The paper „Geometric amplification and suppression of ice-shelf basal melt in West Antarctica“ by Jan De Rydt and Kaitlin Naughten aims to quantify the connections and feedbacks between changes in geometry and the basal melting of ice shelves in the Amundsen Sea. Results are based on a suite of regional, coupled model simulations and a reference simulation with fixed present-day geometry. The model is forced through restoring temperature, salinity and velocities at the open ocean model boundary. The authors conclude that "newly identified geometry-melt feedbacks play an important role [in enabling] a sustained negative net mass balance of the ice shelves, which drives further grounding line retreat and ice-sheet mass loss“ [quite from Conclusions].

General comments:

The paper leaves the reviewer with a mixed impression. The study design and methodology are state of the art, figures are well crafted, and the presentation is mostly clear. There is, however, still some room for improvement.

First, the definition of „cavity transfer coefficients“ gives the study a quantitative touch. That’s nice, but while the coefficients are defined by the equations in which they appear, the paper is very short on information on how they are actually computed from the model results. Furthermore, the coefficients are not used to their full potential, and actually in two ways so. Firstly, the authors provide time series of all coefficients for all ice shelves for many of the experiments, but we do not learn much about the processes that shape the feedback in different ways for different ice shelves. There is a lot of potentially interesting analysis hidden in the discussion of coefficient time series (e.g. on how sub-ice circulation responds to changes in geometry), but the discussion goes coefficient-by-coefficient where going ice shelf-by-ice shelf would have made it easier to actually learn things about the real world. The authors are therefore encouraged to re-organize this section. 

Secondly, the authors motivate their study by going „towards bridging the gap between [two] incomplete modelling approaches“, namely fixed-geometry melt rate modelling on one side, and ice sheet modelling with weakly constrained (and not always physics-based) melt rates on the other side. Given that coupled ice sheet-ocean models do exist (one is in this paper) but with the currently available computational resources cannot even dream of fully covering the long timescales relevant to ice sheet processes, there is clear scope for guiding ice sheet modelling communities towards how to represent geometry-melt rate feedback with more physics than typically used in ice sheet models now, which creates a strong motivation for a study like the one presented – but unfortunately the authors do not follow up on this aspect. Calling for a full set of guidelines that modellers can follow would be asking too much, but a little bit of perspective on how exactly the results of this study can „be considered in future projections of Antarctic mass loss“ [quote from the abstract] would make this paper a lot stronger.

Last but not least, the conclusion could be a bit stronger in summing up the main findings – also to make sure that the authors‘ intentions are actually matched.

Specific and technical comments, corrections, etc:

l. 15 „widespread dynamic thinning“: Some ice research groups restrict the use of the term „dynamic thinning“ to the meaning „thinning by flow field divergence“. This does (to first order) not contribute to sea-level rise. I assume the authors use „dynamic“ for „rapid, and with a lot of variability“. Unless the authors do imply the strict meaning, I suggest to remove the word.

l. 19: „imbalance“ is a euphemism for „mass loss“ here, isn’t it? I suggest to be clear and use „mass loss“

l. 21: It should be either „For future decades to centuries,“  or the „in future decades to centuries“ needs to be shifted behind „indicate that“

l. 26: no comma after „ice shelves“

l. 41: no comma after „centuries“

l. 46: no comma after „cavities“

l. 85: no comma after „interface“

l. 104: given the boundaries of the regional model, I wonder whether „10 vertical levels with 40 m resolution down to 2000 m“ actually exist (to that depth). Besides, the leves are horizontal, not vertical (it’s the vertical coordinate, not vertical levels).

l. 105: is the representation of ice and bottom topographies piecewise linear (=shaved cells) or piecewise constant = stepwise (=partial cells)?

l. 114: Given the potential effects of deep convection on the continental shelf  (outside the cavity), the assumption of „thermohaline properties in the deepest part of the cavities“ to be „largely unaffected by changes in surface waters“ seems daring. Isn’t the (variability of) processes on the continental shelf one aspect of what coarse-resolution ice sheet models using melt rate parameterizations consistently tend to miss? I suggest to remove the sentence.

l. 156: please add a word or two on how the two time slices for hi_melt and lo_melt were chosen and how the respective data sets were created. Reasoning is clear and convincing, just what exactly has been done could be explained a bit more.

l. 165: inform _on_ the future evolution

Section 3: stating that all variables are time-dependent and removing the „(t)“ in all equations would improve readibility. If all variables were defined as also space-dependent except for averaged properties, which could be denoted with an overbar, the „(x)“ could go too. Whereever possible (including in all the text obviously), the use of K instead of °C is encouraged, because K is an SI unit, while °C is not.

