Review of Antarctic sensitivity to oceanic melting parameterizations

This paper uses the recently produced extended ISMIP6 Antarctica oceanic and atmospheric forcings which now extend to year 2300 (as opposed to year 2100 previously) and investigate the impact of these forcing on Antarctica using the YELMO ice sheet model. In their study, the authors not only consider the impact of the model when forced with both the oceanic and atmospheric forcings but also when forced by only either of them. They concluded that oceanic forcings greatly affect the West Antarctic Ice Sheet (WAIS) and not so much the East Antarctic Ice Sheet (EAIS) and vice versa for the atmospheric forcing.

This paper gives more insights on the capability of the YELMO ice sheet model in performing future projection experiments.

While analyses on the impact on the ice sheet in separating the oceanic and atmospheric forcing from the ISMIP6 is not well published, previous work done on oceanic forcing impact on the AIS that could have been discussed in greater length in this manuscript.

In addition, the discussion of the initialization strategy and its impact on the simulation deserve to be expanded on many levels from its details and result demonstrations to the impact on the forcing interpretations.

Very similar work of this nature has been done before by Lipscomb et al. (2021) and Berdahl et al. (2023) and the authors did not contrast or conform their work with theirs (I will be specific on this matter in my comments afterwards). Also, I believe the authors missed a few opportunities in this manuscript: they could have added more analyzes focusing more on the impact of heat at the grounding line compared to under the ice shelves (which they hinted at in Fig1 but never really pinpointed thereafter); they could have compared their results to model running at higher resolution and making the point that a 16km-resolution model performs relatively well in these types of experiments compared to model using higher resolutions. Regarding the latter point, they could have analyzed the impact of the first 100 year forcing on their model and compare with the results from Seroussi et al. (2020) since they did not participate in this community experiments.

Finally, this manuscript uses forcings that were created for a community effort, the ISMIP6 AIS-2300 which is an extension of the original ISMIP6 that only provided forcing and simulations until year 2100. To the best of my knowledge, the extended dataset is not yet publicly available and should be used with restrain. While the wiki page for the extended experiment is available for now, it only describes the extended protocol but does not provide a link to the data.  I do not see reference credited for these forcing. The manuscript of the extended community experiments was not submitted at the time of this article’s submission and was, in fact, just submitted recently. In the absence of citation, and as a courtesy and recognition of the work into generating the dataset, the authors should reach out to the core ISMIP6 organizing group on how to best cite or acknowledge the extended dataset. Please, be mindful about these types of details in the future. For now, please refer to Nowicki et al. (2020), Jourdain et al. (2020), Barthel et al. (2020) (since these papers show the forcing to 2100 and the same scripts were used to prepare the data) and Seroussi et al. (2024) (under review). For further information on ISMIP6 AIS dataset, I would refer the co-authors to the ghub page: https://theghub.org/groups/ismip6/wiki/MainPage/ISMIP6Projections2300Antarctica

and the section “how to obtain the 2300 AIS ISMIP6 dataset” for more information.

I would support the publication of this manuscript after major revisions.

Specific comments

As mentioned above, a lot of this work feels like an extension of Lipscomb et al. (2021) and Berdahl et al. (2023) who carried out similar experimental setup and parameter sensitivity studies using a very similar initialization procedure. Using the original ISMIP6 oceanic forcing only dataset Lipscomb et al. (2021) not only showed the accelerated retreat behavior before year 2100 also shown in this study, they also show very similar patterns of retreat. Since they are using all the ocean melt calibrations from ISMIP6, please discuss your results with respect to theirs as they also ran to year 2500, like you. These additions can be done in the discussion sections and in the introduction (lines 75-84).

Berdahl et al. (2023) extended Lipscomb et al. (2021) by doing a sensitivity study on thermal forcing (TF) and basal friction parameters. The thermal forcing parameter they varied in their study was \gamma0, just like you and they concluded that not only ocean thermal forcing played a more important role than basal friction law in future forcing simulations, but that increasing gamm0 led to more melting and grounding-line retreat, just like what you find. Please discuss your finding in comparison to theirs.

