The impact of increases in South Asian anthropogenic emissions of SO 2 on sulfate 1 loading in the upper troposphere and lower stratosphere during the monsoon season and 2 the associated radiative impact 3

14 The Asian summer monsoon plays a key role in changing aerosol amounts in the upper 15 troposphere and lower stratosphere (UTLS) via convective transport. Here, we use the 16 ECHAM6–HAMMOZ global chemistry–climate model to investigate the transport of 17 anthropogenic South Asian sulfate aerosols and their impact on the UTLS. Our experiments 18 (ten-member ensemble) with SO2 emissions enhanced by 48 % over South Asia, based on an 19 Atmos. Chem. Phys. Discuss., https://doi.org/10.5194/acp-2019-81 Manuscript under review for journal Atmos. Chem. Phys. Discussion started: 1 March 2019 c © Author(s) 2019. CC BY 4.0 License.

Reply(1): We agree these are important points that need to be clarified. The focus of our manuscript is to understand the convective transport of Asian sulfate aerosols during the monsoon season. Therefore free simulations (10 member ensemble) were performed for the year 2011 with a one-year spin-up for the year 2010. The analysis is presented for the year 2011. These experiments are canonical in design as their aim is to understand the radiative impact of Asian sulfate aerosols. The model results do not include the influence of QBO, which has a periodicity of 22-24 months. Also, the QBO is not internally generated in the model. We now clarify this in the manuscript at L236.
We thank the reviewer for the valuable suggestion about analyzing the role of the quasibiennial oscillation (QBO) in understanding the large-scale isentropic transport from the tropics to the extra-tropics. QBO can be generated in the model by the external nudging. To understand the influence of enhancement of sulfate aerosols on QBO, we have now performed external nudging experiments for the years 2008 -2016 (CTRL and Ind48 simulations). Our model simulations show that the enhancement of sulfate aerosols slows down the QBO propagation ( Figure 1a-b below). There is interannual variability in transport sulfate aerosols by the phases of QBO (Figure 1c). It affects the transport out of the tropics (Figure 1d). Since the focus of the present paper is to highlight the seasonal transport and associated radiative impacts, we plan to provide detail analysis of the interaction of QBO and sulfate aerosol in a separate paper which will focus on "Influence of sulfate aerosol on QBO: implications on Asian summer monsoon convection".
Following the reviewer's suggestions, we have now added a discussion about sulfate export from the tropical reservoir during E/W phase of QBO (section 6 in the manuscript).
Some recent studies have focused on this topic, looking at model simulations for SO 4 aerosols from sulfate geoengineering (e.g., Aquila et al., 2014;Visioni et al., 2018). It is true that in this latter case, as well as for aerosols from major tropical volcanic eruptions (e.g., Pinatubo; Trepte and Hitchman, 1992) the aerosols are located a few kilometers above those convectively uplifted from the surface, but the QBO impact on the latitudinal transport of aerosols in the lower stratosphere should be significant, anyhow. The link between tropical UTLS sulfate (convectively uplifted from South Asia) and its poleward transport is mentioned in several places in the manuscript (lines 22-23, 77-79, 309-314, 321-325, 344-345, 398-399, 450-453) and is one of the key points in the discussion. For this reason, need to be addressed.
Reply (2): Thank you for this important point. As mentioned in the reply(1), the QBO is not internally generated in the model (L231-232). To see the influence of QBO on the transport in our model, nudge simulations need to be performed. However, the model simulations used in the manuscript are 10-member ensemble free runs. These runs show poleward transport of the Asian sulfate aerosols by the lowermost branch of the Brewer-Dobson circulation.
We have also added a discussion about sulfate export from the tropical reservoir during East-West phases in discussion section 6.

3)
Proper acknowledgment of previous works in the literature is needed. The authors cite the review paper of Kremser et al. (2016), but they should do the same for the SPARC assessment of stratospheric aerosol properties (ASAP, 2006), as well. Here, in the uncertainties section of Chapter 6, a detailed discussion is made on the potential impact of future trends of stratospheric sulfate aerosols due to increasing anthropogenic sulfur emission in South Asia. A citation to SPARC-ASAP would be appropriate, for example, at lines 76-77 and line 287.
