up to 1 pptv of I_y is transported to the tropical free troposphere where, due the local changes in ozone abundance and temperature, a pronounced partitioning shift lead to the formation of the tropical rings of atomic halogens (Saiz-Lopez and Fernandez, 2016). A fraction of this tropospheric inorganic iodine is washed out, and the remaining fraction is injected to the stratosphere as inorganic PGs through the tropical tropopause region with an almost negligible contribution from SGs (see Fig. 4). In contrast, Fig. 8f shows a gradual enhancement in Cl_y occurs in the middle stratosphere, which arises due to the longer photochemical lifetime of chlorinated VSLS which are mostly transported unaltered to the lower stratosphere (Fig. 4a). For bromine, Figs. 7d and 8g highlight that both organic SGs and inorganic Br_y PGs contribute to the total stratospheric bromine loading due to SLH.

Figure 9 summarizes the global mean (90° N–90° S) vertical profiles of the sum of all organic and inorganic halogen species from the Earth's surface to the middle stratosphere for the CC-SLH [2×2-ndg] and CC-SLH [2×2-hst] experiments. The figure distinguishes the contributions from long-lived chlorine and bromine species, as well as the total halogen fraction transported as VSL SGs and inorganic PGs, consistent with Figs. 7 and 8. The conversion of unreactive organic SGs to reactive PGs is found to be very similar between the two meteorological representations (free-running vs. nudged dynamics) among model compsets. We note that these global and annual mean averages lumped together continental and oceanic domains within the boundary layer, as well as tropospheric and stratospheric parcels due to the latitudinal variation of the tropopause (see dashed horizontal lines in Fig. 9). However, three main differences characterizing the vertical distribution for each halogen family are distinguished. First, the tropospheric chlorine abundance is dominated by long-lived species, with minor contributions from VSL chlorine and Cl_y. Due to the long lifetimes of CFCs and CH₃Cl, the crossing point between LLH_Cl and Cl_y representing the photochemical conversion from the organic to the inorganic fractions occurs in the middle stratosphere (∼ 40 hPa, see Fig. 9a). Second, the VSL bromine contribution is comparable (although smaller) to the contribution from long-lived halons and CH₃Br, and therefore the initial enhancement in Br_y abundance, as well as the crossing point with LLH_Br, occurs at lower heights in the stratosphere (∼ 75 hPa, Fig. 9b). Third, given that there are no long-lived iodine species and the short photochemical lifetimes of VSL_I (particularly CH₃I), both the tropospheric and stratospheric abundance is dominated by inorganic I_y. This is in line with the tropospheric and stratospheric mean burdens shown in the stacked bars of Fig. 3g–i.

Given that the atmospheric impact of SLH over different atmospheric components depend on the overall abundance of reactive halogens, whose initial step is the release of a halogen atom through Reactions R1a and b, left panels in Fig. 10 shows the Cl, Br and I atom release from the different organic and inorganic sources as a function of height. Here note that only emissions or processes that lead to net inorganic halogen production are considered. In addition, right panels in Fig. 10 show the chemical partitioning between all halogen species constituting each inorganic halogen family (Xy) which interconvert between one and the other following gas-phase and heterogeneous phase reactions involving other atmospheric components such as O₃, OH and NO₂ (see summarized Reactions R1–R13 scheme in Sect. 1 and Tables S1–S3).

As summarized in Table 8, SSA-dehalogenation is the dominant source of bromine and the second largest source for chlorine. However, Fig. 10a and c highlights that these sources are primarily confined to the lower troposphere where most of the washout takes place. For chlorine, the contribution of the acid-displacement HCl release dominates, increasing surface Cl_y particularly in the NH mid-latitudes and coastal locations (Fig. 11d). The photochemical degradation of both VSL_Cl and VSL_Br dominates the release of Br and Cl atoms throughout the free troposphere, although the biogenic ocean flux is the sole source of bromine, while for the case of chlorine the contribution from anthropogenic VSL_Cl is at least two orders of magnitude larger than the natural oceanic source. In addition, the release of Cl atoms from long-lived CH₃Cl accounts for an equivalent source as that arising from anthropogenic VSL_Cl (Fig. 10a). In contrast, bromine release from long-lived CH₃Br only surpass that arising from VSL_Br in the stratosphere (Fig. 10c). For iodine, ocean HOI/I₂ release dominates the surface emissions, while the photochemical degradation of CH₃I alone controls the organic to inorganic iodine conversion throughout the troposphere (Fig. 10e).

