|The manuscript mainly addresses three aspects: (i) dry day occurrence frequency over the Indo-Gangetic Plains and the northwestern and southeastern parts of India during the middle period of Asian summer monsoon and their long-term variations, (ii) role of aerosols, specific humidity, ENSO and Sunspot cycle in influencing the long-term trend in dry day occurrence frequency at the above regions, and (iii) the predicted occurrence of dry days till the year 2100. Overall, the topic is scientifically important and highly relevant. The first and third objectives are almost achieved. However, the role of aerosols in influencing the long-term trend in dry day occurrence frequency (which is the main theme and novelty of this work) could not be achieved satisfactorily. The major comments and suggestions on this manuscript are given below.|
1. The primary data on aerosol characteristics (aerosol optical depth (AOD) and contributions to AOD by black carbon (BC), organics, PM2.5 dust, sulphates, and sea salt) are taken from MERRA-2. Several papers published earlier have shown the overall usefulness of MERRA-2 AOD. However, accuracy of the contributions from the individual aerosol components at different geographical regions is not clear, though this product has been compared against observations at a few locations. In order to address this issue, in the present study, the monthly mean near-surface level BC concentration at Kolkota (a polluted region in the Indian subcontinent) has been compared with that of the BC contribution to AOD obtained from MERRA-2 at the same location (Fig.S2). (Here I assume that the contribution of BC to total AOD is the Y-axis of Fig.S2; it is not clear from the text or figure caption). How to validate the columnar contribution of BC to AOD by comparing with the surface values of BC concentration? The inference from Fig.S2 is that the two values are well correlated (which is also expected), but it does not validate the BC contribution to AOD or provide information on the absolute accuracies (bias, slope, uncertainty). Further, the monthly mean values of MERRA-2-derived BC themselves show scatter of +/-0.01 to +/-0.02 (about +/-33% of the mean BC AOD) for the same surface concentration of BC (Fig.S2). Hence, Fig.S2 does not provide information on the absolute accuracies or reliability of the individual aerosol species in MERRA-2. Also, as seen from Fig.S2, the variation of surface BC concentration (varying in the range of 4 to 25 microgram/m3) with columnar BC AOD is linear. What about the non-linear effects due to large variations in effective aerosol single scattering albedo and large changes in multiple scattering contributions for such wide range of BC concentration?
2. It may be noted that MERRA-2 assimilates AOD estimated from from MODIS, AVHRR and MISR satellite data and insitu AOD observations from AERONET network. While MERRA-2 globally compares well with the AOD observations (especially over the marine regions), it cannot correct for the deficiencies existing in terms of the missing emissions (e.g., Buchard et al., 2017). More importantly, the satellite-derived AOD used in MERRA-2 was only from AVHRR till 1999, while the subsequent period has seen a major increase in the assimilated satellite data, including MODIS and MISR (see Fig.3 of Buchard et al., 2017). This will enhance the quality of MERRA-2 AOD data during the post-1999 period compared to the period before. The present study focuses on the period between 1980-2015 (Line no. 353). What is the bias or accuracies in MERRA-2 AOD during the periods before and after 1999? This is especially important when the AOD is apportioned into different chemical components.
3. Even if the reliability of the individual chemical compositions from MERRA-2 is acceptable and the comments-1 and 2 given above are ignored, the analysis carried out in the present analysis does not show that the observed increase in dry day frequency (DDF) is caused by aerosols. There is a clear association between the increase in DDF and AOD in several cases (and the individual contributions by some of the species). But this increase in AOD and individual chemical species can (and most probably) be due to increase in dryness, which increases the aerosol production as well as their residence time in the atmosphere, both of which contribute to increased accumulation of aerosols in the atmosphere.
