Referee #1
I think the paper has improved considerably since the first version. In particular, I am happy that the authors have dropped the part about the wavelet analysis.

However, I still have some comments that the authors should address before I can recommend that the paper is accepted for publication.

Major points:

1) I am confused about the statistical significance and the way it is shown in the figures. In line 161 it is stated that the hatched areas are not significant. This choice is also mentioned in the caption to Fig. 3. But then in Fig. 3 there are areas with little change (shown with white) that are significant while some areas with large change (blue) don't seem significant. The same holds for the other figures except for Fig. 4 where it makes sense that the 'SAI - CTL' is not significant. In line 208 the differences in Fig. 5 are described as significant across the whole region while almost everything is hatched in that figure.

I also think the t-test should be described in more details. For example, what is the number of independent points used. And I think it would make more sense to hatch the significant regions instead of the insignificant regions.

2) I am also in doubt about the interpretation of the correlations in Fig. 9 and Table 3. If the time-series are not detrended you will mix the signals from trends and from faster variability. The correct way would be to make correlations from detrended series and see if these correlations 'predicts' the size and sign of the trends. I am also in doubt of what and how you conclude about the drivers of the larger reduction in dust in the SAI compared to RCP8.5.


Minor points:

l20: while for -> except for.

l150: Is the meaning of the last part of this sentence that you also focus on dust hotspots. I think the sentence is unclear.

Fig. 3: In the title to panel f it says 'Seasonal trend'. I think it should be 'Seasonal time-series'. It is not the trend that is shown.

Fig. 4: In the caption it writes 'd) The annual mean of the dust concentration ..' I think you mean the surface temperature.

Fig. 5: There are no numbers on the colorbar.

L209: I think there should be a full stop before 'seasonal cycle plots ..

I am unsure if the hatching in Fig. 9 shows the significance of the correlations or the regions with large dust concentrations as it says in the caption. The authors should definitely show where the correlations are statistically significant.

Referee #2
The expectation was that this second version of the manuscript would be much improved, and the addition of uncertainty limits on all the parameters is a big improvement. Unfortunately there are problems with a number of figures and with the interpretation of many of these figures, leading to statements not supported by the figures.

Major comments:

The most surprising result of this work is how little difference there is between the two scenarios RCP8.5 and SAI for most of the parameters considered, Figs. 2, 6, 7, 10, 11; except temperature and TLAI, and temperature is a given considering the added stratospheric aerosol. This is an important point and should be emphasized more.

There is ambiguity with the use of the word significant. In statistics it has a very specific meaning, which is used in this paper some of the time. When describing differences it may be used, e.g. a significant difference, to imply a large difference. This is confusing in many places below and only very late in this review did I realize that the authors may be using the first definition, while I was assuming the second.

This leads to the complication of describing differences amongst the scenarios for the different parameters. Words like significant, considerable, large, small, … are subjective and readers and authors may differ on their meaning. In contrast words like larger, smaller, higher, lower, or more preferably percent differences, are objective and not as prone to misinterpretation. That is part of the reason, to suggest in my first review, that the authors present their results in terms of percent difference/change. They did this for some variables, but not all. The whole text should be gone through to make the comparison sentences more objective. Many of the comments below are related to confusion about these words, and disagreements with the authors interpretation/description of the figures.

18-21 “This reduction in dust over the MENA region is stronger under the SAI scenario, while for the dry season (e.g., summer with the strongest dust events), more reduction has been projected for the global warming scenario.” What is this statement based on, which figures? If anything Fig. 3e, d, g, h) might just suggest the opposite.

110-115 The amount of annual sulfur injections under the SAI scenario are not specified. Are they known? Text should be added how the SAI scenario is used to maintain 2020 temperature conditions. Is sulfur dioxide injected into the model? Or is the model just artificially nudged to keep the temperature at 2020 levels? If that is the case then aren’t the SAI scenarios and the control close to the same, or how do they differ? The authors should explain more fully how the SAI scenario is created.

180-181. Incomplete sentence. Maybe delete “that”.

183 Northeastern

Figures 3, 5, … When gray and white are included in the color scheme, and then stippled over in gray, it is difficult/impossible to separate the color contours from the stippling. For example Fig. 5n), is most of the region the central color, white, or is there some gray factored in? Also on Fig. 5, TLAI, there is no quantitative information on the color bar. What for example indicates no change?

207 “The TLAI under the RCP8.5 scenario shows some significant reduction compared to the CTL across the whole region,…” Here is a case in point from the comment above. What constitutes significant? The gray shades are difficult to separate from the stippling, compare Figs 5d) and e). Without numbers on the color bar how is the reader to know that gray is significant? Why doesn’t Fig. 5p) support the above statement. This panel shows that RCP8.5 and CTL are the same except for Jan – May which seems inconsistent with Figs 5g) j), unless the white is not apparent in these figures.

