the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
On the 2011 record low Arctic sea ice thickness: a combination of dynamic and thermodynamic anomalies
Abstract. The sea ice thickness is recognized as an early indicator of climate changes. The mean Arctic sea ice thickness has been declining for the past four decades, and a sea ice thickness record minimum is confirmed occurring in autumn 2011. We used a daily sea ice thickness reanalysis data covering the melting season to investigate the dynamic and thermodynamic processes leading to the minimum thickness. Ice thickness budget analysis demonstrates that the ice thickness loss is associated with an extraordinarily large amount of multiyear ice volume export through the Fram Strait during the season of sea ice advance. Due to the loss of multiyear ice, the Arctic ice thickness becomes more sensitive to atmospheric anomalies. The positive net surface energy flux anomalies melt roughly 0.22 m of ice more than usual from June to August. An analysis of clouds and radiative fluxes from ERA5 reanalysis data reveals that the increased net surface energy absorption supports the enhanced sea ice melt. The enhanced cloudiness led to positive anomalies of net long-wave radiation. Furthermore, the enhanced sea ice melt reduces the surface albedo, triggering an ice–albedo amplifying feedback and contributing to the accelerating loss of multiyear ice. The results demonstrate that the dynamic transport of multiyear ice and the subsequent surface energy budget response is a critical mechanism actively contributing to the evolution of Arctic sea ice thickness.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(2961 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on tc-2020-359', Anonymous Referee #1, 25 Feb 2021
tc-2020-359
On the 2011 record low Arctic sea ice thickness: a combination of dynamic and thermodynamic anomalies
Xuewei Li, Qinghua Yang, Lejiang Yu, Paul R. Holland, Chao Min, Longjiang Mu, Dake Chen
Overview:
The authors investigate the 2011 Arctic record low sea ice thickness minimum by examining satellite-derived ice thickness, area and volume, CMST modeled sea ice thickness and drift, ice transport through the Fram Strait, and the thermodynamic and dynamic processes which contributed to the record minimum.
Utilizing AWI and ESA CryoSat-2 ice thickness data, coupled with ice age data to differentiate between FYI and MYI, they examine the Arctic sea ice area, extent, and volume during the period of 2010-2020. They note the ice thickness minimum in Oct 2011, the third lowest observed ice area minimum in Fall 2012, and the decline of MYI from 2010-2012.
The CMST dataset generated from the MITgcm model which assimilated SMOS ice thickness and SSMIS ice concentration is used to analyze ice transport through the Fram Strait for the period of Oct 2010-September 2016. They find that declining trends in MYI volume in October 2010, January 2011 and March 2011 were associated with dynamic processes.
Dynamical and thermodynamic processes are evaluated during the study period to examine the role of ice thickening, advection, convergence, and residual (melting/freezing). These fields are examined for two periods: October 2010-April 2011, and May-September 2011. Anomalies of these terms are made by subtracting the 6-year mean for each month. They found a strong thinning of the ice cover along the Canadian Archipelago and portions of the central Arctic with enhanced melting (compared to 6-year mean) during the period of May – September 2011.
The Arctic Oscillation (data provided by NCEP) analysis in conjunction with the ERA5 reanalysis is used to examine the monthly variability between the AO, positive ice export anomalies through the Fram Strait, surface air temperature anomaly, and surface net heat flux anomaly. They find a positive AO from Feb-April 2011, a maximum ice export through the Fram Strait in March 2011 which coincided with strong air surface temperature and surface net heat flux anomalies in March 2011.
SLP anomalies from the ERA5 reanalysis for the period of October 2010-January 2011 showed a peak positive anomaly over southern Greenland, and a gradual divergence of sea ice along the eastern Arctic toward the northern CAA. Enhanced transpolar advection (Fig. 5a) showed the transport of sea ice from the Beaufort Sea along the coast toward the Fram Strait.
Lastly, they examine the radiative fluxes for June, July, and August 2011 to investigate the impact of cloud cover, albedos and net surface longwave radiation anomalies on the enhanced melt of primarily FYI during this period.
This is a well written paper which investigates the contributing factors which led to the record 2011 Arctic sea ice thickness minimum. A combination of satellite-derived products, the MITgcm-based CMST and the ERA5 reanalysis are used in this study. Figures and table are clear and easy to understand. All references appear to be correct. I recommend a minor revision. See comments below.
