Everything in the paper seems to be correct, but it is not an exciting paper.  The paper does not have interesting new scientific discoveries.  Rather, it reads like a long technical report.  The descriptions of how each variable responds are overly comprehensive.  This would make a nice introductory paper for a special issue where the results of the experiment they describe are carried out by multiple models.  The results are based on only one, albeit excellent, model, but I wonder how model-dependent the results are.  And I wonder how forcing created with WACCM will interact with other models with different physics, chemistry, and resolution.  By the way, the paper does not describe the resolution of the model simulations.

There are small edits for the authors to consider in the attached annotated manuscript.

The main goal of the paper is to present the CCMI-2022 simulation setup senD2-sai as an alternative to running a fully coupled model with Stratospheric Aerosol Injection (SAI), and to justify its validity by showing WACCM6 senD2-sai output and comparing it to a fully coupled simulation (SSP2-4.5-SAI), which is also used to generate the aerosol input. The main conclusion is that the setup is valid to isolate the direct effects of SAI on stratospheric and tropospheric climate that are not mediated by changes in the ocean. However, the differences with the fully coupled simulation highlight that the setup is not valid to study surface climate, as the response in surface temperature and precipitation is very different. The stratospheric response is isolated in senD2-sai as a strengthened polar vortex in both hemispheres, which impact surface climate by inducing a positive NAM and SAM. The ocean-mediated response is El Niño-type including its known teleconnections affecting wave activity in the troposphere and stratosphere. Therefore, the conclusion is that the setup is useful to study the effects of SAI on the stratospheric dynamics and chemistry in a homogenized framework, but that doing this has non-negligible limitations due to the missing ocean feedbacks.

The manuscript includes noteworthy and novel insights on the separation between the stratospheric only effects and the effects including the ocean feedbacks. The analyses are robust, the figures clear. However, the large number of simulations and having to remember their technical description and slight differences somewhat difficults following the paper and obscures the main points. I have some questions and suggestions to improve the readabilty and facilitate interpretation of the results. I also include some specific comments on the text below.

General comments:

For instance, could the number of simulations used be reduced? For the reader it is diffucult to remember the differences between all the simulations, so I suggest keeping a smaller number of simulations in the main results description, and discuss the complementary runs in a separate discussion section. For example, I understand the interest of comparing with refD2 from CCMI-2022, but can be in a separate section after the main results are presented.

If I understand correctly, the only difference between SSP2-4.5 and refD2 is the implementation of eruptive sulful injections and the evolution of ODS. In this sense, the cleanest comparison would have been to create a simulation exactly equal to refD2 but including the SAI. However, this is not available and that is why SSP2-4.5 and SSP2-4.5-SAI have been used. Is this correct? Please explain the reasoning for having both simulations in the manuscript in Section 2.2.

Also, Figures 3, 5 and 6 show the differences between SSP2-4.5 SAI in 2080 – 2099 and the control experiment (SSP2-4.5) in 2020 – 2030. Is it really necessary to use control simulations, SSP2-4.5 and senD2-fix, for the present period? Could you compare the SAI simulations in the later period to the earlier period of the same simulation, when SAI should not have an impact yet? This is typically done in climate change studies and would simplify the interpretation.

The senD2-sai setup is not coupled to the ocean and thus the SST and sea ice need to be prescribed. The proposed setup is to prescribe climatological SST and sea ice from the present. However, approach largely dampens the climate variability in these simulations (as seen in Fig. 1), and this is important because the type of responses that are analyzed are modes of variability (NAO, SAM, ENSO). Is there an alternative way to construct these simulations? For example, prescribe the SST from a coupled model SSP-4.5-SAI simulation, similar to what is done for refD2? This would likely improve the total model response to SAI. Is this not proposed to avoid imposing an ocean response which is suspect of being model dependent? Has this been discussed in previous literature? This discussion should be included in the paper.

Since an important motivation for the study is the need to increase the number of models that perform SAI experiements, it would be good to add the information of how many CCMI-2022 models do not have a coupled ocean currently.

Specific comments:

L47: Keeble et al. 2021 looks at CMIP6 models.

