The atmospheric component of the Community Atmosphere Model version 5 (CAM5) (Lamarque et al., 2012) is the atmospheric component of the Community Earth System Model, version 1 (CESM1.2.2, Hurrell et al., 2013), with a top at around 45 km. CAM5 has a horizontal resolution of 1.9° latitude × 2.5° longitude, utilizing the finite-volume dynamical core (Lin and Rood, 1996). The model has 56 vertical levels, with a vertical resolution ∼ 1 km in the upper troposphere and lower stratosphere. The modeled winds and temperatures were nudged to the 3 h Goddard Earth Observing System 5 (GEOS-5) reanalysis dataset (Molod et al., 2015) every time step (30 min) by 1 % (i.e., a 50 h Newtonian relaxation timescale). The aerosol is interactively simulated using a sectional aerosol microphysics model, the Community Aerosol and Radiation Model for Atmospheres (CARMA, Yu et al., 2015). The model uses the Model for Ozone and Related Chemical Tracers (MOZART) chemistry that is used for both tropospheric (Emmons et al., 2010) and stratospheric chemistry (English et al., 2011; Mills et al., 2016). The volcanic emissions from continuously degassing volcanoes use the emission inventory RCP8.5 and FINNv1.5. No volcanic eruptions except the Hunga 2022 eruption are included.

The initial volcanic injection altitude and area are determined by validating the water and aerosol transportation in months shown in Fig. 1 following the tests in Zhu et al. (2022), Wang et al. (2023), and Zhang et al. (2024). In these simulations, the H₂O is injected at 25 to 35 km altitude, and SO₂ is injected at 20 to 28 km altitude. The injection latitude ranges from 22 to 14° S, and longitude ranges from 182 to 186° E (Zhu et al., 2022). The initial injection of H₂O is 150 Tg, with ∼ 135 Tg left after the first week following the eruption.

The Center for Climate System Research/National Institute for Environmental Studies-Model for Interdisciplinary Research on Climate version 3.2 Chemistry Climate Model (CCSRNIES-MIROC3.2 CCM) (Akiyoshi et al., 2023) was developed based on version 3.2 of the MIROC atmospheric general circulation model (AGCM), incorporating a stratospheric chemistry module that was developed at the National Institute for Environmental Studies (NIES) and the University of Tokyo. The model has a horizontal resolution of T42 (2.8° latitude × 2.8° longitude) and 34 vertical levels, with a vertical resolution ∼ 1 km in the lower stratosphere–upper troposphere and ∼ 3 km in the upper stratosphere and mesosphere. The top level is located at 0.01 hPa (approximately 80 km).

The chemistry in the CCSRNIES-MIROC3.2 CCM is a stratospheric chemistry module including 42 photolysis reactions, 142 gas-phase chemical reactions, and 13 heterogeneous reactions for multiple aerosol types (Akiyoshi et al., 2023). Tropospheric chemistry is not included, but the stratospheric chemistry scheme is used for both the troposphere and the mesosphere.

In the CCSRNIES-MIROC3.2 CCM, only Exp2b can be performed. The atmospheric temperature and horizontal winds are nudged toward Modern-Era Retrospective analysis for Research and Applications Version 2 (MERRA-2) reanalysis (Gelaro et al., 2017) with a 1 d relaxation using instant values at 6 h interval (Akiyoshi et al., 2016). The HadISST data are used during the simulation.

The CCSRNIES-MIROC3.2 CCM does not have any microphysics scheme for volcanic aerosols. The surface area and spectral optical parameters of extinction, single scattering albedo, and asymmetric factor for Hunga aerosols were prescribed in the model from the GloSSAC version 2.22 aerosol data (Jörimann et al., 2025). H₂O was injected instantly on 15 January 2022 at the 12 grids of the model in the region 181.4–187.0° E in longitude, 14.0–22.3° S in latitude, and 12.0–27.6 hPa in pressure level. A uniform number density of 1.709×1015 molec. cm⁻³ H₂O was injected in each of the 12 grids, which amounts to ∼ 150 Tg.