General Comments:
Use consistent use of shortwave and longwave (not short-wave, long-wave) throughout the paper. For example see lines: 202, 205, 215. Figure 6 caption uses correct form.
Lines 86-88: Gate positions at 82°N between 12°W and 20°E and 20°E between 80.5 and 82°N are defined. Which position(s) are included in Table 1?
Although not a major contributor to Arctic ice export, please comment on the role of the absence of ice arches (or bridges) in 2008-2009 and 2009-2010 (Ryan and Munchow, 2017) on the potential impact of ice export through the Fram Strait in those years.
Technical Corrections:
Line 74: Should be CryoSat-2
Line 82: What is the source of the sea ice concentration used?
Line 130: Do you mean CAA instead of CA?
Line 145: Same comment as above
Line 170: Include years October 2011-April 2012
Line 233: I assume you mean “net surface shortwave”?
Line 380: Include years of 2010, 2011 in caption.
Line 385: Why not show 4-month averages for all 3 plots shown: Oct-Jan, Feb-May, June-Sep? Any particular reason why Fig. 5g, 5h. 5i encompass a larger area than the two panels above?
Table 1: Include 6-year mean in bottom of table
Ryan, P. A. and A. Munchow (2017), Sea ice draft observations in Nares Strait from 2003 to 2012, J. Geophys. Res. Oceans, 122, 3057–3080, doi:10.1002/2016JC011966.
-
AC1: 'Reply on RC1', Qinghua Yang, 24 Apr 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Qinghua Yang, 24 Apr 2021
-
RC2: 'Comment on tc-2020-359', Anonymous Referee #2, 12 Mar 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-RC2-supplement.pdf
-
AC2: 'Reply on RC2', Qinghua Yang, 24 Apr 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Qinghua Yang, 24 Apr 2021
-
CC1: 'Some thoughts on the altimetry aspects of this paper', Robbie Mallett, 12 Mar 2021
I read this paper with great interest as a user of CryoSat-2 data, and have a couple of thoughts regarding the authors’ use of this data in the spirit of TC discussion.
It seems that a headline result of this study is that “a sea ice thickness record minimum is confirmed occurring in autumn 2011” (I hope this is fair to say given that it’s the second sentence of the abstract and first of the Summary/Discussion). I think to fully make this claim there should perhaps be a deeper consideration of the nature of this metric and the uncertainties in altimetry-derived SIT (particularly over thin ice).
To state the obvious, SIT is a local property of sea ice, whereas extent and volume are global properties of the ice pack. The impact of this is that mean SIT is sensitive to the area over which it’s averaged (unlike the other two metrics). The approach in this paper is to just average over the “area of actual ice coverage” (L37). The decision to average over this area this has many implications.
For instance, are the authors including the sub-Arctic seas like the Baltic and Okhotsk Seas and Baffin and Hudson Bays? I believe the AWI product includes SIT values for these. If the SIT of these regions contributes to the ‘mean SIT’ statistic, then how relevant is their interpretation of the 2011 minimum in terms of the dynamic/thermodynamic budget which is only produced for the Arctic Ocean (and also only with reference to 2011-16).
Next the authors should probably acknowledge that CryoSat-2 doesn’t do a good job of retrieving the thickness of thin ice (<0.5m; see Ricker et al., 2017). This is because this ice protrudes above the waterline by less than 5cm, and even less with snow cover. But here I think the authors are averaging over quite a lot of thin ice to generate their statistic. The merged CS2-SMOS product was developed with this limitation in mind, and (in my opinion) should be the product of choice for this calculation. I particularly think this because the CMST model assimilated this product, so it should perhaps be used for consistency’s sake anyway. A related issue is that CS2 simply can’t measure ice below a certain thickness. By just taking the average in places where it can measure, the authors are likely biasing their mean SIT statistic high. The size of this bias will depend on the extent of sea ice with unmeasurably low freeboard. I think they should state what fraction of the total sea ice area (as measured by a scatterometer or radiometer) is covered by the altimetry data under consideration. It’s possible that this fraction is very high and my concern isn’t warrented, but I think it is relevant.
Finally I’m not really sure what the whole-Arctic mean SIT minimum is supposed to tell us. I can imagine a year where there’s an early freezeup and therefore a lot of very thin FYI coverage. So volume could be up, but mean SIT down. (but what if this weren’t measurable by CS2?) But I can also imagine a scenario where we’re low on MYI, so volume could be down as well as mean SIT. So does whole-Arctic mean SIT mean anything? I think some more consideration of the relationship between the metric and sea ice volume is warranted.
