I have not managed to get anyone very closely associated with the model to do a review. Some claimed conflict of interest and others were too busy. So, I wonder if you could get a user of the model who is not an author to look through the paper for any minor errors, and to indicate any places in the manuscript where the clarity could be improved? The thing is that I am having trouble believing than any of the peer review team have checked the many details in this paper thoroughly, and yet I am aware that it is exactly the publication of these details that is so valuable to the community. Therefore, it is really important that we do our best to minimise the possibility of any errors.

Although the code is not publicly available, I must hold you to the same peer review standards as other papers submitted to GMD. I requested the code and was granted access. This consisted of 3 separate downloads (Socrates, Jules, the UM). Socrates contained instruction on how to compile into the UM, but there were no similar files in Jules or in the UM. In order to have a chance of attaining scientific reproducibility, you need to provide instruction for how to compile exactly the version of the coupled code described in the paper. Supplements are not archived to the same standard as the main text of the paper so this would probably be best presented as an Appendix. However, I understand if you prefer to add it to the technical information in the supplementary information and then reference it in the main body of the text (perhaps in the code availability section).

Thanks for taking the time to find a user/developer to peruse the manuscript. I'm happy to accept the paper now.