Nicola et al. use the BedMachine bathymetry product and ISMIP6 dataset (observations + extrapolation of temperatures (T) and salinities (S) around Antarctica and its continental shelf) to discuss some features of these datasets, for example the relevance of different troughs in delivering heat to the grounding lines (for BedMachine) and the spatial differences in ocean temperatures and salinities (for ISMIP6 dataset).The authors then calculate ice shelf basal melt rates using the box model PICO, where the T/S input is a) conditions at the calving front and b) conditions at the continental shelf. The latter is presented as the "upper limit of melt rate changes".

General:

I have to admit that I have struggled with the aims and the novelty of this manuscript. The first half of the paper introduces new terminology of 'oceanic getaways' and 'critical access depths' but it is really just talking about bathymetric features, specifically troughs that have received a lot of attention in the past decades as the sea floor around Greenland and Antarctica have become better mapped. The same flood-fill algorithm that the authors use here was also used to produce and extrapolate the ISMIP6 dataset, so I don't think that part is anything new. The next bit of the paper that discusses ocean properties at the calving front vs the continental shelf doesn't do much more than stating the differences in the fields in ISMIP6 dataset and some generalities.

The section "Oceanic gateways to major Antarctic ice shelves" is a mixture of literature review, and speculation about potential impacts of high melt rates (as calculated by using shelf break instead of calving front temperatures) - but the impacts are not modeled here so just a brief mention in discussion would be enough for that part. For the literature review - a lot of this refers to studies about ocean circulation beneath the specific ice shelves, but all that is ignored in PICO, so I don't see why that is reviewed here, since the box model doesn't know anything about horizontal circulation. Other references serve to show that bulk present day PICO melt rates are reasonable, but that was already tuned elsewhere in previous publications, so not sure why that is necessary here again.

I think that especially with the simple box model it is really easy to produce large increase in melt rates for any given ice shelf, all that is needed is a change in input temperature that is fed into box 0. Since there are no oceanic processes accounted for that would be allowed to mix or divert away this change, the model essentially by construction contains a tunnel that conducts outer T and S directly to the grounding line. So the result that large change in input (which is chosen here but not really physically justified) causes large change in output, which is the result of this manuscript, is definitely not a surprising one. The question is whether the numbers produced here for the increased melt rates are realistic or otherwise useful in some way. I don't think the authors have even tried to make a case for either usefulness or realism of these high melt rates. The only argument that was provided in the paper is that this approach here is "straight-forward and easy to run" but without it being realistic or useful, simplicity on its own, is not enough of an argument for publication.

A general characteristics of this paper is that the authors state assumptions but don't justify them. A good example of an unjustified assumption is the one that grounding lines are always accessed via 'prominent getaways' - that clearly doesn't hold in present day for the cold ice shelves Ross and Filchner-Ronne and others, yet this inconsistency is not at all addressed. Also there is some misuse of terminology. For example the temperature the authors have chosen for the CSB T and S is quite arbitrary, yet they call their perturbed melt rate result "upper limit". Surely not everywhere is this arbitrary point the temperature max along the shelf break, so even higher melt rates could be reached with PICO. I don't think the term warm-water intrusion is accurate for the use in the context of a long term, large scale and lasting change. Intrusion is an intermittent oceanographic feature. The sensitivity study here assumes that oceanographic conditions within the cavity change, that is warm water from the open ocean comes across the continental shelf break and stays and that is something very different and more difficult to establish than an intrusion which would largely mix in with other water masses on its way to the grounding line and become much cooler and fresher by the time it comes in contact with the ice.

The PICO model has some clear biases compared with the observations of Adusumilli et al. 2020, namely it overemphasizes a melt rate pattern of high melt at grounding line and low melt towards the front and does not take into account the 3D structure of the circulation, which results, for example, in omitting mode 3 melting features near ice shelf fronts. Accordingly, the melt rates 'assuming warm water intrusion' have the same biased melt rate pattern as the original PICO melt rates except now the melting is higher. Can you comment on the bias and its implications? For example in the context of Reese et al 2018 - if grounding line are most sensitive to melt rate change, overestimating melting there could be problematic, yet it is probably happening since the bulk melt rates are tuned to agree with observations and mode 3 is absent - resulting in freezing or low malt rates near the front - positive bias in grounding line melt rates is clearly visible in sectors 3-5.

Specific:

Access depth seems to be a key concept here but it is not clearly defined (I think the language is the problem). Figure 1 doesn't help - it is stated in the text that access depth is a field defined everywhere (and provided on a certain discrete grid) but the figure only points to a single point in the image, which is confusing. Further on Fig one - what is the -1800 m in the image showing horizontal distance? Shouldn't that be depth for the purposes of your continent definition?

Similarly g is not clearly defined. From the paper it is sort of clear what the authors mean from the context but that is relying on the reader being on board with the writers.

Sign convention of z vs depth needs to be consistent.

Fig 7 and similar - x axis needs to be labeled on each subplot to make clear what distance is meant for each case

other specific comments are in the attached pdf

General comments:

Nicola and coauthors identify key oceanic pathways in the major Antarctic glacier basins and estimate current and future melt rates based on the present water properties, assuming that warm waters from the shelf break reach the grounding lines. I found it interesting to see the range of access depths and ocean water properties around the Antarctic Ice Sheet in one simple figure, and it is interesting to quantify the upper limit of future melting assuming warm water intrusion through these pathways. The data were clearly visualized, making it easy to synthesize the wealth of information presented. The analyses and profiles shown for the major ice shelves were especially nice to see and will make this study of interest to a range of groups who study various components of the Antarctic ice/ocean system.

I do not have any major concerns about the approach or the conclusions, as the caveats associated with this methodology are clearly noted in the discussion. I have several comments mostly relating to the presentation of the study, so I recommend this study for publication after minor revisions. I hope the authors find these comments helpful.

Specific comments:

1. The motivation of the study could be more clearly laid out. Ocean gateways are important for ice shelf melting, but why does this analysis help the scientific community? Rather than giving a lengthy overview of why troughs and gateways are important and then explaining the approach, I suggest making a more concise overview of the importance, state (more clearly) what the critical uncertainties are, what the approach is, and what the hypothesis is.
2. What are the criteria for classifying something as a ‘prominent gateway’? In L145 the authors state that “large portions” of the GLs need to be reached, but how was this determined?
3. Section 3.3.4 (Amundsen Sea) – the main gateway is identified as the Abbot Cosgrove Trough. Is this distinct from the ‘Eastern Trough’ commonly referred to in studies about the Amundsen Sea (e.g., Dutrieux et al., 2014 already cited in this study)? Does it feed into the PIG-Thwaites Trough shown in Fig 9a, or is it separate? It would be helpful to clarify which of the oceanic gateways identified are/aren’t in agreement with the main pathways identified in previous studies.
4. The mismatch in melt rates between the approach in this study and the melt rates of Adusumilli need to be addressed, as that is critical for gauging how reliable the estimates of future melt rates are. Some regions do better than others, perhaps due to different reasons, so I’d suggest explaining this for each of the major ice shelf regions.
5. I found Section 3.3 rather hard to read. Rather than explaining the results and then reviewing various relevant literature, I’d suggest reviewing the findings from the literature and then presenting the new results within that context. It feels very scattered in its present state.

Other suggestions (to improve the presentation, not necessary for publication in my opinion):

1. In Figure 2, it would be nice to see the temperature relative to the pressure melting point.
2. Several figures likely need larger text to for readers to see the numbers in each of the small boxes – probably fine for PDFs, but not printing. Perhaps the numbers are not critical for understanding the figures, as the colors also reflect the values.