(for instance just before the freeze-up there is less sea ice volume than afterwards. But thin FYI proliferates after the freeze-up. So the effect of the freeze-up (I think) is that sea ice volume goes up, but mean sea ice thickness goes down sharply. But then volume and mean-SIT both grow together). So the minimum occurs after freezeup but before the ice starts thickening in earnest. Does the day of minimum mean-SIT therefore occur before CS2 starts working? I’d be interested to know on what day of the year piomas predicts the minimum in whole-Arctic mean SIT, and for that matter what year has the lowest minimum in that data.
A couple of narrower points:
The authors state: “The mean sea ice thickness within the area of actual ice coverage in October 2011 reached the lowest record for that calendar month in any year of the satellite records”. The authors should point out that they have not examined the whole satellite record, which includes pan-Arctic SIT snapshots from ICESAT (2003-2010), and coverage up to 81.5 degrees by Envisat. Would be perhaps worth confirming that 2011 is a record low when also compared to ICESAT derived thickness? In particular I'm thinking about winter 2007-8 after the SIE minimum.
I’m also not sure that it’s right to cite the NSIDC Kurtz & Harbeck data as an ESA product. Given I think both Kurtz and Harbeck were at and still do work at NASA? Could be wrong about this though.
Figure 4(b): It looks a lot like FYI export from the FS is negative for almost all months here? So it’s flowing Northwards? Maybe I have the sign convention wrong, but in that case isn’t MYI flowing backwards? I think some explanation is warranted about why it looks like there’s an ice-type-dependent flow direction.
Figure 5): "Wind anomalies". Does the length of the arrow represent the magnitude of the velocity vector anomaly? Or the magnitude of the wind speed anomaly? These can be quite different. If it's the first then a large arrow can represent wind of the same speed going in a very different direction. If it's the second, then a large arrow can represent wind blowing in the same direction but at a different speed. I suppose it must be the first, because you can't have a negative arrow size? Or maybe it could be, because the arrows could then point backwards. Worth clarifying.
Lastly since this work presents data from CS2 altimetry and a model (which assimilates a related product: the CS2-SMOS data), as a reader I'm interested to know how independent the model and the altimetry are. Does the SIT data 'force' the model behaviour? Or is it a weak influence that can be relatively ignored?Citation: https://doi.org/10.5194/tc-2020-359-CC1 -
AC3: 'Reply on CC1', Qinghua Yang, 24 Apr 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-AC3-supplement.pdf
-
AC3: 'Reply on CC1', Qinghua Yang, 24 Apr 2021
Interactive discussion
Status: closed
-
RC1: 'Comment on tc-2020-359', Anonymous Referee #1, 25 Feb 2021
tc-2020-359
On the 2011 record low Arctic sea ice thickness: a combination of dynamic and thermodynamic anomalies
Xuewei Li, Qinghua Yang, Lejiang Yu, Paul R. Holland, Chao Min, Longjiang Mu, Dake Chen
Overview:
The authors investigate the 2011 Arctic record low sea ice thickness minimum by examining satellite-derived ice thickness, area and volume, CMST modeled sea ice thickness and drift, ice transport through the Fram Strait, and the thermodynamic and dynamic processes which contributed to the record minimum.
Utilizing AWI and ESA CryoSat-2 ice thickness data, coupled with ice age data to differentiate between FYI and MYI, they examine the Arctic sea ice area, extent, and volume during the period of 2010-2020. They note the ice thickness minimum in Oct 2011, the third lowest observed ice area minimum in Fall 2012, and the decline of MYI from 2010-2012.
The CMST dataset generated from the MITgcm model which assimilated SMOS ice thickness and SSMIS ice concentration is used to analyze ice transport through the Fram Strait for the period of Oct 2010-September 2016. They find that declining trends in MYI volume in October 2010, January 2011 and March 2011 were associated with dynamic processes.
Dynamical and thermodynamic processes are evaluated during the study period to examine the role of ice thickening, advection, convergence, and residual (melting/freezing). These fields are examined for two periods: October 2010-April 2011, and May-September 2011. Anomalies of these terms are made by subtracting the 6-year mean for each month. They found a strong thinning of the ice cover along the Canadian Archipelago and portions of the central Arctic with enhanced melting (compared to 6-year mean) during the period of May – September 2011.