The Canadian Middle Atmosphere Model (CMAM) is based on a vertically extended version of CanAM3.1, the third-generation Canadian Atmospheric Model (Scinocca et al., 2008). Compared to the standard configuration of CanAM3.1, for CMAM the model top was raised to 0.0006 hPa (approximately 95 km), and the parameterization of non-orographic gravity wave drag (Scinocca, 2003) and additional radiative processes important in the middle atmosphere (Fomichev et al., 2004) have been included. The gas-phase chemistry includes a comprehensive description of the inorganic O_x, NO_x, HO_x, ClO_x, and BrO_x families, along with CH₄, N₂O, six chlorine-containing halocarbons, CH₃Br and, to account for an additional 5 ppt of bromine from short-lived source gases, CH₂Br₂ and CHBr₃ (Jonsson et al., 2004). A prognostic description of, and associated heterogeneous chemical reactions on, water ice PSCs (PSC Type II) and liquid ternary solution (PSC Type Ib) particles is included, although gravitational settling (dehydration/denitrification) is not calculated, and species return to the gas phase when conditions no longer support the existence of PSC particles.

The simulations for the HTHH-MOC simulations were performed at T47 spectral resolution (approximately 3.8° resolution on the linear transform grid used for the model physics), with 80 vertical levels giving a vertical resolution of approximately 0.8 km at 100 hPa, increasing to 2.3 km above 0.1 hPa. The CMAM does not internally generate a QBO, so the zonal winds in the equatorial region were nudged towards a dataset based on observed variations up to December 2023, constructed using the method of Naujokat (1986) and extended into the future by repeating a historical period that is congruent with the observed QBO in late 2023. Water vapor from the Hunga eruption was added as a zonally averaged perturbation to the model water over 5 d from 00:00 UTC on 20 February 2022. The spatial distribution of the anomaly was designed to reproduce the water vapor anomaly observed in mid-February by the Atmospheric Chemistry Experiment – Fourier Transform Spectrometer (ACE-FTS) (Bernath et al., 2005) satellite (Patrick Sheese, personal communication, 2022), with a maximum value of 13.3 ppm at 17° S and 25.5 km and producing an anomaly of ∼ 150 Tg H₂O in the stratosphere.

The chemistry–climate model EMAC (ECHAM5/MESSy Atmospheric Chemistry) consists of the European Centre Hamburg general circulation model (ECHAM5) and the Modular Earth Submodel System (MESSy) (e.g., Jöckel et al., 2010). Here we use the version of Schallock et al. (2023) in horizontal resolution T63 (1.87° × 1.87°) with 90 levels between the surface and 0.01 hPa.

Vorticity, divergence, and temperatures between surface and 100 hPa are nudged to the ERA5 reanalysis of ECMWF (Hersbach et al., 2020), as well as surface pressure. SSTs and sea ice cover are prescribed by ERA5 data. The model can generate an internal QBO, but for comparison with observations it was slightly nudged to the Singapore data compiled by the Free University of Berlin and Karlsruhe Institute of Technology (Giorgetta et al., 2006).

The model contains gas-phase and heterogeneous chemistry on PSCs and interactive aerosols. Surface mixing ratios of chlorine- and bromine-containing halocarbons and other long-lived gases are nudged to Advanced Global Atmospheric Gases Experiment (AGAGE) observations. The microphysical modal aerosol module contains four soluble and three insoluble modes for sulfate, nitrate, dust, organic and black carbon, and aerosol water (Pringle et al., 2010). The instantaneous radiative effect by tropospheric and stratospheric aerosols can be calculated online by multiple calls of the radiation module. Volcanoes injecting material into the stratosphere are considered as in Schallock et al. (2023) using the perturbations of stratospheric SO₂ observed by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) and aerosol extinction observed by OSIRIS. This method, based typically on data of a 10 d period, distributes the injected SO₂ over a larger volume than typical point source approaches using the same integrated mass (see also Kohl et al., 2025). For Hunga this method has the disadvantage that H₂O and SO₂ are not co-injected since H₂O is injected in 12 h in a slab consisting of four horizontal boxes and a Gaussian vertical distribution centered at 21.5 hPa. For Exp2a we continue the 30-year transient simulation presented in Schallock et al. (2023) with and without Hunga Tonga. The simulated H₂O perturbation is consistent with Fig. 1. The SO₂ injection is derived based on the extinction from the OSIRIS observation averaged over about 10 d (Fig. 3) (Bruehl et al., 2023).

