the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Exploring the sensitivity of the Northern Hemisphere ice sheets at the last two glacial maxima to coupled climate-ice sheet model parameters
Abstract. Simulations of past periods are useful for testing the ability of numerical models to simulate ice sheet changes under significantly different climate conditions to present day. This can help improve projections of future sea level rise made by these same models and avoid over-tuning to particular (e.g. modern) stationary climate conditions. The Last Glacial Maximum (LGM; ~21 thousand years ago (ka)) has been extensively used for this purpose since it is relatively well constrained by empirical evidence. However, less is known about the Penultimate Glacial Maximum (PGM; ~140 ka) and why the vast ice sheets covering much of the Northern Hemisphere (NH), differed to the LGM. The answer likely lies, at least in part, in the different orbital configurations between the two periods, and the resulting impact on climate-ice sheet interactions.
Here, we perform and compare the first large ensembles of coupled climate-ice sheet (FAMOUS-BISICLES) simulations of the LGM and PGM to better understand how NH ice sheets interact with the climate and quantify how sensitive the simulations are to the choice of uncertain model inputs, including physical parameter values. Specifically, we vary 12 uncertain parameters that control the model representations of ice sheet albedo, ice dynamics and climate. The ensembles are evaluated against palaeo-evidence of global mean temperature, ice volume and extent to calibrate the model and find combinations of parameters that simultaneously yield plausible ice sheets and climates for both periods. The sensitivity of the North American ice sheet and the Eurasian ice sheet during the LGM and PGM, to each of the 12 parameter values, is explored using Gaussian Process emulators to perform a Sobol sensitivity analysis. From the whole ensemble, we find two simulations that meet our evaluation constraints for the LGM ice sheets. The parameter values that influence the albedo of the ice sheet have the largest influence on the resulting ice sheet volumes, but several other parameters display different sensitivity indices depending on the ice sheet (North American versus Eurasian) and time period (PGM versus LGM). This includes parameters that affect the cloud liquid water, lapse rate, basal sliding and downscaling elevation heights.
- Preprint
(5198 KB) - Metadata XML
-
Supplement
(501 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-3896', Anonymous Referee #1, 15 Mar 2025
Review of Patterson and others:
“Exploring the sensitivity of the Northern Hemisphere ice sheets at the last two glacial maxima to coupled climate-ice sheet model parameters”
Summary:
The authors investigate the sensitivity of key parameters in a new coupled climate-ice sheet model, FAMOUS-BISICLES, for simulating the Northern Hemisphere ice sheets at two glacial maxima. They conduct large ensemble simulations and evaluate the results against paleo-evidence of global mean temperature, ice volume and ice extent. To achieve this, they use Gaussian Process emulators and perform a Sobol sensitivity analysis. They finally identify two simulations that satisfy their evaluation constraints for the LGM and PGM.
General comments:
The study demonstrates valuable improvements in using the coupled ice sheet-climate model, FAMOUS-BISICLES, to better simulate the Northern Hemisphere ice sheets during various glacial periods, and highlights key parameters. However, beyond the tuning process and model parameter study, I do not see significant new findings. Therefore, please investigate further the new insights related to simulating the last two glacial maxima. Otherwise, the work may be more suitable for a model development journal.
The structure and logic of the study need to be improved for greater clarity, and the sentences could be more concise. The Introduction and Methods sections are somehow redundant. Moreover, I would suggest separating the Results and Discussion section to improve clarity and organization.
In the introduction, the authors discuss various uncertainties in reconstructing the PGM ice sheets and the possible reasons for the differences between the PGM and LGM ice sheets. However, it is unclear how these discussions are linked to the final conclusion, as the reason for the differences between the PGM and LGM ice sheets are not investigated in this study.
I’m not sure if it is appropriate to use “coupled climate-ice sheet” in the title, given that the ocean is prescribed, and the active components are only the atmosphere and the ice sheet.
Specific comments:
Line 16-18: The background introduction in the abstract is somewhat redundant and somewhat off-topic. The authors suggest that the answer likely lies in “the different orbital configurations between the two periods” and “climate-ice sheet interactions”. However, the manuscript does not address the differences in orbital configurations or climate-ice sheet interactions.
Line 20: “better understand how NH ice sheets interact with the climate”. However, the manuscript does not examine the interaction between the ice sheets and the climate. Instead, the parameters investigated in the model mainly focus on downscaling or factors that influence ice sheet surface/basal processes.
Line 26-30: What does the statement “… we find two simulations …” indicate? Does it suggest that the parameter criteria for simulating the LGM/PGM ice sheets are too strict? Otherwise remove this sentence. Additionally, please clarify the following sentence by specifying which parameters are more sensitive to which ice sheet or time period.
Line 180 and beyond: It is somewhat odd that you claim BISICLES is a model well-suited to simulate the past evolution of marine ice sheets, yet in the experiment setup, the ocean is prescribed and not investigated.
Figure 1a-b: Please also include the difference in SST between the LGM and PGM.
