General comments

Richaud et al. present an analysis of record lows in Arctic and Antarctic sea ice extent using mass balance decomposition in an atmosphere-forced ocean–sea ice simulation. Their findings highlight that interactions at the ice–ocean interface are common to the four case studies examined, and that the role of basal melt in particular is increasing with climate change.

The main novelty of this study is that different sea ice low events from both hemispheres are analysed in a self-consistent framework, which is distinguished from previous literature examining single cases using a variety of observational data sources and/or model configurations. I agree this approach has merit. While there is not much new insight into the individual cases (which have been studied extensively), the more generalised understanding and comparison of sea ice lows makes up for this. The manuscript is structured logically overall, and the figures are very clear. I like the design of Figs. 4 and 5 in particular, reflecting nicely the use of “anatomy” in the manuscript title.

I would appreciate some clarification on how anomalies in the mass budget terms are calculated. This may reflect a misunderstanding on my part or a potential issue in the methodology. Lines 196–204 explain how the climatological seasonal cycle is calculated, but it does not seem like the long-term trend is removed from the data. This would cast doubt on the magnitude and sign of the anomalies presented in Fig. 4 (and, to a lesser extent, Fig. 5), and hence the interpretation, depending on the strength of the trends in each term. For example, Fig. 1.b has clearly not been detrended. In that case it doesn’t really matter for the purpose of the plot, but if the corresponding time series for the mass budget terms looked like this then everything after roughly the mid-point of the time series would be a negative anomaly. Despite all this, Fig. 4 clearly show anomalies of both sign, and the resulting interpretation is consistent with previous studies, which suggests either (1) the trends are sufficiently small for this to not matter too much or (2) the trends have actually been removed. I would have thought the trends should be removed (and, if only stylistically, also for Fig. 1b, for consistency with the description of a sea ice low on L38 as "when the sea ice extent becomes significantly lower than the trend line").

Also, while the writing is clear enough, there is a somewhat common thread of vague statements that should be clarified, expanded upon, or removed (see specific comments). If the authors are able to address these among other (overall minor) concerns, I would be happy to see the study published in The Cryosphere.

Specific comments

L10: "The Antarctic 2022 event was partly driven by a strong interplay between dynamic and thermodynamic processes": this is vague.

L13–14: "highlights the potential of the ice mass budget decomposition […]": do you mean in terms of applying such a decomposition to observations specifically (linking to the final paragraph of the discussion)? Not clear what this means otherwise, as mass budget decomposition is a fairly standard technique.

L37–39: this is not a definition: what is "significantly lower" or "noticeable"? I suggest removing "defined here as", e.g.: "[…] occurrence of sea ice lows—instances when the sea ice extent becomes […]". Then the next sentence then follows naturally as-is, without having to give a specific definition of "sea ice low" at all.

L44: "was quickly ruled out (Schweiger et al., 2008)": I think it is important to note that this is according to a modelling study.

L50–51: again, I think it is important to note that this is according to modelling studies (as in both references here).

L53–54: I suggest adding the month in brackets, as done earlier in the sentence for spring.

L74–77: (i) I’m not sure it can be both "striking" and "expected"; I would just go with "striking". (ii) I would suggest rephrasing the first part to emphasise that both cyclonic and the opposite anticyclonic conditions are plausible drivers. (iii) Remove "etc.".

L71: Suggest starting a new paragraph with "While come causes […]", as you are now discussing both Arctic and Antarctic sea ice low events whereas the current paragraph is just discussing the latter.

L100: "It also highlights […]": this sentence summarises findings, whereas the previous sentence explains methodology. Suggest rephrasing this to, e.g., "Our analysis reveals […]".

L106: "Section 5 […] provides a perspective […] on the potential for future sea ice lows": there is no such perspective given in section 5. The only part of section 5 relevant to this is the reiteration of the trends in mass budget terms (L481–484), but there is no discussion on the potential of future sea ice lows. In my view, this is a key aspect missing from the discussion section and should be added (e.g., a paragraph, perhaps beginning with the information from L481–484, and linking in the results of section 3).

L120: "lateral melting through ice floe size parametrisation": suggest citing Lüpkes et al. (2012) for this.

