The authors analyzed model simulation geoengineering results of G6solar, G6sulfur, and GLENS to examine effect of stratospheric aerosol injection (including sun-shading) on global drought and associated socioeconomic exposure. G6solar and G6sulfur results are from multi-model simulations of equatorial SO2 injection to bring down global mean temperature from SSP5-8.5 level to that of SSP2-4.5. GLENS results are from CESM simulations that inject SO2 at multiple locations to meet multiple temperature stabilization goals at current-day level under the background scenario of RCP8.5. The authors found that the frequency of global extreme drought is reduced by about 2% for both G6solar and G6sulfur, as well as GLENS. Also, the drought impact from geoengineering on economic and population exposure would be unevenly distributed for developing and developed countries.

This study provides some useful information on the potential effect of stratospheric aerosol injection geoengineering on global drought and economic consequences. However, the useful scientific insights provided is limited, and I think the overall focus of this manuscript is not suited for ACP.  

As stated on the journal website:  “Atmospheric Chemistry and Physics (ACP) is a not-for-profit international scientific journal dedicated to the publication and public discussion of studies investigating Earth's atmosphere and the underlying chemical and physical processes. ACP publishes studies with important implications for our understanding of the state and behavior of the atmosphere and climate, including the troposphere, stratosphere, and mesosphere.”

As stated (also my opinion), the core research area published in ACP should be the underlying chemical and physical processes of the atmosphere. The impact study is encouraged, but should not be the main theme of ACP. The analysis of socioeconomic exposure, which is more policy relevant and much less atmospheric physics relevant, should not be the focus of this study (for publication in ACP). More focus should be given on the drought response to stratospheric aerosol injection and the underlying mechanisms.

The above point is clear by reading the abstract. The abstract has little scientific findings relevant to atmospheric physics except for the one-sentence description of drought response. In my opinion, the study framed here is more suited for publication in interdisciplinary journals such as Environmental Research Letters.

A more detailed comparison between the effect of G6sulfur and GLENs on extreme drought and possible underlying mechanisms should strengthen the physical science component of the manuscript.   

Specific comments:

Abstract: The abstract is poorly written with a lot of unclear sentences.

For example,

“SAG implementation reduces this increase by 1.99% (1.80% in GLENS), primarily due to its cooling effects.”

by 1.99% relative to what ? (current-day? SSP5-8.5?).

“SAG-induced rainfall deficits” deficits compared to what condition?

“Countries with less development experience smaller reductions, or even increases”

Reductions or even increase of what? And, compared to what?

Throughout the text, the authors used SAG to represent stratospheric aerosol geoengineering. A more commonly used word would be SAI (stratospheric aerosol injection). Thus, I use SAI for the following comments.

The writing is sloppy. For example, the first sentence, line 38. (Song et al), year is missing.

Lines 65-67: This description is not accurate. For example, whether SAI would reduce rainfall depends on how SAI is employed and what reference scenario is compared (compare with historical baseline state or high-GHG world without SAI?). It is not right just to state SAG could reduce rainfall without context. 

Lines 68-71: The results mentioned here would all be scenario dependent. In particular, regional climate responses to SAI would strongly depend on the specific SAI scenarios  (e.g. amount of SAI-induced cooling, location of injections, background GHG scenarios). Therefore, it is just misleading to state something like SAI-induced cooling could offset about 90% of extreme drought risks in Cape Town, South Africa.

The latest literature the authors cited in the Introduction are from about four years ago. Many latest developments in the research field of climate effect of SAI are missing.

Line 91: What does linear injection mean?

Line 119: What are the latitudes of SO2 injection? Please specify. This is important.

Line 176: How is scPDSI calculated from PET? This is not described. 

2.6 The heading (and corresponding texts) is misleading. These are just offline calculations, not numerical experiments.

Lines 256-258: Why use 46.5% for G6solar and -1.84 Celsius for G6sulfur?  

Lines 246-270: The description of temperature and precipitation response to G6solar and G6sulfur is too lengthy. These have been shown in previous studies and should only be briefly discussed. Also, for comparison, climate response in GLEN simulation is only described in one sentence (268-270). Why?  

Line 274: Increase by 7.33%. What period compared with what period? This should be clear.

Lines 281-283: frequency of extreme drought events is reduced by 2.12% globally, mitigating 28.9% of the increased drought stress under the SSP5-8.5 scenario.

This sentence is not clear. ‘reduce by 2.12%’, this reduction is relative to what? (SSP5-8.5 or historical baseline?). Also, what does 28.9% mitigation mean? Actually, the description of SAI effect as a whole is not clear. I am not clear whether the SAI effect is compared with historical baseline or SSP5-8.5. 

Lines 285-286: With G6sulfur, a similar reduction of -1.99% is predicted for global drought extremes

In the above, it stated that in response to G6solar, drought is reduced by 2.12%, but it stated here that in response to G6sulfur, drought is reduced by -1.99%. The sign is opposite (I believe the negative sign is not needed). This is just one example of the sloppy writing. 

Lines 292-293：What does mitigation efficacy mean?

