Thanks for the revised manuscript. Referencing the line numbers for edits to the paper would have enabled a quicker response.

Nearly there I hope. A few comments.

1.
Reviewer:"9. Figure 1: How do we define polar amplification? as a mean of ratios or a ratio of means? 15 Hind et al., Scientific Reports (2016)."

Response:"How to quantify polar amplification in a multi-model ensemble is no doubt important for some applications but we do not think it is necessary to include it here. We prefer to keep our message simple and use this figure to show that polar amplification is robustly simulated in response to increases in CO2."

I think this is incorrect. There is no way of analysing an ensemble without adopting some method to do so. As someone who analyses ensembles, I'd very much like to know the most meaningful metric for the polar amplification community. I think what you have done for the Figure is actually something different from either of the two suggestions of the reviewer. The figure shows a mean of normalised fields, so if you took the polar to equator ratio it would be a ratio of normalised means...?

2.
The general idea is that a modeller, armed with a state of the art climate model and a nice big computer, should be able to set up and run the experiments using only the paper and permanently archived open access resources referenced in the paper. I think you are quite close but...

A. The Data request section is a bit difficult to understand. A modeller needs to know what the minimum set of requirements is. I think this might be the DCPP set plus Table 3, but I am not sure. Please clarify.

B. This is probably my stupidity, but I like to think that the aforementioned modeller should be allowed to be almost as thick as me. I could not find the boundary conditions for the experiments on input4mips. Either some more details need to be included in the appendix to help the modeller find what they need, or you need to upload the data to the database.

Thank you for your revised manuscript.

1.
Your response to the question about the definition of how to define polar amplification,
"there is also diversity in the literature about whether polar
amplification is defined by polar warming versus the global average, or versus the
tropical average etc; as a ‘ratio of trends' or a 'trend of a ratios'; which variable to use
(surface temperature or something else); whether it is appropriate to calculate
'regional amplification' based on a limited longitude range, etc. We do attempt to
resolve these issues in this paper, which is intended to document the protocol for
multi-model experiments. "

I am a bit confused, but assume you meant to write "do not" rather than "do" in the last sentence above...? If I understand that correctly, my response is that in a paper devoted to the comparison of polar amplification between models it is necessary to define what you are talking about. Therefore, please can you add a paragraph which discusses the different definitions, with references to the literature.

2.
Boundary conditions for model experiment description papers in GMD must be available before acceptance of the manuscript. If you are not able to get them on to input4mips in a reasonable timeframe, please upload them to an alternative public repository and provide a DOI for the upload in the revised version of the manuscript. (Ideally, a model experiment paper includes evidence of model output which demonstrate that the experiments work as expected. Thus the boundary conditions should actually have been finished - fully tested and ready to go - prior to submission of the manuscript!)

When I went to the Zenodo page for the boundary conditions data I saw, "You may request access to the files in this upload, provided that you fulfil the conditions below. The decision whether to grant/deny access is solely under the responsibility of the record owner."

The data should be made fully publicly available. There is little point promoting an experiment for the climate community to use if the boundary forcing data is not available! Is there a reason why you cannot make the data available?

It is appreciated that you make the point that the version on input4mips may not be identical to the version discussed in the paper. However, rather than state "Preliminary forcing data for testing " ... are available on Zenodo, you should clearly state that the exact version of the boundary forcing discussed in the paper is the one that is available on Zenodo. Your wording suggests that you are not confident that the data are correct. This is concerning and I am inclined to hold the paper until the boundary conditions are finalised. It is generally required that experiments described in GMD papers should have already been performed by at least one model. Please can you confirm whether this is the case for your experiments.

Whatever the answer to the above, it would be appropriate to create a version number for the current boundary conditions and to refer to this in the paper, and also on the Zenodo page. This is the norm for generic GMD experiment description papers, but it has not been enforced for all the CMIP6 experiments. However, as you seem to have a high expectation that the data will change, I think it would be a good idea here.

Thank you for the revision. The paper is now very close to being acceptable. The only remaining issue is the accessibility of the data for the boundary conditions.

Despite being hosted on Zenodo, the data are restricted with access only being provided upon approval by the first author of the paper. I have exchanged a few emails with the first author and it appears that the authors are choosing not to make the data freely available, because they are concerned that it may be misused. Unfortunately this is not a valid reason to not make data available at GMD. We make our code and data open access in the same way that we make the scientific content of the paper open access, and the publication is incomplete if the code and data are not provided. Technically, there is a very simple solution to this problem and this is to remove the requirement for access to the data to be approved at Zenodo. At the same time, more information about the data should be added to the Zenodo page, as at present it is rather confusing and off putting. It seems to suggest that the data are restricted to those who fulfil certain criteria but does not state what those criteria are! It is fine for authors to keep a record of who accesses the data by requiring registration, but access should then be automatic. I tried to gain access myself, and the link I was sent did not work, and further human intervention was then required. I do not like such a system as I have quite a lot of experience with attempting to gain access to code or data that is supposedly available, but that requires human intervention. It is still too common in the community in general that requests for access are ignored.

I am happy to report, however, that I did finally gain access and that all the files were easily readable in netcdf format, and that the data look basically reasonable.

If the boundary conditions change before they are uploaded to the CMIP input4mips database, the authors should write a short update paper outlining the changes. They could also require that people running the experiments state in the data upload which version of the boundary conditions they used.

Note that every paper in GMD is assessed on its own merits against the peer review criteria, and not in reference to other papers already published. Special issue papers in GMD are reviewed in the same way, under the same criteria, and by the same editors as all other papers in the journal. Thus the authors should simply be aiming to publish an excellent GMD experiment description paper.