the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Jan 2025
Interactive coupling of a Greenland ice sheet model in NorESM2
Abstract. On the backdrop of observed accelerating ice sheet mass loss over the last few decades, there is growing interest in the role of ice sheet changes in global climate projections. In this regard, we have coupled the Norwegian Earth System Model (NorESM) with the Community Ice Sheet Model (CISM) and have produced an initial set of climate projections including an interactive coupling with a dynamic Greenland ice sheet. Our focus in this manuscript is the description of the coupling, the model setup and the initialisation procedure. To illustrate the effect of the coupling, we have further performed one chain of experiments under historical forcing and subsequently under high future greenhouse gas forcing (SSP5-8.5) until 2100 and extended until 2300. We find a limited impact of the dynamical ice sheet changes on the global response of the coupled model under the given forcing and experimental setup when comparing to a standard CMIP6 simulation of NorESM with a fixed ice sheet.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3891 KB)
-
Supplement
(219 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3891 KB) - Metadata XML
-
Supplement
(219 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
- RC1: 'Comment on egusphere-2024-3045', Anonymous Referee #1, 05 Feb 2025
-
RC2: 'Comment on egusphere-2024-3045', Anonymous Referee #2, 18 Feb 2025
Review of Goelzer, Langebroek, Born, Hofer, Haubner, Petrini, Leguy, Lipscomb, and Thayer-Calder: “Interactive coupling of a Greenland ice sheet model in NorESM2." (GMD, Paper: egusphere-2024-3045)
The manuscript of Goelzer and others describes the coupling between the Norwegian Earth System Model (NorESM) and the Community Ice Sheet Model (CISM), where the latter represents the Greenland ice sheet. The focus of the paper is the implementation of the coupling, a short description of the initialization, and a brief analysis of performed simulations starting in 1850 and ending in the year 2300, focusing on the future warming scenario following an extended SSP5-8.5 scenario. They briefly show the influence of the ice sheet interaction on the climate and the ice sheet evolution. I'm particularly intrigued by the related paper in preparation (Haubner et al.) and look forward to a more in-depth analysis.
It was an absolute pleasure to read the well-structured and prepared manuscript. The figures are of good quality, necessary, and informative. This work is highly relevant given the number of groups working on the forefront topic of the interaction between Earth System models and dynamical ice sheet models. This work is also intriguing for the general ice sheet modeling community and those using ice sheet model projections to determine the future sea level.
I recommend the publication of the manuscript after minor corrections.
General comments
The manuscript is well-organized and written.
In the coupled model, the Greenland ice sheet has a limited impact on global climatic conditions if we ignore the contribution of a declining ice sheet to the sea level. Is it characteristic of the used model system, or would the authors generalize these results? If we turn towards the sea level response, do the authors detect different sea level responses between simulations where CESM interacts with NorESM and where it "only" receives the forcing from an uncoupled NorESM simulation – without providing feedback? The latter would be comparable with typical ISMIP standalone simulations.
I'm unsure about GMD’s standards, but you may please check the consistent use of “e.g.” versus “e.g.,” as well as “ … and …” versus “…, and …” in lists.
Specific comments
Main document
Page 3, Line 97 (P3, L97): Since you are using an unequal on a variable-thickness sigma coordinate system, I wanted to ask if you have limited the vertical resolution of the lowest layer to avoid numerical issues. Furthermore, what is the typical and minimum lowest layer thickness?
P3, L99: I’m confused about the conflicting information about a domain on a polar stereographic projection while the horizontal resolution shall be 4 km (e.g., P4, L21). Please clarify.
P4, L15: Does the SMB calculation allow for positive and negative sublimation?
P4, L33: What justifies the uniform lapse rate of -6 K km-1? Please also check the provided unit "°K/km," which might contain a mixture of Degrees Celsius and Kelvin.
P4, L41–42: You write, "… for lower snowpack depth, the accumulated snow does not directly contribute to the SMB." What is meant by direct contribution, and what would be an indirect contribution? Please clarify.
