the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Atmospheric drivers of Antarctic sea ice extent summer minima
Bianca Mezzina
Hugues Goosse
François Klein
Antoine Barthélemy
François Massonnet
Abstract. Understanding the variability of Antarctic sea ice is still a challenge. After decades of modest growth, an unprecedented minimum in the sea ice extent (SIE) was registered in summer 2017, and, following years of anomalously low SIE, a new record was established in early 2022. These two memorable minima have received great attention as single cases, but a comprehensive analysis of summer SIE minima is currently lacking. Indeed, other similar events are present in the observational record, although minor compared to the most recent ones, and a full analysis of all summer SIE minima is essential to separate potential common drivers from event-specific dynamics, in order to ultimately improve our understanding of the Antarctic sea ice and climate variability.
In this work, we examine sea ice and atmospheric conditions during and before all summer SIE minima over the satellite period up to 2022. We use observations and reanalysis data and compare our main findings with results from an ocean-sea ice model (NEMO-LIM) driven by prescribed atmospheric fields from ERA5. Examining SIE and sea ice concentration (SIC) anomalies, we find that the main contributors to the summer minima are the Ross and Weddell sectors. However, the two regions play different roles and the variability of the Ross Sea seems to explain most of the minima, with typical negative SIE anomalies about twice the ones in the Weddell Sea. Furthermore, the distribution of SIC anomalies is also different: in the Weddell Sea, they exhibit a dipolar structure, with increased SIC next to the continent and decreased SIC at the sea ice margin, while the Ross Sea displays a more homogenous decrease. We also examine the role of wintertime sea ice conditions before the summer SIE minima and find mixed results depending on the period: the winter conditions seem relevant in the most recent events, after 2017, but marginal for previous years. Next, we consider the influence of the atmosphere on the SIE minima, which appears to play a major role: after analyzing the anomalous atmospheric circulation during the preceding spring, we find that different large-scale anomalies can lead to similar regional prevailing winds that drive the summer minima. Specifically, the SIE minima seem to be associated with dominant north-westerly anomalous winds in the Weddell Sea, while a south-westerly anomalous flow prevails in the Ross Sea. Finally, we investigate the relative contribution of dynamic (e.g. ice transport) and thermodynamic (e.g. local melting) processes to the summer minima. Our results suggest that the exceptional sea ice loss in both the Ross and Weddell sectors is dominated by thermodynamic processes, while dynamics are also present but with a minor role.
- Preprint
(3550 KB) - Metadata XML
-
Supplement
(4015 KB) - BibTeX
- EndNote
Bianca Mezzina et al.
Status: final response (author comments only)
-
RC1: 'Comment on tc-2023-45', William Hobbs, 14 Jul 2023
The authors analysis is a valuable contribution to the current work on understanding recent Antarctic sea ice extremes, and I applaud the attempt to make sense of extreme low sea ice events as a whole rather than focussing on individual events; I think this is valuable and important. The paper is mostly clearly-written with appropriate references, the figures are of high quality and the analysis mostly supports the conclusions. I have uploaded an annotated PDF with specific comments, but I do have some general suggestions/comments
1) if possible I think some updates to include summer 2023 would be worth the (hopefully not too much) extra effort in terms of impact; I acknowledge that it may not be possible to extend the model simulation though
2) I was a little suprised that the model performs quite poorly in the Weddell sector, I think it would be worth checking the ERA5 surface temperature to see whether the model is the problem or the surface forcing. Either way, I think the authors need to be a bit more rigorous in expaining how this bias might effect the analysis. Currently it's a bit dismissive, stating the model is a good representation, but with figures that don't really support that. I think this could be addressed with some carefully calibrated text though.
3) This is my most serious concern, (and sorry, also my most negative) - I think the area budget analysis in section 3.5 is fundamentally incorrect, and is giving the wrong answer. There are 2 key reasons:
- a) almost all the sea ice melt, even in spring, is basal melt, not surface melt (e.g. Grodon 1981), so atmospheric thermal advection is unlikely to have a big impact. (If the model diagnostics output the separate melt components then the authors can check this for themselves, or even prove me wrong!);
- b) in spring, the heat source for that basal melt is solar radiation collected in leads/open water. Hence, because of the albedo feedback, the dynamic and thermodynamic terms are intimately related - move ice out of the way, the suface ocean can warm, and you get more melt. By integrating over very large areas you lose this relationship - the only dynamic contribution mathematically can be movement in or out of that sector, BUT you lose all the information about how the melt is modulated by dynamics
As serious as this concern is, I think it could be resolved fairly easily - rather than spatial integrals, just show maps of the tendency terms' anomalies, and I think that co-dependence should be evident. I note that the climatology maps are shown in the supplement (and indeed show an inverse co-dependence between the dynamic/thermal terms), but I think the anomaly maps are key as well.And if possible from the diagnostics compare the surface and basal melt components (this can actually be done correctly as an area integral)
Gordon, A. L., 1981: Seasonality of Southern-Ocean Sea Ice. J Geophys Res-Oceans, 86, 4193-4197, DOI 10.1029/JC086iC05p04193.
