Fram Strait sea ice export variability and September Arctic sea ice extent over the last 80 years

Smedsrud, Lars H.; Halvorsen, Mari H.; Stroeve, Julienne C.; Zhang, Rong; Kloster, Kjell

doi:https://doi.org/10.5194/tc-11-65-2017

Articles | Volume 11, issue 1

https://doi.org/10.5194/tc-11-65-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/tc-11-65-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 11, issue 1

Research article

|

13 Jan 2017

Research article |

| 13 Jan 2017

Fram Strait sea ice export variability and September Arctic sea ice extent over the last 80 years

Lars H. Smedsrud, Mari H. Halvorsen, Julienne C. Stroeve, Rong Zhang, and Kjell Kloster

Download

Final revised paper (published on 13 Jan 2017)
Preprint (discussion started on 02 May 2016)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review of manuscript “Fram Strait sea ice export variability and September Arctic sea ice extent over the last 80 years” by Lars H. Smedsrud, Mari H. Halvorsen, Julienne C. Stroeve, Rong Zhang, and Kjell Kloster', Anonymous Referee #1, 03 Jul 2016
- AC2: 'Response to Anonymous Reviewer #1', Lars H. Smedsrud, 15 Aug 2016
RC2: 'Review of Smedrud, et al. Fram Strait Ice Flux', Anonymous Referee #2, 19 Jul 2016
- AC1: 'Response to Anonymous Reviewer #2', Lars H. Smedsrud, 08 Aug 2016
EC1: 'Editor Summary', Jennifer Hutchings, 10 Aug 2016
- AC3: 'Response to Editor', Lars H. Smedsrud, 16 Aug 2016

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Lars H. Smedsrud on behalf of the Authors (16 Sep 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (26 Sep 2016) by Jennifer Hutchings

RR by Anonymous Referee #1 (23 Nov 2016)

Suggestions for revision or reasons for rejection

Review of manuscript “Fram Strait sea ice export variability and September Arctic sea ice extent over the last 80 years” by Lars H. Smedsrud, Mari H. Halvorsen, Julienne C. Stroeve, Rong Zhang, and Kjell Kloster

This is my second review of the manuscript. I therefore will concentrate on, if my initially mayor concerns were addressed appropriately. I thank the authors for providing a revised version of the manuscript. The text was changed at many occasions and many specific comments were addressed adequately. The analysis, however, was not changed or extended significantly.

My mayor concerns for the first version of the manuscript were

1) Before 2004 the authors only use the sea level pressure difference across Fram Strait (mSLP) as a proxy for the sea ice area export and it was suggested to combine the mSLP derived ice drift with ice area data from Walsh et al., which is used for other purposes in the analysis.
- The authors did not follow the suggestion to call the mSLP time series a proxy time series (neither ice drift nor ice area are observed directly here) but now clearly state that it is only mSLP based and clarified that several times in the manuscript, which is fine. They also clarified the description in the data section (which now raised new questions, see below).
They, however, did not combine their mSLP time series with the ice area time series. This would add variability in sea ice area to the the so called “sea ice area export time series”, which actually before 2004 only is a (indirect) ice drift speed time series. Ice area information is not included. How variable the ice area in Fram Strait can be is, e.g., shown in Kwok, 2009 (including a negative trend counteracting the positive trend found here).
The author’s argument for not using the ice area data is: “It is not straight-forward to combine it with the SLP observations to make a new and more ‘realistic’ ice export because it only provides a mid-month ice concentration field, and many of the winter months have values based on spatial and temporal interpolation.” I cannot follow that argument in sections 2.1 and 2.2 it is described that (a) even for the SAR drift data monthly ice concentrations are used and (b) the mSLP time series is a monthly dataset. The Walsh et al. mid-month is the best information we have and is used elsewhere in the manuscript. Actually, Fram Strait is one of the better observed areas in the Arctic for the pre-satellite area because of ship logs.

2) Uncertainty of SAR derived ice drift time series.
- More information was added to the manuscript and some estimates of expected uncertainty presented. Thanks. (I had a problem to find the Kloster and Sandven (2014) reference, see below)

3) Two very different time series are merged, the mSLP proxy based and the higher resolution SAR based, which also includes ice area variability. The complete mSLP based time series should be presented in addition to the merged one including statistics to let the reader judge by themselves about the differences.
- The authors stick to presenting “the best possible merged time series” only. Instead they present in their reply (for reasons not completely clear to me) the Walsh et al. 1935-2013 sea ice concentration time series, which actually just proves my last point that there is quite some variability in sea ice area in Fram Strait and including it in the ice area flux time series should be beneficial.
For the mSLP to SAR time series they refer to Smedsrud et al. (2011). Fig. 4 there shows the difference between an NCEP SLP based time series and the SAR one but not the observed SLP based time series used here. I would have loved to have such a comparison plot also in this manuscript. Fig. 4 there actually shows, in my opinion, quite high deviations between the two datasets. In summer deviations can exceed 50% in certain years and in winter deviations still can be at the 25% level. Fig. 2 in this manuscript only shows the mean seasonal cycle including standard deviations, which does not allow any conclusions on the differences, i.e., error of the mSLP based time series if we assume the SAR based one to be the truth.