P4, L44–45: You mention "a horizontal bilinear interpolation and a linear vertical interpolation between" the models. It raises the question of whether the order of these interpolations influences the results. If so, how big is the difference?
P5, L47: In the model setup, masks are used "for the accumulation region, and one for the ablation region." It appears these are computed for each year. Are these masks changing within a year? Please clarify.
P5, L67: You "update the topography every five years" in your simulations. I understand that the commonly small changes justify these five years. Nevertheless, would it be better to tricker updating the topography once an orography change exceeds a given threshold, e.g., |Δh(x,y)| > hthreshold?
P5, L77–78: Do the authors mean "Solid ice fluxes are cumulated and passed to the ocean annually" as a mean flux?
P9, L59: What are the starting carbon dioxide emissions in the year 2300 when the emission starts to decline linearly? Therefore, the authors may write: "... reduced linearly starting from XX Gt year-1 in 2100 to … .”
P10, L80–81, L82, L84: You may help the reader to link the provided information in the text to the related subfigure. For instance, the authors may write:"… directly by NorESM2-EC (Fig. 4a, NorESM ...) compared ... (Fig. 4b, NorESM2-MAR, Fettweis et al., 2017). This … for the same period (Fig. 4C, ERA5-MAR),… ."
P13, Figure 7c: If I understand correctly, a changing ocean forcing is not implemented. Therefore, what drives the elevation reduction in northeast Greenland at the mouth of the 79-glacier (nioghalvfjerds)? Might it be related to the tuning of the ice sheet's basal conditions?
P14, L94–95: I'm with you that "Reconstructions of the climate and ice sheet states further back in time … would be very useful in this context." Is the work of Kjær et al. (2012) and Bjørk et al. (2012) relevant in this respect? Could the authors please provide information on what they would like to obtain from the community?
P14, L104: Is the result that "freshening due to ice sheet meltwater fluxes has little additional effect" on the AMOC a consequence of the analysis of Mikolajewicz and Maier-Reimer (1994)?
P16, L65: Please check the indent of citations.
Figure
Figure 1: What do the arrows represent? Please clarify. Also, the authors should consider skipping all arrows that are not active instead of having one disabled arrow, e.g., the strikethrough arrow.
Figure 2: In subfigure c), the color bar indicates that the value range does not exceed surface elevation differences of ±250 m. If not, please adjust the colorbar or mention it in the figure caption. Also, check this issue for the remaining figures, e.g., four (4) and seven (7).
Figure 4: Since the caption says that all "fields are masked to the modeled ice sheet area in NorESM2 at the end of the year 2014," it raises the question of whether the ice sheet has retreated in some areas. If so, please indicate lost ice.
Figure 5: What does the gray-shaded area mark? Please clarify it in the figure caption.
Figure 6: In the figure caption, please add the information that the air temperature and surface salinity are global, e.g., "a) global 2-m air temperature" and " d) global sea surface salinity."
