The authors present the multi-model mean (MMM) of ice shelf melt rates and other parameters that determine ice shelf melt rates. They claim that the MMM provides a useful comparison between different models and serves as a guideline for observations and modeling. However, I believe that the simulated ice shelf melt rate is a parameter that can be easily tuned by selecting appropriate coefficients, and the multi-model comparisons presented here are somewhat overstating and misleading. I believe more analyses can make this manuscript much more useful for the community. I suggest a major revision.

 

Major Comments

1. The authors discuss, for example, the comparison of simulated total melt with satellite-based estimates (e.g., between lines 325 and 337 and in Table 2). However, in global simulations like this, integrated melt rates are easily tunable parameters. I question the significance of this study's comparison of model outputs that are easily adjustable, rather than focusing on parameters that are more challenging to tune and simulate. I suggest that the authors add a paragraph explaining how the drag coefficients and heat and salt transfer coefficients are determined in all simulations.

2. To enhance the usefulness of the comparisons, I suggest addressing the following aspects (a-d). In the current version of the manuscript, the authors discuss bottom or vertically integrated ocean hydrography, which is not the best metric for determining ice shelf melt rates. I recommend presenting thermocline depths and vertical sections under major ice shelf cavities to illustrate how variations in hydrography lead to differences in simulated melt rates. Additionally, it is important to examine interannual and seasonal variability, discussing the extent of differences in these aspects.

(a) Integrated ice shelf melt rates for each individual ice shelf.

(b) Hydrographic conditions, particularly vertical sections within the ice shelf cavity or at the ice shelf front.

(c) Temporal variability of thermocline depth and its relationship to ice shelf melt rates.

(d) The processes governing ice shelf melt rates, such as thermocline depth variations, temperature changes, and other factors influencing ocean velocity.

Minor Comments

1. I find that Antarctica's wide mean melt rate is not very intuitive. I suggest adding an integrated melt rate in Gt/yr.

2. Lines 79-82: Considering that the Antarctic ice shelf melt rate is an easily tuned parameter, I feel this is somewhat overstating the presented findings here "allows us to constrain both present and future ocean-driven impacts on Antarctica, and provide much needed evaluation with both satellite-derived and in situ estimates of ice shelf basal melt rates". 

3. Lines 403–412: One of the main conclusions of this multi-model comparison is the strong dependence of both thermal driving and friction velocity on the melt rate—concepts that have already been well established in numerous previous studies. While this study represents a first attempt at presenting a multi-model mean, I do not believe that there are more useful comparisons metrics that are not presented in the current version of the manuscript. 

General Comments

This study compares several circum-Antarctic ocean simulations that include sub-ice-shelf cavities to highlight inter-model differences and evaluate results against satellite-derived melt rates. The approach is to take the ocean simulations “as is,” without following a particular protocol. Both model outputs and satellite-derived melt rates are remapped to a common grid to facilitate comparison. This represents the first attempt to provide a Multi-Model Mean estimate of Antarctic ice shelf basal melting, while also highlighting areas of agreement and divergence among models. The work has the potential to serve as a valuable reference for future modeling efforts and to help guide targeted observational strategies.

However, I have a major concern that the comparison is overly focused on melt rate and its primary drivers (friction velocity and thermal driving) while giving limited or no attention to the underlying ocean and sea ice state. These parameters are often highly tunable, which reduces the robustness and generalizability of the findings. As a result, the comparison has limited utility for informing future modeling or observation-based studies. A broader focus on key oceanographic and sea ice metrics would strengthen the study’s relevance and impact.

Here are several suggested comparisons that I believe would significantly strengthen the manuscript. While I do not expect the authors to include all of these, incorporating at least some would enhance the study's depth and utility:

1. Cross-sections (y–z) of mean temperature and salinity in key regions, particularly where Circumpolar Deep Water intrusions occur;
2. T/S diagrams over individual ice shelves, including adjacent continental shelves;
3. Heat transport across individual ice shelf fronts;
4. Mean mixed layer depth (e.g., July to September), based on a clearly defined criterion;
5. Annual cycle of sea ice area and volume;
6. Maps of sea ice concentration and sea ice thickness.

For metrics where observation-based estimates are available (e.g., sea ice properties), I encourage the authors to include these in the comparisons as well.

The manuscript is overall well-written and clear. However, in addition to my main concern noted above, there are also significant issues with several figures, as outlined below. For these reasons, my recommendation is major revision.

Specific Comments

Table 1: I suggest including (1) a column indicating whether the model is “regional” or “global”; (2) a column specifying the atmospheric forcing used to drive the model (e.g., JRA-55, CORE); and (3) for regional models, a column identifying the product used to drive open boundary conditions.

Lines 142–143: “The thermal driving…” – Please modify this sentence to reflect that thermal driving can be negative (i.e., T* < 0), in which case the water is colder than the freezing point.

Section 2.2 “Remapping approach to a common grid”:

I was not familiar with the Voronoi tessellation approach, which I believe is not commonly used in the geophysical sciences. What are the advantages of this method compared to more widely used techniques such as bilinear or nearest-neighbor interpolation? It is mentioned that this approach results in “lower total melt estimates than previous studies” (lines 199–200), and later that “discrepancies likely arise from differences in the areas used to compute melt rates” (lines 252–253). Is this because the Voronoi tessellation is non-conservative? Would it make more sense to convert melt rates to a mass flux, apply a conservative remapping to the common grid, and then convert the flux back to a melt rate? This might avoid the bias toward lower melt estimates. Lastly, please include other specific reasons for selecting this approach over more conventional alternatives.

Table 2: Are the mean melt rates area-weighted? If so, please clarify this in both the table caption and the main text.

Figure 1: I suggest that the comparison with satellite observations be moved from the supplementary material to the main text.

Lines 316–317: “The Multi-Model Mean (MMM) estimate of Antarctic ice shelf basal melting surpasses the performance of any individual model in our ensemble.” The rationale for this statement is unclear. No analysis is provided to support it. Please clarify the basis for this claim or include supporting evidence.

Line 399: “Comprehensive” may be too strong in this context. I suggest using a softer term such as “extensive.”

Code and Data Availability: Please make the code used in this study publicly available to enhance transparency and reproducibility.

Editorial / Typographical Comments

Line 55: “IUGG, WCRP, SCAR” – Please define these acronyms in full for readers who may not be familiar with them.

Line 60: Please define MISOMIP and MISOMIP2.

Line 83: “...the project” – Since the RISE Project is introduced in the next paragraph, I suggest changing “the project” to “this work” or a more neutral phrasing.

Lines 90–91: Please define ICESat, ApRES, NECKLACE, and SOOS.

Table 1: The phrase “...reference for each model” is redundant. I suggest deleting “for each model.” Also, please define u* in the caption.

Lines 149–150: “Thermal driving …” – This is unnecessarily repetitive, as thermal driving was already defined in line 142. I recommend combining the two definitions to avoid duplication.

Line 228: Correct “at teh” → “at the.”

Figure 1: Please define “ISMR” and “SD.” The coordinates are difficult to read; consider increasing the font size of the x and y axes.

Figures 2 and 3: Same comments as for Figure 1.

Line 261: Correct “is suggests” → “suggests.”

Line 375: Fix formatting error in “e.g.,[” – likely a LaTeX code issue.

Supplementary Material: In some figures, the legend text is difficult to read. Please ensure all text is legible at standard PDF viewing size.