This manuscript describes a regionally refined (Arctic) version of E3SM and compares to the RASM model both in JRA forced sea ice-ocean configuration. Overall this is well-written and easy to understand. However, I do have some thoughts here regarding the details of the model and also the originality of the work. I will also note that the categories of "Scientific Significance" and "Scientific Quality" do not really apply here. I believe that the manuscript only provides information about model performance with an without regional refinement and really says nothing about Arctic science. This is probably fine for GMD, but why are these a part of the evaluation?

1. My first major comment is that this is kind of a rehashing of Petersen et al. 2019. While these simulations are forced with JRA as opposed to CORE, the model details are the same in both papers as I see it. The regionally refined aspect is new, but it would be helpful for the authors to contrast the model details from this paper and Petersen et al. 2019. Perhaps a table would be helpful here.

2. One of the challenges of regionally refined grids is the scale aware parameterizations. I would like to see more details here about how the parameterizations adapt from the coarse resolution part of the domain to the finer resolution part of the domain. Again, this would be an interesting thing to contrast with Petersen et al. 2019. 

3. I am curious why the authors chose 3 cycles of JRA? I see the salt and temperature trends are still very strong here and particularly above 1000m depth. This seems to accelerate in the 3rd cycle. Are you using some kind of temperature or salinity restoring? This is often required in a data atmosphere forced mode. I see now where the restoring is 1 year. This is fairly weak. A discussion of how the restoring would impact the drift would be a good addition here.

4. With the drift, it is sort of difficult to compare the third cycle of the E3SM simulations to observations or only to one cycle of RASM. Perhaps it would be better to compare the first cycles.

5. It is interesting that the MOC seems to be fairly well equilibrated (Figure 5b) after 1 cycle despite the heat and salt trends. Also, the MOC max transport is significantly higher in the E3SM-Arctic simulation. It would be interesting to contrast with Bryan, Gent, and Tomas (2014) here in which they found that differences in parameterizations at one degree and 0.1-degree led to differences in the poleward heat transport and hence sea ice differences.

6. Figures 7, 8, and 9 are interesting. The Nares strait is not resolved in the LR simulations as mentioned. I would like to see more discussion on how this impacts the downstream transport and the MOC. Is there a shift in transport from the Canadian Arctic to Fram Strait?

7. Figures 10, 11, and 12 are helpful in understanding the vertical distribution differences. How much do these change between cycles? That is, how are they impacted by the drift.

8. Figure 14 and 15 show the central Arctic profiles of temperature and salinity. This is one of the areas where I am worried about the comparison of the third cycle in the E3SM runs to the first cycle in the RASM runs. The drift in the model causes these profiles to move further from the initialization from PHC. This is always an issue with ocean models. That said, are the RASM runs initialized differently? What is going on here with the RASM temperature profiles? Is vertical resolution playing a role here?

9. In addition to the above, some of the most interesting seasonal differences in 14 and 15 are bove 100m. Instead of density, perhaps you could zoom in on the top 100m instead. The surface temperature and salinity biases are indicative of some ice-ocean coupling issues. Do the salt and fresh water fluxes assume 4 psu in the sea ice?

10. Figure 16 is interesting, but might be a candidate for removal. Perhaps a discussion and leading to another paper. I feel like there could be so much more here and the discussion is fairly limited.

11. It is nice to see that the sea ice volume and area are consistent across the three cycles for the most part. They impacted a bit by the upper ocean drift. There is something funny in the volume before 1995 and after. The trend is faster than PIOMAS before 1995 and then slower after 1995. I worry this is an artifact of the JRA forcing.

12. On the JRA forcing. Are you doing corrections over the sea ice? There are definitely some biases in the JRA in the polar regions. While there were corrections added some biases remain and different than the CORE forcing.

13. What are the lateral boundary conditions for the sea ice and ocean in the RASM runs?

14. In Figure 19, I find it interesting that all of the model simulations have a thickness bias in the Beaufort Sea. It might not be fair to compare longer periods in the model to such a short period of IceSat. There could be some substantial interannual variability here. The biases near the CAA might just be something in the IceSat data. I'm not sure this figure is helpful. Also, I wonder if the "bias" in the Beaufort Sea thickness could be related to freshwater content in the Beaufort gyre? Maybe adding the RASM thickness in Figure 18 instead?

15. Back to my earlier point about scale aware parameterizations. Does this feed into the discussion and conclusions about the regionally refined configuration versus the low res configuration? Also, the JRA biases are key here. Ultimately, even though one improves resolution in the sea ice, the thickness distribution in particular is generally set by the wind patterns as mentioned earlier in the manuscript.

Minor comments:

1. On line 65, please delete "occasionally".

2. How are the sea ice albedo parameters set/tuned?

3. I see this is the B-grid versions of the models. Any thoughts about C-grid? How does the sea ice compare to Turner et al. (2021) which is in review?

4. Some more details of the RASM simulation would be helpful.

Review of:" An evaluation of the E3SMv1-Arctic Ocean/Sea Ice Regionally Refined Model." by Milena Veneziani et al. 