Bibliography
Bjørk, A. A., Kjær, K. H., Korsgaard, N. J., Khan, S. A., Kjeldsen, K. K., Andresen, C. S., Box, J. E., Larsen, N. K., & Funder, S. (2012). An aerial view of 80 years of climate-related glacier fluctuations in southeast Greenland. Nature Geoscience, 5, 427–432. https://doi.org/10.1038/ngeo1481
Kjær, K. H., Khan, S. A., Korsgaard, N. J., Wahr, J., Bamber, J. L., Hurkmans, R., van den Broeke, M., Timm, L. H., Kjeldsen, K. K., Bjork, A. A., Larsen, N. K., Jorgensen, L. T., Faerch-Jensen, A., & Willerslev, E. (2012). Aerial Photographs Reveal Late-20th-Century Dynamic Ice Loss in Northwestern Greenland. Science, 337(6094), 569–573. https://doi.org/10.1126/science.1220614
Mikolajewicz, U., & Maier-Reimer, E. (1994). Mixed boundary conditions in ocean general circulation models and their influence on the stability of the model’s conveyor belt. Journal of Geophysical Research, 99(C11), 22633–22644. https://doi.org/10.1029/94JC01989
- AC1: 'Final author comments on egusphere-2024-3045', Heiko Goelzer, 15 Apr 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3045', Anonymous Referee #1, 05 Feb 2025
Goelzer et al. present the inclusion of a dynamic ice sheet model (namely CISM) into the NorESM2 general circulation model. They describe the methodology of the coupling and their initialisation procedure. They finally show coupled ice sheet – climate model experiments for the historical period and into the future, until 2300. The paper is very clear and nicely written. However, it feels a bit short and does not provide much comparison with previous works and does not provide a lot of material to fully grasp the limit of the approach chosen. Also the main finding is that the interactive ice sheet does not matter much but in fact this been looked with very simple integrated metrics where it would have been useful to discuss spatial patterns of Northern Hemisphere climate change induced by the ice sheet retreat. All this information would be useful for other groups aiming at doing a similar work with alternative climate / ice sheet models. More details in the following.
Major comments
- Discussion with respect to CESM2-CISM previous coupling. The authors acknowledge that they share various important components and methodologies with the CESM2-CISM model. To my understanding, the coupling strategy is identical and only the initialisation strategy is different. Is this correct? If yes I think it should be clarified stronger to put more weigh on the things that are really different. At the moment it is not very clear for example if the elevation class methodology shows subtle differences with Muntjewerf et al. (2021) or if it is exactly the same. P2L70 for example we can read “we describe our novel coupled modelling framework” but then, reading the text I cannot find any difference with the CESM2-CISM coupling. I understand that CESM2 and NorESM2 are two different models, with different oceanic components and different climate sensitivities, so I support the paper. But I think it would be more useful to stress on what you have kept, what you have not, and why.
- Discussion on the initialisation strategy. I understand that it was a practical choice to impose an ice sheet mask and perpetual calving for floating points in order to maintain a good agreement with respect to present-day observed ice sheet topography. This point is clearly explained in the methods and it is shortly discussed in the discussion section. However I think it deserves further attention. As the authors acknowledge, the climate model can present large atmospheric biases (see also my next comment) and these biases are somehow compensated by ice sheet dynamics. The way I see it is that an overall overestimation of SMB (particularly true for North and East of the ice sheet) will translate in reduced basal drag in order to push the excess ice outside the observed ice mask, where this excess ice is simply removed independently from the simulated SMB / oceanic characteristics. It is possible that the “good” match with present-day topography hides efficiently the cold bias of NorESM. In turns, it is also possible that this limits the ice sheet model response to future warming. I wonder if it would have been possible to use some king of bias correction to evaluate this? Even a simple one? Right now I do not find the coupling very convincing since it relies heavily on this “masking” method. Again, I understand why it has been done this way, but at least I expect to see clear and documented justification of it. For example even a control (continuation of pre-industrial) coupled simulation without any masking to show the unwanted ice extension would be useful. I think this problem is common to every modelling groups and I think it is nice to document how far we are and what we can do and we cannot.
- Presentation of model biases. It would be useful to present the climate model biases in terms of regional temperature (annual and summer), precipitation, but perhaps also short/long wave etc. At present there is only a comparison with MAR SMB, restricted over the ice sheet but it would be interesting to also know what happens outside the ice sheet (tundra).
- Impact of interactive ice sheets on the simulated future climate. It is a bit overlooked I think since only simple integrated metrics are shown. I think that a map of regional temperature change w/o the coupling would be useful for instance.
Minor comments / questionsP2L50. Climate biases and climate downscaling are two different problems. We can do a very neat downscaling but if the biases are strong we will not have a high-quality surface mass balance.
P4L15. Why not total precipitation instead? Rain is also contributing to SMB via refreezing.
P4L33. Why this value? Why not the model vertical lapse rate? Do you have runs with different values of this and check its importance on SMB?