L53: ‘stratospheric aerosol distribution or optical properties’. What does this ‘or’ mean? Is it an option or are both things equivalent?

L80: ‘SSP2-4.5 and refD2’ . The name SSP2-4.5 is a bit confusing, since refD2 follows that same IPCC scenario. Could this be called perhaps Ref_coupled, and the other one SAI_coupled or something similar?

L119: Is it correct that you changed the interhemispheric temperature gradient condition by an interhemispheric symmetry aerosol distribution condition? It could be stated more clearly.

L150-151: the globally averaged precipitation change is half as large in the senD2-sai compared the coupled simulation, and this is not clear from the text, which only states ‘the feedback from the atmosphere and land changes in the globally averaged precipitation in senD2-sai is similar to that in SSP2-4.5 SAI’.

L187-189: might this explain the strange change in values in Figure 3 top row central and right panels, at 50N/S near the tropopause?

L243: correct typos ‘stratosphre’ and ‘coupld’

L253: ‘Compared to the present day and independent of SAI’ does this mean in scenarios that do not implement SAI? It could be stated more clearly.

L275: Here and at other places the difference in wave driving between senD2-sai and SSP2-4.5SAI is mentioned, but it is not shown. Could you include a figure of the changes in Eliassen-Palm flux divergence, at least in the supplementary material?

L285: Here an El Niño like response is mentioned but this has not been shown yet in the paper, so please include a mention that it will be shown later.

L293-294: ‘the discussed differences between the prescribed and interactive experiment are similar’ The phase ‘differences are similar’ sounds a bit strange, I suggest rephrasing to ‘changes are similar’.

L320-321: ‘The symmetric injetion in both hemispheres using SAI has been shown to counter the weakening of the AMOC due to climate change’. This seems at odds with the fact that a strong cooling is observed in the North Atlantic in Figure 9 central panels (SSP2-4.5).

L352: ‘dominated the response in the region’. Suggested: ‘in the Pacific region’

L363-367: ‘This austral summer negative SAM and weakening of eddy-driven jet are therefore likely connected to the equatorial Pacific and the El-Nino-like response resulting in changes in tropospheric winds and Kelvin wave fluxes into the stratosphere to contribute to the winter and spring stratospheric easterly response.’ I do not understand this sentence. First, it is unclear how Kelvin waves propagating in the stratosphere influence the SAM. L’Heureux and Thompson (2006) do not mention Kelvin waves. Second, the beginning of the sentence refers to the austral summer SAM response, but the end of the sentence refers to the ‘winter and spring response’.

L455: modeling groups (add ‘s’)

General comments

This paper deals with the Stratospheric Aerosol Intervention, a geoengineering technique proposed to mitigate global warming. In this article, a new simplified Stratospheric Aerosol Intervention (SAI) experiment (senD2-sai) is presented as an alternative to run a fully coupled model (SSP2-4.5-sai) for assessing the SAI impacts on both the Troposphere and Stratosphere in the context of the Chemistry-Climate Modeling Initiative (CCMI-2022). In order to prove the overall equivalence of senD2-sai and SSP2-4.5-sai, a detailed comparison between the simulations is presented. It is argued that the results regarding the SAI effects on the stratosphere are overall similar and that the senD2-sai configuration can be used to evaluate the impacts of SAI on stratospheric chemistry and dynamics for multi-model comparison studies. On the other hand, for the tropospheric and surface climate response, the two setups show very different near-surface temperatures and precipitation patterns. Therefore, senD2-sai is not suitable for investigating near-surface climate. Still, results concerning the SAI-induced changes at the surface are discussed as well. For example, the senD2-sai experiment allows the authors to attribute part of the cooling in the North Atlantic near Greenland in SSP2-4.5-sai simulation to the strengthening of the polar vortex due to the SAI impacts on the Stratosphere in the Noth Hemisphere, which can propagate down to the surface, resulting in a positive phase of the NAO.

