the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Aug 2025
How polar-to-midlatitude atmospheric teleconnections depend on regional sea ice fraction and global warming level
Abstract. The climates of the polar and mid-latitude regions are linked through teleconnections. The regional details of these relationships, and how they may change with global warming, are however still uncertain. Using two large ensembles of coupled climate model simulations (CESM2, ACCESS-ESM1.5) and a composite analysis, we investigate the statistical relationships between sea ice variability and atmospheric circulation patterns, and how they evolve with sea ice retreat for both poles, including sensitivity to sea ice region in the Arctic. We find that relationships between sea ice amount and sea level pressure (SLP), the North Atlantic jet stream, and surface air temperature (SAT), depend on the region where sea ice varies. For instance, the North Atlantic jet shifts southwards when sea ice is low in the Labrador sea, but northwards or weakens (strengthens) for low Okhotsk (Chukchi-Bering) sea ice. We also investigate the circulation patterns associated with changes in Antarctic sea ice. For the Arctic, circulation patterns tend to persist with global warming, until around 3 or 4 °C, when the ice edge has retreated substantially. In the Antarctic, patterns are sensitive to warming also at lower global warming levels for some seasons and variables, but are otherwise often persistent across warming levels. Lagged analysis suggests that the instantaneous relationships mostly reflect the atmospheric conditions contributing to low sea ice, with weaker or altered patterns when sea ice leads. Our results emphasize the importance of regional heterogeneity, and on using large ensembles or other statistically rich datasets, for assessing influences of polar climate change on mid-latitude weather patterns today and in a warmer climate. The overall persistence of teleconnection patterns between sea ice change and atmospheric circulation with global warming is encouraging, as it indicates that the main conclusions from current literature will be applicable also in a future, warmer world with less sea ice.
- Preprint
(4182 KB) - Metadata XML
-
Supplement
(6475 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-4115', Anonymous Referee #1, 28 Sep 2025
-
RC2: 'Comment on egusphere-2025-4115', Anonymous Referee #2, 30 Sep 2025
Review for “How polar-to-midlatitude atmospheric teleconnections depend on regional sea ice fraction and global warming level” by Iles, Samset and Lund
Summary
In this manuscript, the authors use composite analysis with two climate model large ensembles to investigate the teleconnections associated with Arctic and Antarctic sea ice interannual variability and how these relationships evolve under different levels of global warming. The lagged composite analysis shows that the instantaneous relationships mainly reflect atmospheric conditions contributing to low sea ice, while the patterns become weaker or altered when sea ice leads. The authors also demonstrate that the teleconnections are generally robust across warming levels.
Overall, the topic fits well within the scope of Earth System Dynamics, and the authors provide a solid review of relevant literature. However, I have fundamental concerns regarding the methodology, particularly the use of instantaneous composites, and the interpretation of results derived from this approach. I would be happy to recommend publication once these issues are satisfactorily addressed. I hope my comments below will help strengthen the manuscript.
Major comments:
1. The reliance on instantaneous composites does not provide sufficient scientific insight. The authors build on Delhaye et al. (2024), who performed a similar analysis using CMIP6 preindustrial simulations. However, that study explicitly examined both the precursors of regional sea ice loss and the potential atmospheric impacts of sea ice anomalies, particularly in the Barents–Kara Seas. In contrast, the present manuscript emphasizes instantaneous composites, which by themselves lack a clear physical interpretation because they combine both forcing and response signals. As a result, much of the analysis describes patterns without clarifying the underlying mechanisms. This represents a key weakness of the study and could be considered a major limitation.
Related to this point, it is difficult to reconcile the results with much of the previous literature, which focuses on the atmospheric response to sea ice forcing. For example, the discussion section is problematic because the identified teleconnections appear to reflect circulation anomalies that precede sea ice retreat, yet they are compared to studies that investigate the atmospheric response to imposed sea ice loss.
To address this issue, I encourage the authors to place greater emphasis on the lagged analysis, similar to Delhaye et al. (2024). Doing so could help separate atmospheric precursors of sea ice variability (e.g., lag –1 or 0) from the potential impacts of sea ice anomalies on the atmosphere (e.g., lag +1). Such a framing would provide stronger physical interpretation and closer alignment with prior work.