P4L35. Do you have a justification for no-change in relative humidity? It seems not really driven by observations... Again, do you have experiments where you impose a vertical gradient for humidity (specific may be more appropriate). Perhaps it would have made more sense to me to keep the same specific humidity in the vertical but recomputing the relative humidity (hence producing precipitation eventually).
P5L67. It might useful to run some additional sensitivity tests with different update frequencies. Is there a limit from which we observe unwanted jumps for some climatic variables?
P5L60-69. Do the model includes some kind of ice mask where the albedo cannot reach too low values. Have you tried to separate the effects of changing the ice mask and changing the albedo for the future simulations?
P5L78. Homogeneously distributed within the year?
P7 Figure 2. The agreement is really nice. Can you provide an estimate of the drift for this? You could show the map of thickness difference with respect to observations at the end of the cControl experiment.
P7 Figure 2. In the supplement you could show the inferred basal drag coefficient with some details on the law you used. Also a map of observed velocities with respect to simulated ones will be useful, since the climate model biases are compensated by ice dynamics.
P9 Figure 3. Temperature and precip: is it a spatial average? Over which domain?
P10 Figure 4. I imagine that the snow albedo can be tuned somehow in the model. Can you provide some info on how it is computed? Have you tried to tune it? How good is it with respect to satellite observations, and/or MAR?
P12L36. Does NorESM2 includes some freshwater flux scenarios when performing future projections? Do you still have these when you activate the coupling? How large is the scenario compared to your estimated fluxes?
P12 Figure 6. T2m, SSS: global average?
P12L36. Since you have a strong change in AMOC I think it would be useful to have more discussion on the freshwater flux and their impact. Do you some different spatial structures with/without the coupling when looking at the North Atlantic surface and subsurface temperatures? If yes do they come from topography/ice mask changes or freshwater flux?
P13L49. I think it is not enough to simply mention the paper in preparation. Since AMOC is the only change you have with respect to the standard uncoupled simulation, it would be nice to have some plots of regional oceanic changes.
P13 Figure 7. c1850: should be cControl instead?
P14L04. But this might simply results from the limited ice sheet response in these simulations.
Technical correctionsP1L33. Surface albedo is meant?
P12L40. Not useful to cite this since there is no preprint available.
Citation: https://doi.org/10.5194/egusphere-2024-3045-RC1 -
RC2: 'Comment on egusphere-2024-3045', Anonymous Referee #2, 18 Feb 2025
Review of Goelzer, Langebroek, Born, Hofer, Haubner, Petrini, Leguy, Lipscomb, and Thayer-Calder: “Interactive coupling of a Greenland ice sheet model in NorESM2." (GMD, Paper: egusphere-2024-3045)
The manuscript of Goelzer and others describes the coupling between the Norwegian Earth System Model (NorESM) and the Community Ice Sheet Model (CISM), where the latter represents the Greenland ice sheet. The focus of the paper is the implementation of the coupling, a short description of the initialization, and a brief analysis of performed simulations starting in 1850 and ending in the year 2300, focusing on the future warming scenario following an extended SSP5-8.5 scenario. They briefly show the influence of the ice sheet interaction on the climate and the ice sheet evolution. I'm particularly intrigued by the related paper in preparation (Haubner et al.) and look forward to a more in-depth analysis.
It was an absolute pleasure to read the well-structured and prepared manuscript. The figures are of good quality, necessary, and informative. This work is highly relevant given the number of groups working on the forefront topic of the interaction between Earth System models and dynamical ice sheet models. This work is also intriguing for the general ice sheet modeling community and those using ice sheet model projections to determine the future sea level.
I recommend the publication of the manuscript after minor corrections.
General comments
The manuscript is well-organized and written.
In the coupled model, the Greenland ice sheet has a limited impact on global climatic conditions if we ignore the contribution of a declining ice sheet to the sea level. Is it characteristic of the used model system, or would the authors generalize these results? If we turn towards the sea level response, do the authors detect different sea level responses between simulations where CESM interacts with NorESM and where it "only" receives the forcing from an uncoupled NorESM simulation – without providing feedback? The latter would be comparable with typical ISMIP standalone simulations.
