Review of “Ice sheet model simulations reveal polythermal ice conditions existed across the NE USA during the Last Glacial Maximum” by Cuzzone et al.

I apologise for the late return of this review, and any inconvenience it may have caused the authors or editor.

Cuzzone et al. here present simulations of the Laurentide Ice Sheet, using small regions of high horizontal resolution to explore the basal thermal regime in areas of variable topography. Simulations at a continental scale require such coarse resolution that the thermal effect of mountainous regions may be excluded. In this paper, extremely high resolution simulations demonstrate a polythermal bed and sharp thermal gradients in regions across the eastern Laurentide Ice Sheet.

Overall, the authors present some impressive simulations with interesting results. The work is well-described and presented, with some minor corrections to the figures suggested below. The research is clearly a good fit for the journal, and the work is worthy of publication.

Major points

1. A mini-ensemble is presented, with simulations using boundary conditions from 5 different climate models, but I am left wondering if we are confident that the variation between climate model output is certainly the first-order uncertainty to be explored. Perhaps this is a reasonable assumption, but I think this should be discussed in the text. If varying other parameters (such as elements of the PDD scheme of the basal sliding law) would be produce larger variation, plots like figure 3 may provide a false confidence in the certainty of the results. A subpoint is that it would be useful to have a sense of the computational expense of these simulations, both to help the planning of future studies and perhaps to further justify the limited size of the ensemble.

1. A broader justification of the work would be useful in both the Introduction and Conclusion. Understanding the thermal effect of small-scale topography may be important for understanding the past evolution of LGM ice sheets, and future evolution of Greenland and Antarctica. But that is my interpretation, and it would be better to state your justification more explicitly. There is mention of models providing geologic constraints across a wide area, but I’m clear on what this would be useful for (because of my own ignorance, I’m sure!).

1. There are a few points where you compare your results to geologic evidence or other empirical evidence (some specific examples are highlighted below). A more robust quantitative comparison may be insightful, or at least a discussion of why this might not be possible.

Minor points

Figure 1: Could the model nesting be incorporated into this figure, rather than a separate figure in the SI?

Line 167: Adjustment for clarity? “…independent of the thermal state but this has been explored…”

Line 167: I’m not sure I’m following the final sentence of this paragraph. The differences between simulations with and without thermal-friction coupling are measurable but small? Could this be quantified?

Line 201: Is “downscale” the accepted word here? I (rightly or wrongly) interpret that as a statistical process.

Line 207: It seems odd that these locations are partly informed by state boundaries that didn’t exist during (and have no impact on) glaciation. Would a purely geomorphic apolitical process have identified the same location?

Line 228: And is this climate data also used as the surface temperature flux to calculate the ice sheet thermal state?

Line 228: Given the focus of the paper I think it is entirely reasonable PDD values are prescribed and unexplored, as long is it produces a SMB field that serves the simulations. An additional justification may be useful for some readers, though.

Section 2.2: Could this process be summarised in a diagram or table in the SI? I found myself drawing in my notebook to help keep track of the process.

Figure 2: Could colour-blind friendly ramps by used on ice velocity figures? The use of white stippling is nice.

Figure 4: This figure contains a lot of useful information but it takes some work to fully appreciate! To improve readability, you could include a location panel to save readers bouncing between figures? The position and scale of various axis and legend labels could be refined to increase readability. The figure is a bit small, but panel B doesn’t add much to panel E. If you removed B the remaining panels could be enlarged. A scale bar would be useful alongside the graticule.

Figure 5: As above.

Figure 6: And again.

Line 417: “Our results agree with geologic interpretations”. Is it possible to provide a more robust statistical or graphical comparison between your results and geological interpretations?

Line 424 onwards: It would be interesting to see a comparison of the simulated velocity field at different scales? If these regions are important I suppose it would have a measurable impact on the flux across the domain?

This is a really interesting manuscript that models the thermal status of the now-vanished Laurentide Ice Sheet (LIS) in the northeastern United States and Canada. Model results provide insight into the thermal conditions at the base of the ice sheet which can be compared to both geologic observations of glacial erosion and isotopic/geochemical measurements of proxies such as cosmogenic nuclides.

I am not a modeler so have focused most of my comments on the less technical side of the text although reading the modelling section did bring up several ideas that I have made comments about. Given my lack of expertise in the modelling domain I posed these as questions that hopefully the authors can address because many of the readers of this work may have similar questions.

I attach a marked up PDF from my reading and summarize below what I consider to be the most important points to address in revision of the manuscript. I believe addressing these suggestions will make the manuscript more readable to a broader audience and have tried where possible to offer specific guidance to the authors. If the authors have questions about my review, I encourage them to contact me. Paul Bierman

1. The manuscript should be more inclusive in its referencing of prior work. Some earlier papers are not yet cited and there are claims made that are without citations. I’ve made suggestions on the PDF and the authors can find more citations to add.
2. The abstract could be much stronger and more informative. There are ambiguous statements and words that leave the reader uncertain. I suggest a significant and careful edit on the abstract after the paper is revised. See specific suggestions on PDF.
3. While adding references, I suggest the authors look more deeply into the geomorphic and soils literature of New England. There are papers relating to preserved saprolite that are relevant as well as to striations at and near the peaks of most ranges suggesting that at some point the ice was indeed warm-based. Similarly there is striation mapping near coast. Adding a more inclusive discussion of the landscape will provide deeper context for their results. It will also illuminate the importance of transience something that must be key for the LIS but is not yet discussed (and should be). I allude to this in a question on the ms about model spin up.
4. The ms currently moves between spatial changes (plan view) and elevation changes but not in a coherent and predictable fashion. The story will be easier to follow and will benefit from more clarity and an organization structure that clearly and separately considers the elevation effect vs latitude effect on basal thermal regime.
5. From an ease of reading perspective, some paragraphs are really long and contain multiple foci. Breaking into shorter paragraphs with topic sentences will make the paper easier for the reader to understand. Similarly, there are places (I have tried to call them out in the text) where wording is inexact and thus could confuse readers. I have done my best in the PDF to suggest places it could be improved for clarity. There are sections of the discussion that read like methods and likely should be moved to the methods section or removed.
6. The long time of model relaxation seems to violate known ages and geologic history of ice in New England. This should be addressed explicitly. See extended note on PDF.
7. The precision of elevation boundaries to the nearest meter seems to overstate the actually precision of the modelling and the topographic data. Suggest rounding to at least the 10s if not the hundreds of meters given all of the uncertainties.
8. I agree strongly with the first reviewer that more direct comparison between the model and geologic/isotopic data sets would be very useful and strengthen the paper.