the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Aug 2026
Considering ocean conditions within ice-shelf cavities moderately reduces projected mass loss from the Antarctic Ice Sheet
Abstract. Ocean-forced melt of ice shelves is a key determinant of the stability of the Antarctic Ice Sheet (AIS). Currently, most Earth-system models (ESMs) do not simulate ocean circulation in the cavities beneath ice shelves, and ice-sheet models must infer ice-shelf melt rates from ocean conditions external to the ice shelf. A recent configuration of the Energy Exascale Earth System Model, E3SMv2.1, resolves ocean conditions within ice-shelf cavities with an eddy-permitting mesh in the Southern Ocean. Using projected ocean temperatures from E3SMv2.1 that follow prescribed historical and SSP3-7.0 emission scenarios, we parameterize ice-shelf melt using 1) simulated thermal forcing (TF) within cavities and 2) extrapolated TF from the ice-shelf calving front, similar to the community-standard approach when ice-shelf cavities are absent. We then force the MPAS-Albany Land Ice (MALI) ice-sheet model with the parameterized melt from both methods and project AIS mass change to 2100. Results show that using simulated TF from ice-shelf cavities yields a smaller sea-level contribution (SLC) from the AIS through 2100, by ~4 mm (~13 %), compared to the experiment that uses extrapolated ocean conditions from the calving front. In the context of other ice-sheet model uncertainties, an SLC projection error of this size may be acceptable. If open ocean conditions become warmer than simulated cavity conditions beyond 2100, extrapolation risks overestimating the SLC by a larger margin. Our results demonstrate the value of using cavity-resolving ocean models for multi-century ice-sheet projections.
Competing interests: At least one of the (co-)authors is a member of the editorial board of The Cryosphere.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(1554 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 30 Sep 2026)
- RC1: 'Comment on egusphere-2026-4556', Clara Burgard, 20 Aug 2026 reply
-
RC2: 'Comment on egusphere-2026-4556', Anonymous Referee #2, 15 Sep 2026
reply
“Considering ocean conditions within ice-shelf cavities moderately reduces projected mass loss from the Antarctic Ice Sheet” by Shafer et al. quantifies the impact of open ice shelf cavities on projections of sea level rise from Antarctic mass loss. The ice sheet simulations are run offline and forced by thermal forcing from a recently developed cavity-enabled Earth system model. The thermal forcing is extrapolated into missing regions using community-standard methods, once considering the ice shelf cavities, and once where cavities are excluded to mimic a typical ESM with closed cavities. The sea level rise projections by 2100 are 13% higher (ensemble mean) when cavities are excluded, because the extrapolation method does not adequately represent the complex processes determining temperature within a cavity. Furthermore, the indirect impact of basal melting on shelf hydrography likely means that the discrepancy would be larger for a typical ESM with no cavities.
General comments
I enjoyed reading this paper and found it to be a nice illustration of how developing cavities in ESMs, often a long and difficult task, can have a large-scale impact on quantities of broad interest. The paper was easy to read, well informed on the state of the field, and had a carefully considered methodology.
The authors attributed the differences between the simulation to inadequacies in the extrapolation method. However, I think this could also be explained by an important process they have not mentioned: the latent heat sink of ice shelf melting, which acts as a negative feedback on thermal forcing, but crucially is strongest where it is generated (in the cavities) and becomes weaker further away on the shelf. You could demonstrate this with a map of changes in thermal forcing in and around the cavities - I suspect it would show less warming within the cavities than on the surrounding shelf. The authors showed this effect for Totten, but I suspect it is more general than that.
I would have liked to see a figure showing grounding line change between the different MALI experiments (control, CAV, OO); this was only shown for Totten where the differences between CAV and OO were very striking. I am also curious about the forced grounding line retreat (CAV-control or OO-control) compared to the unforced (control-initial). It seemed a strange omission when basal melt flux, ice thickness, and VAF were all discussed in detail.
Some of the figures could be improved for visibility, and I have added specific comments below where I had ideas for how to do so.
Specific comments
Abstract: If you have any space, it would be nice to briefly summarise or at least hint at the reasons behind the differences in the two simulations.
Line 23: “If open ocean conditions become warmer than simulated cavity conditions beyond 2100” - this confused me because cavities are already cooler than the surrounding shelf, in general. Perhaps you mean “If the open ocean experiences more warming than the cavities”?
Line 27: “one of the largest drivers of mass loss” - I don’t think you need the qualifier here, ocean-driving melting is the single largest driver, no? What else could compete?
Line 28: “has a small, direct effect” would be clearer reworded to “has only a small direct effect”.
Line 41: The development of cavities in ESMs began well before CMIP6 in some cases, even if it was not completed in time.
Line 100: How do you treat iceberg meltwater in E3SM?
Line 108: Does the terrain-following layer lead to any pressure gradient errors at the ice front? Why couldn’t you just restrict it to the cavities only, to prevent such errors (i.e. have the z-levels crop into the ice cliff)?
Line 112: Mcwilliams should be McWilliams; strangely this is correct in the references, so maybe it is a more subtle latex problem.
Line 150: Instead of “marine-terminating margins” can you just say ice shelves? I was confused on first read, thinking of marine-terminating glaciers in Greenland.
Line 162: Can you summarise these “necessary modifications”? Are they parameter choices or bias corrections?