I'm unsure about GMD’s standards, but you may please check the consistent use of “e.g.” versus “e.g.,” as well as “ … and …” versus “…, and …” in lists.
Specific comments
Main document
Page 3, Line 97 (P3, L97): Since you are using an unequal on a variable-thickness sigma coordinate system, I wanted to ask if you have limited the vertical resolution of the lowest layer to avoid numerical issues. Furthermore, what is the typical and minimum lowest layer thickness?
P3, L99: I’m confused about the conflicting information about a domain on a polar stereographic projection while the horizontal resolution shall be 4 km (e.g., P4, L21). Please clarify.
P4, L15: Does the SMB calculation allow for positive and negative sublimation?
P4, L33: What justifies the uniform lapse rate of -6 K km-1? Please also check the provided unit "°K/km," which might contain a mixture of Degrees Celsius and Kelvin.
P4, L41–42: You write, "… for lower snowpack depth, the accumulated snow does not directly contribute to the SMB." What is meant by direct contribution, and what would be an indirect contribution? Please clarify.
P4, L44–45: You mention "a horizontal bilinear interpolation and a linear vertical interpolation between" the models. It raises the question of whether the order of these interpolations influences the results. If so, how big is the difference?
P5, L47: In the model setup, masks are used "for the accumulation region, and one for the ablation region." It appears these are computed for each year. Are these masks changing within a year? Please clarify.
P5, L67: You "update the topography every five years" in your simulations. I understand that the commonly small changes justify these five years. Nevertheless, would it be better to tricker updating the topography once an orography change exceeds a given threshold, e.g., |Δh(x,y)| > hthreshold?
P5, L77–78: Do the authors mean "Solid ice fluxes are cumulated and passed to the ocean annually" as a mean flux?
P9, L59: What are the starting carbon dioxide emissions in the year 2300 when the emission starts to decline linearly? Therefore, the authors may write: "... reduced linearly starting from XX Gt year-1 in 2100 to … .”
P10, L80–81, L82, L84: You may help the reader to link the provided information in the text to the related subfigure. For instance, the authors may write:"… directly by NorESM2-EC (Fig. 4a, NorESM ...) compared ... (Fig. 4b, NorESM2-MAR, Fettweis et al., 2017). This … for the same period (Fig. 4C, ERA5-MAR),… ."
P13, Figure 7c: If I understand correctly, a changing ocean forcing is not implemented. Therefore, what drives the elevation reduction in northeast Greenland at the mouth of the 79-glacier (nioghalvfjerds)? Might it be related to the tuning of the ice sheet's basal conditions?
P14, L94–95: I'm with you that "Reconstructions of the climate and ice sheet states further back in time … would be very useful in this context." Is the work of Kjær et al. (2012) and Bjørk et al. (2012) relevant in this respect? Could the authors please provide information on what they would like to obtain from the community?
P14, L104: Is the result that "freshening due to ice sheet meltwater fluxes has little additional effect" on the AMOC a consequence of the analysis of Mikolajewicz and Maier-Reimer (1994)?
P16, L65: Please check the indent of citations.
Figure
Figure 1: What do the arrows represent? Please clarify. Also, the authors should consider skipping all arrows that are not active instead of having one disabled arrow, e.g., the strikethrough arrow.
Figure 2: In subfigure c), the color bar indicates that the value range does not exceed surface elevation differences of ±250 m. If not, please adjust the colorbar or mention it in the figure caption. Also, check this issue for the remaining figures, e.g., four (4) and seven (7).
Figure 4: Since the caption says that all "fields are masked to the modeled ice sheet area in NorESM2 at the end of the year 2014," it raises the question of whether the ice sheet has retreated in some areas. If so, please indicate lost ice.