4. For argument, let us assume that the increase in DDF is due to aerosols, as they can cause changes in clouds, radiation balance of the Earth’s surface and atmosphere as well as make atmospheric thermodynamical changes. In Region-1, during the long dry phase (LDP), about 67% of AOD is contributed by sulphates, while the organics, BC, dust and sea salt contribute ~14%, ~6%, ~8% and 5% respectively (this is a rough calculation made from the median values shown in Fig.3). On the contrary, during the short dry phase (SDP), the AOD contribution by sulphates reduces to ~53%, while the organics, BC, dust and sea salt contribute ~21%, ~7%, ~12% and 7% respectively. Overall, the BC contribution remains ~6-7% of the AOD. How does this compare with the aerosol chemical measurements carried out over this region, reported in the literature? The sulphate AOD increased by ~0.15 between the SDP and LDP in Region-1. How such major contribution of sulphates prevails and can contribute to long dry phases? How such a small fraction and weak increase (in terms of magnitude) of BC (with BC AOD of 0.014-0.025) can cause the atmospheric heating or radiation budget changes required for LDP (spanning for 2-3 weeks)? Note that such values of BC (often more, typically 10% by BC mass fraction) prevail over most of the Indian landmass (including southwest India).
5. Similar scenario prevails over Lucknow (Fig.4). The BC AOD fraction is ~6-7% for SDP and LDP. The increase in BC from SDP to LDP is only ~0.007 (here comes the accuracy of MERRA-2; is this outside the uncertainty limit?), while the AOD increase is ~0.11, of which the contribution of BC is only ~6-7%. Unlike in Region-1a, Lucknow (which is a city located in Region-1a) does not show an appreciable increase in sulphate AOD. Why is this feature distinctly different, when the aerosol residence time can be 3-7 days (during dry phases it can be even more). Observations reported in the literature suggest that the day-to-day variability or variations within a few days in AOD over most of the Indian region can be significantly more than ~0.1. Further, AODs in the range of 0.3 to 0.8 often prevails over Region-1 and at several other regions in the Indian subcontinent. The median value of AOD prevailing over Lucknow during SDP is ~0.27, which is rather clean. The important question is, can such a relatively small AOD or such small variations (reported here) in AOD or BC AOD cause short or long dry phases? If so, what are the thermodynamical changes or radiative impact produced by such variations? This is not studied in the present work. On the contrary, it is very much possible that dry spells can cause accumulation of aerosols to produce the observed variations.
6. Long-term trends in cloud occurrences shown here are very interesting. However, most of the aerosols being limited to the lower atmosphere, the increase in AOD (which is proposed to have produced the dry spells) should have first affected the low level clouds. In contrast, Fig.5 does not show any increase in low level clouds in Region-1 or 1a, but produced significant increase in high level and total clouds. However, the low level clouds did show a weak increase over Lucknow. Why only at Lucknow? Overall, the increase in middle, high and total clouds is much larger than that in the low level clouds. How does this happen? Is it possible that the increase in high level clouds and consequent increase in greenhouse warming has also contributed to the dryness occurrence? (The variations between low level clouds and aerosols itself can be an interesting study).
7. In Region-3, AOD is quite high (median values ~0.55) and comparable for SDP, MDP and LDP. However, based on the average or median values, the total contribution from the individual species (BC, dust, OC, sea salt, sulphate) contributes only ~60% of the total AOD. What are the other components that contribute ~40%? The increase in BC AOD (median) between SDP and LDP is ~0.004 while that of dust AOD is ~0.04. In contrast, sulphate (and OC by a negligible magnitude) has decreased by ~0.03 between SDP and LDP. Why is this contrasting behaviour and how does it compare with the Region-1 (though the precipitation mechanisms at these two places are different, why the role of aerosols is contrasting?).
8. I presume that the data on aerosol characteristics presented in Figs. 3, 4 and 6 are for the respective days of different dry phases (SDP, MDP and LDP). On the contrary, the analysis on AOD versus DDF shown in Fig.S7 (which is interesting) considers AOD during 16-31 July and DDF during 1-15 August. This may have a basis: aerosols may cause the atmospheric thermodynamical and circulation changes, which may result in dry spells subsequently. In that case, why the analysis shown in Figs.3,4, and 6 used simultaneous data for AOD and DDF?
9. In summary, this study shows that there is an association between the increase in dry day frequency and AOD (including some of the individual species) in some regions. As stated earlier, the increase in AOD can be a result of increased dry day frequency as well. This paper does not provide any evidence to show that the increase in dry day frequency is caused by aerosols. Even if it is so, the important question remaining is whether the observed magnitude of increase in aerosols (and individual components) is sufficient to produce SDP, LDP, etc.