Line 413-416: Are the differences in the mean values for the different ice sheets at the LGM or PGM likely due to the different initial conditions? The same question applies to the spatial pattern (Line 418-421). Additionally, please elaborate on what these calculations indicate.
Figure 6-7: Maybe use a different colormap to display the ice thickness pattern. Currently, it is difficult to identify the simulated margin.
Line 517-518 and beyond: Please explicitly indicate what “tgrad” and “drain” refer to, and explain what their indications are. The same applies to the rest of the manuscript (e.g., Line 558, 544, 562…)
Citation: https://doi.org/10.5194/egusphere-2024-3896-RC1 -
RC2: 'Comment on egusphere-2024-3896', Anonymous Referee #2, 28 Mar 2025
Review of Exploring the sensitivity of the Northern Hemisphere ice sheets at the last two glacial maxima to coupled climate-ice sheet model parameters, by Patterson et al.
Summary
The authors examined the sensitivity of several parameters in the FAMOS-BISICLES Atmosphere-Ice Sheet model, applying their system to produce simulations of the NHIS at last two glacial maxima. While I find that the study provides an interesting parameter study and an extensive evaluation of the results against paleo-evidence, there are several comments which should be addressed before this paper can be considered for publication.
General Comments
- I find the language of the manuscript to be rather verbose. A simplification would be beneficial.
- While comparing simulated ice sheets to previous studies, the authors lack discussion of their broader applicability beyond their modeling framework. It’s important to consider if ice sheet volume and extent sensitivity to albedo parameters is consistent across models. While it would be beyond the scope to turn this into a multi-model comparison, explicitly discussing limitations and potential translations to other approaches would strengthen the manuscript.
- Importantly, the use of the term ‘coupled climate-ice sheet’ in both the title and the text is misleading and fundamentally incorrect. The model lacks a fully interactive ocean component, with only the atmosphere and ice sheet being active. This omission means it does not qualify as a true climate model, and the terminology is inappropriate. Referring to it as such overstates the model’s capabilities and misrepresents its actual functionality. The authors do, however, mention this limitation and that this shall be the subject of future work.
- The Methodology section needs some reorganization. Several auxiliary experiments that describe the sensitivity of ice dynamics are presented, but these are ice-sheet-only experiments and have only a tangential impact on the coupled system, where different parameters are evaluated. While these additional experiments are undoubtedly necessary and prudent to properly contextualize the ice sheet simulations within the coupled ice/climate system, I believe they could be moved to an appendix. This would allow the reader to focus on the main findings related to the coupled system.
Specific Comments
- Introduction Section: I find that this could use more focus. The introduction and background suggest that orbital configuration and climate/ice interactions are responsible for the differences seen in the PGM and LGM ice sheet geometries. However, neither orbital differences nor climate-feedbacks are explicitly discussed in the remainder of the manuscript. I would here rather focus on what you are parameterizing.
- Methods: While it’s evident that this isn’t the primary focus of your study, I find it unfortunate that the potential influence of the Antarctic ice sheet on the climate during the two glacial maxima isn’t discussed at all.
- Lines 145-160: It is unclear from the description how surface fluxes of ice mass (meltwater) are handed back to the climate model. It seems that the mass fluxes calculated by the atmosphere are interpolated onto the ice sheet grid and used for SMB, however, the coupling step on the way back to the climate only considers ice extent and orography. Is this correct?
- Line 204: What about basal melting discharged over land at the edge of the ice sheet? This is simply removed? This would imply that the model has a mass leak.
- Lines 562-621: I would be curious in this discussion to see some mention of implications of the (fairly coarse) atmospheric model resolution. The authors mention that precipitation is not downscaled in their approach, which may be a reasonable assumption to make. It might be beneficial to see how this effect plays out with an atmosphere model with finer spatial resolution (or alternatively, references to literature describing these implications)
- Conclusion Section: The authors switch tenses. For example: “We ran ensembles…” coupled later with “Through Gaussian Process emulation… we find…”. I believe this should instead stay consistent throughout the text.
- Figure 1: An anomaly plot vs. PI for SST might be clearer, as well as a relative difference between the two glacial states
- Figure 2: This could be combined into a single map for the entire Northern hemisphere. Presenting two separate panels gives the false impression that you have two distinct, separate ice sheet simulations running (one for each domain). I also think it would be more consistent to plot the coastlines as seen in the model here, rather than modern ones.
- Figure 4: A colormap with discrete level boundaries would be nicer here, in my opinion.
- Figure 5: The white-gray colormap is difficult to read and understand, I would suggest a different colormap here.
Citation: https://doi.org/10.5194/egusphere-2024-3896-RC2
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
154 | 51 | 8 | 213 | 31 | 5 | 7 |
- HTML: 154
- PDF: 51
- XML: 8
- Total: 213
- Supplement: 31
- BibTeX: 5
- EndNote: 7
Viewed (geographical distribution)
Country | # | Views | % |
---|---|---|---|
United States of America | 1 | 83 | 39 |
United Kingdom | 2 | 30 | 14 |
Germany | 3 | 17 | 8 |
France | 4 | 12 | 5 |
China | 5 | 8 | 3 |
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
- 83