Given that the T*s are temperature differences by unit and definition, the term „thermal driving“ or „thermal forcing“ may not be ideal. How about „melt forcing temperature“ (with unit K again) or similar? The term „friction velocity“ is widely used, despite the fact that it is not a „real“ velocity, so there is a parallel. In any case, there should be ONE term defined and then consistently used and no switching between „thermal forcing“ and „thermal driving“.

caption to Figure 3: „oceanic boudary layer“ may be more appropriate than „oceanic mixed layer“ (because the important and correct-for-sure thing is that the layer is a boundary layer, while it being well-mixed may be something we can be less sure of)

Formal structuring of Section 4 is not ideal. It is recommended to add a subsection heading „4.1 Evolution of melt rates and cavity geometry“ (or similar) right between lines 291 and 292. The following subsection headings need to be renumbered then, obviously.

l. 292: no komma after „cavities“

Table 2.: Some discussion of the physics behind the different values for the four coefficients seems desirable. Anything to be learned from the differences and consistencies between them? Also, how can µ become >1?

Caption to Figure 4: Is the GL color really magenta? Not red?

l .312: The „in turn“ at the end of the sentence may be unnecessary.

l. 327: occurs

l. 353: replace „ice draft“ by „ice base“

l. 369: suggest to replace „mixed layer“ by „boundary layer“

l. 373: same, also for l. 376, l. 386, and l. 442

l. 374: delete „years“

l. 391/392: If a transfer coeffficient (here epsilon_T) is determined by the forcing, does that not mean that the transfer coefficient is not well designed to characterize the transfer process it is supposed to describe? Please discuss. Also the mu>>1 calls for a bit of discussion. How can a transfer increase a signal? 

l. 417/418: This is an interesting finding that could well re-appear in the Conclusions.

l. 419-439: This is all plausible and interesting, but why is the x-y circulation adressed so nicely and the classical y-z ice-pump overturning circulation not at all?

l. 463: same here: Why (only) look at the barotropic streamfunction and/but not at the overturning?

l. 469: replace „modulate“ by „dominate“?

l. 476: This is one of the few places where the av_melt experiment appears. Is it needed at all for the point of this paper?  (This is an open question, not a disguised suggestion.)

l. 485/486 „recently“ is actually with the Ice2sea project and the papers spawned from that. Which was in 2012, not 2022. 

l. 486-488: The authors are obviously well aware that coupled ice sheet/ocean models with varying cavity geometry do exist, so their claim that „the interplay between […] far-field ocean variability and geometrically-driven changes in melt rates has not been included in numerical simulations“ seems a bit odd. This needs to be modified to more precisely match what the authors intend to day.

l. 503: replace „orange“ by „red“

l. 503/504.: While the statement in „Results demonstrate that“ is plausible, it is not actually demonstrated (in Fig. 8).

l. 524: Is the reference at the end of the sentence actually needed for the stamement made?

l. 534: solid and dashed

l. 536: „have a much lower amplitude“ –> except for the jump for PIG, no?

l. 589/590: are the „sudden or gradual increases in grouding line discharge“ created by enhanced ice flow or by grouding line migration?

l. 609: The phrase „[…] offsetting shifts in grounding line discharge“ needs to be clarified

l. 613: replace „represented“ by „as indicated“ ?

General Comments

The manuscript „Geometric amplification and suppression of ice-shelf basal melt in West Antarctica“ by Jan De Rydt and Kaitlin Naughten presents innovative research of considerable significance in glaciology. The methodology is robust, employing state-of-the-art coupled model simulations to quantify the connections and feedbacks between changes in geometry and the basal melting of ice shelves in West Antarctica. The discussion and interpretation of results are thorough and with appropriate consideration of related work. Explanations are detailed and the manuscript maintains a logical flow. Despite the complexity, the overall presentation is mostly clear, enhanced by well-structured and colorblind-friendly figures. Certain minor aspects could benefit from refinement for enhanced clarity and depth.