- AC1: 'Reply on RC1', Bianca Mezzina, 10 Oct 2023
-
RC2: 'Comment on tc-2023-45', Anonymous Referee #2, 10 Aug 2023
Review of ‘Atmospheric drivers of Antarctic sea ice extent summer minima’ by Mezzina et al.
Overview: This paper analyzes multiple summer (JFM) Antarctic sea ice minima to discover if there are similarities between the events, and to better understand the main mechanisms causing (and leading up to) the minima themselves. It finds that there are many regional differences, with the Ross and Weddell Sea sectors being the most important for total sea ice minima. Across the events, there are many different causes, with the study suggestion a similar wind pattern (even with different atmospheric circulation patterns) in both sectors, and a dominance of thermodynamic (melting) effects over dynamic (primarily advection).
The paper is an important study, and when improved, can be a valuable contribution to the field. However, there are some things that can be tightened before the paper should be formally accepted, and as such I’m suggestion a revision that would fall between a major / minor revision (some new analysis, mostly new text).
Main comments:
- I think the role of ice advection, including the preconditioning of ice anomalies in the preceeding winter, can be improved substainally. In the comments below, I’m asking the authors to redo section 3.3 with a Hovmöller diagram (with SIE anomalies plotted in longitude and time) to better see any connections with the preceeding winter to the summer minima, and allowing for ice anomalies to move in / out of sectors. I think this will also better guide the interpretation of the dynamic terms, which if I am understanding appropriately, are only calculated at locations of SIC anomalies, and not upstream where the ice anomalies may have originated.
- The ice-ocean model has some rather serious limitations in my view – incorrect sea ice anomalies in general, and a rather poor representation of the patterns of ice loss in the Weddell Sea. The authors mention this originally, and again in the conclusions, but there needs to be more text and insight provided when using the model to understand causes for the SIC anomalies in light of what information the model can actually provide (and how this compares to observations). See some specific mentions in the comments below.
Minor comments / line-by-line comments:
Fig. 1- what is striking to me are times when the model produces a minima below 1 sigma, but this is not in the observations (which often show positive SIE, albeit less than +1 sigma, looks like 1991, 1999 as examples from Fig. 1 for total SIE). What causes these extreme minima in the model, and does this limit the usefulness of the model (i.e., is the model producing the right negative sea ice conditions for the wrong reasons in the -1 sigma observations / -0.5 sigma model comparisons?
L124-126: There is no Fig. 1f, I think you mean Fig. 2f that shows the sea ice sectors?
Eq 1 (near L140) – Is this a total derivative, or a partial derivative? In atmospheric sciences, at least the dSIC/dt = total derivative (Lagrangian, following the motion, so no advective terms), and (SIC)/t = partial derivative, Eulerian, which is a local tendency that has advective terms, which is what I would expect for the equation referenced?
Fig. 2, why not multiply by 100 and show these anomalies as a percent?
L218-219: I would add another reference to Table 1 here when discussing the overall good agreement with the model and observations – although really the agreement is only good in my view in the Ross sector.
Fig. 4 - I think a Hovmöller style plot would be more informative here – sea ice can move from one sector to the others and this plot fails to show that, and therefore may miss the connection of winter minima in other sectors nearby that can lead to a minima in the Ross (especially) but also the Weddell in summer.
L238 – I suspect you mean center and right columns, but probably a moot point since I’m suggesting a revision of this section / figure. It does seem to suggest though for an extreme minima like we have seen in the last few years, a winter preconditioning seems to be important. From what I recall, the recent events in 2022 and 2023 also had an early peak in maximum extent sometime in August, which is worth mentioning I think.
Fig. 5 – I’m wondering the influence of season mean conditions (as shown in Fig. 5) vs. the impact of strong extremes, as in several strong storms that can break up, quickly move and redistribute ice (as shown in Turner et al. (2022)), but may be masked by the use of the seasonal means in Fig. 5. The authors should at least comment on this impact. I suppose extremes could rapidly expand ice (or reduce its retreat), but this seems to be discussed much less in the literature.
Fig 6 caption – more details are needed here, for example, what are the X markings indicating? Is this for a total SIE <-1, or just a regional one, or something else entirely?
Fig. 6 – suggest changing ‘E’ in the figure to ‘eastward’, since this is for a westerly wind that is moving toward the east. I think the arrows are meant to indicate the wind motion, but expanding on this would be helpful, especially since the text talks about wind direction, not where winds are going (and in that regard, why did you change the orientation of Fig. 6 to represent where the winds are going, not where they are from?)