Figure 10b shows that the dominant inorganic chlorine species throughout the troposphere is HCl, which represents between 63 % and 85 % of the total Cl_y partitioning. The Cl atom and ClO partitioning are generally 4 and 1 to 2 orders of magnitude smaller than the global mean Cl_y, respectively, although due to rapid heterogeneous recycling their contributions on the regional scale can increase, particularly in coastal locations (see Fig. 11). For bromine (Fig. 10d), the dominant species in the lower troposphere is HBr both during day and night (Fernandez et al., 2014). However, during reactive transport to the upper troposphere, HBr is heterogeneously converted on ice-crystals first to HOBr and finally to the dominant BrONO₂ fraction in the stratosphere (see Tables 2 and 4). The vertical profile of iodine partitioning in Fig. 10f shows that HOI is the dominant iodine species throughout the troposphere and therefore its washout controls the transport and abundance of inorganic iodine from their dominant surface sources to the stratosphere. It should be noted that the chemical partitioning from reservoir to reactive halogen species is the largest for iodine, followed by bromine and last by chlorine (i.e., IO_x/I_y > BrO_x/Br_y > ClO_x/Cl_y). This is associated with the much faster photochemistry of iodine compared to bromine and, to a larger extent, chlorine, which in turn controls the efficiency of each halogen family in altering the composition and oxidative capacity of the troposphere (see Sect. 4.3.2).

Figure 11 shows the spatial distribution of carbon-bonded VSLS (Fig. 11a–c), total inorganic halogens (Fig. 11d–f) and the most relevant reactive fraction (Fig. 11g–i) of each halogen family averaged over model levels with P≥850 hPa and for the time period 2015–2020. Note that depending on the halogen family considered, the most relevant reactive species for each family are shown: i.e., Cl atom for chlorine, BrO for bromine and the sum of I + IO = IO_x for iodine.

Driven by the prevalence of anthropogenic sources in the NH, VSL_Cl exhibits a pronounced South–North latitudinal gradient between ∼60 and ∼260 pptv (Fig. 11a), with peak values exceeding 500 pptv over China. In contrast, VSL_Br shows a more homogeneous spatial distribution with maximum abundances observed over the equatorial regions due to the dominant source from the tropical oceans (Ordóñez et al., 2012). Minimum, maximum and mean global VSL_Br mixing ratios reach 2.9, 12.9 and 6.8 pptv, respectively. The global mean VSL_I is ∼1.0–1.2 pptv (see Table 8). Inorganic chlorine also maximizes over the NH, particularly over coastal locations where the mixing of halogen-rich SSA and HNO₃/N₂O₅ plumes results in an enhanced chlorine source. The highest Cl atom concentrations occur over the coastal regions of Europe, US and East Asia, with maximum values smaller than 0.4×10-3 pptv and a global mean average of 4×10-5 pptv. These distributions are consistent with maximum and mean Cl_y levels of ∼640 pptv and ∼47 pptv, respectively (Fig. 11d). Similarly, modelled Br_y also maximizes over coastal regions co-located with high SSA abundance (Fig. S2). However, the inter-hemispheric bromine enhancement is less pronounced than for chlorine because SSA-debromination is not controlled by HNO₃ and N₂O₅, but instead by shifts in Br_y partitioning between BrONO₂, BrNO₂ and HOBr (see Table 1). Maximum Br_y peaks in the NH are up to 2 times larger than in the Southern Hemisphere (SH). However, the corresponding maximum BrO abundances does not present such inter-hemispheric asymmetry (Fig. 11h), highlighting the strong dependence of reactive bromine partitioning on the background atmospheric composition of NO_x and HO_x. For iodine, global I_y distributions is primarily controlled by oceanic HOI/I₂ emissions that peak within the tropics. The mean modelled I_y and I + IO surface abundances reach ∼1.25 pptv and ∼ 0.15 pptv, respectively, with corresponding volume mixing ratios spanning between (0.05–21) pptv for I_y and (0.008–0.90) pptv for I + IO. Notably, the global mean IO_x / I_y ratio is ∼0.12, whereas in regions with the highest I_y, abundances it remains below ∼0.04.