Lines 293-295: The comparison between G6sulfur(G6solar) and GLENEs should be discussed more extensively, especially at the regional scale. G6sulfur and GLENs have distinctly different strategies of SO2 injection, but in the manuscript, the response to GLENs is only very briefly discussed.

Lines 312-313: “Relative to SSP5-8.5, SRM-induced cooling reduces extreme drought frequency by 3.44% in G6solar (Fig 2e) and 3.42% in G6sulfur (Fig 2h).”

So the numbers in line 281 (2.12%) and line 286(1.99%) refers to the change relative to the historical baseline?

Line 297: Why not include the similar analysis for GLENS? Is it because the lack of required data?

Line 330: This part builds on the effect of SAI on global drought. It is just an impact assessment of the drought analysis. Not sure to what extent the reported specific numbers here are useful. These numbers in terms of GDP and population exposure are likely to change substantially for different SAI scenarios, in particular at the regional scale.

Lines 370-371: What does asymmetric responses in temperature and precipitation mean?

Lines 368-385: This paragraph mostly discussed previous findings. What is the point here? Also, all result regarding climate response to SAI would depend on the specific SAI scenarios used in that study, but the authors failed to acknowledge this important point. For example, statement like “SAG overcompensates for the greenhouse gas-forced expansion of the Hadley circulation and offsets the poleward shift of storm tracks in mid-latitude of NH” would strongly depend on how SAI is implemented.  

Line 405: In GLENS, SAI is not designed to offset all the GHG-warming, but just to keep temperature at current-day level.  

Lines 409-410: The SAI strategies investigated here are just two specific SAI scenarios, and thus the statement “that the choice of injection strategy does not substantially alter the major conclusions” is just not right. One can certainly design many SAI scenarios that ‘substantially’ alter major conclusions here.  

Fig. 1. TAS and PR in figure are not defined.

There are some interesting things in this study.  The authors have computed a drought index and collocated it with demographic/economic data, which I have never seen done before.  I do question whether this is the best suited for ACP, but I’ll leave that up to the Editor to decide.

I am not entirely sure what to recommend for this paper.  There are many descriptions that are lacking and many others that need to be changed.  The next version of this paper will be so different that it’s hard to make a determination.

My biggest complaint is that there are numerous places where the description is not careful, leading to some over-claims.  For example, while the results are potentially important for climate justice, actual considerations of justice should take numerous variables into account, as well as adaptive capacity (like irrigation, or countries that experience less drought could pay countries that experience more drought).  Lines 331-332 are another example – you’re looking at GDP and population correlated with the Human Development Index.  You are not actually looking at “human societal development”.  It would be very easy to argue about these different data sources and how relevant they are for society, and entire fields of study do exactly that.  The scenarios and strategies that you looked at are quite narrow and not designed to affect drought indices, yet you make broad conclusions about what SAG would do in general.  The point being, you should be careful to stick to discussing what you actually did.

Another major concern is that you only looked at the frequency of extreme drought (lines 139-140).  There are other forms of drought that are still important.  While this doesn’t need to change your analyses, your descriptions should be more careful.

The reporting of the results is quite biased.  One of the major conclusions of the paper is that low HDI countries have a large increase in drought exposure under SAI.  But looking at the results in Figure 4, the error bars are enormous.  This indicates a strong model-dependence of the results, which is hardly mentioned by the authors.  The authors could easily be accused of cherry-picking evidence to make SAG look bad.

And finally, the writing needs to be substantially improved.  Many descriptions are simply glossed over, for example how the scPDSI is defined.  Also, your descriptions of percentage change are confusing – are these relative or absolute changes?

I will also encourage the authors to ensure that all of the elements in the supplemental material are referenced in the main text.

Lines 26-27:  These numbers are scenario-dependent (as you say later in the paper).  You need to be more specific.

Lines 30-31:  See my general comment above.  You looked at one index in three scenarios.  It is very difficult to conclude anything about what people “should” do to promote climate justice.

Lines 67-70:  These are not the only regional effects.  Why are you focused on these regions?

Lines 73-74:  Yes/no questions are not particularly interesting.  Without doing any research, I can tell you that the answers to the questions as posed are yes and no, respectively.  Can you rephrase these?

Line 82:  You explore _some_ uncertainties.

Lines 219-221:  This is only true if the equation is linear.  Otherwise your description needs to be more careful.

Line 233:  Models?

Lines 255-256:  The cooling should not be uniformly distributed.  I think this is an artifact of your choice of color bar.

Lines 333-336:  You need to be much more specific about your assumptions.  Very very small differences in the population growth rate can lead to differences of billions of people by the end of the century.

Lines 357-358:  Exposure is not risk.  https://www.ipcc.ch/site/assets/uploads/2021/02/Risk-guidance-FINAL_15Feb2021.pdf

Lines 371-372:  This is a strawman argument.  These scenarios were not designed to alleviate drought extremes, so it’s unsurprising that they have limited effectiveness in alleviating drought extremes.

Lines 373-374:  This isn’t necessarily rainfall enrichment.  It’s a rainfall increase.  The increase might be in the form of severe storms, which isn’t good either.

Lines 409-410:  This narrow set of simulations does not cover all possible strategies.  You can’t make this conclusion.