|The revision is an improvement over the original manuscript and is close to being publishable. There are many grammatical errors and confusing passages as well as an overabundance hyperbole and misleading statements. There are three sections that contain too much description of unnecessary and uninteresting details, two of these (Sec. 3.1 [up to Sec. 3.1.1] and 3.2 [up to Sec. 3.2.1) should be condensed (Sec. 3.1 could easily be summarized in a paragraph) and the third (last two paragraphs of Sec. 3.1.2) could be deleted along with Fig. 6. The most serious and important criticism is that the manuscript ignores (or barely acknowledges) how the design of the CLaMS experiments affect the interpretation of the results. |
The first of these criticisms concerns potential weaknesses of CLaMS. Transport models, such as CLaMS, have many weaknesses that affect scientific results; these models depend on reliable data to initialize constituent concentrations, reliable wind data to properly transport air parcels, and process models to alter constituents along parcel pathways. A major problem with CLaMS is its dependence on resolved winds from ERA-interim to simulate convective transport. It is possible to mitigate the impact of these weaknesses with careful analysis (for example, through time and ensemble averaging), but they cannot be simply ignored as is the case in this paper. What is particularly troubling about this paper is the statement (line 8, page 13) that suggests that deficiencies in MLS CO data could lead to the disagreement between CLaMS and MLS. There is no doubt that observational error contributes to the disagreement, but to ignore the impact of model error and, therefore, imply that only observational error is at fault is worse than misleading. The authors must discuss how model weaknesses might be affecting their results.
The second criticism concerns an issue I mentioned in my original review, that the authors have addressed, but not with enough substance. The issues concerns the ability of CLaMS to faithfully represent regional tracer concentrations (Fig. 8) in the anticyclone during summer when those tracers were initialized only after May 1st while transport times from the boundary layer to the 380 K isentrope can exceed 90 days. The problem is nicely demonstrated in Fig. 3 of the response to review #1 (which should be included in the manuscript) in which it is shown that the boundary layer contribution to air in the anti-cyclone does not exceed 50% until early August. That, however, is only part of the problem. Boundary layer air from different regions have different transport times. So, this ‘spin-up’ problem (as I called it in my previous review) affects different regions differently – so that the relative contributions from different regions are not faithfully represented in Fig. 8. In the revision, the authors do mention the fact that regional tracer concentrations do not account for air parcels in the boundary layer before May 1 and they do acknowledge that it takes time for concentrations of boundary layer tracer to build up, but that is not enough. They must explicitly acknowledge that this phenomenon affects the interpretation of their results and, if possible, discuss how it affects their results. If this was my work, I would perform an additional CLaMS experiment that starts 3 months earlier to see if the results change form the May 1 experiment and I would examine boundary layer to 380 K transport times from the different regions – regions with longer transport times will be more affected by model initialization.
Note, the line numbers are messed up on my copy of the manuscript. I used the number as marked until the numbers on the manuscript reset to 0. At that point I continue using numbers that are not reset. For example, the last line on page 17 is labeled number 5 – I count that as line number 34.
Page 2, lines 16-17: Remove the clause ‘and is therefore more complex than hitherto believed’. One would have to be ignorant of atmospheric research from the past 50 years to not believe dynamics of the anti-cyclone to be this complex.
P.2 L.21-23: Remove the sentence ‘Our findings … the Asian monsoon anticyclone.’ This is a trivial result. Also ‘memorized’ is not grammatically correct in this context. ‘Imprinted’ or ‘recorded’ would be grammatically correct.
P.2 L.24-27: The sentence ‘Air masses … of the anticyclone’ should be rephrased. It is awkward and confusing.
P.6 L.9,11,14: Change ‘Sect.’ to ‘Sec.’
P.7 l.29-30: Reword the phrase ‘tracer-tracer correlations were used similarly as for the initialisation besides CO’. Its meaning is unclear.
P.8 L.5: Regarding the sentence ‘Water vapour is replaced by ECMWF water vapour in lower model levels’. I assume you mean that you use ECWMF water vapor for lower levels because satellites do not observe those levels. Regardless, rephrase to make it clear what you mean.
P.8 L.11: Remove ‘further the’.
P.8 L.15: Remove ‘to quantify’.
Sec. 3.1 [P. 9 L.24 to P.12 L.7]: This section contains a lot of description of unnecessary details. Please consider replacing most of the content with a single, to the point, paragraph.
P.9 L.10-11: Regarding ‘At each time step of the model (every 24 hours)’. The CLaMS model time step can’t possibly be 24 h (probably closer to 20 min).
P.10 L18: Replace ‘here’ with ‘when’.
P.10 L.20: The sentence ‘On 12 September 2012 conditions during the late-phase of the monsoon and on 7 October 2012 the situation after the breakup of the anticyclone occurring end-September are shown’ is awkward and seemingly has no important content.
P.10 L.23-25: Rephrase the sentence ‘The northern … equatorial easterly jet’.
P.12 L.21-22: Regarding the phrase ‘are normalised so that the maximum value of each trace gas is equal one’. You can delete this phrase. It is not necessary to normalize by the maximum value; the correlation calculation does its own normalization (by the standard deviation – after the mean is removed).
P. 13 L.4: Regarding ‘These high correlations’. What constitutes a high correlation depends on context. Here, since the fields are supposed to be identical – the correlations should be 1.0. The model data only explains 50-65% of the observed variability. It would benefit you to examine correlations between weekly or monthly averages to see how the correlation is affected by data sampling.