2. While the lag analysis helps distinguish circulation patterns that force sea ice variability, it remains difficult to fully separate forcing and response signals. I have two suggestions that might help strengthen the analysis. While I cannot guarantee they will resolve the issue, they may provide additional insight into the two-way interactions:
- The authors emphasize the importance of coupled model simulations for capturing the full range of sea ice–atmosphere–ocean interactions. However, atmosphere-only simulations can still provide valuable perspective on the atmospheric response to sea ice anomalies. For example, CESM2 offers an 11-member atmosphere-only ensemble over 1950–2014 as part of CMIP6, and an additional 10-member ensemble covering 1880–2019 is available at https://gdex.ucar.edu/datasets/d651010/. Since the teleconnections appear robust across global warming levels, these simulations may be used to isolate the atmospheric response to polar sea ice loss, at least for one GSAT level.
- It may be possible to more clearly separate circulation patterns that precede sea ice retreat (e.g., lag –1) from those that respond to sea ice anomalies (e.g., lag +1). Yook et al. (2012) applied a congruence method to distinguish forcing and response patterns in the context of Kuroshio–Oyashio variability, and a similar approach could potentially be adapted here.
3. The manuscript currently contains an excessive number of figures, with many panels (often 20–40 per figure). However, most of these figures do not substantially advance the physical understanding of the processes in question, particularly regarding the forcing versus response of sea ice loss. I encourage the authors to carefully consider whether it is necessary to show all five lags (–2, –1, 0, +1, +2) across all five warming levels (0–4 °C). Streamlining the presentation would make the paper more accessible and focused. Moreover, Figures 7 and 10 do not appear to add significant value in their current form, unless they can be more directly linked to explaining the mechanisms of regional sea ice loss in the Arctic and Antarctic.
4. Instead of constructing composites based on GSAT warming levels, an alternative would be to first remove the ensemble‐mean forcing signal and then repeat the analysis over moving 50-year windows across the historical and SSP periods. I am not suggesting that the current approach is wrong, but I wonder if there is a specific reason the authors chose to base the analysis on GSAT. One potential limitation is that GSAT can be influenced by internal variability, such as the Interdecadal Pacific Oscillation (IPO) and Atlantic Multidecadal Variability (AMV), which may complicate the interpretation of GSAT-based composites.
5. Have the authors examined surface energy flux anomalies in the lagged analysis? In principle, the forcing and response phases should show opposite-signed anomalies (i.e., downward fluxes for the forcing circulation, upward fluxes for the atmospheric response). Including such an analysis could provide additional physical insight and help distinguish between precursor and response signals.
Reference:
Yook, S., D. W. J. Thompson, L. Sun, and C. Patrizio, 2022: The Simulated Atmospheric Response to Western North Pacific Sea Surface Temperature Anomalies. J. Climate, 35, 3335–3352, https://doi.org/10.1175/JCLI-D-21-0371.1.
Citation: https://doi.org/10.5194/egusphere-2025-4115-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 939 | 44 | 12 | 995 | 25 | 20 | 21 |
- HTML: 939
- PDF: 44
- XML: 12
- Total: 995
- Supplement: 25
- BibTeX: 20
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Summary
This manuscript investigates how atmospheric teleconnections between polar and mid-latitude regions depend on regional sea ice variations and global warming levels. Using two large ensemble climate model simulations (CESM2 and ACCESS-ESM1.5), the authors employ various statistical analysis to examine relationships between sea ice variability and atmospheric circulation patterns (surface air temperature, sea level pressure, and jet streams) across different Arctic regions and Antarctic conditions. They found strong regional dependency and persistence of teleconnection patterns till 3~4C warming.
Major comments
1. The composite analysis - major analysis method used - cannot cleanly separate cause and effect, as acknowledged by the authors. While lagged analysis is attempted, the high autocorrelation in sea ice makes interpretation challenging.
2. The 30th/70th percentile thresholds are too arbitrary. If the other values are used, then the result changes?
3. Model validation: I could not find much of information about how well two models perform in terms of teleconnection.
4. Antarctic result: The Antarctic analysis feels somewhat tacked on, with less thorough investigation than the Arctic, e.g,, no regional analysis.
5. This is somewhat related to #1 and #3. Monthly teleconnection between polar and mid-latitude has quite a bit of debate. There are quite a bit of work on daily dataset to tackle on teleconnection and its future (e.g., Kug et al. 2015, Wu et al. 2022, Hong et al. 2024). It is okay to focus on monthly time scales. However, issues like existence of such teleconnection, model validation, and causality need to discussed.
6. While comprehensive as the target, many individual findings confirm previous studies. I really have to ask what are the key findings here?
Overall, this manuscript has really wide range of the scope, which is good. However, there are numerous issues. For example, one could ask perform the Granger causality analysis in response to #1. Or, sensitivity for those threshold and detailed model validation could be suggested. Then, the analysis could be even more comprehensive without clear focus.