Thanks for preparing a thorough and complete response to the referee comments. I'd like to pass the revised manuscript back to them for their consideration before publishing.

It's a little difficult to get a full impression from the track-changed version, but I think it would still benefit from a final copy-editing to tidy up the language - the referees could skip such issues and concentrate on the scientific content

thanks for your comprehensive responses to the Referee comments. They suggest a number of technical corrections and one minor matter of content (referee#1 suggests that a few sentences on how the flow approximation might impact the results are included in the discussion). If you could address these then we'll be good to go.

Figs 2 and 7: panels are labelled a=G4000, b=G750, caption says they are the other way round. It would be worth checking the labelling of panels in figs 3, 12 and 13 too, although their captions do not say which should be which and they look right to me.

line 9,11,12: in an abstract, for clarity and reproducability in whatever system may be indexing it, I would write out "approximately" instead of using "~", and "below" instead of "≤". The typesetter may disagree

line 7: it would be clearer if this had something up front to clearly explain that you're talking about a climate forcing for those who might be unfamiliar with climate model names, eg "We run the simulation based on the ISMIP6 core climate forcing from the MIROC5 AOGCM [...]" (although neither of the acronyms GCM nor AOGCM have been explained in this abstract, and should be if you're going to use them)

----------optional

Reviewer 1 commented
Line 450: The grid dependence of the different models must also depend on the flow approximation in the models [...] I suggest that a few sentences on the flow approximation is included in the discussion.

and you replied
We add a short paragraph on this topic [...] However, we kept this paragraph very general

This addition isn't really a paragraph if we're honest, and I for one would find it really useful if you documented some ideas about how different approximations might behave in different ways as their grid resolutions change. After all, sensitivity of the modelling to grid-resolution is the primary topic of your study and if you can help more people see how your results might relate to their modelling setups then that's good for everyone. I'll leave it up to you to decide whether to expand any further here.