Please see PDF document.

Summary

Sun et al. investigate ice-climate interactions during the Late Ordovician using an asynchronous ice sheet and Earth system model approach. The Late Ordovician glaciation is particularly interesting due to the inferred coexistence of relatively high atmospheric CO2 and expansive ice sheets in the Southern Hemisphere. While previous studies have investigated climate impacts on Late Ordovician ice sheet growth, most studies are limited by one-way coupling and limited representation of climate-ice sheet interactions. Sun et al. make a great advancement by incorporating asynchronous coupling of an Earth system model and ice sheet model and expand upon a similar study (Pohl et al., 2016) by investigating mechanisms of climate-ice sheet interactions in more depth. The authors reveal a key positive feedback loop where ice sheet growth initiates katabatic winds, which further expand ice sheets, providing a physical explanation for the coexistence of widespread ice sheets in a relatively warm climate. Overall, the authors provide a very interesting paper with detailed analysis of the ocean-atmosphere dynamics associated with ice sheet expansion, which is a valuable contribution to the field. However, I do have some concerns/questions that I detail below. In particular, I would like to better understand how the potential cold bias associated with the treatment of vegetation in the CESM simulations may be influencing the simulated expansion of Late Ordovician ice.

Minor Comments

• Line 78: “systematically analyze” and “, which is a gap”
• Line 81: CESM is an “earth system model” not a “general circulation model”

Major Comments

Abstract

• Lines 9-19: I think you should incorporate the coexistence of widespread ice sheets and high CO2 in this climate; it is a very interesting aspect of this time that you are attempting to explain.

Section 2.1

• Lines 90-99: I would highlight other deep-time paleoclimate studies that have used CESM1.2, demonstrating that it is a reliable model to use for this purpose
• Lines 98: The CNDV model in CLM4 traditionally has issues growing vegetation in paleoclimates. In Li et al. (2022), they mention a cold bias introduced by the used of CNDV because the vegetation is underpredicted thus much of the land has a very high surface albedo. Can you show a plot of the simulation’s vegetation distribution and leaf area index? Could this cold bias from unrealistic vegetation be allowing more land ice to grow under high CO2?
• Lines 108-110: Are any ice sheets prescribed in the Li et al. 440 control simulation? What does the polar surface cover look like?

Section 2.2

• Line 127-128: You say “Despite its simplicity, this scheme has been demonstrated to reproduce large-scale ice sheet behaviour effectively and efficiently”, do you have a citation for this?

Section 3.1

• Lines 172-173 You say “Lowering pCO2 might reproduce an ice sheet extent in the offline simulations that is consistent with the geological records (Pohl et al., 2016).” The choice of 6 PAL CO2 based on the approach of Li et al. (2022) and Scotese et al. (2026) is likely overestimated, as Li et al. state in their paper: “CNDV indeed causes cold biases and leads to an overestimation of the CO2 concentrations by about 1.8 times as the ECS of 2.9 °C per doubling atmospheric CO2 is considered.” How do you think that lowering CO2 and prescribing vegetation (greening the surface) may impact the results?
• Figure 2: Where are the glacial deposits from? Please provide citations here.

Section 3.4

• Figure 5: It is difficult to read the vectors in these plots, I would suggest thinning the vectors and/or increasing the size of the vector heads. The lat/lon labels should be simplified to prevent overlap. Also, the “cplSM” label for the coupled simulation should be defined in the figure caption and used consistently in the text (when do you use “cplSM” vs “150ka” vs “50ka” for coupled simulation).
• Figure 6: Similar suggestions to Figure 5.

Supplement

• Figures B5 and B6 are almost impossible to understand due to resolution of the graphics and small plot sizes. I would suggest remaking these figures with improved image resolution and/or without lat/lon labels so that panels can be enlarged.