The NASA Goddard Earth Observing System Chemistry-Climate Model (GEOSCCM) is based on the GEOS Earth system model (Reinecker et al., 2008; Molod et al., 2015). For the HTHH-MOC experiments the model is run on a cubed-sphere horizontal grid at a C90 resolution (∼ 100 km) with 72 vertical hybrid-sigma levels from the surface to 0.01 hPa (∼ 80 km). Dynamics are solved using the finite-volume dynamical core (Putman and Lin, 2007). Deep convection and shallow convection are parameterized using the Grell and Freitas (2014) and Bretherton and Park (2009) schemes, respectively, and moist physics is from Bacmeister et al. (2006). The turbulence parameterization is based on the non-local scheme of Lock et al. (2000). Shortwave and longwave radiative fluxes are computed in 30 bands using the Rapid Radiative Transfer Model for GCMs (RRTMG, Iacono et al., 2008).

Stratospheric chemistry and tropospheric chemistry are from the Global Modeling Initiative (GMI) mechanism (Duncan et al., 2007; Strahan et al., 2007; Nielsen et al., 2017), updated here to include reactions for sulfur species. The GMI mechanism in GEOSCCM has been extensively evaluated for its stratospheric ozone-related photochemistry and transport in various model intercomparisons, including Stratosphere-troposphere Processes and their Role in Climate (SPARC) Chemistry Climate Model Validation (CCMVal), CCMVal-2, and the CCMI (SPARC, 2010; Eyring et al., 2010, 2013; Morgenstern et al., 2017). Aerosol species are simulated by the Goddard Chemistry, Aerosol, Radiation, and Transport, second-generation (GOCART-2G) module (Collow et al., 2024), which includes a sectional approach for dust (five bins), sea salt (five bins), and nitrate (three bins) and a bulk approach for sulfate (dimethyl sulfide, SO₂, methanesulfonic acid, and SO42-) aerosol and carbonaceous species (hydrophobic and hydrophilic modes of “white” and “brown” organics and black carbon).

For the GEOSCCM simulations performed with the GOCART-2G module we use the nominal GOCART-2G sulfate mechanism, updated here to use the online hydroxyl (OH) radical, nitrate (NO₃) radical, and hydrogen peroxide (H₂O₂) from the GMI mechanism instead of climatological fields provided from offline files (Collow et al., 2024). While not a full coupling to the GMI sulfur cycle it nevertheless allows the GOCART-2G sulfate mechanism to have the impact of the Hunga water vapor perturbation on the oxidants. A second “instance” of the GOCART-2G sulfate mechanism is run that is specifically for the volcanic SO₂ and resultant sulfate from the Hunga eruption. This allows us to track the eruptive volcanic aerosol separately from the nominal sulfate instance that sees mainly tropospheric sources. We assign this volcanic instance optical properties consistent with SAGE retrievals of the sulfate aerosol properties, using an effective radius of 0.4 µm. We find that 750 Tg of H₂O is needed in the initial injection to provide a residual ∼ 150 Tg of water in the stratosphere after a week. All other injection parameters follow the protocol. The model spin-up was performed by “replaying” to the MERRA-2 meteorology (Gelaro et al., 2017), and is used throughout the Exp2a results. A MERRA-2 2012–2021 climatology of SST and sea ice fractions are used based on Reynolds et al. (2002).

A second configuration of the GEOSCCM, coupled to the sectional aerosol microphysics package CARMA, also simulated the eruption (GEOS/CARMA). This configuration is the same as above except for the aerosol package and its coupling to the GMI chemistry mechanism. For this version of GEOSCCM, we use the configuration of CARMA described in Case et al. (2023). This configuration uses 24 size bins, spread logarithmically in volume between 0.25 nm and 6.7 µm in radius, and simulates the nucleation, condensational growth, evaporation, coagulation, and settling of sulfate aerosols in these simulations following the mechanism of English et al. (2013). For these simulations, CARMA is fully coupled to the GMI sulfur cycle by the production (i.e., oxidation of SO₂, evaporation of sulfate aerosols) and loss (i.e., nucleation and condensation of sulfate aerosols) of sulfuric acid (H₂SO₄) vapor. Optical properties for the CARMA aerosols are calculated based on the interactively calculated aerosol size distribution. The same injection parameters for GEOSCCM described above are used by this configuration. This model configuration contributed to Exp2a and “replayed” to MERRA-2 meteorology as above.