Reply(3) : We agree. As suggested citation of SPARC-ASAP is added at L81 and L315.

4)
Minor points Both in the abstract (line 19) and in the conclusions (line 447) the authors write: ". . .experiments with SO 2 emissions enhanced by 48% over South Asia. . .". For the reader, it is not clear (mainly in the abstract) with respect to what the emissions are enhanced by 48%. Later on in the text this is made clear (lines 210-213). Reply(4) : As suggested, SO 2 emissions enhancement by 48% over South Asia in the model experiment is made clear in abstract and conclusion (L18-21 and L567-570).
Reply(5) : The above-said sentence is removed from the manuscript.
(8) Line 340: is likely to be caused.
Reply (8) : Above mentioned phrase is not used in revised manuscript.
(9) Lines 370-371: Ozone absorption of the increasing diffuse radiation by sulfate aerosols may also play a role.
Reply(9): The above sentence is now removed from the revised manuscript. Reply (14) : The above reference is added at L315, L834.

Replies to Anonymous Referee #2
I commend the authors for their effort for addressing their issues with the original manuscript.
( 1 ) regarding a possible explanation for the spatial maximum in transmitted solar radiation at the surface in west India at L389 : ( 2 ) for including AERONET AOD in this version of the manuscript .
( 3 ) reducing the number of terms in the OMI SO 2 trend model.
I have a few remaining major issues with this paper: Reply: We thank the reviewer for the positive comments on our study and all valuable suggestions for improving the paper. In the revised version of the manuscript, we have addressed all the reviewer's comments through additional experiments and clarifying some key statements.
Most atmospheric chemistry-climate models do not agree well with satellite remote sensing observations. In the past, several papers have been published stating that majority of CMIP5 models underestimate AOD due to several reasons (Sanap et al., 2014;IPCC;2013;Sockol et al., 2017) e.g., dust optical depth is underestimated in the model (Pu and Ginoux, 2018). There are uncertainties in model estimates of sea salt emission and parameterization (Spada et al., 2013).
We now include additional details in the revised manuscript at L279-282. (2) Because of this disagreement and the vertical transport of aerosols in the Asian monsoon to the UTLS, I think the authors need to have much larger uncertainties on the warming of the lower stratosphere by these (sulfate) aerosols: the abstract states that the warming reaches 0 .6±0.25 K. I would expect that the uncertainty on the warming should probably be comparable to the magnitude of the warming. From the uncertainty on the aerosol abundance available to reach the lower stratosphere (see Fig. 1e), I would expect a relative uncertainty of at least 75 % on the warming (assuming a linear relationship between local aerosol concentration and local temperature) The authors should remember that their warming estimate is obtained by the difference between two simulations (one being the control) and that differencing amplifies uncertainty if there is any kind of (numerical) noise in the model. Also, the method to determine the uncertainty of the warming estimates is not provided. The relative uncertainty on the SO2 trend of 20 % ( = 0 . 9 7 % / 4. 8 % ) could be added ( in quadrature ) to any other independent source of uncertainty.
Reply (2): Yes, as mentioned we have obtained warming estimates from the difference between the two simulations. Each of the CTRL, Ind48 and Ind48Chin70 simulations are ensemble mean of 10 members and deviations within 10 members is represented as uncertainty. As the number of members increases, this uncertainty reduces. In the past, to obtain statistically robust results ten member simulations are performed by the numbers of researchers e.g. The uncertainty in SO 2 trends obtained from OMI data is also given at a 99% confidence interval level. It is mentioned at L166. 3 ) The uncertainty on the SO2 trend from OMI got small compared to the earlier version of the manuscript, i .e . 4 .8 ± 0 .9 7 % / yr versus 4.8 ± 1 .3 % / y r. How did that happen? The uncertainty should have gotten larger with fewer terms.
