The manuscript presents the results of a fully coupled Northern Hemisphere ice sheet—climate model applied to the last two glacial terminations. The manuscript is well-written and nicely illustrated. The description of the model, coupling and sensitivity analysis is mostly clear but could benefit from some minor additions. Overall, I enjoyed reading this paper and I am sympathetic to the aims. I am not suggesting the authors conduct additional experiments. I hope my comments help in improving the manuscript.

Comments

Alternative PGM ice sheet geometry:

The manuscript would benefit from a more detailed explanation of how the alternative ice sheet geometry has been applied. In the methods section, it is only briefly mentioned in L169 and in the results (L345 to L348). It would be valuable to explore the regional and large-scale impacts on the climate resulting from this new ice sheet configuration as well as its implications on the timing and on the deglaciation history during the TII.

Atmospheric resolution:

It would be beneficial to include a  discussion on the limitations due to the climate resolution. The simulations are based on the intermediate complexity climate model iLOVECLIM, with an atmospheric resolution of T21. Previous studies have established the implications of coarse-resolution climate models in the modelling during the last glacial maximum and the deglaciation (eg. Lofverstrom et al., 2018; Lohmann et al., 2021).

Other concerns:

To make the paper more accessible to a broader audience, including non-modellers, it may be helpful to explicitly state that the primary aim is not to precisely replicate the timing and pattern of deglaciation but rather to explore the model's sensitivity throughout both terminations. This clarification can aid in ensuring that readers from various backgrounds can appreciate the study's objectives and outcomes.

Technical comments:

L231. “In ?”

L245. its written “kyrs” while in some other parts of the text is written “kyr” (eg. L283). Moreover, in other parts is written “ka” (L292). Please check.

Figs. 1 - 13. It is written “kaBP” while in Figs 14 and 15 “ka BP”.

Fig 7. Keep the design of the other figures

Fig 12. Replace “rhe” for “the”

Fig. 13. Include legend 

Fig 14 and 15. Keep the design of the other figures

Lofverstrom, M., & Liakka, J. (2018). The influence of atmospheric grid resolution in a climate model-forced ice sheet simulation. The Cryosphere, 12(4), 1499-1510.

Lohmann, G., Wagner, A., & Prange, M. (2021). Resolution of the atmospheric model matters for the Northern Hemisphere Mid-Holocene climate. Dynamics of Atmospheres and Oceans, 93, 101206.

Summary:

The study by Quiquet and Roche analyzes various aspects of the climate and ice-sheet evolution in the last two glacial terminations using the intermediate complexity model iLOVECLIM with an interactive ice sheet component. Experiments are presented in which the model is integrated forward from the glacial maximum state (LGM and PGM) through the deglaciation and the interglacial periods. Sensitivity experiments that isolate the influence of individual forcings (e.g., meltwater fluxes, insolation changes, greenhouse gas variations, etc.) are also conducted. The main conclusions are: (i) the Last Interglacial was warmer and had a higher sea-level than the Holocene; (ii) insolation variations is the main driver of glacial retreat during both interglacial periods; (iii) the Atlantic overturning circulation is found to be more sensitive to collapse under Last Interglacial forcing.

The main novelty of the manuscript is the side-by-side comparison of the last two deglaciations in a coupled model setting. However, it is not clear from the presentation what the truly new results are and in what way this study is advancing our understanding of the last two deglaciations. There are several reasons for this, but most importantly because (i) the manuscript does not include a dedicated discussion section where the results are contrasted with the established literature; (ii) the model is quite simplistic and may not be the most appropriate choice for this type of study; (iii) some of the results are undoubtedly model dependent as they contradict previously published results using other models.

I recommend major revisions before this manuscript can be accepted for publication.

Major comments:

No discussion section:

The lack of a dedicated discussion section makes it hard to get a sense for how the results compare to the established literature and what the potential shortcomings of the study are. You do cite several papers in the results section, but these are primarily used to quantify (and to a certain extent justify) your results. A dedicated discussion section is essential for any study, and this manuscript would certainly benefit from having one as well.

QGPV model at low resolution:

I wonder how appropriate the model choice is for this study. From reading the model description in Quiquet et al. (2021), the atmospheric component of iLOVECLIM is a spectral, quasi-geostrophic potential-vorticity (QGPV) model that was run at a nominal 5.6-degrees (T21) horizontal resolution. It seems to me that this model choice is potentially problematic for at least two reasons:

(i) Several studies have shown conclusive evidence that the numerical convergence of both dry and moist dynamical cores breaks down somewhere between the T31 and T21 resolutions (e.g., Polvani et al. 2004; Lofverstrom and Liakka, 2018), and that resolution can have a substantial influence on the simulated climate (Lohman et al., 2021). The reason for this breakdown is (most likely) that the grid spacing becomes comparable to, or even exceeding the Rossby deformation radius in midlatitudes on sufficiently coarse model grids. This means that baroclinic waves are not appropriately resolved, which are one of the main drivers of the large-scale atmospheric circulation, including the distribution of temperature, precipitation, and wind in mid and high latitudes. While I recognize that it may not be feasible to run the simulations at a different resolution, this potential shortcoming should at least be acknowledged and discussed in the manuscript.

