Deglaciation and abrupt events in a coupled comprehensive atmosphere&ndash;ocean&ndash;ice sheet&ndash;solid earth model

Mikolajewicz, Uwe; Kapsch, Marie-Luise; Schannwell, Clemens; Six, Katharina D.; Ziemen, Florian A.; Bagge, Meike; Baudouin, Jean-Philippe; Erokhina, Olga; Gayler, Veronika; Klemann, Volker; Meccia, Virna L.; Mouchet, Anne; Riddick, Thomas

doi:https://doi.org/10.5194/cp-2024-55

Preprints

https://doi.org/10.5194/cp-2024-55

Preprints

07 Aug 2024

| 07 Aug 2024

Status: a revised version of this preprint was accepted for the journal CP.

Deglaciation and abrupt events in a coupled comprehensive atmosphere–ocean–ice sheet–solid earth model

Uwe Mikolajewicz, Marie-Luise Kapsch, Clemens Schannwell, Katharina D. Six, Florian A. Ziemen, Meike Bagge, Jean-Philippe Baudouin, Olga Erokhina, Veronika Gayler, Volker Klemann, Virna L. Meccia, Anne Mouchet, and Thomas Riddick

Abstract. During the last 20,000 years the climate of the earth has changed from a state much colder than today with large ice sheets in North America and Northwest Eurasia to its present state. The fully-interactive simulation of this transition represents a hitherto unsolved challenge for state-of-the-art climate models. We use a novel coupled comprehensive atmosphere–ocean–vegetation– ice sheet–solid earth model to simulate the transient climate evolution from the last glacial maximum to preindustrial times. The model considers dynamical changes of the glacier mask, land–sea mask and river routing. An ensemble of transient model simulations successfully captures the main features of the last deglaciation, as depicted by proxy estimates. In addition, our model simulates a series of abrupt climate changes, which can be attributed to different drivers. Abrupt cooling events during the glacial and the first half of the deglaciation are caused by Heinrich-event like ice-sheet surges, which are part of the model generated internal variability. We show that the timing of these surges depends on the initial state and the model parameters. Abrupt events during the second half of the deglaciation are caused by a long-term shift in the sign of the Arctic freshwater budget, changes in river routing and/or the opening of ocean passages.

Received: 31 Jul 2024 – Discussion started: 07 Aug 2024

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Status: final response (author comments only)