4) Ice area fluxes presented here are significantly different from previous estimates based on satellite radiometer data. Conclusions drawn from these time series would be opposite than what is presented here. The quality of the time series presented here was not quantitatively assessed agains previous estimates. Conclusions therefore carefully be drawn based on the presented time series and other possibilities and explanation be mentioned.
- The authors give an extensive answer on this concern, which is appreciated. However, not much of it made it to the manuscript. The van Angelen paper is now mentioned and one short paragraph (p5, l38-41) was added at the end of section 2. The additions actually do not discuss differences to previous area flux estimates and no critical discussion was added to the results or conclusion section. While the authors in the first paragraph of their reply “certainly agree that “no particular record is the truth” and that the errors “are not well defined””
The plots in Fig.1 of the reply are actually interesting. I agree that there is no significant trend in ice area for 1935-2013 but there seems to be a clear negative trend after 1955 and also for the 1979-2013 period discussed in more detail here. In the reply Fig. 2 the authors present the mSLP based time series together with some ice concentration variability. The conclusions from this are not completely clear to me besides that the two time series do not co-vary. The only analysis that made it in the new manuscript as far as I can see is that the ice concentration variability only makes up 10% of the ice area flux variability, which is true and was shown before. However, that does not mean that changes in sea ice area cannot offset or reduce a 6% trend in ice area flux.

5) Very far reaching conclusions about the fate of the Arctic ice cover are drawn based on a very uncertain time series and very strong simplifications of physical processes and their interaction in the coupled climate system.
- The authors shortened the discussion section, added more references but rebutted that the simplifications might lead to overstated consequences. I still think the conclusions are too general and not clearly enough marked as hypothetical because they are (a) based on a uncertain time series with contradicting results found by other studies and (b) left out many physical processes, e.g., ice deformation and many others. I would prefer an approach where first the quality of a new time series is assed rigorously and discussed critically before far reaching conclusions are drawn or the limitation are clearly mentioned in the discussion. Anyway, this is probably also a matter of personal taste and one can agree to disagree.

The text of the manuscript has improved. I also appreciate the shortening of discussion section 4.5 and 4.6 while I still don’t agree on the, in my view, over-simplifications in assumptions applied here. However, most of my main concerns (see above) were not addressed. In addition, the improved new method section raised another question about the method and a potential bias. I do not have the feeling the authors included mayor revisions to address the two reviewers and editors concerns. It is the right and has my full understanding that authors rebut reviewers comments they don’t agree on. But I hope I could clarify why I cannot agree on certain points. I therefore stand with my initial assessment that the manuscript needs mayor revisions before publication.

Additional mayor comments

p3, l15-16: If I understand the procedure in the last paragraph correctly you first calculate a monthly meal sea ice drift speed value for the complete Fram Strait, i.e., you have on ice speed value per month. In a second step you now multiply this one ice drift speed value with one monthly mean sea ice concentration value for the complete strait.
Your figure 3 shows that drift speeds are much higher in the east than in the west of Fram Strait. Ice concentrations are, however, are much lower in the east. You have to multiply your high drift speeds in the east with low ice concentrations and your low drift speeds in the west with high ice concentrations to get the correct ice area flux. Your method will yield a positive ice area flux bias, which might explain some of your higher flux estimates compared to previous studies. If this bias is constant in time is not known and therefore can, potentially, also influence the area flux trend.
This issue should be investigated.
In your replies to my comments you write on p5 that you use daily ice concentration data in contrast to what is written here. Such contradictions leave open questions to me about the method applied.

Minor comments

p2, l15-20: Another recent study does not find a significant Fram Strait Sea ice area flux trends for 1988-2012 (Positive trend in summer, negative in winter).

Bi, H., K. Sun, X. Zhou, H. Huang & X. Xu (2016): Arctic Sea Ice Area Export Through the Fram Strait Estimated From Satellite-Based Data:1988–2012, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 9, 3144-3157, doi:10.1109/jstars.2016.2584539.