P.13 L.8: Regarding ‘Reasons for that could be deficiencies in MLS CO’. Please do not place all of the blame for the disagreement on observational error. If you mention problems with the observational data, then you must also mention problems with the model (which are substantial and their impact on your results should be acknowledged whether or not you mention observational error).
P. 13 L.14 and L.16: ‘A remarkable good correlation’ and ‘a very good proxy’ are both overstatements (two examples of the hyperbole mentioned above). Statements regarding the utility of the model as a proxy should be confined to those that help the reader understand which aspects are well represented by the model, which aspects are not, and why. In this context ‘very good’ is not specific enough and, therefore, meaningless. Also, while the correlations are important, whether or not a proxy is good can only be tested using predictions (e.g., via linear regression) and how good those predictions are depends on the context. There is no context given in this manuscript from which to judge the quality of the proxy.
P. 13 L.22: Change ‘the the’ to ‘to the’.
P.13 L.26: Remove ‘also’.
P.13 L.26: Change ‘correlation’ to ‘correlations’.
P.13 L.31: Change ‘similar as for’ to ‘similar to’ or ‘as were’.
P.14 L.11: Change ‘increases’ to ‘increase’.
P.14 L.13: The wording of ‘the missing convection in Asia occurring during’ is awkward. Suggestion: ‘the lack of convection in Asia during’
P.14 L.28: ‘very good proxy’ is an over statement – both that statement and ‘good correlation’ are meaningless without a well-defined context.
P.14. L.26 – P.15 L.20: I would remove these paragraphs and Fig. 6. They contribute nothing meaningful to the paper.
Sec. 3.2 [P. 15 L.23 to P.18 L.25]: This section contains too much description of unnecessary details. Please condense.
P.15 L.28: Improper use of ‘in principle’. Consider a change of wording such as ‘show primarily the same …’ or ‘primarily show the same …’ of ‘show, in general, the same …’
P. 16 L.12: Change ‘are in addition found’ to ‘are also found’.
P. 16 L.15: Change ‘in the same order’ to ‘of the same order’.
P.16 L.17-19: Change ‘Similar as for 1 July 2012, on 2 August 2012 in the mid-phase at 380 K (Fig. 7, 2nd row, left) the distribution of the emission tracer for North India is also confined within the Asian monsoon anticyclone’ to ‘Similar to distributions on 1 July 2012, the distribution of the emission tracer for North India on 2 August 2012 in the mid-phase at 380 K (Fig. 7, 2nd row, left) is also confined within the Asian monsoon anticyclone’.
P.16 L.20-21: Change ‘Contrary to the 1 July at the beginning of August, the highest percentages of emission tracers within the anticyclone are …’ to ‘In contrast to 1 July, the highest percentages of emission tracers within the anticyclone at the beginning of August are …’
P. 16 L.29: Change ‘Sect.’ to ‘Sec.’
P.17 L.7 (actually, the first line): change ‘having large contributions’ to ‘which have large contributions’
P.17 L. 27: Change ‘parcel’ to ‘parcels’.
P.17 L.29: Change ‘evident in the strong gradients’ to ‘as evident in the strong gradients’ or ‘as evidenced by the strong gradients’.
P.17 last line: Change ‘A subsequently isentropic’ to ‘The subsequent isentropic’.
P.18 L.22: The use of ‘flooded’ is great imagery, is provocative, and has stylist power. Unfortunately it is another example of hyperbole. Please change it.
P.19 L.21: Change ‘tracer’ to ‘tracers’.
P.20 L.11: Change ‘In principle’ to ‘In general’.
P.21 L.7 (second line): Change ‘However then’ to ‘However, when’.
P.21 L.14: Change ‘Sect.’ to ‘Sec.’
P.21 L.12-23: This paragraph might work better at the beginning of the Section.
P.21 L.24-33: This paragraph should be replaced with a discussion that explicitly acknowledges that the initial condition of this experiment (i.e., starting on May 1) might affect the results. I recommend the following: include Fig. 3 from the response to Review #1, an additional CLaMS experiment that is initialized earlier in the year (e.g., Feb. 1), and calculate transport times for the different regional tracers as a way to estimate the impact of initial conditions on the results.
P.22 L11-12: Remove ‘remarkable’
P.22 L.18-21: Remove the sentence ‘The good agreement … Asian monsoon anticyclone’. Your work does not predict values of PV using tracer distributions and so you do not provide enough proof nor define a context in which to claim that tracers are a good proxy. Furthermore, you don’t use those distributions as a proxy, which makes the statement unnecessary.
P.22 L25-27: Change ‘After the breakup of the anticyclone in early-October, still high percentages for the emission tracer for India/China are located over India, China and the Pacific Ocean’ to ‘High percentages for the emission tracer for India/China located over India, China and the Pacific Ocean are still found after the breakup of the anticyclone in early October’.
P.22 L.28: Change ‘In this paper, we could answer the question of what is the impact of different boundary layer sources in Asia to the Asian monsoon anticyclone in summer 2012’ to ‘In this paper, we address the question: what is the impact of different boundary layer sources in Asia to the Asian monsoon anticyclone in summer 2012?’
P.22 L.31: Change ‘have an impact on the composition of the Asian monsoon anticyclone in contrast to all’ to ‘have a larger impact on the composition of the Asian monsoon anticyclone than all’.
P.23 L.22: The statement ‘Therefore the processes are more complex than hitherto believed’ does not follow logically from the previous statements and, furthermore, is undoubtedly wrong. Please remove it.
P.24 L.8 (first line): Change ‘flooding’