The authors present an analysis of the effects of stratospheric aerosol injections in the upper stratosphere, in comparison to more typical lower stratospheric injections, as simulated in the WACCM-MAM3 climate-chemistry-aerosol model. While the results are generally presented in a clear way and could be interesting I think that the analysis lacks depth in several respects. My major concerns are listed below.

The comparison of emissions at 50 km is only done with respect to equatorial emissions at lower altitudes. However, there are earlier studies comparing different lower tropospheric emission strategies and also reporting some benefits in comparison to equatorial emissions.  I think the authors need to put the potential benefits of their strategy into the perspective of these other emission strategies.

I think that the model validation in 3.1 is too superficial concerning the upper mesosphere. I agree that it is useful to show an AOD comparison for the Hunga Tonga eruption where the sulfate also reached the upper troposphere. However, a distinct difference between emissions at 50 km and lower altitude emissions is the reported meridional distribution. This will depend on the representation of the overturning circulation in the upper stratosphere which for which an important feature is the semiannual oscillation near the stratopause. How well is that represented in the model. I don’t think it is sufficient to say “temperature at 100 hPa, QBO strength, and polar vortex strength” show reasonable agreement with reanalysis data, because these are all features evaluated in the lower to middle stratosphere. Of course, evaluations of near-stratopause circulation are more difficult due to the lack of observations, but I think this needs to be discussed.

I disagree with the title. The study doesn’t provide evidence that SAI emissions near the stratopause “minimize” side effects. There could be other strategies reducing the side effects further that haven’t been tested. Who knows what effects would be caused by an emission in the upper mesosphere?

At the end of Section 3.2 the authors write “The distinct latitudinal and vertical distributions of aerosols in SAI50 enhance climate cooling benefits while minimizing negative impacts of climate intervention.” Besides the issue with the word “minimizing” mentioned in connection with the title, I also think this statement is not sufficiently backed up, at least not at this point of the manuscript. Possibly the sentence is meant as an announcement for the following two subsections, but it sounds like a summary.

More in general the manuscript suffers from the lack of the definition of a goal for the SAI. Without such a goal, without defining a metric it is impossible to compare which strategy performs best. The goal could (but doesn’t have to) be to produce a climate as similar as possible to an unengineered climate of the same global temperature at lower greenhouse gas levels. In this sense, it is not clear if the stronger Arctic amplification simulated for SAI₅₀ than for SAI₂₅ is actually a desired effect. How strong is the Arctic amplification in greenhouse gas caused warming in WACCM? Which injection strategy is counteracting the amplification more exactly?

Concerning Arctic amplification, the authors write: “In SAI50, the simulated 22% greater global mean surface cooling compared to the 10% increase in global mean AOD (Fig. 1a), is primarily attributed to Arctic amplification effects (Barnes and Polvani, 2015), with a minor contribution from the reduced stratospheric water vapor enrichment (Fig. 2c-d).” I don’t understand this statement. Arctic amplification, depending on the mechanism which causes it in WACCM, should be part of the temperature response in both strategies. Shouldn’t part of the difference between SAI50 and SAI25 be due to the different aerosol distributions. Is the idea that polar aerosols create a larger forcing than low-latitude aerosols? Would that be related to the surface-temperature dependence of stratospheric aerosol forcing as discussed by Hegde et al. (2025). Or to aerosol forcing being more efficient at high than low latitudes? Anyhow, I think it is necessary to physically explain the relatively strong additional global cooling for a relatively weak AOD increase.

Figures 3c and 3d show simulated annual cycles of the high-latitude cooling signals. In the Arctic there is a pronounced seasonal cycle, while it is negligible in the Antarctic. This behaviour is just stated but not explained. To develop trust in such signals it is important to explain the physical mechanism causing this difference. Moreover, with respect of the “cooling benefits” discussion it would be important to discuss if these different annual cycles just offset different annual cycles of high-latitude greenhouse gas warming or if seasonal cycles are strongly modified.