The NASA/Goddard Space Flight Center two-dimensional (2D) chemistry–climate model (GSFC2D) has a domain extending from the surface to ∼ 92 km (0.002 hPa). The model has 76 levels, with 1 km vertical resolution from the surface to the lower mesosphere (60 km) and 2 km resolution above (60–92 km). The horizontal resolution is 4° latitude, and the model uses a 2D (latitude–altitude) finite-volume dynamical core (Lin and Rood, 1996) for advective transport. The model has detailed stratospheric chemistry and reduced tropospheric chemistry, with a diurnal cycle computed for all constituents each day (Fleming et al., 2024). The model uses prescribed zonal mean surface temperature as a function of latitude and season based on a multi-year average of MERRA-2 data (Gelaro et al., 2017). Zonal mean latent heating, tropospheric water vapor, and cloud radiative properties as a function of latitude, altitude, and season are also prescribed (Fleming et al., 2020).

For the free-running simulations, the model planetary wave parameterization (Bacmeister et al., 1995; Fleming et al., 2024) uses lower boundary conditions (750 hPa, ∼ 2 km) of geopotential height amplitude and phase for zonal wave numbers 1–4. These are derived as a function of latitude and season using (1) a 30-year average (1991–2020) of MERRA-2 data for the standard yearly repeating climatological-dynamics simulations (“Clim-NoQBO”) and (2) individual years of MERRA-2 data (1980–2020), randomly rearranged in time to generate interannual variations in stratospheric dynamics (“ensemble1”, “ensemble2”, … “ensemble10”). For the inter-annually varying dynamics simulations, the model includes an internally generated QBO (Fleming et al., 2024).

For experiments that include the Hunga volcanic aerosols, the simulations go through the end of 2023, using prescribed aerosol properties for 2022–2023 both from the GloSSAC dataset and from the OMPS-LP data (Taha et al., 2021, 2022). For experiments that include the Hunga H₂O injection, Aura/MLS observations are used to derive a daily zonal mean Hunga water vapor anomaly in latitude–altitude, which is added to the baseline H₂O (no volcano) through the end of February 2022. This combined water vapor field is then fully model computed starting 1 March 2022 through the end of 2031.

For Exp2b, the model zonal mean temperature and transport fields are computed from the MERRA-2 reanalysis data. These are input into the model and used as prescribed fields (no nudging is done).

The Copernicus Atmosphere Monitoring Service (CAMS) provides daily global analysis and 5 d forecasts of atmospheric composition (aerosols, trace gases, and greenhouse gases (GHGs)) (Peuch et al., 2022). CAMS is coordinated by the European Centre for Medium Range Weather Forecasts (ECMWF) and uses, for its global component, the Integrated Forecasting System (IFS), with extensions to represent aerosols, trace, and GHGs, being called “IFS-COMPO” (also previously known as “C-IFS”, Flemming et al., 2015). IFS-COMPO is composed of IFS(AER) for aerosols, as described in Remy et al. (2022), while the atmospheric chemistry is based on the chemistry module as described in Williams et al. (2022) for the troposphere (IFS-CB05) and Huijnen et al. (2016) for the stratosphere (IFS-CBA). The stratospheric chemistry module of IFS-COMPO is derived from the Belgian Assimilation System for Chemical ObsErvations (BASCOE, Errera et al., 2019). IFS-COMPO stratospheric chemistry is used since the operational implementation of cycle 48R1 on 27 June 2023 (Eskes et al., 2024).

The aerosol component of IFS-COMPO is a bulk aerosol scheme for all species except sea salt aerosol and desert dust, for which a sectional approach is preferred, with three bins for each of these two species. Since the implementation of operational cycle 48R1 in June 2023, the prognostic species are sea salt, desert dust, organic matter (OM), black carbon (BC), sulfate, nitrate, ammonium, and secondary organic aerosols (SOAs).

For Exp2a, cycle 49R1 IFS-COMPO has been used, which will become operational for CAMS production in November 2024, at a resolution of TL511 (∼ 40 km grid cell) over 137 model levels from surface to 0.01 hPa. Cycle 49R1 IFS-COMPO integrates a number of updates of tropospheric and stratospheric aerosols and chemistry. The most relevant aspect for this work concerns the representation of stratospheric aerosols, which has been revisited with the implementation of a coupling to the stratospheric chemistry through a simplified stratospheric sulfur cycle including nucleation/condensation and evaporation processes, as shown in Fig. 4. Direct injection of water vapor into the stratosphere is expected to enhance the nucleation and condensation of sulfate through the reaction with SO₃ and production of gas-phase H₂SO₄.