Reply (3): Although terms were reduced in equation 2 the level of confidence was also reduced. Now the uncertainty in SO 2 trends obtained from OMI data is also given at a 99 % confidence interval level. Therefore, now it is given as 4.8 ± 3.2 % yr -1 over India and 7.0 ± 6.3 % yr -1 over China. It is mentioned at L166.
4) The paper focussed on the monsoon season because convective transport in this season lifts boundary layer aerosols to the UTLS. But this is also a season with a strong rainout, and this might be leading to model bias since the rain out will be difficult to simulate accurately. Furthermore, by limiting this paper to only the monsoon season, the conclusions are less interesting than a paper that providing the same analysis over a full year. Reply(4): As suggested we have provided analysis for a full year from additional simulations for India and China SO 2 emissions perturbed according to the respective trends based in OMI observations during 2006 -2017, (1) SO 2 emission over India increased by 48 %, (2) SO 2 emission over India increased by 48 % (Ind48) and decreased over China by 70 % simultaneously (Ind48Chin70). 5) Over shooting convection (L313) is suggested as a pathway for aerosols to reach the stratosphere, but would't the aerosols grow in to cloud droplets with so much humidity and such strong convection? I suppose this is addressed by L75 of the introduction.
Reply (5): It is now mentioned at L86. 6) Why examine the SO2 emissions from India only? The emissions from China have changed by the same magnitude but opposite sign (L83). The net effect might be very small if emissions from India and China were both perturbed according to the respective trends (based on OMI). Chinese emissions are very relevant to the Asian monsoon region. Based on these issues (particularly # 1), I feel the quality of this paper is some what low and the work is of limited scientific interest (see the previous paragraph and issue # 3). I do not find the model simulations convince me in terms of being realistic and yet the stated uncertainties are rather small. I suppose I could accept this paper if realistic and adequately described uncertainty calculations were included. Reply(6): As mentioned in reply (3) we have now included results from additional simulations where SO 2 emissions over India and China are perturbed according to the trends observed by the OMI satellite.
The uncertainty estimates are replaced by 99 % confidence interval levels.
As mentioned in reply 1-3, we have shown uncertainties within 10 members of simulations.
Other general comments (7) The model does not include nitrate, although it is acknowledged that it is an important aerosol species (L94 , 97, 100, 102, 258 ) .
Reply (7): The uncertainty due to absence of nitrate aerosol is mentioned in the manuscript at L277 as "Inclusion of nitrate aerosol may affect the distribution of the AOD." (8) The reference list needs to be improved. A consistent format is needed. Capital letters should not be used in common nouns in the article title, except for the first word of the title.
Reply (8): As suggested reference list is improved and capital letters are removed from the common nouns in the article title.
See specific comments below. . I believe the statement regarding g 50 % of the aerosols being located above 4 k m is not present in the 2010 paper. Also, this statement cannot be generally true ; the fraction will be much less than 20 % in polluted places and perhaps this value is appropriate for a remote observing location in the Himalayas where the surface is at an altitude of 2 k m .
Reply(25) : The above sentence is re-phrased as: Dumka et al. (2014) has documented that in AERONET observations the aerosols above 4 km contribute 50 % of AOD at Kanpur (in Indo-Gangetic plains). See the snapshot below: (26) L 2 6 4 : If there are models that perform better than ECHAM6 -HAMMOZ in terms of accurately simulating A O D , the authors should justify their decision to prefer ECHAM6 -HAMMOZ .
Reply (26) : As mentioned in reply(1), we have added discussion and reasons for regional under-estimation or overestimation of AOD in the model (L275-282). The model underestimates AOD over India while it overestimates over China in comparison with MISR. While it is underestimates over both the regions in comparison with AERONET. The above discussion is pertaining to the monsoon season while the current manuscript discusses all the seasons. Therefore it is now removed from the manuscript. We have added supplementary figure 1 depicting regions of convection during different seasons using OLR from NCEP; Ice-crystal-number concentration and cloud-droplet number concentration from ECHAM6-HAMMOZ. (34) L 3 2 9 : " occurs on a daily scale " -> " it is of short duration ( i . e . , days) and is episodic.