Specific Comments

The authors use a new configuration of the Úa-MITgcm model, with a mutually evolving dynamical ice sheet and 3D ocean. They provide a detailed description of the coupled ice-ocean model and experimental design, including the definition of “cavity transfer coefficients”. The entire analysis and the overview of the three numerical experiments are clear and concise. I have only 3 suggestions:

1. When describing the model's domain and specifications (in “2.1. Coupled ice-ocean model setup”), consider adding a brief explanation of why these specific parameters and configurations were chosen. It could enhance the clarity for less experienced readers and it would provide context for the approach. For example, is the resolution and number of layers selected purely based on the compromise between computational feasibility and accuracy or was the nature of specific physical processes considered as well?
2. The authors focus on the different cavity transfer coefficients in order to diagnose the feedbacks between changes in basal melt, imposed ocean boundary conditions and cavity geometry. This deserves to be highlighted and it could be beneficial to provide more information on how these cavity transfer coefficients are computed. Also, the findings in this section are quite interesting, and it certainly could be useful to discuss their interpretation further in the conclusions.
3. Overall, the summary & conclusions part could benefit from a slightly more detailed overview of the findings. It would be good to highlight the implications more and to mention specifically how their work overcomes the limitations of current ocean modeling approaches in the context of the Amundsen Sea's basal melt rates. The results and interpretations of the findings are well documented and highlighting them in the conclusions would provide a clearer and more impactful closure to the paper.

Technical Corrections

Minor typographical errors are present and there are a few instances of complex sentences that could be simplified for clarity.

1. Line 21-24: “In future decades to centuries, numerical mass balance projections indicate….” -Consider restructuring the sentence (or splitting it into two) for clarity.                                                                                                                                                                                                    For example, "In future decades to centuries, numerical mass balance projections indicate that the Amundsen basin is likely to remain Antarctica’s dominant contributor to sea level rise. This persists despite significant uncertainties in climate forcing and poorly represented physical processes, such as ice-shelf melting and temporal changes in ice rheology, basal sliding, and ice front location."
2. Line 35-38: “In recent decades ice-shelf thinning rates have decreased...” -Consider restructuring the sentence for clarity.      
3. Line 41: “The rate at which glaciers in the Amundsen basin will continue to lose mass over the next decades to centuries, is controlled “ -the comma after "centuries" is unnecessary.
4. Line 46: “A third, as-of-yet poorly understood process, is the potential feedback between changes in the geometry of ice-shelf cavities, and the ocean dynamics ….” – the comma after "cavities" is unnecessary.
5. Line 58: "(Edwards et al., 2021, e.g.)," -the use of "e.g." seems unnecessary.
6. Line 87: "(Patmore et al., 2019, e.g.)" - the use of "e.g." seems unnecessary.
7. Line 114-116: -Consider assessing and re-writing the entire sentence “thermohaline properties in the deepest parts of the cavities, which are thought to be largely unaffected by changes in surface waters.” The decision to ignore surface fluxes is a valid approach in order to focus mainly on the interactions between cavity geometry and basal melt. However, the statement “thermohaline properties in the deepest parts of the cavities, (which) are thought to be largely unaffected by changes in surface waters” doesn’t seem correct.
8. Line 326-327: "High-frequency fluctuations at monthly timescales are predominantly caused by Eddy activity at the ice front, which occur irrespective of the changes in cavity geometry..." –adjust the verb to “occurs” if the sentence is grammatically consistent.
9. Line 369-370: Please explain the meaning of “changes in the mixed layer speed” as I might have misunderstood the usage here.
10. Line 404-405: "The fourth and final coefficient to control average melt rates..." -The phrase might be more grammatically correct as "The fourth and final coefficient controlling average melt rates..."
11. Line 419: "Whilst basal gradients do locally impact on the buoyancy of the mixed layer..." -Consider changing "impact on" to "impact".
12. Line 480 and 492: The phrase "geometrically driven" could be hyphenated as "geometrically-driven" for clarity.
13. Line 489: Small typo in “Embayement” -Consider changing it also to Amundsen Sea Embayment.
14. Line 494-495: Consider changing “a” to “an” in “over a 18-year period”
15. Line 534: It should be “and” in “(solid an dashed blue lines in Fig. 9)”
16. Line 580: “and grounding line retreat is either slowed down, or prevented” - the comma after "down" is unnecessary.