Figure 5: What does the gray-shaded area mark? Please clarify it in the figure caption.
Figure 6: In the figure caption, please add the information that the air temperature and surface salinity are global, e.g., "a) global 2-m air temperature" and " d) global sea surface salinity."
Bibliography
Bjørk, A. A., Kjær, K. H., Korsgaard, N. J., Khan, S. A., Kjeldsen, K. K., Andresen, C. S., Box, J. E., Larsen, N. K., & Funder, S. (2012). An aerial view of 80 years of climate-related glacier fluctuations in southeast Greenland. Nature Geoscience, 5, 427–432. https://doi.org/10.1038/ngeo1481
Kjær, K. H., Khan, S. A., Korsgaard, N. J., Wahr, J., Bamber, J. L., Hurkmans, R., van den Broeke, M., Timm, L. H., Kjeldsen, K. K., Bjork, A. A., Larsen, N. K., Jorgensen, L. T., Faerch-Jensen, A., & Willerslev, E. (2012). Aerial Photographs Reveal Late-20th-Century Dynamic Ice Loss in Northwestern Greenland. Science, 337(6094), 569–573. https://doi.org/10.1126/science.1220614
Mikolajewicz, U., & Maier-Reimer, E. (1994). Mixed boundary conditions in ocean general circulation models and their influence on the stability of the model’s conveyor belt. Journal of Geophysical Research, 99(C11), 22633–22644. https://doi.org/10.1029/94JC01989
- AC1: 'Final author comments on egusphere-2024-3045', Heiko Goelzer, 15 Apr 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 904 | 86 | 18 | 1,008 | 41 | 26 | 36 |
- HTML: 904
- PDF: 86
- XML: 18
- Total: 1,008
- Supplement: 41
- BibTeX: 26
- EndNote: 36
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Petra M. Langebroek
Andreas Born
Stefan Hofer
Konstanze Haubner
Michele Petrini
Gunter Leguy
William H. Lipscomb
Katherine Thayer-Calder
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3891 KB) - Metadata XML
-
Supplement
(219 KB) - BibTeX
- EndNote
- Final revised paper
Goelzer et al. present the inclusion of a dynamic ice sheet model (namely CISM) into the NorESM2 general circulation model. They describe the methodology of the coupling and their initialisation procedure. They finally show coupled ice sheet – climate model experiments for the historical period and into the future, until 2300. The paper is very clear and nicely written. However, it feels a bit short and does not provide much comparison with previous works and does not provide a lot of material to fully grasp the limit of the approach chosen. Also the main finding is that the interactive ice sheet does not matter much but in fact this been looked with very simple integrated metrics where it would have been useful to discuss spatial patterns of Northern Hemisphere climate change induced by the ice sheet retreat. All this information would be useful for other groups aiming at doing a similar work with alternative climate / ice sheet models. More details in the following.
Major comments
- Discussion with respect to CESM2-CISM previous coupling. The authors acknowledge that they share various important components and methodologies with the CESM2-CISM model. To my understanding, the coupling strategy is identical and only the initialisation strategy is different. Is this correct? If yes I think it should be clarified stronger to put more weigh on the things that are really different. At the moment it is not very clear for example if the elevation class methodology shows subtle differences with Muntjewerf et al. (2021) or if it is exactly the same. P2L70 for example we can read “we describe our novel coupled modelling framework” but then, reading the text I cannot find any difference with the CESM2-CISM coupling. I understand that CESM2 and NorESM2 are two different models, with different oceanic components and different climate sensitivities, so I support the paper. But I think it would be more useful to stress on what you have kept, what you have not, and why.