L121: "level-ice melt ponds": suggest adding a citation (Hunke et al., 2013) for this too as it may not be obvious to all readers that "level-ice" refers to a specific parameterisation choice.

L140: I am satisfied with the justification that the bias in the forcing, and hence simulation, can be overlooked by removing the mean/trend and examining anomalies. However, I wonder why the authors decided to use ERA5 as atmospheric forcing despite known issues (the authors themselves cite a study from 2019) with temperature biases in the polar regions specifically. What is the benefit of using ERA5 over another reanalysis, for example? It would be good if the authors could add a bit more here to justify the choice of ERA5, even if it is just on practical grounds such as data availability or suitable resolution.

L164–167: I think this important justification for using a model analysis would make more sense at the beginning of section 2.1 or perhaps somewhere towards the end of section 1, because at this stage the model has already been introduced/described.

L183: do you mean "become clearer in Section 3" (this seems to be discussed there, on L232–233, L282–283).

Figure 2: I’m not sure this figure adds much and is potentially a little confusing. The arrows point upwards for positive terms which corresponds to increasing ice mass. This works visually for, e.g., basal growth (mass moves from the ocean to ice), but it does not work for the snow–ice term (the arrow is consistently drawn upward for the sign convention, but visually this is like mass moving from ice to snow). The transport term makes sense visually but does not match the upward/downward = positive/negative flux. I realise this is explained in the caption, but overall the diagram itself does not consistently illustrate the sign convention or the physical processes (the latter being the more intuitive use of such a figure). I would suggest removing as the budget decomposition is fairly standard and in any case it is intuitive enough which terms increase or decrease sea ice mass. As for the sign convention, this is stated clearly enough in the main text and helpfully repeated in the captions of Figs. 4–5, so again I do not think the figure is needed.

Figure 3: is it possible to arrange this horizontally, as in Fig. 4?

L225: "slightly higher": some estimation/quantification should be given (e.g., is it a few percent?)

L226–227: it might also be related to the absence of atmospheric feedbacks? Anyway, this is a relevant limitation of the methodology which should be mentioned in the discussion with some consideration of potential impacts on the results.

L246–247: could you expand on/clarify this comparison? The study cited is examining the sensitivity of a similar model’s mass balance to the discretisation of the ice thickness distribution. The statement that "some quantitative discrepancies exist" is vague.

L248–251: This paragraph comparing the Arctic and Antarctic is quite short. If there is not much else to say, I would suggest merging it with the end of the previous (Antarctic) paragraph. Alternatively, it could be expanded by drawing some conclusions, such as expecting to see more sea ice lows in certain regions of the Arctic but more uniformly around Antarctica. Also, why is it interesting to note similarity of the Greenland Sea sector with the Antarctic? Presumably, it means we might expect similar mechanisms for sea ice lows occurring in those sectors.

L311–313: Figure S1 is plotting May, not March.

L334: "consistent with observations": I think a reference is warranted here to justify this statement (as is done later for the Antarctic case, L410).

L350: "Yet, no significant influence of the cyclone is visible in the model outputs (not shown)": this is vague and unsubstantiated, and also slightly confusing given that the next few lines discuss indirect effects on the ocean heat content over the Beaufort sea. Perhaps, in the quoted line, the authors meant "direct", rather than "significant", impacts? Some comment on what impacts on the model outputs were checked but not seen should be added here.

L375: "match well with satellite observations of sea ice concentration anomalies": but the plot is of sea ice thickness? Similar for L409–410.

L401: in section 4.3, Fig. 5.b is not cross referenced anywhere (in fact, Fig. 5.b is not referred to anyway in the manuscript).

L425–426: The phrase "more difficult to analyse and understand" should be removed as it is vague (no reason is given) and the next few sentences seem to provide analysis of this sector anyway. Also, I’m not convinced that the anomalies there are "small compared to the other sectors", assuming the units of each inset panel of Fig. 5.b are the same: for example, the total mass flux (black bar) is about 130 Gt, certainly larger in magnitude than that of the Ross–Amundsen sector (less than 100 Gt), and the snow-to-ice formation (pale blue bar) is larger in magnitude than any of the other sectors.