Overall the study can be a valuable reference for future studies about the effect of SAI in the stratosphere using a simpler model setup that in turn can attract more modeling groups, which do not have the possibility to run a fully coupled model, to participate in the CCMI. Therefore, I consider it worth publication.  However, there are several caveats that prevent the publication at this stage, which I list below as major comments. Further, I have multiple questions and suggestions aiming to improve the clarity of the paper. Moreover, some technical comments are also given in chronological order.

Major comments:

1) The manuscript is not acceptable in its current form as a research article, because it does not report any substantial new results. Also, the current title and abstract look rather like a technical report. Technical note would be a more suitable format and further changes to the manuscript should be made with this goal. However, note that GMD is the recommended journal for description of numerical experiments.

2) The main message of the manuscript is not clear (if not the technical note). What I am particularly missing is a formulation of some scientific questions in the introduction that would be answered in the conclusions based on the results presented. Right now, the conclusion section is a very incoherent mixture of summary and discussion of technical aspects.

3) I am doubtful about the proposed experimental set-up of the senD2-sai experiment because the prescribed aerosol distributions from WACCM will be non-native for other models (different tropopause heights, tropical widths, BDC structure..). It would make more sense if the prescribed distributions were derived individually for each model.

4) At many places in the manuscript, linearity and hence separability of the responses is often implicitly assumed, leading to statements as "...this experimental setup allows us to isolate the stratospheric (top-down) impact of SAI on the troposphere and removes any potential (bottom-up) influences from an uncertain ocean response.". It has to be acknowledged that in reality there will be non-linear interactions between the mechanisms and that the "isolated" stratospheric impact is in fact (arguably highly, but as your result show definitely significantly) idealized.

Specific and technical comments

To increase the clarity of the model description, it would be beneficial to include Table 1 right after paragraph 2.1 in order to have all the differences among the simulations at hand instead of including it in the appendix. Furthermore, if I understood correctly, the control simulation SSP2-4.5 is only needed since the refD2 experiment was not available at the time of the SSP2-4.5-sai simulation.  That is the reason why both are described in section 2.2. For completeness, since at L316-L317, it is said that SSP2-4.5 looks almost identical to refD2, it would also be better to include the plots as in Figures 9 and 10 (first column) for the refD2 simulation in the supplementary material (appendix).

In paragraph 3.1 on p. 8, it is stated that "in the interactive stratospheric aerosol simulation, aerosols are transported and sediment into the troposphere and deposited at the surface. In contrast, for the prescribed stratospheric aerosol experiment, the model zeros out the prescribed aerosol distribution and SAD below the tropopause." Why are the prescribed aerosol distribution and SAD below the tropopause set to zeros in the simplified simulation (senD2-sai)? If I understood correctly, in senD2-sai instead of using interactive stratospheric aerosols the provided aerosol properties by the SSP2-4.5-sai experiment are used. Are these properties not available for the troposphere? Is the aerosol presence in the troposphere negligible?

L25: G6 experiments not introduced or referenced.

L25: typo SSP5-85, it should be SSP5-8.5.

L26: Would not be better to cite Gidden et al. (2019, https://doi.org/10.5194/gmd-12-1443-2019) or Meinshausen et al. (2020, https://doi.org/10.5194/gmd-13-3571-2020) where the SSP5-8.5 scenario is described?

L45: change “Co-authors” with “et al.”; this variation appears also in lines L46 and L100, look carefully if other occurrences are present and change them accordingly.

L81: as in L26, would not be better to cite Gidden et al. (2019, https://doi.org/10.5194/gmd-12-1443-2019) or Meinshausen et al. (2020, https://doi.org/10.5194/gmd-13-3571-2020) where the SSP2-4.5 scenario is described?

L84: in the parenthesis “(Figure 1, top and panel, green lines”, do you mean “(Figure 1, top and middle panels, green lines)”?

L87: before using the acronym “AOD” (Aerosol Optical Depth) state its meaning.

L91: typo “SSP2.4-5”, change it with “SSP2-4.5”; this typo is present throughout the paper, look carefully and change it accordingly.

Fig. 1 and L96 - Statistical significance not assessed in Fig. 1 (missing information on the standard deviation or spread of mean values). What is the meaning of dashed lines in the upper two panels?