Figure 12 presents the seasonal variation of surface organic and inorganic halogen species averaged within different latitudinal bands, distinguishing between the tropics (20° N–20° S) and the extratropical domains in the NH (20–90° N) and the SH (20–90° S). The stacked bars shown in Fig. 12 represent the total halogen abundance decomposed into the reactive (XO_x) and reservoir species, as well as organic VSL_X species for each experiment. The corresponding zonal average seasonality of VSL_X halocarbons, as well as the geographical Xy (reactive plus reservoir) seasonal cycle over the boundary layer are shown in Figs. S5 and S6, respectively. We first describe the seasonality in the NH extratropics and then highlight the main similarities and differences with the SH extratropics and the tropical mean.

The NH extratropical Cl_y presents a pronounced seasonal cycle with approximately doubled values during the boreal winter (∼90–100 pptv) compared to the summertime (∼45–50 pptv). This is coherent with the seasonal variation of the heterogeneous recycling efficiency that reduces the conversion of both chlorinated (e.g., ClONO₂) and nitrogen (e.g., N₂O₅ and HNO₃) reservoirs to more reactive species. For the same reason, the organic VSL_Cl and VSL_Br abundances also peak during the winter when the photochemical decomposition is smallest (see Fig. S5). In contrast, the iodine seasonal cycle shows maximum abundances during the summertime, both for I_y and VSL_I. This is explained by the dominant contribution of oceanic sources for iodine, which increase associated with the larger fraction of ice-free ocean in summer. The inorganic Br_y abundance is lower and maximizes during the spring for CC-SLH [2×2-ndg] in comparison with CC-SLH [2×2-hst], which shows larger values and a maximum peak during the summer, due to the larger efficiency of SSA-dehalogenation recycling in CC-SLH [2×2-hst] (see Table 8). Here we recall that sea-ice polar halogen emissions are not yet considered in CESM2-SLH, and therefore, the modelled seasonal cycle observed for the different halogen species differs from those described in the NH high- and mid-latitudes as well as near Antarctica in previous works (Fernandez et al., 2019, 2024).

The predicted seasonal cycle of iodine over the SH extratropical regions presents a similar pattern to that in the NH, although with a 6-months shift, mostly for VSL_I. However, and due to the much lower influence of anthropogenic ozone pollution over the SH, the I_y abundance remains less variable throughout the entire year. In contrast, VSL_Br show larger abundances during the austral winter due to the reduced photodecomposition. The same seasonal pattern is also observed for VSL_Cl, although we note that total mixing ratio are remarkably lower in the SH compared to the NH extratropics. Larger Br_y abundance is predicted during austral spring and summer following the seasonal changes in sea-salt distributions particularly over the Southern Ocean, resulting in higher Br_y enhancement for CC-SLH [2×2-hst] than for CC-SLH [2×2-ndg] (see Fig. 11h). Is worth noting that over the SH extratropical latitudes, the Cl_y seasonality follows that of Br_y, which is explained by the fact that within this region, the SSA-dehalogenation flux for chlorine (which is linked to that for bromine, see Table 1) dominates over the acid-displacement reaction controlling the total chlorine flux over the NH.

Finally, the tropical VSL_I abundance is the lowest compared to the NH and SH extratropics due to the higher photochemical decomposition close to the Equator. In contrast, the I_y abundance in the tropics is larger than in the NH and SH extratropics due to the dominant tropical HOI/I₂ source (Fig. 1f). On the other side, surface VSL_Cl and VSL_Br tropical surface abundances lack of a pronounced seasonal cycle, although a regional seasonality is predicted in SGI_Br due to the changes in convective transport to the tropical tropopause layer (Fernandez et al., 2014, 2021). The double peak annual cycle in reservoir Xy observed within the tropics is attributed the seasonal changes in atmospheric rainout during the wet and dry seasons that affect the washout of inorganic halogens.