(ii) I would like to see a thorough discussion on the appropriateness of using a QGPV model as the atmospheric component in a coupled, global model configuration. QGPV is a decent first-order approximation of the synoptic and planetary scale circulation in mid and high latitudes, but it is not an appropriate description of tropical and subtropical circulation where ageostrophic processes dominate because of the smallness of the Coriolis parameter near the equator. Can we really trust a coupled atmosphere-ocean model that is largely incapable of representing the low-latitude atmospheric circulation with even first order accuracy?

No discussion about atmospheric circulation changes:

Previous studies have shown that the large-scale atmospheric circulation is strongly influenced by both the height and spatial distribution of the Northern Hemisphere ice sheets (e.g., Lofverstrom and Lora, 2017; Kageyama et al. 2021). Importantly, it has been shown that the North American ice sheet affects the temperature and precipitation distribution (i.e., the surface mass balance) over the Eurasian Ice Sheet (e.g, Liakka et al., 2016).

I think this study would be more convincing if the authors also included figures showing changes in the atmospheric circulation. Not least since the ice-sheet mass balance (i.e., the deglaciation) is to first order driven by changes in the temperature and precipitation distribution, and the QGPV atmospheric model is quite simplistic and may not capture some of the main circulation changes identified in numerous other studies using more comprehensive models.

Model dependence:

It is compulsory to discuss potential model dependence on results and conclusions in any modeling study. You mention model dependence in a few places in the text, but it would be good to consolidate this in a dedicated discussion section. One of your main conclusions is that insolation is more important for deglaciation than vegetation changes. I agree that this is what your results shows, but it appears to be contradicting the results in, e.g., Sommers et al (2021), who argued that vegetation changes are at least equally important, if not more important than insolation changes for the deglaciation of Greenland in the Last Interglacial. This is just one example of potential model dependence of your results that should be acknowledged and properly discussed in the manuscript.

General experiment design:

I am confused by the experiment design. The introduction states that the Northern Hemisphere ice sheet distribution in the PGM and the LGM were quite different, where the former had comparatively more ice in Eurasia relative to the LGM, and vice versa in North America.

However, the experiments presented here use the same ice sheets as initial conditions for both the LGM and the PGM. What is the reason for this choice since this appears to be a substantial deviation from reality? Would a different ice sheet initial condition alter the results in any way, for example through differences in the large-scale atmospheric circulation?

I recognize that spinning up the ice sheets for the PGM is a major task that may be computationally unfeasible. Therefore, I am not necessarily recommending that you re-run the simulations with more appropriate ice sheets for the PGM, but a discussion of the potential influence of these types of deviations from reality should at least be recognized and appropriately discussed in a dedicated discussion section.

References:

Kageyama et al.  (2021). The PMIP4 Last Glacial Maximum experiments: preliminary results and comparison with the PMIP3 simulations. Climate of the Past, 17 (3), 1065–1089. doi: https://doi.org/10.5194/cp-17-1065-2021

Lohmann, G., Wagner, A., & Prange, M. (2021). Resolution of the atmospheric model matters for the Northern Hemisphere Mid-Holocene climate. Dynamics of Atmospheres and Oceans, 93, 101206.

Liakka et al.: The impact of North American glacial topography on the evolution of Eurasian ice sheet over the last glacial cycle, Clim. Past, 1225–1241, https://doi.org/10.5194/cp-12-1225-2016, 2016

Lofverstrom, M., and Liakka, J. (2018). The influence of atmospheric grid resolution in a climate model-forced ice sheet simulation. The Cryosphere, 12(4), 1499-1510

Lofverstrom, M., Lora, J.M., 2017. Abrupt regime shifts in the North Atlantic atmospheric circulation over the last deglaciation. Geophys. Res. Lett. 44, 8047–8055. https://doi.org/10.1002/2017GL074274.

Polvani, L. M., Scott, R., and Thomas, S.: Numerically converged solutions of the global primitive equations for testing the dynamical core of atmospheric GCMs, Mon. Weather Rev., 132, 2539– 2552, 2004.

Sommers et al.: Retreat and regrowth of the Greenland ice sheet during the Last Interglacial as simulated by the CESM2-CISM2 coupled climate-ice sheet model. Paleoceanography and Paleoclimatology 36, 2021