10. On the contrary, the manuscript does not present the role of changes in atmospheric circulation pattern, atmospheric thermodynamics, radiation balance or surface temperature variations among SDP, MDP and LDP, all of which are expected to be important (all of which can be also produced by aerosols and other climate forcing mechanisms) in producing dryness and increase in its occurrence (Example, Raman and Rao (1981) on the relationship between blocking highs and droughts; Krishnamurti et al. 2010, etc).
11. Overall, the manuscript (text) is too lengthy for conveying the message presented in it. At the same time, some of the very important information are not provided: e.g., the individual species contributions (like BC, dust PM2.5, OC, etc) are the contributions to AOD, proper, incomplete Fig/Table captions (e.g., Fig.S1, S2,S4, S11 (even axes titles are missing), Table-S2, Table-S3, Tables 1,2 and 3). Readability of the manuscript has to be significantly improved.
1. Table-1: Why is the range of dry days for short dry phase (SDP) different for different regions? Same is the case with MDP and LDP.
2. In Fig.2(a), correlation coefficients between DI and DDF are positive. Then, why there is a decreasing trend in Fig.S4 (R1, M9; R2, M9; R3, M9; R1, M8; R3, M7, etc)?
3. Lines 245-250: What is the validity of this assumption when the region experiences intra seasonal oscillations and active and break spells?
4. Fig.S4: DI has both positive and negative values even when the number of dry days in a month is 29-30 (e.g., R3, M6; R3, M8; R1,M6). What could be the mechanism?
5. Lines 412-414: Regarding sulphate AOD - This is incorrect and against what is seen in Fig.3. It showed a distinct increase from 0.1 to 0.25 between SDP and LDP.
6. Lines 450-452: “The distribution analysis on total aerosol AOT shows much larger values over Lucknow than in region 1a and also the variability of the median values with the quartiles and whiskers are also far more deterministic here …”. This statement is incorrect as is evident in the AOD variations shown in Figs. 3 and 4. In fact, distinct increase in AOD (median and distribution) between SDP and LDP is better seen in Fig.3 (Region 1a).
7. Line-470: “hence the dependence of dry days can be primarily associated with urbanization”. This statement is not supported by the facts at this stage. Or refer to Fig.8 here while making this statement.
8. Lines 503-504: “Figure 5 reveals that region 1 has a weak but discernible increase from 5 to 15 days in last 60 years”. Is this correct? The mean trend line shows an increase from 9 to 13 days only. Similarly, the trend line for Lucknow shows an increase from 9 to 17 (and not 4 to 25 days as given in Line 511). The mean long-term trend over the 60 years and the scatter in the values (year to year variability) should be stated unambiguously.
9. Line 529: “… reduction in cloud particle size …”. This is not shown in the manuscript.
10. Lines 548-550: “Another periodicity is expected to lie at ~1-2 years which represents the year-year varying component of urbanization.” What is the year-to-year varying component of urbanization; how urbanization can have a periodicity of 1-2 years?
11. Unnecessary words may be avoided (words like ‘now’, ‘next’’, etc are used unnecessarily at several places; Also see usages like ‘methodically introduce’ (line-13), ‘crucial concern’ (line 28), ‘third and final’ (Line-171), ‘considerable conditions’ (line-225), ‘whether daily’ (line-257), ‘a set of various components’ (line 275), ‘over mentioned regions’(subtitle, Line 290), ‘all throughout in 1a’ (line -335), ‘which demand primary importance throughout the study’ (lines-339-340), ’ENSO oscillations’ (line 355), ‘previous attempts taken’ (line-357), ‘the distribution of total aerosols start increasing‘ (Lines 368-369), ‘rain cloud’ (Line 399; avoid ‘rain’ here, as the cloud burning effect due to aerosol absorption is possible for non-precipitating clouds as well), ‘an attempt has been progressed over Lucknow… ‘ (Line 477), ‘slant rise in dry days’ (Line 621), ‘rlilpl’(Line 675), ‘A series of investigations are progressed which infer …”(Lines 722-723). This is not a complete list.
12. Missing reference: Tyalagadi et al. 2015
13. Lines 329-331: What does this mean?
14. Line 430: Modify as : “… 0.542, 0.129, … and 0.124 for BC, dust … and sulphates respectively”. Similar is the case in all places where MLR coefficients are given in this manuscript.