Cold Extremes during Dansgaard-Oeschger Oscillations

del Amo, Ignacio; Ditlevsen, Peter

doi:10.48550/arXiv.2609.03664

Preprints

https://doi.org/10.48550/arXiv.2609.03664

Preprints

09 Sep 2026

| 09 Sep 2026

Status: this preprint is open for discussion and under review for Earth System Dynamics (ESD).

Cold Extremes during Dansgaard-Oeschger Oscillations

Ignacio del Amo and Peter Ditlevsen

Abstract. This paper studies the statistics of extreme cold air surface temperatures in a climate that experiences abrupt changes. We employ a CCSM4 simulation of Last Glacial Maximum conditions that exhibits rapid switching between stadial and interstadial states as data. Non-stationary linear Generalized Extreme Value (GEV) distributions are fitted to find the regions that show the most prominent changes and associate them with physical processes. The results are then compared with a non-linear model, which gives a more detailed picture of how the parameters change as a function of the AMOC strength. While different regions of the world show different extremal behaviours, many regions show an approximately linear relationship between the parameters of the GEV distributions and the strength of the AMOC within the stadial and interstadial states, with some non-linear oscillation or jump where the transition between them takes place. Physical processes such as the expansion and retreat of the sea ice and the relative changes in the strength of the currents are shown to impact the magnitude and variability of the extremes, with significant changes observed in the three parameters of the GEV. They also create teleconnections that are compared whenever possible with various proxies for the temperature of the air and the water. We show how mapping the parameters of the GEV distributions into the AMOC strength gives a way to compare results between different climate models and different climate states. Comparisons, however, need to pay heed to the dynamical characteristics of the state and the location to be meaningful.

Received: 04 Sep 2026 – Discussion started: 09 Sep 2026

Ignacio del Amo and Peter Ditlevsen

Status: open (until 10 Dec 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-5368', Anonymous Referee #1, 24 Oct 2026 reply
The authors study the extreme value statistics of a low resolution CCSM4 simulation with Last Glacial Maximum conditions, containing Dansgaard-Oeschger (DO) events. To model the minima the authors employ a non-stationary generalized extreme value (GEV) distribution, where the location, scale and shape parameters are expressed using linear predictors of the AMOC strength as a covariate. Different configurations of this model are fitted independently to all local time series, and the dependence of the parameters on AMOC strength is further explored by fitting a stationary GEV model on a rolling window. Notable regions provide insight into the relationship between the AMOC strength and the distribution of annual minimum temperatures, and are highlighted and discussed.
The manuscript presents a thorough analysis of how cold extremes respond to abrupt climate transitions, an interesting topic that is well within the scope of ESD. However, I have one main concern regarding the applicability of extreme value theory to the dataset, and I therefore recommend a major revision before the manuscript can be considered for publication. My main concern, together with some smaller comments and technical corrections, is detailed below.
While the dataset used is almost 8000 years long, it is sampled at monthly resolution. This means that the number of observations in each block is 12, which is very small. In addition, seasonal effects would likely make annual minima cluster near the coldest winter months, further reducing the effective number of observations in each block. My primary concern is therefore that extreme value theory might not be appropriate for this dataset, since the GEV distribution is an asymptotic result.
If it is feasible, I would recommend repeating the analysis using a CCSM4 simulation output with higher temporal resolution (e.g. daily). Otherwise, I believe it is necessary to validate the GEV assumptions, for example by showing which months tend to provide the annual minimum across the domain (to check the effective block size) and by providing goodness-of-fit diagnostics (e.g. Q-Q plots) for a selection of grid points, and to compare the GEV with alternative distributions such as the normal or skew-normal (e.g. by AIC or another appropriate criterion). If the GEV model does not fit well, it might be necessary to consider alternative models, and to reinterpret the results.
Other comments:
I would like some details on how the GEV models have been fitted. Furthermore, the code to replicate these results has not been made available, making it harder to check for potential mistakes in the implementation.

The introduction of the manuscript suggests that cold spells play a part in motivating the research. Since the CCSM4 output consists of monthly averages, such cold spells would likely be smoothed out.

Section 5.2 on Extrapolation might feel a little outside the overall scope of the rest of the manuscript. Removing it, summarizing it as a small part of another section or moving it to the appendix, would make the manuscript more focused.

The window length used for the rolling window fits is not stated.

Inconsistent confidence levels. 99.999% in Section 2.1 (This could make sense given the high number of grid points, but a comment explaining why this significance level was chosen would be clarifying), 99% in screening step of Appendix B, 99% in Section 4.1.1, 99.5% in Figure 10.

Why were the lower and upper thresholds chosen to be 12 and 23.5 Sv, respectively?

Technical corrections:
Several ‘a’/’an’ errors:
«an strengthening», on page 15, 2nd paragraph of Section 4.2,

«a AMOC» 2nd sentence of Section 5.1, page 15

«a RCP 4.5» first paragraph, 2nd sentence of Section 5.2, page 17

«a AMOC» twice in paragraph 2 of Section 5.2, page 17

«a approximately», 3rd paragraph of Section 6, page 19

«a infinite», end of paragraph 5, Section 6, page 19.

Missing space «data.However» near the bottom of page 15 (Section 5.1)

«That» and «Than» are both swapped in the penultimate sentence of Section 5.1, paragraph 3 (page 16): «This suggests than the transitions between stadial and interstadial are fundamentally different that just a collapse...»

«Figure 2 shows that a linear trends are preferred», 3rd paragraph Section 5.2, page 17. Sentence uses plural and singular at the same time

«this criteria», beginning of penultimate paragraph, Section 5.2, page 18. Singular of ‘criteria’ is ‘criterion’.

3rd paragraph of Section 6, page 19: «thus the statistics encompass data points...». This is expressed in a slightly confusing way. The statistics are computed based on windows that encompass data points, but I am sure most readers will understand the intended meaning.

«specially» should be «especially», 1st sentence, 2nd paragraph, Section B.3, page 24

«this biases» should be «these biases», beginning of the last paragraph of Section B.2, page 24

«not-detected» should not be hyphenated, 2nd sentence, 2nd paragraph, Section B.2, page 24

«That can also produce that an opposite trend...», 4th sentence, 2nd paragraph, Section B.2, page 24

Reply
Citation: https://doi.org/10.5194/egusphere-2026-5368-RC1

Ignacio del Amo and Peter Ditlevsen

Viewed

Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.

Total article views: 252 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
249	0	3	252	0	0

HTML: 249
PDF: 0
XML: 3
Total: 252
BibTeX: 0
EndNote: 0

Views and downloads (calculated since 09 Sep 2026)

Month	HTML	PDF	XML
Sep 2026	152	1	153
Oct 2026	76	1	77
Nov 2026	21	1	22

Cumulative views and downloads (calculated since 09 Sep 2026)

Month	HTML	PDF	XML
Sep 2026	152	1	153
Oct 2026	76	1	77
Nov 2026	21	1	22

Viewed (geographical distribution)

Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.

Total article views: 252 (including HTML, PDF, and XML) Thereof 252 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 09 Nov 2026

Short summary

The last glacial period on Earth was subject to rapid changes in temperature. These changes were transitions between two different climate states, which included important changes in the strength of the currents in the Atlantic Ocean. We employ a simulation of these conditions to study how temperature minima changed across the oscillation between states, how the different physical processes affected the extreme events and if we can project this knowledge into the future.


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