(35):L 3 3 9 -3 4 1 : The authors speculate in many places in the paper ( search for " may " or " likely " ) . I find it excessive and in this case, I wonder if the anomaly is significant ( ie . , real ) . L 3 4 4 : More speculation with little support … . An equally " likely " explanation, in my opinion, is the much greater concentration of sulfate in the lower troposphere between 50 -70  N .
Reply(35): Now, above sentences are written as affirmative.
Reply(37): t is removed from the revised manuscript.
(39) L 3 7 9 ( S e c t . 5 . 2 ): I believe the first paragraph here could be quite confusing for readers. The trend in radiative forcing from Ramanathan et al. is provided ( but with the wrong units, should be W/m2/y r ) and then the next sentence presents the magnitude of the anomaly due to sulfur emissions found in this work. But the numbers are really not comparable since Ramanathan et al. were looking at a much longer and earlier period. I realize the authors may be trying to simply cite this related work here but I fear that readers will be believe they should somehow compare the radiative forcing anomaly simulated in this work to the trend from Ramanathan et al . and / or Padma Kumari et al . This can be remedied by simply starting the sentenceat L382 with " While not directly comparable to these previous studies , … " . Reply (39): As suggested we have added the line "While not directly comparable to these previous studies" at L 450-451.
(40) L 384: The authors are cherry -picking the evidence . Figure 6 b does not show a very coherent pattern . While the tendency is for negative solar radiation at the surface in northern India , west India is not the only exception ( e . g . east India ) .
Reply (40): We agree that this needs to be clarified and thatnk the reviewer for pointing this out. The above sentence is re-written at L455-463 as "We estimate the changes in net solar radiation at the surface for four seasons from the Ind48 and Ind48Chin70 simulations. Figure  7i-l shows that the Ind48 simulations have produced negative anomalies in net solar radiation (SR) at the surface (~-0.5 to -3 W•m −2 ) over India and parts of China (where sulfate aerosols are transported) due to the enhanced sulfate aerosol layer reflecting back solar radiation. In general, the seasonal mean distribution of anomalies in net solar radiation at the surface is similar to the distribution of the anomalies in RF at the TOA. Reduction of Chinese SO 2 emissions along with an increase of SO 2 emissions over India (Ind48Chin70) has produced a reduction of solar radiation over India while there is a significant increase over China (1 -5 W•m −2 ) (see Fig 7 m-p)".
(41) L 3 8 7: " connecting the boundary layer of the ASM region to the UTLS " sounds poetic, but it is not demonstrated in this paper. I believe the reduction in surface radiation is mainly due to aerosols in the boundary layer and the aerosols in the UTLS have a very minor contribution to the received short wave radiation at the ground. This can be tested by removing the aerosols from the UTLS and looking at the change in short wave radiation at the surface.
Reply(41) : The above sentence is removed.
Reply (42) : We have now given uncertainties at a 99 % confidence level.
(43) L 4 0 5 : subsidence is not discussed in section 5 . 3. I suggest that "and subsidence" is removed here. Reply(43) : As suggested it is removed.
Reply(44) : We have removed 'the strong subsidence" and re-written it as 'upper tropospheric cooling and enhanced stability may suppress the rainfall' at L501-502.
(46) L 4 3 4 : " anomalies are negative " does not belong in this sentence. Please reword so that this is a proper sentence.
(48) L 4 6 2 : There is not " good " agreement between the off-line calculations and the model. Also, " minor " is absolutely not acceptable in the next sentence .
Reply (48): This sentence is re-written as "These values are comparable with results of the ECHAM6-HAMMOZ simulations, with the minor differences likely due to the implicit dynamical impacts in response to enhanced south Asian SO 2 emissions in ECHAM6-HAMMOZ not being represented in the offline model". L584-587.
Reply (49) (58) L 8 2 1 ( F i g . 3 ) : The black vertical bars are not described in the caption and should be removed because they block the colour contour plot .