Section 3.5 / Fig. 7 : does the model have Ekman induced sea ice changes, whereby westerly winds could expand ice through an equatorward ocean movement (from my understanding this was a large contributing factor to the expansion of ice through 2016 via the increased westerlies / positive SAM phases).
Section 3.5 / Fig 7: another question on the interpretation of these results- given the model is prescribed the winds but doesn’t often get the right magnitude of sea ice loss (Figs. 2,3 and Table 1), wouldn’t the dynamic term be under-represented by the model (it would have a much weaker value than expected since it has the correct winds but incorrect sea ice anomalies)? At the very least, it isn’t likely getting the dynamic term right given the winds are prescribed by the sea ice anomalies are wrong. This needs to be mentioned, and some estimate of this bias / error given to better interpret these results.
L317-318: Wouldn’t the dynamic terms be stronger not at the location of SIC anomalies, but upstream, where the ice originated? Also, how do should the magnitudes of either terms be interpreted here given that the model’s ice loss / anomaly is often incorrect (especially in the Weddell)? I’m not sure with the model biases and the way this analysis is presented that it is possible to conclusively state the thermodynamic term is often larger than the dynamic. These results should be linked to the Hovmöller diagram I am suggesting.
Citation: https://doi.org/10.5194/tc-2023-45-RC2 - AC2: 'Reply on RC2', Bianca Mezzina, 10 Oct 2023
-
RC3: 'Comment on tc-2023-45', Anonymous Referee #3, 21 Aug 2023
This manuscript attempted to summarize sea ice concentration (SIC) minima events for the Antarctic region and find out universal mechanisms that apply to all events. Results from this study provide important information for understanding the occurrence of Antarctic SIC minima events which might happen more frequently in the future in the context of climate change. The manuscript is overall well organized, with comprehensive analyses and clear interpretations of the results. There are some major issues to be addressed for this work to be published in The Cryosphere as follows.
- While this study has made lots of efforts in revealing the control of major climate modes on the SIC minima events, as shown in the results, it is hard to attribute the occurrence of these events to a universal anomalous pattern of any single climate mode or a combination of climate modes. The authors finally attributed these events to north-westly wind anomalies in the Weddell Sea and south-westerly wind anomalies in the Ross Sea. These conclusions are to some extent useful, but probably not so helpful if we want to predict future SIC minima events. In fact, in addition to SAM and ASL, there are other climate modes that can affect the sea ice anomalies in the Ross Sea and Weddell Sea, such as ENSO, PSA, PSA2, zonal wave 3, etc. While there are interactions among these climate modes, I still suggest the authors to further examine the patterns of climate modes other than SAM and ASL and see if a systematic anomalous pattern of these modes or their combinations can be found for the SIC minima events. If such a pattern can be found, the information would be much more useful for the scientific community to understand the future occurrence of SIC minimum events.
- Lines 140-141: Separating the processes controlling the tendency of SIC into a dynamical term and a thermodynamical term is a simple way. Though the authors mentioned more detailed terms in the texts that are included in the dynamics and thermodynamics (Lines 145-140), it is better to analyze these terms in Section 3.5 (Sea ice budgets), so the readers could know which specific terms are dominant as well as the physical processes behind these terms.
- As there exist notable differences between the modelled and observed SIC anomaly patterns in the Weddell Sea and the Ross Sea (Figs. 2 and 3), the authors should discuss how the model performance would affect the sea ice budget analysis in the discussion section.
Specific comments
- Lines 137-139: It is hard to understand the two criteria for selecting SIC minima events in the Weddell Sea, especially why different thresholds much be chosen for observations and model results for either criterion, and the authors should provide more explanations.
- Lines 145-146: In my mind divergence results from advection, and the two terms should not be treated separately, though I do see such separations in other literatures. I hope this can be clarified here.
- Line 280: ENSO is not examined in this study in a straightforward way so this sentence needs to be revised. Meanwhile, though ENSO can have influence on the ASL, I still suggest the authors to examine ENSO separately.
- Lines 283-284: Southwest wind anomalies can actually bring colder air masses from the Antarctic continent to the Ross Sea and increase the ice freezing, rather than causing “thermodynamic melting” mention here. So how to understand the ice melting?
- Lines 314-315: Any explanations for southerly wind in 2017 over the Weddell Sea, which is different from the wind patterns in other years?
- The legend or caption of Fig.7 should also explain the cross symbols in the two panels.
Citation: https://doi.org/10.5194/tc-2023-45-RC3 - AC3: 'Reply on RC3', Bianca Mezzina, 10 Oct 2023
Bianca Mezzina et al.
Bianca Mezzina et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
437 | 198 | 28 | 663 | 58 | 12 | 11 |
- HTML: 437
- PDF: 198
- XML: 28
- Total: 663
- Supplement: 58
- BibTeX: 12
- EndNote: 11
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1