Summary comment:

My name is Benjamin Rabe and I have been asked to review the manuscript "Arctic sea level variability from high-resolution model simulations and implications for the Arctic observing system" by Guokun Lyu and coauthors.

Overall this is a very interesting paper, and the use of high-resolution ice-ocean simulations to evaluate observational coverage and variability seen in advanced analyses of both in-situ and satellite data is likely to improve knowledge in the community and enhance future analyses of Arctic sea level and freshwater content / salinity. The paper is generally well written in terms of structure and language.

I see a few issues with the current version of the manuscript: there are several citations important for this topic that have not been considered, notably studies on in-situ observations. Further, there are open question as to the optimal analysis method described in the appendix. Data citations are almost entirely missing, which needs to be corrected to comply with FAIR principles.

Please consider my detailed suggestions given as comments in the "supplement" PDF. Overall I would consider this manuscript publishable subject to the modest corrections I suggest.

Lyu et al. use two setups of MITgcm to investigate sea level variability of the Arctic Ocean on different time scales. Comparison with observations shows acceptable quality of model simulations that allow authors to look at different time frequencies of variability and specify main physical contributors. They proceed with an attempt to give an estimate of capability of the current observing system to monitor freshwater content variability.

The paper will be interesting for Arctic Ocean modelling and observational communities, and can be published after some moderate revision.

General comments:

When you are talking about anomalies, please specify the time period of the reference mean.

Overall I would work on the general quality of the figures, especially ones with arrows on them. Most of them are hard to read. 

Minor comments:

L71 Please provide more info on vertical grid spacing, especially in the upper levels.

L78 Please identify the source of river runoff climatology

L80-81 Please provide more details on the procedure of initialising the high resolution model from the low resolution fields. Which tracers are interpolated? Did you interpolate the velocity field as well?

L81 Can you specify what variables are outputet (especially 3D ones).

L106 Please define Pb

L142 I would say that the word “matches” is a stretch, especially for the Beaufort Gyre. The intercomparison results would be more convincing if you could provide some error estimates for observations. On the other hand from Fig.3 it is clear that the amplitude of variability is underestimated in the models.

L154-155 “Relative more significant” → “Relatively stronger”? I don't think you mean significance in a statistical sense here, so better not to use this word to avoid confusion. 

L156-157 I would love to see more support for this statement, for example a figure in the Support materials with Low/High-pass filtered data for frequencies you are working further (<30 days, seasonal cycle, decadal).

Fig 5. Letters A and B should be made bigger. Why 2004? Anything particular about this year?

L196-197 It is clear why you use NwAC box, but please justify the selection of the box in ESS (I guess it’s close to maximum RMS variability, but this should be stated).

L207 It is really hard to see anything on this figure, especially some correlation along 1000 isobath. The figure should be made much clearer. Or maybe just make it bigger.

L210-212 How should I interpret the correlation arrows? Is it the direction that has maximum correlation? What does the length of the arrows mean?

L213 I guess you mean A and B from Fig 5? Then I believe the pictograms on Fig. 6 do not coincide with placement of resigning on Fig.6, at least for the region A.

Fig.6 You need to provide legend for arrows (maybe it’s the arrow over Greenland, but it’s not the proper place for the legend) and better explain how to interpret them. Magenta pictograms are practically invisible. Try using a pointing arrow, or highlight them in another way. Make sure the regions are right. Overall the figure is really messy and has to be made much clearer. 

Figure 8. Why 500m? Usually the boundary is set at 200 m.

Figure 9. Anomalies with respect to what (specify time period)?

L282-306 Does those results expand results of Koldunov et al., 2014, obtained earlier for the same time scales on the low resolution version of the model? Or are they similar?

Figure 11. Please put units on axes.

L377 Can you clarify how you distribute the profiles? Do you take positions and timing of real world profiles (I guess so), or do some random sampling? One clarification sentence would be beneficial.