Your discussion on the initialization deserves additional details:

1. What do you use for calving?
2. What data are you using for Surface Mass Balance (SMB) and air temperature?
3. What do you use for your geothermal forcing?
4. What data are you using for comparing your ice velocity results?
5. In many places you are referring to Bedmachine. If the climatic and ice velocity dataset you are using are similar than the ones used in Bedmachine, please cite them separately as they were not produced by Bedmachine!
6. How does varying \gamma0 influence the inverted parameters?
7. You mention that your spin-up last 20 kyrs, during which the first 15 kyrs are used to invert for observation. What do you do for the remaining 5kyrs?
8. Please show a map of \deltaT_sector resulting from your inversion process. This map can contain the values for all 3 \gamma0. Please discuss in more length the implication of this coefficient. In particular, if you need to cool down your basin by several degrees to obtain a stable ice sheet state, it means (at least one interpretation could be) that the thermal forcing used in the inversion are too warm to obtain a steady state with current observations and the melting of ice shelves is underway.
9. As you are separating WAIS and EAIS in your analysis, and you are not showing 2-d representation of the forcings, I would suggest you add a figure which separates the averaged time series of your forcing for these 2 regions (or mention that the mean forcing evolutions are very similar in both regions).
10. It is highly unlikely that the thermal forcing provided by the dataset will always coincide with an ice draft point in a cavity. Please describe your interpolation/extrapolation process in this case.
11. In Figure 4, please add the grounding line location on your plot. Also, since you are focusing on WAIS in your results with ocean TF, I would appreciate a closer look at the WAIS which better shows your grounding line locations and your fit to observation. Currently, it is difficult to distinguish.
12. Your inversion procedure results in a lot more ice shelves than currently observed. Please discuss the impact of this feature on your results. For example, in the Amundsen Sea Embayment (ASE) your model predicts an ice shelf that could provide more buttressing compared to current observations.
13. In your text, you refer to specific location with TF warming (e.g., ASE and Bellingshausen line 226) but point to an averaged time series as “proof”. For this reason, I would suggest adding a figure showing this TF in 2015 and 2300. For practicality, you could show the depth averaged TF between depth raging 200 m to 800 m, say (where the groundling line is located mainly).
14. Please show in one of your figure or mention in your text the behavior of your control run over the length of your experiment. You mention that you are running one, but do not discuss its evolution (even one sentence could suffice).

In Fig. 1, you go through the trouble of differentiating the TF at the grounding line from the one in cavities. While for most model the TF at the grounding line is always warmer compared to in the cavity by a fraction of a degree, the TF from CESM2-WACCM is warmer by a couple of degrees in the cavities. Could you infer any importance on this behavioral difference from your results?

In my (limited) experience about the use of YELMO in publication, it is more frequently used in paleoclimate studies as opposed to century forecasting ones. In this study YELMO is run at a resolution of 16km which is a resolution many people would argue is too coarse to help reduce uncertainties in future projections. If you can, I would suggest you try to make the point in arguing that your results are encouraging at that resolution. I think you started hinting at this with your table 4. I would also suggest adding the results from Lipscomb et al. (2023) to this table and add the resolution of each model for their study. Also, as hinted above, you could succinctly (or not) write a section about your results regarding the ISMIP6 experiment until 2100 and compare YELMO to the output from Seroussi et al. (2020). This would be of limited extra work since these would be a subset of your current runs.

I am reiterating the significance of \deltaT_sector in your results interpretation. Particularly, when interpreting the results for CCSM4 (e.g., Sect. 4, lines 924-296). (Since I have not seen the values for \deltaT_sector, this comment is somewhat speculative but with a point to be addressed nonetheless). If your inversion leads to a \deltaT_sector=-3 degC in the ASE and the TF for CCSM4 is about 4 degC, your model will actually feel only 1degC there. This means that the slow response of say Thwaites, is due to your model calibration as opposed to the forcing itself. This is a particular and important point to make to not undermine the importance of a 4 degC TF suggested by CCSM4. Please clarify this point in your discussion. As a parallel, I would refer to Lipscomb et al. (2021) and their 2 degC synthetic experiment in the ASE for that exact reason.

The beginning of section 3.1 refers to the initialization procedure’s results. I would move it at the end of section 2.3 or create a new section 2.5. In any event it is out of place in a section that is supposed to introduce sensitivity experiments.

Throughout your manuscript you refer to sea level equivalent in your figures. I did not see anything indicating otherwise, therefore I presume that the values were converted solely from the ice mass above floatation. I would like to bring to your attention that this conversion is not as simple (Goelzer et al. (2020)) and I would encourage you to document your conversion to avoid future readers’ ambiguity. Along these lines, please add “s.l.e” (sea level equivalent) to “m” when you refer to m of sea level in your units in the text.

Regarding the figures, I found the spatial figures really hard to read and would highly suggest enhancing them by making them bigger for a start.

Data availability: while the YELMO code is open source, please provide a link to your output and configuration files for reproducibility purposes.

Technical comments

Line 52: please cite Edwards et al. (2021) regarding ice-ocean interaction uncertainty.

Line 68: remove the http link and web address from the text.

Line 75: please include results from Lipscomb et al. (2021) in your discussion here.

Line 89: please cite Nowicki et al. (2020), Jourdain et al. (2020), and Seroussi et al. (2024) for the extended ISMIP6 AIS 2300 dataset. (See main comments above.). Remove the http citation.