The volcanic injection takes place between 03:00 and 06:00 UTC on 15 January 2022, with a uniform vertical distribution between 25 and 30 km of altitude, over a rectangular region of 400 km (latitude) × 200 km (longitude) centered on the coordinates of the Hunga volcano. The injected quantities are 0.5 Tg SO₂ and 190 Tg H₂O.

The Institut Pierre-Simon Laplace Climate Modelling Centre (IPSL CMC; see https://cmc.ipsl.fr, last access: 12 August 2025) has set up a new version of its climate model in the runup to CMIP6. Further description of the IPSL-CM6A-LR climate model can be found in Boucher et al. (2020) and in Lurton et al. (2020). New development of the model is now ongoing to prepare the IPSLCM7 version.

The IPSLCM7 climate model is using the general circulation model named LMDZ (Laboratoire de Météorologie Dynamique-Zoom; Hourdin et al., 2006). The LMDZ version used for this study is based on a regular horizontal grid with 144 points regularly spaced in longitude and 142 in latitude, corresponding to a resolution of 2.5° × 1.3°. The model has 79 vertical layers and extends up to 80 km, which makes it a “high-top” model. The model shows a self-generated quasi-biennial oscillation (QBO), whose period has been tuned to the observed one for the present-day climate (Boucher et al., 2020).

The aerosol is interactively simulated in the STRATAER module using a sectional scheme with 36 size bins. STRATAER is an improved version of the Sectional Stratospheric Sulfate Aerosol (S3A) module (Kleinschmitt et al., 2017). It now takes into account the photolytic conversion of H₂SO₄ into SO₂ in the upper stratosphere (Mills et al., 2005). The size-dependent composition of H2SO4/H2O aerosols is now computed iteratively to ensure that the surface tension, density, and composition are consistent in the calculation of the Kelvin effect. The surface tension, density, H₂SO₄ vapor pressure, and nucleation rates are calculated based on Vehkamäki et al. (2002). The version of the LMDZ6.2-LR-STRATAER atmospheric model used in the HTHH Impact project accounts for the stratospheric H₂O source from methane oxidation. The chemistry is simulated using the REPROBUS (REactive Processes Ruling the Ozone BUdget in the Stratosphere) chemistry module that includes 55 chemical species and a comprehensive description of the stratospheric chemistry (Marchand et al., 2012; Lefèvre et al., 1994, 1998).

For Exp2a, the H₂O and SO₂ are injected at 27.5 km altitude using a Gaussian distribution and standard deviation of 2.5 km. The injection latitude ranges from 22 to 14° S, and longitude ranges from 182 to 186° E. The injections of H₂O and SO₂ are 150 and 0.5 Tg, respectively. The SSTs are taken from the IPSL climate-coupled simulation run under the CMIP6 Tier 1 SSP2-4.5 scenario (O'Neill et al., 2016).

The Model for Interdisciplinary Research On Climate – CHemical Atmospheric general circulation model for Study of atmospheric Environment and Radiative forcing (MIROC-CHASER) version 6 (Sekiya et al., 2016) is a chemistry climate model, with a top at around 0.004 hPa. The present version of MIROC-CHASER is built on MIROC6 (Tatebe et al., 2019) and has a spectral horizontal resolution of T85 (1.4° latitude × 1.4° longitude). The model has 81 vertical levels, with a vertical resolution 0.7 km in the lower stratosphere, ∼ 1.2 km in the upper stratosphere, and ∼ 3 km in the lower mesosphere. In the free-running simulations, the model generates QBO internally. The ensemble members have different initial conditions (1 January 2022), which are generated using slightly different nudging relaxation time during the spin-up. The aerosols are interactively simulated using a three-mode modal aerosol module (Seikiya et al., 2016). The chemistry uses comprehensive troposphere–stratosphere chemistry (Watanabe et al., 2011). The volcanic emission from continuously degassing volcanoes uses the emission inventory of Fioletov et al. (2022). For the explosive volcanic eruptions during the spin-up time, explosive volcanic emissions follow Carn (2022).

For Exp1 fixed-SST simulations, the model uses the observed SST from 10-year climatological mean from 2012 to 2021 using the monthly-1deg CMIP6 AMIP SST (Gates et al., 1999).