Finally I see a major issue with the lack of discussion of the additional costs of emitting near the stratopause compared to the lower troposphere. The authors are briefly mentioning the option of using rockets and conclude that the “results clearly indicate that a detailed engineering design study […] is warranted.” I think at least a brief estimation of costs based on existing rockets would be necessary. One could argue that scientifically it is interesting to see the dependence of SAI effects on the injection height. But if feasibility plays no role, why not emit at 70 or 100 km? As the authors claim to “propose a novel SAI approach” I think a minimum effort on estimating feasibility is necessary.

Hegde, R., Günther, M., Schmidt, H., & Kroll, C. (2025). Surface temperature dependence of stratospheric sulfate aerosol clear-sky forcing and feedback. Atmospheric Chemistry and Physics, 25(7), 3873-3887.

Injection Near the Stratopause Minimizes the Stratospheric Side Effects of Sulfur-Based Climate Intervention

by Pengfei Yu, Yifeng Peng, Karen H. Rosenlof, Ru-Shan Gao, Robert W. Portmann, Martin Ross, Eric Ray, Jianchun Bian, Simone Tilmes and Owen B. Toon

Review by Thomas Peter

General comments

This is an interesting manuscript on a new idea for how climate intervention through stratospheric aerosol injection (SAI) could be implemented by injecting at much higher altitudes (~50 km) than previously proposed.  This could lead to significantly less harmful side effects than previous proposals to inject SO₂ into the lower stratosphere.  This could reduce both the strong warming of the lower stratosphere due to IR absorption by H₂SO₄-H₂O aerosol, which affects climate and precipitation zones in the troposphere, and the depletion of stratospheric ozone.  As the authors describe in their manuscript, this new idea could even boost the efficiency of surface cooling.

To my knowledge, this idea is original, and the proposed method could potentially be very important and would fit well into ACP.  The authors are to be commended for developing this concept.  Unfortunately, however, I do not believe that the technical details necessary to justify the feasibility of this novel method are sufficiently developed for publication in ACP.

The differences in aerosol formation at an altitude of 50 km compared to 25 km are considerable and would need to be discussed:

(1) The temperatures are so high that the formation of H₂SO₄-H₂O droplets close to injection altitude seems unlikely.

(2) The air density is so low that the fall velocity of particles, if they form, will be very high, which is a major limitation.

(3) The H₂SO₄ photolysis, which is neglected in this modeling, could massively alter the model results.

These three aspects are not addressed in the submitted manuscript (with the exception of a reference to the fact that H₂SO₄ photolysis is neglected).  However, the issues associated with these aspects are central to such a new proposal and must not be ignored.  Therefore, I do not think that the manuscript can be accepted for publication in its present form.

Specific comments

In the following, I will first explain my concerns regarding these three points in more detail and then provide a list of minor comments and suggestions for improvement of the manuscript.

Regarding (1):  At the stratopause, temperatures range between 260 and 270 K.  At such high temperatures, the normal Junge aerosol can no longer exist.  The H₂SO₄ vapor pressure of aqueous sulfuric acid is approximately 10^-8 hPa (see Fig. 4 of Carslaw et al., Revs. Geophys., 35, 2, p. 125,1997).  At the stratopause this would correspond to 10 ppbv H₂SO₄ in the gas phase, i.e. about 100 times the total mixing ratio normally present in the lower stratosphere during volcanically quiescent periods.  I think the mixing ratio reached by SAI₅₀ would remain lower.  Therefore, the nucleation of sulfuric acid particles probably only begins well below the injection height.  Figure 1C seems to confirm that.  How does the model treat this nucleation?  No information is provided about the microphysical code, which for this proposal should be of major concern.  I suppose the microphysics is treated by a modal approach, and it would be important to see size distributions at various altitudes.