Line 109: please cite Goldberg et al. (2011) as a reference to DIVA.

Line 110: The description of the AIS grid is in a weird location in the text as it precedes the reference to idealized experiments. I suggest moving this detail at a less ambiguous place (maybe end of paragraph (see next comment)).

Line 114: I would start a new paragraph after “Last Glacial Maximum”.

Line 119: I believe you mean Leguy et al. (2014) (their 2013 manuscript was in the TCD and the 2014 version is the final published version).

Line 123: here and elsewhere I would suggest using a more active voice instead of passive. I would replace “has been” by “is”.

Line 143: remove the reference to the wiki page.

Line 152: “bellow the rest” please be more precise.

Line 152-153: “Therefore […]” I find this sentence odd, out of place, erroneous, and not adding any useful information to the manuscript. I suggest you remove it, or you clarify what you are trying to get at. You are trying to compare the output of global climate models (GCMs) that are (commonly) tuned for the pre-industrial time period and then react to a forcing scenario after the historical period. The atmospheric and oceanic forcing used here are a response of the GCM to SSPs and RCPs forcings. When I read your sentence, it almost sound to me that you are trying to say that the atmospheric and oceanic forcings co-evolve independently which is simply not true.

Line 159: replace “On the shelves” by “in ice shelf cavities” (or “In cavities)”

Line 160: what are the “main” cavities of WAIS and EAIS? Please be more precise.

Line 161: see comment for line 159.

Line 161: “The grounding line […]” I don’t understand the point you are trying to make here. Please remove or rephrase.

Line 164: here and elsewhere, I find it clearer to refer to equations as “(Eq. 3)” compared to “(3)”.

Line 169: where is the climatology from? Note that Bedmachine is a dataset for bed topography, ice thickness and surface elevation, not for ice velocity, SMB, or air temperature.

Line 170: what happens during the last 5kyrs of the initialization’s procedure?

Line 179: please provide a map for \DeltaT_sector for the 3 \gamma0 parameters.

Line 188: I would suggest rewriting this sentence with an active voice: “conducted: atmosphere-only runs in which […] thermal forcing are imposed […] simulation and ocean-only simulations with the surface mass balance and surface temperature […]”

Line 189: (regardless of my previous comment) remove “temperature and salinity” since the thermal forcing was derived using these 2 variables.

Line 190: why do you feel the need in mentioning both precipitation and SMB?

Beginning of Section 3.1: see main comments about the discussion of initialization.

Line 193: what dataset are you using for ice velocity comparison?

Line 194: there is no need to keep referring to Bedmachine and its reference. Once is enough.

Line 198: similarly to above, I would suggest a more active voice and rewrite as: “For the medium value of \gamma0, the model overestimates ice thickness at the ice-sheet margins.”

Line 202: please rephrase.

Line 203: replace “the values of” by “varying”.

Line 205: 3.5 m of sea level (or s.l.e after defining the acronym somewhere).

Line 206: replace “but” by “and”.

Line 209: replace “over” by “under”.

Line 210: I believe you meant to refer to Fig 1.c.

Line 211: say more about the difference in temperature and the rate of change in the first century.

Line 212: remove “Furthermore”.

Line 220: the contribution of WAIS and EAIS adds up to 95%. Where are the 5% missing, from the control?

Line 220: replace “1 m” by “\approx 1 m s.l.e”.

Line 221: replace “thus […] even” by “being the major contributor”.

Line 227: Fig 1 is not a spatial figure and for this reason it does not support your claim for the warmer waters of the ASE and Bellingshausen.

Line 236-237: you have not defined which are the main ice shelves of the west (this is ambiguous). Remove “of the main ice shelves of the west”.

Line 284: replace “not allow” by “not lead”. Also, you have not shown spatial patterns of SMB. Maybe it might be worth doing so.

Line 291: the relationship of \gamma0 on sea level contribution was also shown in Berdahl et al. (2023). Please add this to your discussion.

Line 336: update citation and remove the web address.

Tables

Table3: in caption, replace “at the year” by “at year” and “the heat exchange velocity” by “\gamma_0”.

Figures

Figure 1: In your text, you mention that the EAIS and WAIS will be affected differently by the forcing. It might be worth it to add the mean forcing for EAIS and WAIS separately in this figure.

Also:

1. Panel b: add “at the grounding line” to the y-axis label.
2. Panel c: how did you obtain your thermal forcing, did you filter them out to only account for year 2015 cavities or are these averages over each full basins in your Antarctic grid? Please add some details here.
3. For each panel and the figure caption, add that these are mean evolutions.

Figure 2: Please make the figure bigger. Also:

1. Add labels to the colorbars.
2. Add a legend to the figure describing the grounding lines.
3. In figure caption, I believe (b) shows thermal forcing differences.