214-215 “On the contrary, under the SAI scenario compared to the CTL, the TLAI shows a significant increase both spatially and temporally (Fig. 5b, e, h, k, p and q).” Same problem as above. The color under the stippling (indicating significance) seems to be white, right in the middle of the color bar, which should mean 0/no change, but without a quantitative color bar it is unknown. The only place with positive colors are Spain and Eastern Europe. Changes shown as percent may help this presentation.

200-220 TLAI is difficult for any but an expert to understand. Perhaps a few words of explanation would help the reader understand what it means physically. Is it the fraction of a unit area covered by leaves, or??

226 “and in the Middle East with two major dust hotspots, the reduction is even stronger” This seems to be splitting hairs for the contour plots. This reader does not see a stronger reduction in the Middle East comparing the left and center set of panels. Perhaps the authors are emphasizing Figs 6p) and q), but calling these differences strong is a stretch. Plotting the differences as a percent would show a small percent difference, barely exceeding the uncertainty bounds.

222-232 A 0.2 m/s wind change is 5% for a 4 m/s wind. Seems pretty close to the uncertainty limit. Still not sure why percent isn’t used throughout as suggested in the first review.

238-241 “The box plot and monthly mean values of the precipitation from different scenarios (Fig. 7p) show that under the RCP8.5 scenario (compared to the CTL), precipitation is projected to significantly increase during the summer season, and under the SAI scenario (compared to the CTL), this region would experience more precipitation during the spring and summer.” It is difficult to see that this statement is supported by Fig. 7p) where the three means of the scenarios are barely outside the 25th/75th quartiles. Perhaps the percent differences are on the order of 10% in some months. Thus why the claim of significant? What do the authors consider is meant by significant, just that statistically the difference of the means is significant, or that physically the difference is significant? The different intent of this word needs to be clarified by the authors.

250-254. This language is much improved by discussing the differences as “higher/lower, moderate positive trend”, rather than putting a value on the difference, e.g. significant.

258 “over the MEAN region (Fig. 9h) and caption Fig. 9”, and text following. Where is the MEAN region and why is Figure 9h singled out to illustrate it? Fig. 9 shows the same MENA region shown on Figures 3-8.

260 “columnar dust concentration lower than 35 (μg/m3) are depicted with hatch-line in the Fig 9.” Once again the hatching obscures any of the contours except the brightest ones at the ends of the color bars, and in many panels the dust hot spots cannot be identified. Thus the discussion about these hotspots cannot be followed. Plus almost the entirety of each panel is hatched. It would make more sense to hatch the regions where dust concentrations are > 35 ug/m3. Also why is this number chosen? What is its significance?

Fig. 9 and Table 3. If the authors wish to compare these two they should be organized in the same way. Table 3 starts with precipitation and ends with wind speed. Fig. 9 is organized in reverse order to Table 3. Plus the dust hot spots are either not included on Fig. 9 or included in such a way that they cannot be seen. This makes it impossible to confirm, or follow the discussion from lines 260-272.

Fig. 10 The rows of this figure should be organized so that they follow the same pattern as Fig. 9 and Table 3, once they are synced. The order of the parameters at present is somewhat random. Here temperature is at the bottom. In Fig. 9 it is at the top. This just makes more work for the reader to understand the paper.

274-275. There is almost no change in dust in Fig. 10a3). So how do the authors claim just a lesser than strong reduction here?

283 “considerable reduction of dust concentration between the control and the two future scenarios for both regions” This is a reasonable statement for R5, but not R4. Again stating the differences as percent change is a quantitative statement which is objective. Words like considerable and significant are subjective and can be interpreted many ways.

283-290 The authors frame the discussion as cause and effect. The reduction in wind speed is “controlled by the lower wind speed and higher leaf area index.” But that’s more than the model can determine. All that can really be said is that they are correlated.

307-308 “reduction rate of the dust concentration is about 5-40% for the RCP8.5 scenario (compared to CTL), where it is stronger from March to September” Where does 40% come from. The maximum difference in Fig. 2a) is about 4 ug/m3 which is about 16% of the CTL at maximum.

314-315 Here the authors suggest the increasing temperature is the reason for decreasing dust, but again all that can really be said is that they are correlated. The cause and effect is difficult to ascertain.

330 “While this increase in the TLAI is found to be small over the dust hotspots” Why is this considered small? TLAI increases in R4/5 by more than 100% and in R2 by this much for SAI.

