Summary

The manuscript presents simulations of Antarctica to investigate the sensitivity of the modeled thermal state to the boundary conditions and inversion method using the three-dimensional thermomechanical ice sheet model, ISSM. They focus on the influences caused by differences in existing geothermal heat flux maps and the effect of differences in the ice vertical velocity. Both GHF and vertical velocity are poorly constrained in models but are known to affect the thermal state. The authors provide a new set of model simulations with different combinations of GHF maps and vertical velocity parameterizations to generate 3D temperature fields. By comparing their modeled temperature fields to existing borehole temperature profiles, the authors conclude that vertical velocity has a greater influence on the thermal state than GHF. This new contribution is very compelling, since it implies that vertical velocity is critical to constrain in ice sheet models in order to accurately model the thermal state. However, the authors miss the opportunity for some additional analysis and discussion which will further strengthen their findings and narrative.

Major issues:

The main findings are ice-sheet scale conclusions about the effect of boundary conditions and model initialization on the thermal state, while the boreholes that are analyzed to make these conclusions are largely from the Siple coast, which has a unique thermal configuration (Englehardt 2004, Bougamont et al., 2015, Ng and Conway 2004, etc). The stagnation of Siple Coast ice streams (i.e. Kamb >150 years ago) exhibits an interesting thermal regime today that must contain memory of the past slow down. However, this would not be the case for most other parts of Antarctica. I think the paper should be strengthened in two ways.

• The authors should add more discussion on the Siple Coast model/observation discrepancies since it is very interesting. I think there is a missed opportunity to elaborate on the convex vs. concave temperature profiles for models vs. observations for the KIS and ER profiles (Fig. 2). I wonder if the difference in shape is an indication that the model’s lack of the long-term thermal state memory really matters for this region. I know this shortcoming is hinted at in the last paragraph of the discussion, but I think the authors miss the chance to add interesting discussion about what the discrepancies in model vs observed temperature profiles are telling us about the thermal regime. I’m not expecting new results, but I would like to see some speculation about the effects of boundary conditions vs. initialization approach (inversion with present day conditions, paleo spin-up, thermal steady state approximation, etc) in the discussion.
• If the focus of the paper is on the broadscale effects of GHF vs vertical velocity on thermal regime, then there are more borehole temperature profiles, which should be added to the analysis. There are more from East Antarctica such as Dome C, Lake Vostok, Talos Dome, South Pole. There are also more borehole temperature profiles in other parts of West Antarctica such as Kohnen and Byrd (maybe some others I am missing?).

Regarding the naming of “fast flow” and “slow flow” regions, somewhere early on in the manuscript, it should say that it isn’t possible to drill into most really fast flowing regions because deformation etc. prevent drilling. In the paper I would say “fast flow” only defines a unique subset of ice streams (WIS and BIS) where the flow regime supports drilling, so there are borehole temperature profiles for only those regions. Because of this, the “fast flow” conclusions may not apply for other parts of Antarctica. I might even recommend renaming “fast flow” to “Siple coast fast flow” or something like that throughout the manuscript to clarify this point.

There are known difference in 2m air temperature amongst reanalysis products such as ERA-interim, ERA5, RACMO, MERRA, MAR. Why do the authors choose ERA-Interim? The author’s test the effect of the basal boundary condition by changing GHF maps, while the surface boundary condition is never tested. It would be helpful to see additional simulations using different 2m air temperature products to see its effect on the vertical temperature profiles, even if this effect is less significant.

Also on the topic of surface temperature, in Fig. 2, it looks like the surface temperatures between the ISSM simulations and observation are not a great match for some boreholes (e.g. KIS, WIS, ER, Bruce, UC, ER, SD). I see in the text it says that the model surface temperatures are adjusted using an exponential decay function to better match the observations so I would like to know why there is this miss match. Would a different 2m air temperature map provide a better match to observations needing less correction (see comment above)?

How was 10m/yr threshold for surface velocity chosen for the IVz-nosliding experiment? Are the results sensitive to nudging this threshold? I am not necessarily asking for more model simulations here, but I would like to better understand the choice and its likely effect on the results.

Minor Issues:

In the abstract, fast flowing has a velocity threshold definition but not slow flowing. This should also be provided. Is it slower than 50 m/yr or something else? The reasoning behind these choices should also be explained somewhere near the beginning of the manuscript. For example, Dawson et al., 2022 uses 100 m/yr to define fast flowing regions. Are these thresholds a result of the velocities seen at the boreholes and some natural separation in the velocities/profiles?

It would be useful to see observed surface velocities at the boreholes reported in Table 1 so that the reader could see what borehole sites are within the model prescribed no sliding regions (as well as the fast and slow flow groupings). It’s hard to get this information from Fig. 1 right now… perhaps if a 10m/yr contour was drawn on the map then that could also work.

The organization of the subplots in Fig. 2 is somewhat confusing to me. I think the first row are the “linear” borehole profiles and then the rest are the more “concave” profiles. It also took me a while to see that the bottom row of profiles are the ones from the fast flowing regions. I think it would be helpful to see the profiles boxed into a slow flowing group (where IVZ-nosliding fits the observations better) and a fast flow group (where IVZ fits better), like the subtle separation in Table 1.

In Fig. 2, I would also find it helpful to see the boreholes that go all the way to the bed somehow indicated in Fig. 2.

Line 69: HO should be defined as higher order, with appropriate citation given.

Line 79: Be clearer about what the temperature rigidity relation is (e.g. give page # in Cuffey).

Line 99: Give version of BedMachine

For Fig. 4, I could see on the colorbar writing GHF instead of G to be more consistent with the text.

Paragraph starting on line 292: I am confused what is being reported here. I think you mean total melt water volume rather than melting rates. Melting rates should be reported in mm/yr or m/yr (such as the author’s Fig. 4 and Pattyn, Jouquin, Llubes) while total melt water volume is Gt/yr. This paragraph should be clarified what measure is being discussed. It would also be helpful if the authors state what their values are rather than just saying they are lower than Pattyn and higher than Llubes.

On line 308-309, elaborate more on the mass conservation -> melting rates -> understanding subglacial hydrology comment. I’m not sure I understand what this sentence is trying to say. I think the paragraph could use some rewriting to clarify the point.

Regarding the data availability statement, I believe that this paper would have broader impact and community interest, myself included, if the ISSM thermal model results from this analysis were made available as part of this study. I recommend providing a link to download the gridded temperature fields or simply providing the ISSM outputs for each run. This would enable further comparisons and validation of thermal modeling efforts.

Inconsistent with the use of Gt vs. Gton throughout the manuscript.

Writing style in general is mixing past and present tense, which should be resolved.

Technical corrections:

Typo on line 43: incompressbility  

Mistake on line 102: “three” -> “two”

Typo on line 114: exptrapolate

Typo on line 142: extra space after Y?

Typo on line 156: Missing a space (“datasets.Table”)

Typo on line 163: two commas

Typo on line 226: indicates -> indicate

Typo on line 306: delete “a” and “goods” -> “good”

Typo on line 340: velocitiy