For Exp2a, the atmospheric temperature and winds are nudged to MERRA-2 reanalysis with a 12 h relaxation using 3 h meteorology. The observed SST uses the NOAA 1/4° Daily Optimum Interpolation Sea Surface Temperature (OISST) from 2022 to 2023 (Huang et al., 2020).

The initial volcanic injection altitude and area are not tuned but follow the experimental protocol. For Exp1 and Exp2a, the H₂O and SO₂ are injected at 25 to 30 km altitude. The injection latitude ranges from 22 to 14° S, and longitude ranges from 182 to 186° E. The initial injection of H₂O is 186 Tg, with ∼ 150 Tg left after the first week following the eruption. The large initial H₂O injection is necessary to keep 150 Tg in the stratosphere as requested by the experimental protocol because a large amount of ice clouds generates and falls to the troposphere soon after the eruption.

The United Kingdom Earth System Model (UKESM, Sellar et al., 2019, 2020) is the successor to the HadGEM2-ES model (Collins et al., 2011), jointly developed by the UK Met Office and the Natural Environment Research Council (NERC) to deliver simulations to the Coupled Model Intercomparison Project Phase 6 (CMIP6; Eyring et al., 2016). For HTHH-MOC, we run the updated UKESM1.1 system (Mulcahy et al., 2023), which consists of the physical climate model HadGEM3-GC3.1 (Kuhlbrodt et al., 2018; Williams et al., 2018) and has improved tropospheric aerosol processes and aerosol radiative forcings (Mulcahy et al., 2018, 2020). The GC3.1 system comprises the GA7.1 global atmosphere model configuration (Walters et al., 2019), which uses the ENDGAME dynamics system (Wood et al., 2014), at a resolution of 1.875° longitude by 1.25° latitude, with 85 levels extending to 85 km. Specifically the simulations apply the UKESM1.1-AMIP academic community release job (at v12.1 of the Unified Model), as supported by the UK National Centre for Atmospheric Science.

The interactive atmospheric chemistry module UKCA (UK Chemistry and Aerosols) has a number of chemistry configurations, with UKESM1.0 for CMIP6 applying the combined stratosphere and troposphere chemistry (CheST) option (Archibald et al., 2020), essentially a combination of the stratosphere chemistry (Morgenstern et al., 2009) and tropospheric chemistry (O'Connor et al., 2014) UKCA schemes. The UKCA aerosol scheme is the GLOMAP-mode aerosol microphysics module (Mann et al., 2010, 2012; Bellouin et al., 2013), with UKESM1.0 including the initial set of adaptations to GLOMAP for simulating stratospheric aerosol (Dhomse et al., 2014). For all UKESM1.0 integrations for CMIP6, the system was applied with evaporation of sulfate aerosol de-activated, stratospheric aerosol properties enacted from the CMIP6-prescribed zonal mean dataset (Luo, 2017; Jörimann, 2025), but for the integrations here we have applied the system for interactive aerosol across the troposphere and stratosphere, enacting a Hunga emission of volcanic SO₂ following the 0.5 Tg at 25–30 km Tonga-MIP protocols (see Table 6).

For the improved UKESM1.1 version applied here, the other most relevant development, compared to UKESM1.0 used for CMIP6, is the interactive atmospheric chemistry module UKCA (UK Chemistry and Aerosols), which has the updates to heterogeneous chemistry added by Dennison et al. (2019), to represent reactions occurring on the surfaces of polar stratospheric clouds and sulfate aerosol more realistically, with modified uptake coefficients of the five existing reactions and the addition of a further eight reactions involving bromine species. For these simulations, for the first time we have added the equilibrium liquid PSC scheme of Carslaw et al. (1995) to UKESM; this scheme is an interim implementation here coupling the five existing heterogeneous reactions of chlorine activation here then occurring on solid and now also liquid ternary-aerosol PSCs.

For Exp2, UKESM1.1 is run in a specified dynamics configuration (Telford et al., 2008, 2009), the atmospheric temperature and winds nudged to ERA5 every 6 h, the Newton relaxation applied for levels 12 to 80 of 85 (between 1 and 60 km). Sea-surface temperatures and sea ice are prescribed from the Reynolds v2.1 datasets, during both the 2017 to 2022 spin-up period and the 2-year Exp2 period to December 2023. Monthly varying anthropogenic atmospheric chemistry and aerosol emissions were set following the CMIP6 SSP2-4.5 datasets.

