Dear Authors,

The referees have given their feedback on your manuscript, with 2 referees raising substantial criticism and insisting that publication of the manuscript in the present form is not possible. They recommend major changes for reaching publication quality.

While your investigation and the results are certainly interesting for the readers of AMT and especially the remote sensing community, I agree with the more critical reviewers' views that there remain major methodological questions which need to be addressed before the science case can be made. Likewise, the manuscript lacks clarity and is sometimes difficult to follow. Significant changes to the paper are therefore required before publication in AMT can be considered. Consequently, I decide that the paper can be reconsidered after major revisions.

Pleaase take all referee suggestions into account when improving the paper. Scientific significance and scientific quality are two core criteria for publication in AMT. It is therefore important that the methodology is improved and well explained. As outlined by the referees, there are some major questions to be addressed:

1) How representative are the data that you show and do they support the conclusions? Why has the investigation been restricted to only two days at two different sites? Why does one day only cover less than two hours of measurement? Would observations at other network sites lead to the same or similar conclusions? Your study would gain much from including more sites and extending the period of observation and this is urgently recommended. It is difficult to understand why the investigation does not include more sites and a more significant time span in the first place.

2) It is claimed that 'significant' biases between databases are observed, but this conclusion is partly due to ignoring database uncertainties. If I am not mistaken, methane line intensities in HITRAN already have 20% relative uncertainties, for example. Given these high values, it is rather surprising that the agreement between databases is better than a few percent. A comparison of databases, such as in this manuscript, certainly needs that related uncertainties are discussed. Moreover, a bias is significant when it exceeds 2 or 3 standard uncertainties (depending on the desired probability level). Given the TCCON sigma values in Tables 4 and 5 and assuming a similar sigma for the other databases, the observed differences are not statistically significant even if the intensity uncertainties in the databases are neglected. But this is not obvious without attributing actual uncertainties to the observed bias values.

3) While the discussion of the 13-CH4 isotopologue takes a lot of space despite the conclusion that meaningful isotopologie retrievals cannot presently be made, it is not discussed quantitatively how much the inclusion of the new TROPOMI window possibly improves the isotopic retrieval and how the very different averaging kernels impact on determining the delta.

4) Presentation and discussion of the CH4 bias sensitivity (towards other parameters, such as water column, SZA, ...) is lacking clarity.
- Table 8, for example, displays a coefficient of determination and not a linear relationship and I would expect that it is R^2 between the bias (y) and the vapour column (x) (and not vice versa). The use of R^2 as a figure of merit should be explained in more detail. While a value of R^2 > 0.5 indicates that a variability of a parameter y can primarily be explained by the variability of parameter x, a value of 0.2 still indicates that y somewhat depends on x and that there is an interference.
- What does it mean when you write (p 18) that relationships in Table 8 are unlikely to be purely down in water vapour?
- The conclusion at the end of the paragraph is also not clear and the claim that 'As with Table 8 the results from window 1 show the weakest relationships' is not entirely true. It seems that TCCON window 3 has almost always the lowest R^2, indicating that it is most consistent with TCCON window 4 - SEOM window 1 comes next.

How do you distinguish between systematic and undesirable correlations (due to spectroscopy, for example) and accidental correlations (spatial variation of the CH4 column that possibly correlates with SZA, etc.)?
Finally, how do you explain that the bias depends on the water mixing ratio? Is there any hypothesis?


5) The overall presentation should be improved.

Besides these points that need thorough consideration, there are a few minor comments that should be looked at:

- The abstract talks about specific TROPOMI windows (plural form). In the text, there is only one such window, apart from the fact that the same window is given a different name when it concerns another isotopologue.

- Bias values such as in Tables 4 and 5 should be given with sign and with uncertainties (standard deviation or error of the mean). Otherwise the numbers are difficult to interpret, especially since the numbers of digits shown falsely imply a very small uncertainty.

- Why has there been only one hour of measurements been selected for Figure 3? Are there no other measurements on the same day? For studying a SZA dependency days with better coverage would be more favorable. Figure 3 also is the only instance in the paper, where windows 1 to 4 are designated by the centre wavenumber. This is confusing. Also why do many measurements come in pairs with gaps in between?

- Tables 4 and 5: it seems that GEISA yields the highest consistency in the inter-window comparison. This is an interesting point that requires diccussion.

