Dear authors,

Thank you for the detailed responses to the comments of the two reviewers. In general I find the responses adequate and to the point. However, I am not satisfied with the response to one key point raised by the second reviewer (major point 2). The issue is that comparing EM27/SUN data analyzed with PROFFAST versus TCCON analyzed with GFIT is not an apples-to-apples comparison.

The reviewer has a very valid point here. Although differences between individual processing codes are interesting, this paper is about comparing different instruments and especially the influence of low-resolution spectroscopic measurements on the consistency with the high-resolution reference TCCON-data. I understand the importance of comparing standard COCCON measurements (measured with EM27/SUN instrument and processed with PROFFAST) with TCCON, but I also find it very important to demonstrate the impact of lower resolution spectroscopy on the results. In the current paper, instrumental effects cannot be disentangled from effects caused by lower resolution, and some differences are simply caused by different processing algorithms. The manuscript would be much stronger if these effects were separated.

Comparing the HR125HR with the HR125LR results obtained with the same instrument would allow isolating the effects of spectral resolution, but unfortunately, the HR125LR results were obtained with a different processing algorithm. The HR125LR data should be produced with GFIT to maintain consistency with the HR data and the other retrievals. Furthermore, the effect of changing from GFIT to PROFFAST should be demonstrated for the EM27/SUN instrument.
A viable way forward would be to first show only results obtained with all instruments using GFIT, and then in an additional section show the differences between EM27/SUN data processed with GFIT and PROFFAST.

I also support the reviewer's suggestion to discuss the HR125LR together with the comparisons with the other LR instruments (minor point 3 of the referee). In the original paper, comparisons with HR125LR were presented in a separate section, which made the paper unnecessarily long. In the revised version, the comparisons between HR125LR and HR125HR and the comparisons of HR125LR with the other LR instruments have simply been moved to Appendices C and D. However, as mentioned above, the comparison between HR125HR and HR125LR is of particular interest (both processed with GFIT) and should appear in the main body of the text.
Before making any further changes to the manuscript, please let me know how you plan to address these points.
Best wishes
Dominik Brunner

Dear author,

Thank you very much for the revised paper, which got close to being acceptable. However, one of the reviewers is asking for a few further clarifications as listed below and in their comment. Although I consider this being rather a minor than a major revision, I am formally asking for another major revision because I would like to have the possibility to discuss your response and modifications to these rather technical (but in my view relevant) points with the reviewer one more time.

Kind regards
Dominik Brunner

Points that need further clarification:

1. Several other groups successfully process the CO channel in the EM27/SUNs using the
EGI/GGG software. It is unclear what the problem is here, but the authors could reach
out to other EGI/GGG users for help in processing the CO channel.

2. You state that “The intraday scatter generated by GFIT is noticeably higher than that
achieved with PROFFAST.” You cite a private communication with Niki Jacobs to confirm
this observation. However, from what I understand, the PROFFAST software typically
averages forward and reverse spectra, thus decreasing the scatter. Has this
forward/reverse averaging been done here in the PROFFAST retrievals in this
comparison? GGG does not typically perform this averaging.

3. You show the xair diurnal variability for the PROFFAST and GFIT retrievals in Fig 13. The
GFIT retrievals show what you call a bifurcation. Is this bifurcation between consecutive
forward and reverse spectra, or does xair vary smoothly by time of day? If the latter, this
could indicate that the ZPD times are listed incorrectly in the GGG runlog. Please check
and fix the error. Also in this figure, are forward and reverse scans averaged for the
PROFFAST results? It would be appropriate to show (both in Figure 13 and 12) results
that have been filtered identically and averaged identically.

Dear authors,

I am very pleased to let you know that I accept your manuscript for publication "as is".
I would like to thank you very much for hard work on responding to all reviewer's questions and for having resolved all issues and improved the paper significantly until its final form.
This is an important publication!

Best regards
Dominik Brunner