The authors have adequately addressed all reviewer comments in their response. The authors should submit a final manuscript that includes each of the changes as they describe in their responses to the reviewers.

Just a few very minor I found, that at most require a few words to clarify:

Figure 1 fonts are small.

Line 29 – remove comma; remove excess parentheses.

Line 33 – add “and” before land

Line 57 – comma before rather

Line 128 and figure 2. I found this section a bit confusing and had to read a few times. You are comparing the traditional and invariant annual cycles with the actual recorded observations for 2016-2017. Since the traditional and invariant are both versions of the typical annual cycle for the 40-year period, is it not unsurprising that they should not compare well with 2016-2017? The improvement in MSE is surprising, since it is hard to tell the traditional and annual cycles apart from these plots, but I don’t understand why we would expect a particular year to match the “mean” annual cycle all that well, or why we might expect the invariant to fit better. Is it not possible that for some years, the invariant might actually fit worse? It is clear that the APAC matches closely, but the reduction in MSE is less than I might have expected compared to the invariant based on Fig 2. Perhaps it is just that the first half of the year matches quite well for all, so the actual MSE is relatively small in all cases. Maybe just some short statement to this effect would help?

Line 201-204 – isn’t the difference in true vs recorded sea ice extent just due to errors in the retrieval (algorithm, weather, etc), and not real variations such as due to sea ice drift? Perhaps you mean that short term variability between the different passes, or timing of passes can create errors due to ice movement (i.e. when the daily extents are averaged from two daily swaths). I suppose real physical variations in ice conditions due to drift or other factors at the ice edge could lead to small changes in the algorithm’s calculated concentration; when this variation occurs when C ~ 15%, then you could get pixels switching between ice and no ice when binarized for ice extent that would contribute to volatility. My only point here is to state why ice drift effects might lead to variations that are not real. Similarly in the conclusion, you conflate the volatility with ephemeral effects, yet there are ephemeral effects that are real day-to-day changes in sea ice extent. I suggest a parenthetical definition of what you mean by ephemeral when this is mentioned in the conclusion.

Figure 6 – fonts are very small (also, please check font sizes in other figures so they will all be readable when figures are sized for publication).

Line 382 – “writing and editing”