L97: In the parenthesis “(Figure 1, top and middle panels, blue and red lines)”, do you mean “(Figure 1, top and middle panels, blue and green lines)”?

Figure 3 and Figure 6: the control experiment for the senD2 experiment is referred to as “senD2-bg”, should not it be “senD2-fix”?

L102: change “Figure S1” with “Figure A1” as presented in the appendix.

L126-127: "However, differences in the aerosol distribution..." But this is underlaid by the intermodel-differences in atmospheric structure and circulation that will persist and the prescribed SAI distributions will not be in balance with this in other models.

L138: "...the 5-day instantaneous varying stratospheric aerosol distribution.." - Really, you want to prescribe this from WACCM to all models based on the corresponding model time?

L147 - Why climatological SSTs from REFD2 and not from say SSP2-4.5-SAI?

Around L195 - the issue of different tropopause heights - This issue will get far bigger when the prescribed field will be used in other models.

Fig. 3 - Why do you show the results in a vertical reduced domain, not even showing the stratopause region?

Fig. 4 - Why relative differences for H2O and O3? Also, the restriction to a fixed meridional belt does not take into account possible changes to the horizontal structure of the circulation and width of the upwelling region.

L207: after “compared to the control” add a reference to the Figure “(Figure 3, middle row, first and second columns)”, otherwise the sentence appears misleading and one can interpret it as referring to Figure 3, middle row and third column.

L231 "..simulated weakening of zonal winds..contributes to the direct radiative changes" - How can the zonal wind weakening contribute to the radiative changes and warming? I suspect that you have the causality absolutely reverse here.

L236: "..with a somewhat weaker polar vortex.."-> s with a significantly weaker polar vortex

L243: typos, change “stratosphre” to “stratosphere” and “coupld” to “coupled”.

L244-L246: add the reference to Figure 5.

The whole discussion of Fig. 5 - The impact on QBO and the difference between coupled and prescribed in this respect is not mentioned.

L250 "..due to the combined effects.." - You forgot to list tropical upwelling and QBO that are also affected and will play a role.

L251: change “Figure S2” to “Figure A3”, according to the appendix.

L262-L263: add the reference to Figure 4, second panel.

Fig. 7 - What is the meaning of thick lines? No information on the robustness of the mean value is given.

Section 3.2 - I think that this section should be deleted because the definition of the prescribed simulation renders it useless for assessing the tropospheric effects. To my understanding you allow the surface temperatures (not SSTs) to evolve according to radiative and other forcings, which contradicts your claims about isolating the top-down mechanism in this section.

L265 and L267: it is not clear what “(Figure S2)” refers to.

L267-L271: add the reference to Figure 6.

L310: at the beginning of section 2 it is said that CESM2 (WACCM6) is called WACCM6 throughout the paper, hence here just call it “WACCM6” instead of “CESM2 (WACCM6)”.

Figure 8 - Again, not clear what the dashed and solid lines stay for and there is no information on the spread behind the mean.

L385: change “(Figure S3)” to “(Figure A2)”, according to the appendix.

L411-413 - "The constrained experimental setup using a prescribed stratospheric aerosol distribution and fixed SSTs and sea ice successfully isolates the SAI-induced stratosphere-controlled (top-down) processes from ocean processes and their feedbacks with the atmosphere." - I think that this is an unsupported claim. As argued throughout the review, the set-up plus minus works in the same model that has been used to create the aerosol distribution. Besides this, feedbacks can be and are nonlinear and cannot be fully isolated. This should be at least acknowledged. Also, in my eyes, the effect of and on QBO remains unclear.

L426: "..model specific internal adjustements.." What does this mean?  And what about validity of the model specific tuning to other than SAI conditions?

L433: " The setup also removes the potential for an over or under-cooling of models using a prescribed

aerosol distribution from a different model (Zhang et al., 2024a)" - How? I probably missed this important feature of the setup.

L435 " ..are relatively small, as shown here" -  This is a very vague statement. Moreover, it is unclear, whether this will hold also in other models, where you prescribe WACCM distributions.