Line 170: “Ipcc” has the same problem as before with references capitalisation.
Line 185: I don’t understand how you “adjust TF in the deepest level by the depth-dependent freezing temperature”. Do you mean you simply extrapolate TF downwards, since it is already expressed relative to the depth-dependent freezing temperature?
Figure 1:
- Specify “Basal Melt Flux” in the title, otherwise it could suggest all types of glacial meltwater.
- What are the thin and thick lines - monthly values and annual values, or running means? If it’s a running mean, can you smooth over the gap between the orange and red simulations by prepending the monthly data from the parent simulation, and vice versa?
- The control basal melting is a touch high in E3SM. Can you briefly summarise where this is coming from, i.e. which ice shelves and why?
Line 205: Note that melt rates are also a function of depth-dependent hydrography. For example, a thinning ice shelf can shoal into cooler waters and experience a large decrease in basal melting, which wouldn’t be captured by a static ice simulation.
Line 237: “The vertical extrapolation adjusts the TF following the pressure-dependence of the freezing point” - as I noted above, TF is already expressed relative to the freezing point, so why is any adjustment necessary?
Line 238: I don’t understand how the horizontal and vertical extrapolation steps are alternated. Is it (1) creep one cell horizontal, then one cell vertical, and repeat? Or (2) creep horizontally until you can’t go any farther, then creep vertically - and if so, what is the exit condition for creeping vertically?
Line 252: What is the intended benefit of recomputing all extrapolated cells at each iteration? Is it effectively a smoothing filter?
Equation 5: The position of the brackets < > doesn’t match what is in the text on line 262.
Line 273: You could mention that the cavities are cooler because of the latent heat sink being stronger, as per my general comment.
Figure 4:
- Specify “Basal Melt Flux” in the title, as before.
- The pale lines for the individual members makes it really hard to see if the ensembles overlap. Would you consider adding shading? You could still have thin lines in a darker colour.
Line 313: Why do you specify that the basal melt flux starts out positive? Surely it’s not possible for it to be anything else?
Line 322: I agree you need a matching grounded ice mask for the comparison panels, but I think it would be useful to keep the original masks for the single simulation panels (a, b, d, e). Otherwise we cannot see how the grounding line changes over the 21st century.
Line 328: Do Thwaites and Pine Island still show the largest increases when you express changes in melt rates as percentages, rather than absolute changes?
Line 335: What do you mean by “directly calculated” vs “parameterised” in this context? All ways of directly calculating melt rates (eg, 3-equations) are parameterisations.
Figure 5: I find it really hard to see any of the interesting details in this figure, as they are mostly in very small cavities which sometimes are only a few pixels wide. Most of what is shown is the mask. Here are some ideas for how to improve this:
- Choose a different colour of mask - the black is too close to the dark red. Maybe a medium grey?
- Zoom into the regions with small cavities, eg Amundsen Sea and Totten, in the corners of the panels.
- Try a nonlinear colour map so that low melt rates are still visible.
- Show refreezing in a separate figure. It’s not clear if there is any refreezing at all (perhaps in CAV but not OO, unless there is significant ISW outflow onto the shelf in OO which mixes upward?), and the sign conventions get increasingly complex if you consider both melting and refreezing in the same figure (c and f: positive = increase in melting or decrease in freezing, negative = decrease in melting or increase in freezing; g and h: similar; i: my head hurts).
Figure 6: It’s hard to see any change in the grounded ice thickness in most regions, because the colour bar is saturated with the rapidly-thinning ice shelves. A nonlinear colour map would help a lot.
Line 385: I think OO and CAV are the wrong way around.
Figure 7: As before, can you try shading the ensemble ranges instead of (or as well as) showing thin lines?
Line 425: Why is there such a huge range in the expAE0* experiments? Were they perturbed parameters, or something else?
Figure 8: It would be interesting to see this expressed as a percentage as well, perhaps in separate panels.
Line 447: “gamma” is missing latex math formatting.
Line 452: “downward trend in the total AIS SLC” is confusing, because it seems to imply sea level fall. Could you reword to say “downward trend in VAF” or “accelerating trend in SLC”?
Line 463: Dronning Maud Land seems worth a mention here, as it is only just behind Brunt-Stancomb for VAF change.
Line 483: The Ross is the largest ice shelf by area - no need for “one of the”.
Line 488: Specify that ISW is supercooled, not just “cooler” than mCDW.
Line 493: Specify “HSSW” instead of “warm water”, otherwise I think about mCDW.
Figure 9: The colour scales in a-i are all very hard to distinguish differences. Can you play around with limits, colour palettes, and perhaps nonlinear colour maps to make these interesting details easier to see?
Line 507: How much can you generalise the Ross case, given the patterns shown in Figure 6? The combination of OO having more melting in the interior, and less melting at the ice front, seems similar in the Filchner-Ronne and Amery at least (at least for ice thickness as shown in Figure 6).
Line 540: I think you mean a warmer and saltier continental shelf, not fresher?
Line 560: Change “between” to “with”.
Citation: https://doi.org/10.5194/egusphere-2026-4556-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 71 | 45 | 12 | 128 | 11 | 8 |
- HTML: 71
- PDF: 45
- XML: 12
- Total: 128
- BibTeX: 11
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
My comments are in the attached document.