370-372 “We further conclude that, over the coming 80 years, the dust mass concentration generally decreases with an increase in the precipitation, soil water, and leaf area index, and a decrease in temperature and 10m wind speed over the MENA region” Is this the case for both RCP8.5 and SAI. If so this should be stated.

Referee #2
This manuscript can go forward to publication, although there are still problems scattered throughout which should be corrected. The comments below note the ones I found. There may be others. My reading was somewhat quick, but I would not have expected even this much for a third revision. The ellipses below set off quotes from the manuscript.

19 …, except over dust hotspots and for the dry season (e.g., summer with the 20 strongest dust events), which more reduction has been projected for the global warming scenario … This phrase doesn’t make sense from “which”. End this sentence at … events). Then start a new sentence with whatever the which … is to imply. I can’t offer a suggestion as I don’t understand it.

42 … from remote (what) …?

53 … regional projections over West Africa projected … Maybe try, regional predictions …

62 … models which participated …

198 … Overall, in Fig. 2e, the highest dust concentrations (up to 37 μg/m3) are found … What is the basis for this statement. The scale on Fig. 2e only goes to 30, and none of the lines even get close to this maximum. Oops I see the problem. The labels on the figures are in error. The bottom plot is 2f, not 2e as it is labelled now and led to the comment above.

222-223 Doesn’t this suggest that SAI isn’t doing anything if there is no change with respect to the control? This should at least be acknowledged if not commented upon.

Fig. 5 figure caption … total leaf are …??? Change is misspelled in the figure label for the color bar.

Figs. 5q and r. The authors make no comment on the fact that for TLAI RCP8.5 and CTL are at their maximum now and don’t change in the future. While the big difference is that TLAI increases under SAI. This is difficult to understand. Isn’t the CTL holding the climate state as it is in 2010 or so? Why isn’t this reflected in a difference with RCP8.5, Fig. 5p.

260 …The box plot and monthly mean values of the precipitation from different scenarios (Fig. 7p) show that under the RCP8.5 scenario (compared to the CTL), precipitation is projected to almost a 20% increase during the summer season, and under the SAI scenario (compared to the CTL)… These differences are hardly significant. The three scenarios are almost all within their data ranges. Also this may be only true because the precipitation is so low to begin with, making this comment even less significant. The real take home is that for most of the year and throughout the century there is very little difference in the scenarios compared to the CTL.

264 … MENA region will increase by 5% under the SAI and RCP8.5 scenarios… Is this number even measurable considering the natural variation of precipitation? Again there seems little significant differences between the scenarios. That is the important thing, not small, immeasurable differences.

266-275 Same comment as above. The differences pointed out are hardly measurable, using Fig. 8q to discuss trends is a stretch, and the surprising thing is how little differences appear between the trends.

291- … the wind speed is the main parameter that affects dust concentration change … This seems so obvious that it hardly needs to be verified with a model. Likewise precipitation.

Fig. 9 What is the difference between the two columns? One might assume that one is RC8.5 and one SAI but neither the figure caption nor the plots indicate what they are.

Fig. 10 Why isn’t the CTL added to these plots so the reader could compare the impact of the two scenarios, not compared to each other, but compared with the current climate. The main take away from Fig. 10 is that aside from temperature and maybe TLAI the two scenarios hardly differ in their impact on all the other parameters. The CTL appears in Fig. 11 so it is available.

299 … approximately 20% and 10% reduction is projected for the SAI scenario over the R2 and R3 hotspots respectively (Fig. 10a2 and 10a3)… Perhaps this is a valid statement for 10 a2, but not for 10 a3 where differences between the scenarios are imperceptible.

330-335 Or the authors could say that the dust concentrations declined 5-30% under either of the scenarios tested, saving a sentence and drawing attention to the similarities of the two scenarios. But why is this so? According to Fig. 9 dust is correlated with temperature. So why isn’t the dust less under SAI since there is a clear temperature difference between these two scenarios.

337-339 This is the explanation for the previous question that is probably correct and quite interesting. This point should be made early on in the manuscript as it provides a basis for better understanding of the results. And may also contribute significantly to the precipitation changes.

341 Shouldn’t LAI have already been defined by now. It has been discussed since Fig. 5.

363 … R4 dust hotspot will encounter an enhancement in the annual precipitation (i.e., about 100% and 65% under both RCP8.5 and SAI simulations respectively), consistent with an increase TLAI index (i.e., more than 100% under both scenarios)… Why aren’t these differences reflected in Figs. 5 and 7. According to Fig. 7 neither scenario shows more than a few percent difference in precipitation for R4. There is a precipitation increase of 10-20% just north of R4 for some months for SAI, but not RCP8.5. The maximum for the percent difference scale is 25 (30)% for precipitation (TLAI).

372-374. This conclusion seems so obvious that one shouldn’t need a model to make that statement.