- Discussion on the initialisation strategy. I understand that it was a practical choice to impose an ice sheet mask and perpetual calving for floating points in order to maintain a good agreement with respect to present-day observed ice sheet topography. This point is clearly explained in the methods and it is shortly discussed in the discussion section. However I think it deserves further attention. As the authors acknowledge, the climate model can present large atmospheric biases (see also my next comment) and these biases are somehow compensated by ice sheet dynamics. The way I see it is that an overall overestimation of SMB (particularly true for North and East of the ice sheet) will translate in reduced basal drag in order to push the excess ice outside the observed ice mask, where this excess ice is simply removed independently from the simulated SMB / oceanic characteristics. It is possible that the “good” match with present-day topography hides efficiently the cold bias of NorESM. In turns, it is also possible that this limits the ice sheet model response to future warming. I wonder if it would have been possible to use some king of bias correction to evaluate this? Even a simple one? Right now I do not find the coupling very convincing since it relies heavily on this “masking” method. Again, I understand why it has been done this way, but at least I expect to see clear and documented justification of it. For example even a control (continuation of pre-industrial) coupled simulation without any masking to show the unwanted ice extension would be useful. I think this problem is common to every modelling groups and I think it is nice to document how far we are and what we can do and we cannot.
- Presentation of model biases. It would be useful to present the climate model biases in terms of regional temperature (annual and summer), precipitation, but perhaps also short/long wave etc. At present there is only a comparison with MAR SMB, restricted over the ice sheet but it would be interesting to also know what happens outside the ice sheet (tundra).
- Impact of interactive ice sheets on the simulated future climate. It is a bit overlooked I think since only simple integrated metrics are shown. I think that a map of regional temperature change w/o the coupling would be useful for instance.
Minor comments / questions
P2L50. Climate biases and climate downscaling are two different problems. We can do a very neat downscaling but if the biases are strong we will not have a high-quality surface mass balance.
P4L15. Why not total precipitation instead? Rain is also contributing to SMB via refreezing.
P4L33. Why this value? Why not the model vertical lapse rate? Do you have runs with different values of this and check its importance on SMB?
P4L35. Do you have a justification for no-change in relative humidity? It seems not really driven by observations... Again, do you have experiments where you impose a vertical gradient for humidity (specific may be more appropriate). Perhaps it would have made more sense to me to keep the same specific humidity in the vertical but recomputing the relative humidity (hence producing precipitation eventually).
P5L67. It might useful to run some additional sensitivity tests with different update frequencies. Is there a limit from which we observe unwanted jumps for some climatic variables?
P5L60-69. Do the model includes some kind of ice mask where the albedo cannot reach too low values. Have you tried to separate the effects of changing the ice mask and changing the albedo for the future simulations?
P5L78. Homogeneously distributed within the year?
P7 Figure 2. The agreement is really nice. Can you provide an estimate of the drift for this? You could show the map of thickness difference with respect to observations at the end of the cControl experiment.
P7 Figure 2. In the supplement you could show the inferred basal drag coefficient with some details on the law you used. Also a map of observed velocities with respect to simulated ones will be useful, since the climate model biases are compensated by ice dynamics.
P9 Figure 3. Temperature and precip: is it a spatial average? Over which domain?
P10 Figure 4. I imagine that the snow albedo can be tuned somehow in the model. Can you provide some info on how it is computed? Have you tried to tune it? How good is it with respect to satellite observations, and/or MAR?
P12L36. Does NorESM2 includes some freshwater flux scenarios when performing future projections? Do you still have these when you activate the coupling? How large is the scenario compared to your estimated fluxes?
P12 Figure 6. T2m, SSS: global average?
P12L36. Since you have a strong change in AMOC I think it would be useful to have more discussion on the freshwater flux and their impact. Do you some different spatial structures with/without the coupling when looking at the North Atlantic surface and subsurface temperatures? If yes do they come from topography/ice mask changes or freshwater flux?
P13L49. I think it is not enough to simply mention the paper in preparation. Since AMOC is the only change you have with respect to the standard uncoupled simulation, it would be nice to have some plots of regional oceanic changes.
P13 Figure 7. c1850: should be cControl instead?
P14L04. But this might simply results from the limited ice sheet response in these simulations.
Technical corrections
P1L33. Surface albedo is meant?
P12L40. Not useful to cite this since there is no preprint available.