- Section 3 lists a lot of results, but very little conclusions are drawn from the derived quantities. Often the message is missing. What does it mean when you write 'Windows 5 and 6 vary in their sensitivity, with the different spectroscopic databases showing similar sensitivity in window 5, but very different ones in window 6.'?

- I am not convinced that there are 'significant' differences between different windows used for the retrieval of methane isotopologues.
Taking sigma_window of ~ 1.5 - 4 ppb for 13-CH4 as an uncertainty of measurement, an uncertainty on the order of 75 - 200 ‰ can be derived. This is not so much different from the observed variability in Table 7. Again, your definition of significance needs to be explained.

- One of the delta values in Table 7 is not compatible with the bias values shown in the Tables. Please check all values carefully.

- Explain the choice of the 4th TCCON window as a reference. Wouldn't it be more consistent to use some average of TCCON windows as a reference? Or can the other windows be neglected as they are much narrower? The reader would likely be interested to know how each of the three windows contributes to the overall CH4 data product.

- The comparison of Tables 8 and 9 should not start by mentioning a close agreement, because the Ascension data from August are showing a quite contrasting picture. It is striking that the discussion does not discuss this difference.

- The correlation plots C9 etc. show discontinuities, most prominently for the 13C containing isotopologue. This is not really discussed.

- Why is the Tsukuba April 2016 the reference in section 3.5.2 and Appendix C ? It has the smallest number of data points.

- One of the most remarkable results is that inclusion of window 5 seems to reduce the uncertainty of 13CH4 retrieval. Could this improvement somehow be quantified?

- Insignificant digits in Table 4 should be removed.

Dear Authors,

The reviewers have judged the quality and the topic of your paper sufficient to be published in AMT. However, they have raised some important points that need to be considered before publication. The paper can thus be reconsidered after major revisions, which should address these issues in detail.

Dear authors,

I am happy to announce that all referees recommend your article for publication. Unfortunately, the current version does still contain too many glitches and phrases that are difficult to understand. I therefore ask you to correct the following list of remarks and suggestions. Please highlight any changes with respect to the current version. Since there might be oversights from my part, I also would like to ask you to carefully reread your corrected manuscript for further errors. Once you have updated the manuscript, the article can proceed to the next stages.