To evaluate the performance of CESM2-SLH, surface ozone abundance for the CC-SLH [2×2-ndg] and CC-NOH [2×2-ndg] experiments was compared with observations reported in the first phase of Tropospheric Ozone Assessment Report (TOAR-I; https://igacproject.org/activities/TOAR/TOAR-I, last access: 4 May 2026) (Schultz et al., 2017). The central panel in Fig. 15 shows the geographical distribution of annual mean surface ozone for CC-SLH [2×2-ndg] during year 2015, along with 12 surrounding panels showing the seasonal comparison with observational data at coastal or close-to-coastal locations (red points in central panel). The model reproduces the hemispheric asymmetry in surface ozone, with background levels below 40 ppbv in the pristine SH, contrasting with peak concentrations (60–75 ppbv) over continental and coastal Asia. Moreover, the ozone peak on the African continent is mainly linked to biomass burning. These maxima values are consistent with those reported by previous studies using CESM2 (Emmons et al., 2020). Moreover, Fig. 15 also shows a visual comparison of the monthly (lines) and annual mean (markers) surface ozone concentrations between the CC-SLH [2×2-ndg] (blue) and CC-NOH [2×2-ndg] (red) simulations and TOAR-I observations (black) across multiple stations located between extra-polar latitudes (60° N–60° S). Overall, the model captures the observed seasonal ozone cycles across most stations. This agreement is particularly strong at northern mid-latitude stations (>30° N), where the model reproduces both the magnitude and seasonality of ozone, with peaks in spring-summer driven by enhanced photochemical production. Similarly, good representation of ozone seasonality is observed at station (h) located in east Asia as well as in station (l) south of New Zeeland. Most notably, the inclusion of SLH chemistry results in an overall reduction of ozone that reduces the model bias and in most locations is larger than the spread of modelled values between the CC-SLH and CC-NOH experiments. However, discrepancies emerge at some stations like i) in central Chile, where the model successfully reproduces the observed ozone magnitude but fails to capture the seasonal cycle. This divergence likely stems from incomplete representation of ozone sources and sinks in this pristine region, as well as due to the coarse resolution used. Indeed, the coarse resolution of the model is well-known factor affecting the comparison with in-situ observations, as the local ozone values in some observational sites do not necessarily represent the mean background levels predicted over a 1.9°×2.5° pixel size. Nevertheless, Fig. 15 clearly shows that the inclusion of SLH chemistry in CESM2 helps to close the gap between the model and observations.

This section highlights the effect of SLH on ozone destruction by means of computing the odd-oxygen chemical loss (OddOx_Loss) as performed in previous studies, following the definitions presented in Table 5 of Saiz-Lopez et al. (2014) (see Appendix for further details). Figure 16 shows the annual mean OddOx_Loss as a function of latitude and height for CC-SLH [2×2-ndg], where the individual contribution of the different families contributing to ozone depletion are distinguished. Going from top to bottom, in the upper stratosphere, OddOx_Loss is dominated by NO_x cycles (Fig. 16c), while from the middle and lower stratosphere down to the free-troposphere, the dominant OddOx_Loss family is HO_x (Fig. 16b) until O_x becomes the dominant ozone loss channel over the boundary layer (Fig. 16a). Note the halogen-driven OddOx_Loss shown in Fig. 16d–f maximize in the tropical upper troposphere (dominated by IO_x) and global lower stratosphere (both BrO_x and IO_x increase, with a minor role of ClO_x), where they contribute to additional ozone destruction. Most of this halogen-driven ozone loss results from: (i) faster recycling and gas-phase iodine chemistry compared to bromine and chlorine; and (ii) the larger conversion of iodine to reactive product gases occurring in the lower troposphere. In addition, note the important contribution of BrO_x and ClO_x cycles on ozone depletion over the Antarctic lower stratosphere of the southern hemisphere high latitudes (Fig. 16d and e), which maximizes during spring and is responsible for the formation of the Antarctic ozone hole (see Fig. 18).