L544: "direct": did you mean "indirect"? (I would have thought the "direct" effect of the atmosphere would be via surface melt/growth, and the "indirect" effect via eventual influence on basal melt/growth as described in the text).

L551: important to note that the studies cited here are about the longer timescale (multi-decadal) impact of ocean heat transport on sea ice, not about yearly lows.

Figures 4 and 5: these are very nice figures, but a couple of small suggestions: a different colour map for the sea ice thickness might be helpful visually, as it currently clashes with the colour choices for the budget terms. Also, in the captions, I suggest noting that the units of each inset are the same (assuming they are?), e.g., "the scales of the y-axes differ among panels, but the units are always Gt)".

Technical corrections

L31: "By contrast" → "In contrast"

L72: "all those events" → "sea ice low events"

L87: "authors" → "authors'" (add apostrophe); remove "known" and "to date"

L99: "allows to" → "allows us to"

L106: "all those" → "these"

L112: "allowing to" → "allowing it to"

L127: "[…] models, which […]": remove comma

L137: "millions" → million (twice on this line)

L148: "remaining" → "remainder"

L170: a matter of preference perhaps, but is the subscript "i" on these terms necessary (all terms throughout refer to sea ice)?

L174: "in or out of" → "into or out of"

L211: "Apart for" → "Apart from"

L222: "Those results" → "These results"

L245: "The seasonality, and relative […] those terms […]": remove comma and replace "those" → "these"

Figure 3 caption: "during month" → "during the month"

L262: "melts" → "melt"

L294: "can then be used" → "are now used"

L362: remove "therefore" (currently reads like this is a conclusion from the previous sentence about the cyclone)

L373: "are specific": clarify (did you mean "non-specific"?)

L382: "basal and surface" → "the base and surface"

L464: "denominator between" → "factor explaining"?

L566: move comma after "effort" to be after "campaigns" on L567, and replace "incoming" → "upcoming"

L567: "would provide a great framework" → "will provide an ideal framework"?

L571: "those" → "these"

Throughout: this may be fixed at the proofing stage anyway, but there are many instances where a hyphen should be replaced with an en dash for ranges/connections ("1979–2008", "ice–ocean", etc.)

References

Lüpkes, C., V. M. Gryanik, J. Hartmann, and E. L. Andreas (2012): A parametrization, based on sea ice morphology, of the neutral atmospheric drag coefficients for weather prediction and climate models, J. Geophys. Res., 117, D13112, doi:10.1029/2012JD017630.

Hunke, E. C., D. A. Hebert, and O. Lecomte (2013): Level-ice melt ponds in the Los Alamos sea ice model, CICE, Ocean Model., 71, 26–42, doi:10.1016/j.ocemod.2012.11.008.

Richaud et al.: Anatomy of Arctic and Antarctic sea ice lows in an ocean–sea ice model

The authors in this study use the NEMO-SI3 model to investigate extreme lows in sea ice extent in both hemispheres over the historical period by using sea ice mass budget terms, both hemispherically and regionally. These modeled mass budget terms are particularly valuable as they are not available in observational products but they provide insight into the processes driving the low extent years. The authors find that basal processes (melt and growth) become increasingly dominant in the mass budget in both hemispheres. Additionally, they find that during case study low extent events the importance of thermodynamics and dynamics can vary, as can the importance of difference processes by season or hemisphere due to the local ice state and drivers. Overall, this study is well designed and provides insight into how sea ice evolution is changing. I have some major concerns, detailed in the attached document, about the analysis that I would recommend being addressed in a major revision before publication.

I am suggesting major revisions for this manuscript, mostly based on some very confusing and internally inconsistent graphics (Figures 4/5).  While the authors have a clear sign convention (positive fluxes denote ice gain, implying energy loss), the graphics as designed often cause confusion on that convention -- especially for someone who suffers from directional confusion.    The graphics would easy misinform a casual reader using the graphics for a quick summary.   This has, I believe, even lead to the authors making an erroneous statement (once that I have caught).   

There is also considerable incomplete explanation and inconsistent explanation of their purported connection between sea ice thickness anomalies and ocean heat content.  

A full report is attached, Addditional supporting information -- an animated gif and an annonated copy of the manuscript will be forwarded to the assistant editor.