RC1:
'Comment on cp-2024-55', Louise Sime, 08 Sep 2024

Review of “Deglaciation and abrupt events in a coupled comprehensive atmosphere–ocean–ice sheet–solid earth model” by Uwe Mikolajewicz et al.
Manuscript summary: This study introduces a novel comprehensive coupled atmosphere-ocean-vegetation-ice sheet-solid earth model, MPI-ESM/mPISM/VILMA, designed to simulate the last deglaciation with high realism. This model is unique because it includes interactive Earth system components such as ice sheets, icebergs, solid earth, and dynamic river directions, allowing for a comprehensive exploration of deglacial processes. The findings highlight that the model can reproduce abrupt millennial-scale climate events and that the timing of these events is influenced by initial conditions and model parameters rather than solely by external forcing. Additionally, the model reveals that changes in Arctic sea-ice export and freshwater dynamics significantly affect the Atlantic Meridional Overturning Circulation (AMOC) and North Atlantic climate.
I find this study really quite exciting. It provides a breakthrough in understanding how different components of the Earth’s system interact over long timescales, particularly during the complex process of deglaciation. The ability of the new model to simulate abrupt climate events and reveal unexpected dynamics, including the influence of freshwater dynamics and ice sheet surges on ocean circulation, opens up new avenues for exploring past climate changes and understanding past climate tipping behaviours.
The analysis is mostly well-organised and constructed, and the paper is well-written. It effectively conveys the findings, highlights the novelty of the research, and provides a clear and concise overview of the study’s objectives, methods, results, and implications. With the caveats below about it being on the long side, and whether the authors could extend parts of the analysis and split it into two, it is clearly suitable for publication in CP.
Major points:
This is not necessarily a criticism, but I find there is really a lot of material/work in this manuscript. It might be more digestible to most readers if it were split in two, to allow more focus on some aspects. Perhaps with the material on abrupt changes placed into a second manuscript?
The simulations are generally run from 26ka to 1850, allowing a climatological mean PI for each simulation to be specified as the last 1,000 years or so of the run. However, we never see what the PI simulations look like. The manuscript would benefit from an appendix to show how the PI states differ for each version of the model, and some indication of how far they deviate from the observed PI, and a comment on this to be added to 2.2 and 3.1.
The description of the pre-26ka model spin-up seems to imply that the ocean is spun up for just 10 years, and then the first 1,000 years of each simulation are disregarded. This would give only 1,010 years to spin up the ocean. I think this may not be correct because the text also states that the simulations are initialized from pre-existing glacial simulations. However, this is not very clearly explained. Like the PI point above, spin-up is relatively important, so this should be carefully clarified in 2.1.
Some clarification/justification for modifications to ocean mixing/vertical diffusion would be useful, both in the context of expected glacial-interglacial ocean mixing changes (due to bathymetric and stratification effects on the dissipation of internal tides/waves). Alongside this, a strengthened discussion of how AMOC and mixing may be affecting the results from this model, given abrupt changes results are relatively strongly dependent on the strength of the AMOC & mixing i.e. discussion of would we expect other GCMs/ESMs to behave similarly (if they also had the same extra component coupled onto them, and were run for similar experiments), or are these unique to this model?
The section on Abrupt Events is very sensibly laid out; I like the analysis. However, I find it slightly surprising that the focus is solely on simulated abrupt cold events, given there is considerable community interest in the possibility of abrupt warmings too, and some deglacial events are indeed abrupt warmings, rather than solely coolings. If the authors intend to retain the focus solely on abrupt coolings, then it would be helpful to have a sentence or two added to the Introduction and Section to better justify this focus. Otherwise, the focus could perhaps be broadened to also include abrupt warmings, which are also visible in the timeseries provided. See also the first main comment about splitting this manuscript in two.
Minor points:
The model name is a currently a bit of a mouthful, have the authors also considered giving the model a shorter name too MPI-ESM-extended or similar, perhaps?
L18, missing refs, for very different rates of changes
L19-22, greater and lesser ‘volumes’ rather than changes?
L34 abrupt ‘AMOC’ changes, or what events?
L34, ‘the quantification of’ rather than ‘the exact changes’
L54, and after, clarify what ‘CMIP-style’ means. Either CMIP models, or perhaps models that use CMIP-atmosphere (AMIP?) models, or perhaps that are run at common CMIP atmosphere-ocean resolutions? Either way, replace ‘CMIP-style’ descriptions with something more meaningful.
L75, split this into two sentences.
L113, clarify what is meant here by radial directions – by depth, but not by lat or long?
Section 2.2 and 3.1, please see above first two main points.
Table 1 headings, spell out the headings better. Information about spin-up is not clear. Better to replace the last column ‘Parent run’ with a much clearer verbal description. Exp file names can be omitted.
~L125, consider to add a short subsection or para in 2.1 which describes the water tracer/dye methodology, including precise conditions for tracing/dying water, and how tracers/dye is reset (presumably without resetting on exposure, the ocean would eventually saturate?).
L193, maximum?
L200-202, better to replace this with an Appendix that more carefully shows what the PI states look like themselves (not just the LGM-PI anomalies).
Figure 3, obscures more than it shows. It might be more instructive to see the LGM-PI anoms subtracted from the equivalent Anna et al and Osman et al anoms.
Figure 5, I really like these dye/tracer sections.
Figure 5 and Figure 6, and thereabouts in text, please ensure there is sufficient information in the text to reassure the reader that these results are not due to spin-up issues. See also main points, and minor point aboves about better clarification on spin-up procedures and ocean run durations.
Page 6, and generally dye/tracer results. Please clarify how dye/tracer is removed. Is it reset to 0 on exposure to the atmosphere? Or something else?
L295, it is interesting that these coupled model simulations do not help much with this Antarctic ice sheet extent problem, possibly it is worth highlighting that this results supports the idea that the problem is in the representation of ice sheet physics in PISM rather than in the forcing/coupling?
L310-311, rewrite this sentence – it is very hard to understand.
L375, missing punctuation/sentence issue.
Table 2, this table would be easier to digest if the rows were shaded to reflect whether the simulated even occurs earlier, later, or (within the uncertainties) at commensurate with the proxy evidence for a similar event.
L529, salinity twice
L531, ‘varies significantly’ rather than the significantly varies
L581, see first comment on model name
L602, remove ‘also’?
L607-610, there are some odd clause orders in here. Check ordering for English, and improve the sentences.
L632, either clarify what is meant by unexpected, or possibly rewrite this sentence to focus on the successfully model simulation of hereto uncaptured processes? (river rerouting, arctic freshwater sign changes, strait flow impacts, and other ice-sheet change related climate-land surface-ocean related processes.).