Fig. 13: caption says dataset until December 2013, text says until December 2014

References
I could not verify any statements regarding the SAR sea ice drift dataset (e.g. sec 2.1) because I could not find how to access the main reference:
Kloster, K. and Sandven, S.: Ice Motion and Ice Area Flux in Fram Strait at 79N using SAR and passive microwave for Aug. 2004–Jul. 2014., Technical Report no. 322c. Nansen Environmental and Remote Sensing Center, Bergen, Norway, 2014.

If this technical report is not accessible to the general public it should be removed from the reference list.

Hide

ED: Publish subject to minor revisions (Editor review) (26 Nov 2016) by Jennifer Hutchings

AR by Lars H. Smedsrud on behalf of the Authors (06 Dec 2016) Author's response Manuscript

ED: Publish subject to minor revisions (Editor review) (13 Dec 2016) by Jennifer Hutchings

Dear Lars and co-authors,

Thank you for your response to the review and especially for paying attention to the concern regarding the role of varying ice area in interannual variability in Fram Strait export.

The reviewer did express concern about your results being overstated in the discussion and there not being a clear delineation between the direct findings and hypothesizes about how changes in export drive interia Arctic ice pack extent changes in summer. I also find this, and agree there are missing mechanism and processes in your discussion. I agree with the reviewers that it would be best to focus on the data set presented (your proxy time series) and clearly delineate hypothesized implications of your findings. Below are some specific comments where I find you can draw this distinction.

In the abstract your newly added text discusses the relationship between export and opening in the central Arctic, and hence summer ice extent. It would be important here to identify that you only find 18% correlation between export and end of summer ice extent.

Section 4.6 is where your discussion delves into territory outside of the scope of your time series analysis. The paragraph at line 30, page 12 is in particular overstating in how you relate your findings to climate model simulations. These models do not show the recent decline in ice extent, and if they do it may not be for the same mechanisms as we are observing now. There is a wide range of literature on this that I am sure you and your co-authors are aware of. No evidence of a trend in export in models between 1979-2013 should not be compared to your result that there is no trend over a longer time period. And no trend does not imply that the phenomena are purely driven by natural variability. It is possible for interannual and decadal scale variability to increase, due to external forcing, while maintaining no trend. In my opinion this section should be reduced to meet the reviewer's concerns.

You have presented a very simplistic model for how increased export relates to summer ice extent reductions. In general I am a fan of simple mechanistic models that can be tested, provided the limitations of the model are clearly articulated. I would caution you to make sure your text is clear that you are describing the mechanism disregarding some potentially important factors (changing stability in the ocean/atmosphere for example), in the interest of presenting the existence of the link and recognize that quantifying the physical mechanisms involved in the link is not possible.

Specific Edits Required

p1,l34: remove duplicate 'the last'

p2,p13-22: I disagree about not referencing recent papers you were unaware of when writing your manuscript, if they are relevant to your results. Kloster and Sandven (2015) is obviously a relevant study and does enter into the conversation about the large differences in estimates of trends in Fram Strait export.
On this topic, did you consider the differences in location of flux gates in explaining differences between studies?

p4,l26: I realize that you can not identify if there have been changes in seasonality of the fram strait ice drift and export prior to 1979. However if there were changes this will affect your bias correction. It is fair to point out the limitations of your assumptions. p5,l21 might be a relevant place to point this out, or where you introduce the method.

p5,l26: How do you blend the two products? The use of this word makes me think you may be doing more than a simple substition. Please be more precise in your description here.
I am happy that you have discussed the relative impact on results for both the full proxy time series and the blended time series. I do not that you have chosen to present in figures the time series that demonstrates the largest change in the last decade.

p6,l11: "true monthly mSLP". I am not sure what you mean by true here. I would chose a descriptor that clarifies this is the observed pressure. (truth is not a given for weather station data too, any measurement has an error associated with it!)

Best regards,
Jenny

Hide

AR by Lars H. Smedsrud on behalf of the Authors (16 Dec 2016) Author's response Manuscript

ED: Publish as is (19 Dec 2016) by Jennifer Hutchings

AR by Lars H. Smedsrud on behalf of the Authors (21 Dec 2016)

Short summary

Export of Arctic sea ice area southwards through the Fram Strait from 1935 to 2014 is calculated based on satellite radar images and surface pressure observations. The annual mean export is 880 000 km², representing 10 % of the Arctic sea ice area. In recent years the export has been above 1 million km², and there are positive trends over the last 30 years. Increased ice export during spring and summer contributes to more open water in September, and this correlations has increased over time.