The Arctic Oscillation (data provided by NCEP) analysis in conjunction with the ERA5 reanalysis is used to examine the monthly variability between the AO, positive ice export anomalies through the Fram Strait, surface air temperature anomaly, and surface net heat flux anomaly. They find a positive AO from Feb-April 2011, a maximum ice export through the Fram Strait in March 2011 which coincided with strong air surface temperature and surface net heat flux anomalies in March 2011.
SLP anomalies from the ERA5 reanalysis for the period of October 2010-January 2011 showed a peak positive anomaly over southern Greenland, and a gradual divergence of sea ice along the eastern Arctic toward the northern CAA. Enhanced transpolar advection (Fig. 5a) showed the transport of sea ice from the Beaufort Sea along the coast toward the Fram Strait.
Lastly, they examine the radiative fluxes for June, July, and August 2011 to investigate the impact of cloud cover, albedos and net surface longwave radiation anomalies on the enhanced melt of primarily FYI during this period.
This is a well written paper which investigates the contributing factors which led to the record 2011 Arctic sea ice thickness minimum. A combination of satellite-derived products, the MITgcm-based CMST and the ERA5 reanalysis are used in this study. Figures and table are clear and easy to understand. All references appear to be correct. I recommend a minor revision. See comments below.
General Comments:
Use consistent use of shortwave and longwave (not short-wave, long-wave) throughout the paper. For example see lines: 202, 205, 215. Figure 6 caption uses correct form.
Lines 86-88: Gate positions at 82°N between 12°W and 20°E and 20°E between 80.5 and 82°N are defined. Which position(s) are included in Table 1?
Although not a major contributor to Arctic ice export, please comment on the role of the absence of ice arches (or bridges) in 2008-2009 and 2009-2010 (Ryan and Munchow, 2017) on the potential impact of ice export through the Fram Strait in those years.
Technical Corrections:
Line 74: Should be CryoSat-2
Line 82: What is the source of the sea ice concentration used?
Line 130: Do you mean CAA instead of CA?
Line 145: Same comment as above
Line 170: Include years October 2011-April 2012
Line 233: I assume you mean “net surface shortwave”?
Line 380: Include years of 2010, 2011 in caption.
Line 385: Why not show 4-month averages for all 3 plots shown: Oct-Jan, Feb-May, June-Sep? Any particular reason why Fig. 5g, 5h. 5i encompass a larger area than the two panels above?
Table 1: Include 6-year mean in bottom of table
Ryan, P. A. and A. Munchow (2017), Sea ice draft observations in Nares Strait from 2003 to 2012, J. Geophys. Res. Oceans, 122, 3057–3080, doi:10.1002/2016JC011966.
-
AC1: 'Reply on RC1', Qinghua Yang, 24 Apr 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Qinghua Yang, 24 Apr 2021
-
RC2: 'Comment on tc-2020-359', Anonymous Referee #2, 12 Mar 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-RC2-supplement.pdf
-
AC2: 'Reply on RC2', Qinghua Yang, 24 Apr 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Qinghua Yang, 24 Apr 2021
-
CC1: 'Some thoughts on the altimetry aspects of this paper', Robbie Mallett, 12 Mar 2021
I read this paper with great interest as a user of CryoSat-2 data, and have a couple of thoughts regarding the authors’ use of this data in the spirit of TC discussion.
It seems that a headline result of this study is that “a sea ice thickness record minimum is confirmed occurring in autumn 2011” (I hope this is fair to say given that it’s the second sentence of the abstract and first of the Summary/Discussion). I think to fully make this claim there should perhaps be a deeper consideration of the nature of this metric and the uncertainties in altimetry-derived SIT (particularly over thin ice).
To state the obvious, SIT is a local property of sea ice, whereas extent and volume are global properties of the ice pack. The impact of this is that mean SIT is sensitive to the area over which it’s averaged (unlike the other two metrics). The approach in this paper is to just average over the “area of actual ice coverage” (L37). The decision to average over this area this has many implications.
For instance, are the authors including the sub-Arctic seas like the Baltic and Okhotsk Seas and Baffin and Hudson Bays? I believe the AWI product includes SIT values for these. If the SIT of these regions contributes to the ‘mean SIT’ statistic, then how relevant is their interpretation of the 2011 minimum in terms of the dynamic/thermodynamic budget which is only produced for the Arctic Ocean (and also only with reference to 2011-16).
