My comment is uploaded as a PDF

Review of the paper “Coupling the regional climate MAR model with the ice sheet model PISM mitigates the melt-elevation positive feedback, by Delhasse et al.

The paper by Delhasse et al. presents a coupling methodology between the regional atmospheric model (forced by the large-scale CESM2 climate model) and the PISM ice-sheet model over the 1991-2200 period. Different levels of complexity for the coupling are compared: A real two-way coupling, a one-way coupling in which only the melt-elevation feedback is accounted for through an offline correction and, finally, the no coupling scheme where MAR is run with a fixed topography of the Greenland ice sheet and PISM is forced by the surface mass balance and the surface temperature computed by MAR. A focus is made on the role of the melt-elevation feedback and the barrier winds on the evolution of the Greenland ice sheet geometry (and the consequences on ice surface velocities), the surface mass balance and its components (runoff, meltwater production solid and liquid precipitation) and the contribution of Greenland ice sheet to sea level rise. Most often the results are generally presented in a fairly clear manner, although some parts of the paper require further explanation or analyses (see Specific Comments). This study is similar to the one conducted by Le chlec'h et al. (2019) that was based on the coupling between MAR (forced by the CMIP5 climate model MIROC5) and another ice-sheet model (GRISLI), despite a slightly different methodology to extend the warming scenario beyond 2100. I find it very instructive to compare two modelling studies done with similar approaches but different models. The discussion provides a detailed comparison explaining the different conclusions of these two studies. Overall, I find the study of Delhasse et al. deserves to be published after a number of revisions have been addressed.

Specific Comments

In several places, I think the paper would be of better quality if some clarifications or additional explanations were made.

1/ First of all, I think that a more extensive description of the PISM model is required. The description provided in the paper is too technical, if not incomprehensible,for a reader who is not familiar with ice sheet dynamics models. Although a number of references are given in which PISM is described, I think it is important to be able to understand the important points of this section without going to look for the references mentioned. Also, I think it is necessary to define a number of notions, such as SIA, SSA, the Mohr-Coulomb criterion, exponent of the sliding law, flow enhancement factor, etc... This list is obviously not exhaustive and some equations (and their meaning) would allow a better understanding of how the model works. For example, in figure 6 you represent the driving stresses. But how are these driving stresses calculated? Since you mention in the description of PISM the aspects related to the dynamics of the ice, it would be interesting to refer to them more in the analysis of the results, or else to keep in the description of the model only the characteristics used for the analysis of the simulations.

2/ The GRIP record does not extend to 125 ka as the signal was perturbed at the bottom of the ice core. While, I think that it shouldn’t affect much your results, this should be mentioned. Also, the original publication should be provided (Johnsen et al, 1992) instead of Johnson et al. (2019).

3/ The initialisation method of PISM and the coupled model could be better explained with a scheme. For example, you mention that during the first step of your initialisation PISM is forced with the 2D temperature anomalies of the last glacial cycle. Are these temperatures correspond to the internal temperatures? If so, these anomalies are not a forcing but an initial state for the next step of the initialisation. Other clarifications are needed: Explain why do you need to equilibrate the vertical temperature profile. This is not clear in the manuscript. How these anomalies are computed? What is the impact of using anomalies instead of absolute values?

4/ The offline correction method is a key component of the overall paper as is drives the melt-elevation feedback, but the description of the method is very short. Although, I think I understand the basic principles of the method, I found that more details would have been welcome, and possibly a scheme to better illustrate the method. Similarly, I think that the analysis of the results and the comment of Figure 8 are unclear (as well as the conclusions drawn from this figure) are unclear. For example, the authors mention "The dependence is no longer linear" (P12, L20). This just means that the temperature gradients are not the same in the two experiments, if I understood correctly. In other words, the altitude correction is not the same in both experiments. Again, this part of the analysis would be better understood if the authors had given more details about the correction method. Also, I am not sure I agree with the first conclusion of this analysis, namely, "the linear correction is no longer valid in the ice sheet margins". I don't think the analysis leads to this conclusion if the altitude correction is different in the two experiments. Perhaps it would have been better to plot the regressions for each experiment independently (not as anomalies) and to examine the slopes of the regression lines separately. Also, at the ice sheet margins, the behavior is not the same for altitude differences below ~350 m and above. This could be mentioned.

4/ The role of barrier winds seems to be a key element to explain the differences in the melt-elevation feedback between MAPI-2W and MAPI-1W. Could it be mentioned more explicitly in the abstract?