Citation: https://doi.org/10.5194/cp-2024-55-RC1
- AC1: 'Reply on RC1', Uwe Mikolajewicz, 29 Oct 2024
  
  The comment was uploaded in the form of a supplement: https://cp.copernicus.org/preprints/cp-2024-55/cp-2024-55-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/cp-2024-55-AC1
RC2:
'Comment on cp-2024-55', Sam Sherriff-Tadano, 24 Sep 2024

Review of Mikolajewicz et al. “Deglaciation and abrupt events in a coupled comprehensive atmosphere–ocean–ice sheet–solid earth model”
Summary
Mikolajewicz and others present the ensemble simulations of the last deglaciation with a coupled comprehensive atmosphere–ocean–ice sheet–solid earth model. All the simulations reproduce the general characteristics of climate and AMOC at the LGM and the deglaciation within the uncertainties of reconstructions. The authors further show intriguing results, such as the effects of ocean background mixing on the AMOC, the comparison of the dye tracer with the traditional definition of AMOC depth (AMOC=0 line), the regular and chaotic aspects of ice serge and their impacts on the abrupt climate change and effects of the opening of Bering Strait and the river routine on the abrupt climate changes. Particular attention is also given to the freshwater budget analysis in the Arctic when explaining non-surge-related abrupt climate changes..
The results presented in this study with their complex coupled earth system model are all exciting and are of interests of the readers of Climate of the Past. I also believe that the simulations performed in this study will greatly contribute to advance the science of the climate community. While, I recommend an intermediate revision based on the comments listed below, I do feel these results should be published. Thanks for all the effort in conducting this study!
General comment
My main concern focuses on the analysis of freshwater budget over the Arctic region (Fig. 12) and its effect on the AMOC.
The abrupt AMOC changes in the second half of the deglaciation are explained by the balance of freshwater input and sea ice export in the Arctic, however some ambiguities remain. For example, the rapid salinity drop in the Arctic Ocean (Fig. 12a, blue) precedes the timing of the shift in the net freshwater budget defined by the author (Fig 12 black triangle), though this was not explained. Perhaps, gradual changes in oceanic freshwater transport may affect the exact timing, but please clarify this point.
Second, the change in AMOC in response to the opening of Bering Strait is also explained by means of sea ice export. However, the role of changes in oceanic freshwater transport between the Arctic Ocean and the North Pacific is not discussed, which could be important (e.g. Hu et al. 2012). Please add this analysis.
Lastly, I would like to suggest the authors to add a discussion on the stability of the AMOC in this model. In the simulations of deglaciation, the AMOC basically remains stable in its vigorous (or so-called interstadial) mode; e.g. even after a drastic weakening due to substantial freshwater input, the AMOC rapidly recovers to the strong mode once the freshwater flux ceases. This may be influenced by the presence of the North American ice sheet, but if the AMOC remains stable in the strong mode throughout the deglaciation, it may be more challenging to reproduce the Heinrich Events (HE) and BA-like events. Given that the introduction mentions the relationship between freshwater input and BA, it would be valuable to expand on this point in the discussion or conclusion.
Minor comments
L43-44: Not an easy sentence to understand. Could you explain a bit more, please?
L60: Please cite Snoll et al. (2024, https://doi.org/10.5194/cp-20-789-2024)
L90: Would be useful to describe the climate sensitivity of the AOGCM somewhere in this subsection.
L119: Could you elaborate a bit more on how the SMB are given to the ice sheet model? Is it a 10 year average?
L131: Just wanted to make sure, but in D1.2 and D1.3, changes in ocean mixing is applied after the spin-up, right? If so, please clarify it wherever appropriate.
L149: Could you briefly explain the reason of changing the sea ice parameter, please?
L150: Could you briefly describe the reason why the modifications in background mixing is applied only at the upper ocean, please?
L222: Nice results! This might be a difficult question, but why does this model manage to reproduce the LGM AMOC, but not the model version(MPI-ESM) submitted to Kageyama et al. (2021)? Resolution, meltwater from ice sheet or shape of the ice sheet? Please add a sentence on this point.
L269: While it is displayed in Fig. 8, I think it would be extremely useful to summarize the volume of individual ice sheets at the LGM and PI in a independent Table. This would be a great reference for the readers.
L271: This is a common problem in other climate-ice sheet model simulations so it might be a good idea to refer to those studies as well (e.g. Ziemen et al. 2014, Sherriff-Tadano et al. 2024, https://doi.org/10.5194/cp-20-1489-2024)
L360: It’s interesting to see the decoupling of Antarctic ice sheet and NH ice sheets, despite the climate model reproducing the temperature evolution of SH high latitudes (FIg. 6c). What happens to ice calving of Antarctica ice sheet in response to sea level rise due to NH ice sheet melting (e.g. Gomez et al. 2020, https://doi.org/10.1038/s41586-020-2916-2)?
L382: The effect of surface warming on the AMOC can be state dependent. For example, Zhu et al. (2015, https://doi.org/10.1007/s00382-014-2165-x) showed that surface warming from a glacial state can cause an intensification of the AMOC by reducing the amount of sea ice.
L434-443 & Fig. 12: The abrupt reduction of Arctic salinity (Fig. 12a) precedes the shift in net-FW input over the Arctic (black triangle) and the abrupt AMOC weakening (Fig. 12c), which may imply a role from other processes in reducing the surface salinity over the Arctic and in weakening the AMOC. Could you add an explanation on this point please?
L441-443: I’m not entirely convinced by this sentence because the role of sea ice export on the AMOC can be quite complicated. For instance, reduced sea ice export from the Arctic can lead to a freshening of the Arctic Ocean, which may increase the oceanic freshwater transport to the Atlantic and weaken the AMOC. Conversely, reduced sea ice export can also decrease sea ice melting in the North Atlantic, potentially increasing salinity in the DWF region and intensifying the AMOC. Therefore, a more comprehensive analysis of the freshwater budget is necessary to clarify this point.
L444-449: I think this paragraph is missing the perspective of freshwater export from Arctic to North Pacific (e.g. Hu et al. 2008 https://doi.org/10.1175/2007JCLI1985.1). Please rewrite the paragraph including this point.
L538: I think it would be helpful to clarify that Hu et al. (2015) drew their conclusion based on simulations without any ice sheet melting flux involved, while this study included the effect of ice sheet melting in a realists framework. The finding of Hu et al. (2015) was useful to understand the effect of Bering Strait on the net fw input over the Arctic, however as the authors state, the inclusion of ice sheet melting is important in controlling the timing of the shift of the net Fw input over the Arctic during the deglaciation.
Fig.3: Sorry if I’m wrong, but I feel the temperature anomaly and sea ice edges over the North Atlantic look inconsistent in (C) and (D). For example, a larger sea ice expansion is observed over the North Atlantic in (C) than (D), whereas the magnitude of surface cooling is larger in (D) than (C) in the same region. By any chance, the temperature anomalies are mistakenly swop between (C) and (D)?
Fig.A1: In terms of Dye Tracer and AMOC in PI, experiment D1.2 seems the best, but that is the one which does not simulate shallower and weaker AMOC at the LGM. Do you have any comment on that?
Fig. 7: Please expand the area of the NH so that the readers can compare the full extent of the simulated North American ice sheet with reconstructions.
Fig. 12b: Does the blue line include the input from ice sheet melting and river run-off? It seems so from the main text, but it is not explained in the caption.