Next the authors should probably acknowledge that CryoSat-2 doesn’t do a good job of retrieving the thickness of thin ice (<0.5m; see Ricker et al., 2017). This is because this ice protrudes above the waterline by less than 5cm, and even less with snow cover. But here I think the authors are averaging over quite a lot of thin ice to generate their statistic. The merged CS2-SMOS product was developed with this limitation in mind, and (in my opinion) should be the product of choice for this calculation. I particularly think this because the CMST model assimilated this product, so it should perhaps be used for consistency’s sake anyway. A related issue is that CS2 simply can’t measure ice below a certain thickness. By just taking the average in places where it can measure, the authors are likely biasing their mean SIT statistic high. The size of this bias will depend on the extent of sea ice with unmeasurably low freeboard. I think they should state what fraction of the total sea ice area (as measured by a scatterometer or radiometer) is covered by the altimetry data under consideration. It’s possible that this fraction is very high and my concern isn’t warrented, but I think it is relevant.
Finally I’m not really sure what the whole-Arctic mean SIT minimum is supposed to tell us. I can imagine a year where there’s an early freezeup and therefore a lot of very thin FYI coverage. So volume could be up, but mean SIT down. (but what if this weren’t measurable by CS2?) But I can also imagine a scenario where we’re low on MYI, so volume could be down as well as mean SIT. So does whole-Arctic mean SIT mean anything? I think some more consideration of the relationship between the metric and sea ice volume is warranted.
(for instance just before the freeze-up there is less sea ice volume than afterwards. But thin FYI proliferates after the freeze-up. So the effect of the freeze-up (I think) is that sea ice volume goes up, but mean sea ice thickness goes down sharply. But then volume and mean-SIT both grow together). So the minimum occurs after freezeup but before the ice starts thickening in earnest. Does the day of minimum mean-SIT therefore occur before CS2 starts working? I’d be interested to know on what day of the year piomas predicts the minimum in whole-Arctic mean SIT, and for that matter what year has the lowest minimum in that data.
A couple of narrower points:
The authors state: “The mean sea ice thickness within the area of actual ice coverage in October 2011 reached the lowest record for that calendar month in any year of the satellite records”. The authors should point out that they have not examined the whole satellite record, which includes pan-Arctic SIT snapshots from ICESAT (2003-2010), and coverage up to 81.5 degrees by Envisat. Would be perhaps worth confirming that 2011 is a record low when also compared to ICESAT derived thickness? In particular I'm thinking about winter 2007-8 after the SIE minimum.
I’m also not sure that it’s right to cite the NSIDC Kurtz & Harbeck data as an ESA product. Given I think both Kurtz and Harbeck were at and still do work at NASA? Could be wrong about this though.
Figure 4(b): It looks a lot like FYI export from the FS is negative for almost all months here? So it’s flowing Northwards? Maybe I have the sign convention wrong, but in that case isn’t MYI flowing backwards? I think some explanation is warranted about why it looks like there’s an ice-type-dependent flow direction.
Figure 5): "Wind anomalies". Does the length of the arrow represent the magnitude of the velocity vector anomaly? Or the magnitude of the wind speed anomaly? These can be quite different. If it's the first then a large arrow can represent wind of the same speed going in a very different direction. If it's the second, then a large arrow can represent wind blowing in the same direction but at a different speed. I suppose it must be the first, because you can't have a negative arrow size? Or maybe it could be, because the arrows could then point backwards. Worth clarifying.
Lastly since this work presents data from CS2 altimetry and a model (which assimilates a related product: the CS2-SMOS data), as a reader I'm interested to know how independent the model and the altimetry are. Does the SIT data 'force' the model behaviour? Or is it a weak influence that can be relatively ignored?Citation: https://doi.org/10.5194/tc-2020-359-CC1 -
AC3: 'Reply on CC1', Qinghua Yang, 24 Apr 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2020-359/tc-2020-359-AC3-supplement.pdf
-
AC3: 'Reply on CC1', Qinghua Yang, 24 Apr 2021
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
1,011 | 579 | 69 | 1,659 | 65 | 65 |
- HTML: 1,011
- PDF: 579
- XML: 69
- Total: 1,659
- BibTeX: 65
- EndNote: 65
Viewed (geographical distribution)
Country | # | Views | % |
---|---|---|---|
United States of America | 1 | 668 | 41 |
China | 2 | 249 | 15 |
Germany | 3 | 119 | 7 |
South Korea | 4 | 68 | 4 |
Netherlands | 5 | 65 | 4 |
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
- 668