5/ Overall, the paper is written in understandable English. However, I think that the quality of the manuscript would be greatly improved if it were proofread by a native English speaker.

Minor comments and typo errors

Sections 2.1.3 and 2.3.1: Inisialisation à Initialisation

P1, L3: While MAR is able to diagnose the ice sheet surface mass balance thanks to the implementation of snow/ice layers, this is not the case for many atmospheric models. The statement “atmospheric models which can represent the SMB” should be tempered.

P1, L5: Remove “as well as atmospheric models”

P1, L10: corrected to à corrected for

P1, L11: extrapolated à interpolated

P1, L12: avoid à avoids or “prevents from a too expensive coupling”

P2, L26: Remove “as forcing”

P2, L31: “As the coupling is dependent on the used ISM à What do you mean? I guess that you mean that the results of your coupled simulations are model-dependent? Please, clarify.

P2, L35 “assess the offline method” should be changed in “assess the ability of the offline method…”

P3, L1 “feedback” à feedbacks

P3, L1: Replace “as well as “by “ “and which”

Section 2.13: It seems to be that the abbreviation ST has not been defined before. Also, explain why PISM needs to be forced with ST.

P4, L-25: The GRIP record was perturbed at the bottom and did not extent to -125 ka. This should affect your results so much but should be mentioned.  

P4, L26-27: Provide the resolutions in km, not in meters (as in the other parts of the manuscript).

P5, L5: “For the MAR variables, they are interpolated” à “The MAR variables are interpolated…”

P6: L13: MARref forced with CESM2 is also run with PISMsp5 topography (see previous sentence). This sentence is a bit confusing. Please rephrase. Also change PSIMsp5 in PISMsp5.

At different places in the manuscript there is confusion between interpolation and aggregation. Outputs coming from a higher resolution model are aggregated on a coarser model grid. Variables computed with a lower resolution model are interpolated on a finer model grid.

P7, L23: I feel that 10% is not so negligible.

Section 2.4: You should add a comment explaining why you deal sometimes with surface mass balance and sometimes with mass balance. Also, explain (maybe before section 2.4) the difference between both.

P7, L29: Components refer to SMB. Replace “their components” with “its components”

P8, L4: Replace -200 Gt.10^-3 by -200 10^-3 Gt (same thing for L2)

P8, L4-L5: The sentence is a bit confusing as the climate is not stabilized just before 2100.

P8, L12: Remove meltwater.

P9, L10: “synoptic features of the large-scale CESM2 forcing” à Could it be illustrated with a figure (in the Supplement part for example)?

P11, L7: underestimated à underestimates

P12, L1-2: This sentence is not clear. Je ne comprends pas pourquoi les résultats de MAPI-1W sont contraires à ceux de MAR. I think that the sentence should be rephrased.

P12,L2: Figure 1a indicates that the melt-elevation feedback is taken into account via an offline correction. This seems to be in contradiction with the text.

P12, L3 : become à becomes

P12, L17: To which altitude difference do you refer. This is not clear and justifies a more in-depth explanation of the offline correction method.

P13,L6: Remind the link between wind, T2m and SHF, as it appears to be a key mechanism to explain the difference in the melt-elevation feedback in your simulations.

P13, L6-7: This should be illustrated with a figure.

P13, L7-8: Fig9e represents a cross section of the topography and not an evolution of the topography. Please, rephrase. The mitigation of the melt-elevation feedback with elevation lowering is better illustrated with Figs 8a and S4a.

P14, L3: at the ice sheet margins / on the ice sheet margins

P14, L4: I don’t understand what you mean by “inside the ice sheet” in parenthesis

P17, L16: “Do not consider” à Please rephrase. I suggest something like “We must not consider” or another equivalent formulation.

P18, L14: sensibility à sensitivity

P18, L19: oppositely à in opposite ways ?

Figures

Figure 1a : See comment, P12, L2

Fig. 2: The grey band is not visible. Maybe you could choose a darker colour.

Fig 3: Green and red contours cannot be easily distinguished

Fig. 4 caption: (RU, in green) à (RU, in orange). Precise in the figure caption that solid lines correspond to the coupled experiment

Fig.9: Indicate what do the y-axes and x-axes represent (Figs.9a, 9c, 9e). The line thickness of the black lines in Figs.9b and 9d could be slightly increased.