The manuscript compares the Arctic regionally refined E3SM configuration (E3SM-Artic-OSI) for the ocean and sea-ice in a stand alone mode using data-based atmosphere land and hydrology components. 

The author carries out the comparison based on a variety of globally modeled but also Arctic ocean related quantities, such as AMOC strength, horizontal temperature and salinity biases but also vertical section of temperature and salinity of certain key features. Further the author focuses on the comparison of the exchanges through Arctic Gateways and the Arctic freshwater content between the  higher and lower resolved Arctic E3SM configuration. Also the trends in the sea-ice components between both E3SM configurations have been thoroughly analyzed. A variety of observational derived data have been used to evaluate and classify the  improvements in the E3SM-Artic-OSI configuration.

In summary it is shown that E3SM-Arctic-OSI significantly improves the simulated Arctic Ocean and sea-ice when compared to a less resolved Arctic model configuration (E3SM-LR-OSI). In particular the exchanges through the main Arctic gateways, Arctic Freshwater content variability and sea ice trends benefit from the refined resolution. Although other aspects like upper ocean stratification and sea ice thickness need further tuning as it is the case in other models as well. 

To my knowledge, the here presented work is novel and of special interest to the broader E3SM modeling community. The author presents very well the shown results and improvements between the different model configurations. Therefore I would recommend that the paper is accepted after some minor revision.

Comments:

-line 4: ...cost of high – resolution (HR) regular gridded global configurations...

-line 6: “...while employing data-based atmosphere, land and hydrology components….” If I understand well you run MPAS+Sea-Ice more or less in a standalone mode with a prescribed atmosphere. What are in this case the land and hydrology components? Maybe some rephrasing is necessary.

-line 43:  “…Wang et. al 2018…”, Its fully OK here to cite Q. Wangs paper, but since you anyway run your configuration uncoupled it might be worth it to cite also the papers of C. Wekerle et. al 2013, 2017a and 2017b. Since they directly deal with mesh improvements in the Arctic realm (but in an uncoupled FESOM standalone configuration) and their consequences for the local ocean circulation down to an eddy resolving regime (C. Wekerle et. al 2017b).

-line 75: Why only 10km was chosen for the Arctic refinement, considering the rossby radius for high latitudes, at this resolution the Arctic ocean will be barely eddy permitting. For example the standard higher resolved Arctic FESOM configuration goes down to a resolution of 4.5km using around 650k surface vertices. There exists an intentional similar paper to yours within the FESOM community (C. Wekerle et. al 2017a), maybe it’s worth to be cited

-line 88-89: Why is there no background diffusivity utilized? 

-line 110: I would like to know but also the community might like to know at which time-step the high and low resolution configuration perform.

-line 111-118: I haven’t fully got the point why or for which purpose you bring the RAMS configuration into the comparison since it’s also just another model that is not directly related to E3SM. Maybe you can clarify in a couple of sentences, also in the introduction,  what’s the benefit of RAMS in this comparison. 

-line 132-136: For my own curiosity (doesn’t need to be in the paper), can you say something to the “...recent improvements in the MPAS-Ocean eddy parameterisation scheme…”

-line 145: Are there known causes why your AMOC in the high but especially in the low resolution are so weak?

-line 145-148: I would be interested to know what the maximum March (NH) and September (SH) mixed layer depth in both configurations looks like.

-line 211:  …heat flux through Davis Strait …

-line 320, Fig.14 and Fig.15: You compare the 3rd. cycle of your E3SM simulations with the 1st. cycle of the RAMS simulation. In my experience, there are usually considerable differences between the 1st. and the 3rd. forcing cycle. At least this is the case for the Atlantic and Southern Ocean. For the Arctic these differences might be not that large, but nevertheless it might be of benefit also to provide the temperature and salinity profiles of the 1st. forcing cycle of your E3SM simulation in Fig. 14 and Fig. 15 as a dashed line. 

References:

Wekerle, C., Q. Wang, S. Danilov, T. Jung, and J. Schröter (2013), The Canadian Arctic Archipelago throughflow in a multiresolution global model: Model assessment and the driving mechanism of interannual variability, J. Geophys. Res. Oceans, 118, 4525–4541, doi:10.1002/jgrc.20330.

Wekerle, C., Q. Wang, S. Danilov, V. Schourup-Kristensen, W.-J. von Appen, and T. Jung (2017a), Atlantic Water in the Nordic Seas: Locally eddy-permitting ocean simulation in a global setup, J. Geophys. Res. Oceans, 122, 914–940, doi:10.1002/2016JC012121.

Wekerle, C., Wang, Q., von Appen, W.-J., Danilov, S., Schourup-Kristensen, V., & Jung, T. (2017b). Eddy-resolving simulation of the Atlantic Water circulation in the Fram Strait with focus on the seasonal cycle. Journal of Geophysical Research: Oceans, 122, 8385–8405. https://doi.org/10.1002/2017JC012974