Citation: https://doi.org/10.5194/cp-2024-55-RC2
- AC2: 'Reply on RC2', Uwe Mikolajewicz, 29 Oct 2024
  
  The comment was uploaded in the form of a supplement: https://cp.copernicus.org/preprints/cp-2024-55/cp-2024-55-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/cp-2024-55-AC2

Viewed

Total article views: 887 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
537	142	208	887	10	9

HTML: 537
PDF: 142
XML: 208
Total: 887
BibTeX: 10
EndNote: 9

Views and downloads (calculated since 07 Aug 2024)

Month	HTML	PDF	XML	Total
Aug 2024	166	87	7	260
Sep 2024	129	29	34	192
Oct 2024	111	12	93	216
Nov 2024	89	8	48	145
Dec 2024	42	6	26	74

Cumulative views and downloads (calculated since 07 Aug 2024)

Month	HTML	PDF	XML	Total
Aug 2024	166	87	7	260
Sep 2024	129	29	34	192
Oct 2024	111	12	93	216
Nov 2024	89	8	48	145
Dec 2024	42	6	26	74

Viewed (geographical distribution)

Total article views: 885 (including HTML, PDF, and XML) Thereof 885 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 18 Dec 2024

Short summary

A fully coupled atmosphere-ocean-ice sheet-solid earth model was applied to simulate the time from the last glacial maximum to the preindustrial. The model simulations are compared to proxy data. During the glacial and deglaciation the model simulates several abrupt changes in North Atlantic climate. The underlying meachanisms are analysed and described.


Total:	0
HTML:	0
PDF:	0
XML:	0