Figure 3: I would suggest creating 2 numbers, one with your extended ice shelves, and one that uses the same ice extend as Bedmachine.

Figure 4:

1. Add titles to the figures.
2. Create a subplot zooming in on the WAIS.
3. Add the grounding line locations for both model and obs.
4. What data did you use for ice velocity? (Note that Bedmachine used a dataset and if you are using the same, you still need to refer to it separately.)

Figure 5:

1. In the caption, line 2, replace “The values on” by “The values in”.
2. In the caption, line 3, add “at year 2500” after “contribution”.

Figure 6:

1. Place the title from the y-axis to the top of the figure.
2. Add a y-axis title “WAIS” for panel (a) and “EAIS” for panel (b).

Figure 7:

1. Make this figure bigger if you can.
2. It is very difficult to see the grounding lines. It looks like the stronger differences are for WAIS so maybe create a figure that focuses on WAIS.

Figure 10:

1. Do these time series begin with the Yelmo ice extent which is larger than observed?

Figure 11:

1. Make this figure bigger if you can.

Figure 12:

1. Place the title from the y-axis to the top of the figure.
2. Add a y-axis title “Atmospheric forcing only” for panel (a) and “Ocean forcing only” for panel (b).
3. In figure caption, instead of saying “is neglected […]” maybe say something like “the inversion process lead to similar initial ice sheet states regardless of \gamma_0”. (\gamma_0 influences the impact of ocaen state in ice cavities.)

Figure 13:

1. Make the figure and the font size bigger.
2. Refer to the panels in your caption.

References

Barthel, A., Agosta, C., Little, C. M., Hattermann, T., Jourdain, N. C., Goelzer, H., Nowicki, S., Seroussi, H., Straneo, F., and Bracegirdle, T. J.: CMIP5 model selection for ISMIP6 ice sheet model forcing: Greenland and Antarctica, The Cryosphere, 14, 855–879, https://doi.org/10.5194/tc-14-855-2020, 2020.

Berdahl, M., Leguy, G., Lipscomb, W. H., Urban, N. M., & Hoffman, M. J. (2023). Exploring ice sheet model sensitivity to ocean thermal forcing and basal sliding using the Community Ice Sheet Model (CISM). The Cryosphere, 17(4), 1513-1543.

Edwards, T., Nowicki, S., Goelzer, H., Seroussi, H., Marzeion, B., Smith, C. E., . . .1492

Gladstone, R. (2021). Quantifying uncertainties in the land ice contribution to1493

sea level rise this century. Nature, 593 . doi: 10.1038/s41586-021-03302-y

Goelzer, H., Coulon, V., Pattyn, F., De Boer, B., & Van De Wal, R. (2020). Brief communication: On calculating the sea-level contribution in marine ice-sheet models. The Cryosphere, 14(3), 833-840.

Goldberg, D. N.: A variationally derived, depth-integrated approxi- mation to a higher-order glaciological flow model, J. Glaciol., 57, 157–170, https://doi.org/10.3189/002214311795306763, 2011.

Jourdain, N. C., Asay-Davis, X., Hattermann, T., Straneo, F., Seroussi, H., Little, C. M., and Nowicki, S.: A protocol for calculating basal melt rates in the ISMIP6 Antarctic ice sheet projections, The Cryosphere, 14, 3111–3134, 2020.

Leguy, G. R., Asay-Davis, X. S., & Lipscomb, W. H. (2014). Parameterization of basal friction near grounding lines in a one-dimensional ice sheet model.  The Cryosphere, 8(4), 1239-1259.

Lipscomb, W. H., Leguy, G. R., Jourdain, N. C., Asay-Davis, X., Seroussi, H., and Nowicki, S.: ISMIP6-based projections of ocean-forced Antarctic Ice Sheet evolution using the Community Ice Sheet Model, The Cryosphere, 15, 633–661, 2021.

Nowicki, S. M., Payne, A., Larour, E., Seroussi, H., Goelzer, H., Lipscomb, W., Gregory, J., Abe-Ouchi, A., and Shepherd, A.: Ice sheet model intercomparison project (ISMIP6) contribution to CMIP6, Geoscientific Model Development, 9, 4521–4545, 2016.

Seroussi, H., Nowicki, S., Payne, A. J., Goelzer, H., Lipscomb, W. H., Abe-Ouchi, A., Agosta, C., Albrecht, T., Asay-Davis, X., Barthel, A., et al.: ISMIP6 Antarctica: a multi-model ensemble of the Antarctic ice sheet evolution over the 21st century, The Cryosphere, 14, 3033–3070, 2020.

Seroussi, H., et al.: Evolution of the Antarctic Ice Sheet over the next1 three centuries from an ISMIP6 model ensemble, Earth's Future, 2024 (under review)

see attached.