Figure 17 shows the mean vertical distribution of the dominant OddOx_Loss channels averaged globally (Fig. 17a–c) and within the tropical regions (Fig. 17d–f). To highlight the influence of SLH on tropospheric chemistry, we quantify in absolute terms the contribution from each family for the CC-SLH [2×2-ndg] and CC-NOH [2×2-ndg] experiments, as well as the differences between them in the right-most panel. The dominant ozone losses in the troposphere for CC-NOH [2×2-ndg] (Fig. 17a) are due to direct O_x photolysis, followed by HO_x cycles, both of which are significantly reduced when SLH are included, particularly in the lower and free troposphere (note NO_x only contribute significantly to OddOx_Loss in the middle and upper stratosphere). Indeed, for the CC-SLH [2×2-ndg] experiment (Fig. 17b), the total contribution of halogens increases significantly and represents between 10 % and 30 % globally (Fig. 17c), with a variable vertical profile that is dominated by IO_x cycles in the troposphere, and an increasing contribution mostly from BrO_x close to the tropopause. This increase in halogen driven ozone loss is compensated by a proportional decrease in the OddOx_Loss destruction by HO_x and O_x in comparison with CC-NOH [2×2-ndg] (negative change in Fig. 17c and f). Most notably, the largest halogen-driven contribution to total ozone loss is due to iodine chemistry, which represents up to 30 % within the tropical free troposphere in agreement with previous estimates (Saiz-Lopez et al., 2012, 2014), followed by bromine and with a minor contribution by chlorine.

Table 10 summarized the annual mean odd-oxygen loss channels, as well as the net chemical ozone production and net chemical change for the different simulations considered. Globally, the halogen driven tropospheric photochemical ozone destruction reaches 659.4 Tg yr⁻¹ for CC-SLH [2×2-ndg] and 662 Tg yr⁻¹ for CC-SLH [2×2-hst], dominated by IO_x-induced OddOx_Loss (accounting for 76 %–79 %) and with minor contributions for bromine (16 %–18 %) and chlorine (5 %). These values show excellent agreement with previously reported halogen influences obtained with CESM1 (Badia et al., 2021). Table 10 also shows that the tropospheric ozone photochemical production and the gross chemical loss remains almost unaltered (with minor changes below ∼3 %) regardless of the consideration or not of SLH sources and chemistry, consistent with previous studies (Barrera et al., 2023). In contrast, and driven mainly by the larger ozone destruction when SLH are considered, the estimated O₃ dry deposition is reduced by approximately 16 % for CC-SLH [2×2-ndg] and CC-SLH [2×2-hst] compared to their corresponding CC-NOH [2×2-ndg] and CC-NOH [2×2-hst] experiments, which is in line with the 11 % reduction previously estimated in CESM1. The net stratospheric to tropospheric exchange (STE) for CC-SLH [2×2-ndg] and CC-SLH [2×2-hst] reaches 164 and 261 Tg yr⁻¹, respectively. Even though in absolute terms the contribution of STE to the tropospheric ozone budget in CESM2-SLH is smaller than for CESM1 (e.g., STE accounts for 341 Tg yr⁻¹ for CESM1-SLH [SL23]), the inclusion of SLH compared to NOH results in an equivalent halogen-driven reduction in the contribution of STE of approximately 20 %, particularly for the nudging setup (Barrera et al., 2023). Differences between model versions are attributed to dynamical changes in the representation of air subsidence and downward ozone transport across the tropopause between the old and new nudging approaches (Davis et al., 2022) and will be evaluated in future studies.

Table 11 shows the most important chemical pathways contributing to the formation of OH, where the individual contributions have been separated between the dominant primary production (P), which is the production of O¹D by ozone photolysis at hν<330 nm followed by reaction with tropospheric water vapour; and secondary production (S), which accounts for all the remaining processes involving VOCs, NO_x and O_x (Lelieveld et al., 2016), as well as the reactions of hydrogen peroxide and halogenated hypohalous acids (Bossolasco et al., 2025). Due to the larger O₃ burden in CESM2-SLH compared with CESM1, we note that the changes in OH production are also expected to differ between the different model versions.