The Whole Atmosphere Community Climate Model version 6 (WACCM6; Gettelman et al., 2019) is the high-top version of the atmospheric component of the Community Earth System Model, version 2 (CESM2), with a top at around 140 km. WACCM6 has a horizontal resolution of 0.9° latitude × 1.25° longitude, utilizing the finite-volume dynamical core (Lin and Rood, 1996). The model has 70 vertical levels, with a vertical resolution of ∼ 1 km in the lower stratosphere, ∼ 1.75 km in the upper stratosphere, and ∼ 3.5 km in the upper mesosphere and lower thermosphere (Garcia et al., 2017). In the free-running simulations, the model generates QBO internally (Mills et al., 2017; Gettelman et al., 2019). The ensemble members differ in the last date of nudging (from 27 January to 5 February 2022). The aerosol is interactively simulated using a four-mode modal aerosol module (MAM4; Liu et al., 2012, 2016; Mills et al., 2016), in which we used the Vehkamäki nucleation scheme (Vehkamäki et al., 2002). The chemistry uses comprehensive troposphere–stratosphere–mesosphere–lower thermosphere (TSMLT) chemistry (Gettelman et al., 2019). The volcanic emissions from continuously degassing volcanoes use the emission inventory of Andres and Kasgnoc (1998). For the explosive volcanic eruptions during the spin-up time, explosive volcanic emissions follow Mills et al. (2016) and Neely and Schmidt (2016), with updates until 2022.

For Exp1_CoupledOcean simulations, the ocean and sea ice are initialized on 3 January 2022 with output from a standalone ocean model forced by atmospheric state fields and fluxes from the Japanese 55-year Reanalysis (Tsujino et al., 2018). To accurately simulate the early plume structure and evolution, the winds and temperatures in WACCM are nudged toward the Analysis for Research and Applications, MERRA-2 meteorological data (Gelaro et al., 2017) throughout January 2022. After 1 February 2022, the model is free-running to capture fully coupled variability. For the fixed-SST simulation, the model uses the 10-year climatology SST from 2012 to 2021. The SST data are OISSTv2, which is a NOAA High-resolution (0.25 × 0.25) Blended Analysis of Daily SST and Ice (Banzon et al., 2022).

For Exp2, the atmospheric temperature and winds are nudged to MERRA-2 reanalysis with a 12 h relaxation using 3 h meteorology (Davis et al., 2022). The observed SST uses the 10-year climatological mean from 2012 to 2021.

The initial volcanic injection altitude and area are the same as described for Sect. 4.1 CAM5/CARMA.

WACCM6/CARMA only performed Exp2 and used a configuration similar to WACCM6/MAM4 with the same horizontal and vertical resolution, SSTs, and meteorological nudging. Differences compared to WACCM6/MAM4 are the chemistry and aerosol configuration used. WACCM6/CARMA used the middle atmosphere chemistry with limited chemistry in the troposphere and comprehensive chemistry in the stratosphere, mesosphere, and lower thermosphere (Davis et al., 2022). Furthermore, we use the Community Aerosol and Radiation Model for Atmospheres (CARMA, Tilmes et al., 2023, based on Yu et al., 2015 with some updates) as the aerosol module, in which we used the Vehkamäki nucleation scheme (Vehkamäki et al., 2002). CARMA defines 20 mass bins and tracks the dry mass of the particles and assumes particle water is in equilibrium with the environmental water vapor. The approximate radius ranges from 0.2 nm to 1.3 µm in radius for the pure sulfate group that sulfate homogeneous nucleation occurs in and ranges from 0.05 to 8.7 µm in the mixed group that tracks all major tropospheric aerosol types (i.e., black carbon, organic carbon, sea salt, dust, sulfate).

The initial volcanic injection altitude and area are determined by validating the water and aerosol transportation in the first 6 months against MLS and OMPS observations. In these simulations, the H₂O is injected to 25 to 35 km altitude following Zhu et al. (2022), while the SO₂ is injected 82 % of the total mass to 26.5–28 km and 18 % to 28–36 km altitude. The injection latitude ranges from 22 to 6° S, and longitude ranges from 182.5 to 202.5° E.