Dear Dr. van Calcar, Dear co-authors,

Thank you very much for your replies to the reviewer comments. I would now like to ask you to implement the envisaged changes and submit a revised manuscript which will undergo another round of review. I have added some additional remarks below (supplementary file) on some of the reviewer comments which might help in drafting the revised manuscript.

Thank you very much and I am looking forward to receiving your revised manuscript.

Best regards,

Johannes Sutter

Dear Dr. van Calcar and Co-authors,

thank you very much for submitting the revised manuscript incorporating the reviewers comments. We will now send your revised manuscript back to the referees for a second round of reviews.

With kind regards,

Johannes Sutter

Dear Dr. van Calcar,

we have now received two reviews and both reviewers in principle suggest publication providing additional contextualization of your results. The revised manuscript will not be sent out to reviewers again but only undergo review by the editor.

Apart from the reviewer comments on the discussion of the model results there remain stylistic and formatting issues (see both reviewer's comments).

Currently, your manuscript and abstract do not clearly state that your experiments only consider ice sheet model projections where MISI is already under way. Thus, generalization of your conclusions to other modes of ice sheet response (e.g. linear trend, equilibrium) remains difficult. I think it is important for the reader to understand this from the get-go (including the abstract). While you added paragraphs (l367 - 371 & 571 - 578) discussing MISI I do not see “a scenario with slower retreat without a collapse” (l573).

Figure 4 shows the experiments with ongoing MISI and the supplements don’t include the no-collapse case either. Maybe a change in the wording alongside quantification of grounding line retreat paint a clearer picture compared to general terms such as "collapse" vs. "slower retreat".

Sentences such as l575-577 "If the low-emission scenario were to produce less grounding-line retreat than in our current simulations, the region over which ice mass loss occurs cannot be very different from the low emission case" are unclear and not sufficient to support your statement "Hence, the preferred ELRA and 1D GIA models are also expected to remain applicable."

All figures in the supplement show ice sheet configurations with substantial grounding line retreat of Thwaites and PIG and Figure 4 does not seem to include a scenario without ongoing MISI judging from the sea level contributions displayed. Please add the SSP1-2.6 experiments without collapse to the timeseries figures (at least Fig. 4). So far Fig. 5 seems to be the only figure showing results from SSP1-2.6 forcing albeit with ongoing MISI in WAIS and Wilkes.

We are looking forward to receive your revised manuscript.


With best regards,

Johannes Sutter

Dear Dr. van Calcar,

thank you very much for your response to both the reviewer and the editor comments. Below i list a number of remaining minor points which include clarifications on the initial model state/spinup you employ. Thank you for your patience in the review process.

With kind regards,

Johannes Sutter

Review by Editor : Approximating 3D bedrock deformation in an Antarctic ice-sheet model for sea-level projections; van Calcar et al.

L115 I suggest to remove “and it shows notable discrepancies compared to the 3D GIA model it was benchmarked against” as such a statement would require a more robust analysis of these discrepancies. I assume all models of reduced complexity will lead to notable discrepancies compared to 3D GIA over long timescales.

L115-116 “so compute” -> “computing”

L169 “a quadratic local law” -> “a quadratic local parameterization”

L171 I suggest to change “The model does simulate marine ice sheet instabilities, but not marine ice cliff instabilities” to “The model does not include a marine ice cliff instability parameterization”. MICI involves an explicit parameterization while MISI usually emerges as a result of the flow law, bedrock geometry etc.

L443: Regarding your response to the reviewer comment “L.360: PISM and Kori use parameterizations for sub-shelf and grounding-line melting that yield similar sensitivities to those used here. I don’t think this explains the large sensitivity of IMAU-ICE that you observe.”

Your response highlights the impact of initialization, which, as you correctly state, would lead to different basal melt rates even for the same basal melt parameterization. However, your explanation covers the imprint of initialization not the melt parameterization itself. If you want to stick to this response then you would have to remove the part of the sentence stating : “the basal melt scheme, and the melt parametrization at the grounding line”. So far, your paragraph explaining the high SLE in your simulation is very general. There is no information on your initial model state expect that present bedrock topography is used, so the reader cannot evaluate this aspect of your work. Did you e.g. use inversion or a freely evolving spinup? I would suggest either explicitly stating that you did not investigate the reasons for the high model sensitivity (and that it is not a focus of this paper) or providing a more elaborate analysis in the supplements by providing a figure of the initial model state (e.g. thickness differences to present day observations, model drift). If you use an initial state from a previous publication, then it would be helpful to state this somewhere in the methods. If not it, I would suggest adding a short paragraph on the model initialization and a figure showing the present day initial state in the supplements.

L448-450: the forcing e.g. in the Amundsen Sea used in your study could be compared to the forcings used e.g. in ISMIP6. This way you could either remove the sentence “Other climate models provide significantly different thermal forcings in terms of magnitude and spatial distribution, which might not be captured by the two climate models used in this study” or explicitly state the quantitative differences in the thermal forcing and whether they lead to the high SLE contributions in your simulations. I suggest either quantifying the differences in your forcing (e.g. the thermal forcing in our simulations is substantially higher compared to ..) or remove the sentence.

L 450-455 I appreciate the modifications to this sentence, but I do not think you can state “retreat significantly whereas the rest of the West Antarctic Ice Sheet is relatively stable”. Figure 4 shows a continued ice loss until the end of your simulation, i.e. ongoing retreat. Chances are that the remaining parts of the WAIS in SSP1-2.6 will melt as well over time would the simulation(s) run longer. My point being, that stability cannot be inferred from your figures.

L457 “Together, these simulations capture both rapidly retreating and relatively stable drainage basins across
different Antarctic regions.” I think this sentence is valid for many simulations as there will be always relatively stable drainage basins at least for the forcings and timescales considered here. The relevant basins here are however the major marine drainage basins which are associated with MISI. Those are all retreating or collapsing in your model ensemble for both chosen forcings.

L626 “A widely used mantle viscosity for a 1D Earth model is 1021 Pa∙s (hereafter referred to as 1D21) (Gomez et al., 2015; Konrad et al., 2015; Rodehacke et al., 2020; Golledge et al., 2019)”

I am not sure whether the references are fitting here. E.g. Konrad et al 2015 don’t use a single viscosity value, Golledge et al 2015 don’t mention the viscosity they use, but they indeed use 10^21 in previous publications. Gomez et al use a range of viscosities. So, strictly, none of the references show that 1D21 is used widely. Some model studies make an effort to treat this viscosity as a free tuning parameter. I would suggest either referencing a study showing that 1D21 is widely used or referencing studies which explicitly use 1D21.

Fig 5: It is good that you try to be consistent with the color-code for the grounding line and the timeseries plots but maybe cmaps such as cm_crameri.broc or cmocean.cm.tarn allow for readability of both grounding line and cmap. However, as one reviewer pointed out, it could ease interpretation and readability if you would use a different colormap for bedrock response and ice thickness changes. Also, currently only thickness changes of grounded ice at the end of the respective simulation are shown while the majority of the bedrock response is due to the thickness change over all areas (i.e. including previously ice covered regions).