The tropospheric ozone depletion induced by SLH result in a reduction of P from approximately 116 Tg yr⁻¹ for the CC-NOH [2×2-hst] and CC-NOH [2×2-ndg] sensitivity simulations (which represent approximately 47 %–48 % of the total OH source) to approximately 101 Tg yr⁻¹ (44 %–45 %) for the main CC-SLH [2×2-hst] and CC-SLH [2×2-ndg] experiments. Similarly, the secondary ozone-driven pathway (O₃ + HO₂) reduces from approximately 12 % to 10 % due to the inclusion of SLH, inducing minor alterations in the other secondary channels that remain below a few percent (see Table 11). These reductions in P and S are compensated by ∼5 % increase in the contribution from the photolysis of HOCl, HOBr and HOI, which accounts for more than 11 Tg yr⁻¹ of the global annual production. Consequently, there is a continuous shift in the chemical OH production from primary to secondary sources, which implies that once formed, OH has an increased capability of being chemically re-generated in the atmosphere when SLH are considered. Therefore, SLH results in a net reduction of OH abundance while the tropospheric oxidation capacity of OH on a per atom basis increases. This is typically measured by computing the OH recycling probability, r=1-P/(P+S) (Lelieveld et al., 2016), which represents the capability of OH to be regenerated from secondary sources. Most notably, considering SLH in CESM2 increases r from approximately 52 %–53 % to 55 %–56 % (see Table 11).

This section presents the combined influence of SLH chlorine, bromine and iodine on stratospheric ozone in CESM2-SLH. Following previous studies (Saiz-Lopez et al., 2015; Fernandez et al., 2017; Cuevas et al., 2022; Barrera et al., 2023; Villamayor et al., 2023), we distinguish the effects observed over the global lower stratosphere from those affecting the Antarctic ozone hole. Figure 18a shows the annual mean change in O₃ associated with SLH, defined as the difference in O₃ partial column densities (DU) as a function of latitude and height between CC-SLH [2×2-ndg] and CC-NOH [2×2-ndg] (ΔO₃). This diagnostic highlights that the SLH influence on stratospheric ozone maximizes in the lowermost stratosphere, just above the tropopause (Fig. 18a), which is of major importance as the radiative perturbations in this region are highly sensitive to small changes in ozone abundance (Riese et al., 2012; Saiz-Lopez et al., 2012, 2023). Most notably, the SLH influence integrated in the lower stratosphere (from the tropopause to ∼20 hPa) presents a pronounced latitudinal variation, with ozone reductions increasing from the Equator and tropical regions to the mid-to-high latitudes (Fig. 18b and c). This is due to the larger conversion from SGI to PGI as air is transported from the tropical regions towards the poles through Brewer-Dobson circulation, combined with the lower temperatures prevailing at high latitudes that accelerate halogen chemical cycles. Here we note that SLH influence in lower stratospheric ozone is dominated by bromine and iodine chemistry (see Figs. 16 and 17), although the net effect is particularly sensitive to the background inorganic chlorine (Cl_y) levels in the stratosphere (Barrera et al., 2020; Villamayor et al., 2023). Therefore, the total SLH impact is expected to vary over time following the overall stratospheric chlorine loading before and after the peak in ODS observed in the stratosphere following the Montreal Protocol (WMO, 2018, 2022).

Figure 18b and c shows the net SLH-induced O₃ difference (ΔO₃) integrated over the lowermost stratosphere, where most of the SLH influence occurs. The resulting ΔO₃ for the nudged configuration ranges from -2.3 to 3.3 DU within the tropics (20° N–20° S) up to -9.5 to 13.8 DU for the high-latitudes (poleward of 60°). These ozone changes represent, respectively, -1.8 % to 2.7 % and -4.0 % to 8.4 % for the annual mean, depending on latitude. Current impacts are larger than those reported in previous CESM1 studies (Barrera et al., 2020; Fernandez et al., 2021), as those analyses only considered SLH bromine, while the CESM2-SLH simulations assessed here additionally include SLH iodine and anthropogenic VSL chlorine. Finally, Fig. 18d shows that stratospheric ozone destruction maximize during austral spring over the Antarctic polar vortex, where SLH bromine and iodine enhance the ozone hole deficit by more than -20 DU and -14 % during October, in agreement with previous reports based on CESM1 (Fernandez et al., 2017; Cuevas et al., 2022).