With kind regards

Christof JANSSEN

L 17 : suggesting that further
L 20 : verb missing : 'indicates different' instead of 'indicating differing'
L 24 : Delete phrase With the aim of
L 25 : ‰ or % ?
L 27 : What does TEXT mean ?
L 79 : similar studies in the past
L 81 : replace 'seasons. We' by 'seasons we'
L 84 : write 'unlikely to be affected at the'
L 87 : replace '(see Eq. 1)' by 'which is defined as' and move Move eq 1 directly behind
L 94 : Replace 'This is a quantity that has been used' by 'd13C has been used'
L 94 : Do you mean global studies ? If not, than I would prefer to remove this word here.
L 127 : specify 'variations' as 'variations (standard deviations)'
T 2 : For clarity, please replace ±sigma with ± 1 sigma
T 2 : Check unities. They are missing on the sigma values. The best is to add '/ ppmv', etc. in the table header
L 149 : remove 'attempt to'
L 156 : Phrase starting with 'Combined' is incomplete. Join phrases ending and starting in this line
L 168 : remove 'below' as Tables can flaot.
L 172 : Correct grammar. A verb is missing in subordinate phrase starting with 'where'
L 177 : The database included with the updated GGG2020 software, referred to in this study as GGG2020, which includes numerous updates to the GGG2014 spectroscopic parameters. -> The database included with the updated GGG2020 software which includes numerous updates to the GGG2014 spectroscopic parameters and is referred to as GGG202O in this study.
L 180 : 2020 is the current release, write HITRAN2016 release instead of current release HITRAN2016
L 182 : lines shapes -> line shapes
L 189 : differentiate with them -> differentiate between them
L 191 : Some work has been performed previously comparing spectroscopic databases -> Some work on comparing spectroscopic databases has been performed previously
L 202 : parameters, the remaining -> parameters. The remaining
L 205 : speed dependent -> speed dependance
L 206 : Please indicate the spectral bands used by Mendonca et al. The utility of using these parameters likely depends on a particular band.
L 210 : We note the -> We note that the
L 217 : Make sure the equation terminates wih a semicolon, not with a full stop.
L 218 : Remove full stop.
L 221 : Please sepell out w.r.t.
L 229 : impact -> impacts
L 233 : parameters 1) -> parameters: 1)
L 236 : all trace gas fitting is unclear (fiting of all trace gases ?). Please rephrase
L 240 : Please refer to line strengths or line intensities and not cross sections
L 250 : use 'on these bias and precision values'.
L 251 : instead does not seem to make any sense her. You could use 'still' here
L 252 : Remove Fundamentally
L 254 : Insert full stop after '2018, 2019)' and use capital I in in-situ
L 258 : Please explain 0.02 ppb.
L 278 : is more complex region -> covers a more complex region.
L 296 : Start new phase before 'this is likely'
Figure 2 caption. Remove blue colour.
L 303 : write out w.r.t.
L 305 : Firstly without a secondly is considered bad style. Please replace further in line 307.
L 314 & 315 : remove all commas
L 323 : Write Fig. 8
L 323 : Phrase unclear. Do you mean : We also find that the Xair value indicates the retrieval quality with only ... ?
L 325 : see comments on L323 & 303
L 326 : Norm Abs Mean has been abbreviated and defined differently in Eq 4. Please use consistent notations (NAmean)
L 327 : This has been sais before. It is sufficient to write '... reference retrievals. Thus we assume any ...'
L 330 : Use consistent notation Norm Abs Mean. I propose to use NAM throughout, but corect Eq 4 as well.
L 371 : write 'knowledge of spectroscopic parameters in window 1 is not as good as in other windows'
L 375 : replace 2017), however these' by '2017). However, these'
L 377 : Replace 'The implication being' by 'This implies'
L 389 : 'We therefore fit' should start a new phrase.
L 407 : 'and 12. Where' -> 'and 12 where'
L 411 : 'to be the presence' -> 'to indicate the presence'
L 411 : 'First considering' -> 'We first consider'
L 414 : relationship in singular
L 415 : add missing 'of'
L 423 : w.r.t.
L 425 : shows -> show
L 427 : much larger than for/in any other case.
L 427 : write 'especially in the Ny-Alesund retrievals'
L 431 : Fig.
L 434 : The sentence starting with 'with' is gramatically incomplete.
L 439 : remove variation
L 444 : Rephrase, eg Given that both Tsukuba and darwin have large variations ...., it is interesting that ...
L 457 : suggests -> suggest
L 481 : biases, the -> biases. The
L 482 : results, however -> results. However,
L 483 : The other sites, -> The other sites
L 486 & 487 : Three consecutive uses of 'clear'. Please improve
L 492 : We note when -> We note that when
L 493 : indicating these results -> indicating that these results
L 494 : Start new phrase efter combinations.
L 497 : write out w.r.t.
L 500 : the expected result. -> expected.
L 502 : Eq 1), -> Eq 1)
L 505 : write out wrt
L 507 : Tables 4 and 5 indicate
L 508 : Better write 'GGG yields a mean uncertainty of 13CH4 retrievals between ...'
L 510 : by ~x200 -> by about 200 times
L 513 : delete retrievals
L 513 : at remote -> of remote
L 515 : , considering -> and considering
L 516 : The phrase is confusing. Please restructure
L 519 : (at least 2x any other -> (at least twice than for any other
L 537 : Reconsider the phrase starting with 'Overall' as it is not clear.
L 539 : , however, Window -> . Window 4, however,
L 542 : Remove 'in this paper' (we are not at the conclusion of the paper yet) and terminate the phrase after biases.
L 543 : however it -> However it
L 544 & 545 : you probably meant to write 'due to simultaneous variation of a number of ...', Delete simultaneously
L 594 : delete in each case. This seems redundant.
L 610 : analysed is mispelled
L 610 : Terminate phrase after databases.
L 603 : too many commas
L 607 : write out w.r.t.
L 610 : suggesting that TCCON ....
L 611 : bias to -> bias with respect to
L 614 : suggest the GEISA -> suggests that the GEISA
L 620 : show - shows
L 620 : write out w.r.t
L 623 : remove comma
L 624 : Raplace coma by full stop.
L 627 : Please rephrase. What do you want to say here ? What is a unique formation ?

Dear Authors,

Thank you for complying with the suggested changes. I would like to congratulate for your manuscript finally being accepted for publication in AMT. Please prepare the documents for the editorial team by applying the four following technical corrections:

L 92 : move the phrase "where VPDB refers to Vienna Pee Dee Belemnite, an international reference standard for 13C assessment." directly behind eq 1.

Table 2 : Replace ° by °C

L 224 : Remove the whole line.

L 504 : remove the Guillemet before the < sign.