Regarding (2):  The air is so thin at the injection height that particles with a radius of 100 nm sediment by about 10 km within a month (eyeballed from Fig. 2 of Müller & Peter, Ber. Bunsenges. Phys. Chem. 96, p. 353, 1992).  Since this is a fundamental aspect of the proposed injection scheme, it would be necessary to check the model's implementation of this process carefully and to provide arguments, why this fast sedimentation does not invalidate the whole procedure.

Regarding (3):  The photolysis of H₂SO₄ molecules is a central process in this scheme and cannot be ignored without good reason.  A quick test with our own chemistry-climate model shows that the amount of condensed H₂SO₄  in the Junge aerosol layer is 2-4 times higher without photolysis than with photolysis.  I would estimate that the reduction of aerosol mass in the upper stratosphere could more than a factor of 10.  I must therefore assume that in the modeling work shown, a significant portion of the aerosol is formed solely due to the lack of photolysis in the model.

While points (1) and (2) can probably be positively resolved with the existing model runs (showing that the model correctly calculates the partial and vapor pressures of the aqueous sulfuric acid under upper stratospheric conditions and the sedimentation velocity of the aerosol particles formed), point (3) is likely to pose a bigger problem. If photolysis cannot be incorporated into the existing model, at least a very clear warning should be included for the reader that this omission may significantly influence the result and diminish the efficiency of the proposed method.

Technical comments listed by line number

1. L. 19: “SAI using sulfur cools the planet” à “SAI using sulfur has been proposed to cool the planet”
2. L, 20: Better don’t talk about “traditional” SAI, in particular not in the first sentence of the abstract. All SAI is still in the proposal phase and largely unsubstantiated ideas, nothing traditional.
3. L. 38: In addition to Ferraro et al. (2015) and Visioni et al. (2021), another excellent example making this point is Wunderlin et al., “Side effects of sulfur-based geoengineering due to absorptivity of sulfate aerosols”, GRL, 2024.
4. L. 66: I do not understand the “positive ozone chemical tendency”.
5. L. 73: Here I expected more information on the type and characteristics of the microphysical module used in this modelling work
6. L. 79: I do not understand and do not accept that the fact that the column-integrated stratospheric burden of H2SO4 is much smaller that the burden of sulfate aerosols could be used for not having to take the photochemistry of H2SO4 into account. In the warm upper stratosphere, all H2SO4 is gaseous and exposed to H2SO4 + hv.
7. L. 84: “SO2 was continuously injected … with a total rate of 10 Tg per year”. It might be more meaningful to say “with a total rate of 27 Mg per year” to stress the continuous character, or even “with a total rate of 27 tons per year”.
8. L. 87: 5 years of model spin-up plus 15 years of actual model run. This is okay.  But then, does Figure 1 show the 5 years of spin-up plus 10 years of model run?
9. L. 90: I do not understand these scaling factors. Where are they from?
10. L. 103: The “2022 Hunga volcanic eruption”. Okay, everybody knows which volcano this is, yet it is a pretty crude abbreviation.
11. L. 110: The “spread is designed to capture” sounds weird. Are you designing a spread or is the spread the result of your simulations?
12. L. 127: Sentence confusing. For clarity, I would rewrite “… are similar for all lower altitude injections (at 20 km, 25 km and 35 km),...”.
13. L. 129: The word anomaly is used abundantly, also when it is just a “change” or even a total number. For instance, in Figure 1a is it really AOD anomaly or just AOD.  And why is it in Figure 1C simply “SAD”, and not “SAD anomaly”?
14. L. 260: “would be” instead “is”.
15. L. 295: Where is the dip in SAD (Fig. 1c) at 18 km come from?
16. L. 295: The red curve in Fig. 1c would probably look quite different if H2SO4 + hv was taken into account.
17. L. 296: Why distinct?
18. L. 301: “Multiple” is not a verb.
19. L. 305: “from ensembles” is slang. “of the ensemble members” would be appropriate.