the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 Nov 2024
Defining Antarctic polynyas in satellite observations and climate model output to support ecological climate change research
Abstract. Antarctic polynyas are key components of Antarctic marine ecosystems, influencing light and nutrient availability and open water access for marine predators. Thus, changes in the physical characteristics of polynyas can influence how these ecosystems respond to a changing climate. Here, we explore how to identify polynyas using satellite and Earth System Model data, and we assess the impacts of using different polynya-identification metrics (sea ice concentration or thickness). Our results show optimal metrics for polynya definition will depend on the temporal and spatial resolution of the data, as well as the season and region of interest. These results highlight the importance of identifying polynyas on grids of the same type and resolution when comparing polynyas from different data products. We find that sea ice thickness is more suitable for identifying polynyas in model data in winter months in contrast to spring months when both sea ice thickness and concentration may be suitable metrics. We then use the Community Earth System Model Version 2 (CESM2) to investigate ecosystem function within polynyas and find that there is enhanced phytoplankton productivity in modeled polynya features in both hindcast and fully coupled simulations, with springtime polynyas remaining an important control on Antarctic productivity under future climate change.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(6046 KB)
-
Supplement
(2420 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(6046 KB) - Metadata XML
-
Supplement
(2420 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3490', Anonymous Referee #1, 16 Dec 2024
-
AC1: 'Reply on RC1', Laura Landrum, 13 Feb 2025
Thank you to the reviewers for careful and considerate reading of our submission and thoughtful feedback that will make this a stronger paper. Thank you also for your general support of the manuscript and overall suitability for publishing.
After reading all three reviews we agree that we can do a better job of clearly communicating the intent and results of this paper. Our four primary intentions and results in this work are:
- An analysis of the sensitivity of polynya detection in gridded sea ice products (both satellite and model output) to sea ice product types, grids, and polynya definition metrics and thresholds. When comparing polynya areas from different data products it is important to use uniform grids, and optimal metric choices may also depend on season and region of interest.
- A comparison of Antarctic coastal polynyas detected in satellite observations and simulated in a climate model forced with atmospheric reanalysis data. A hindcast version of the CESM2 identifies similar integrated Antarctic coastal polynya areas, although the polynyas identified in the model are individually larger and less numerous than those in the satellite product.
- To assess the internal variability of coastal polynya areas in an ensemble of fully coupled climate model simulations for historical and future time periods. Coastal polynya areas identified in climate model output exhibit significant internal variability in historical and future time periods. The same metrics and thresholds tend to identify higher springtime coastal polynya areas in the hindcast simulation than in the fully coupled CESM2, which may be related to the slightly more rapid sea ice retreat in the hindcast simulation (e.g. Figure 1).
- To test if coastal polynyas identify marine productivity “hot spots” and how both polynyas and sea ice zone NPP may change in the future. Coastal polynyas identify “hot spots” in marine productivity, relative to the rest of the sea ice zone, in the CESM. Average NPP within coastal polynyas does not change significantly in future scenarios, although the contribution of NPP within polynyas relative to the total sea ice zone productivity decreases as sea ice zone NPP increases and total polynya areas decrease. This is particularly true in the spring.
If given the opportunity to rewrite this submission based on reviewer feedback, we will restructure the paper based on reviewer comments regarding lack of clarity and highlight and distinguish between these four primary results. A revised abstract reads:
Antarctic polynyas are key components of Antarctic marine ecosystems, influencing light and nutrient availability and open water access for marine predators. Thus, changes in the physical characteristics of polynyas can influence how these ecosystems respond to a changing climate. Here, we explore how to identify polynyas using satellite and Earth System Model data, and we assess the impacts of using different polynya-identification metrics and thresholds (sea ice concentration or thickness). Our results show optimal metrics for coastal polynya definition will depend on both the data resolution, as well as the season and region of interest. Temporal resolution (daily vs monthly) has an impact on polynya identification in specific regions and seasons but has no significant impact on integrated hemispheric coastal polynya area. Our results highlight the importance of identifying polynyas on grids of the same type and resolution when comparing polynyas from different data products. We find that sea ice thickness is more suitable for identifying polynyas in model data in winter months in contrast to spring months when both sea ice thickness and concentration may be suitable metrics. The Community Earth System Model Version 2 (CESM2) identifies similar integrated Antarctic coastal polynya areas and locations as satellite-based sea ice data, although polynyas identified in the CESM2 tend to be larger and fewer in number than those identified in the satellite products particularly in the winter. We then use the CESM2 to investigate ecosystem function within polynyas and find that there is enhanced phytoplankton productivity in modeled polynya features in both hindcast and fully coupled simulations. Springtime polynyas remain an important control on Antarctic productivity under future climate change, although the relative role of polynyas decreases as polynyas diminish and phytoplankton productivity outside of polynyas increases as sea ice cover decreases.
We will also provide information on modifications that are made within the text to address the reviewer concerns.
Responses to specific reviewer comments:
Reviewer 1
Reviewer 1 points out that did not adequately address potential model biases early on in the paper.
We will add specific caveats in the introduction highlighting that many climate models have large biases in the Antarctic sea ice and that our work is focusing on polynya identification in gridded products and an analysis and application of polynya identification on one climate model. We will add broad discussion (w.r.t. climate models in general) and model-specific (e.g. the CESM) biases in sea ice and how this may also impact polynya detection beyond the identification methodologies analyzed in this submission.
Comments from reviewer 1 highlight that our motivation for using the hindcast simulation (JRA-CESM) was not clear.
Our comparisons between coastal polynyas identified in satellite data and model output are presented primarily using the hindcast version of CESM (JRA-CESM) as this provides the closest analogue to the observations. The hindcast (which is an ocean-sea ice only simulation) is forced by atmospheric reanalysis products. Using the hindcast thereby removes differences between satellite observations and the model that would be due to atmospheric forcing or internal variability in the fully coupled CESM simulations. Thus, much of the simulated ice conditions and variability should be more directly comparable to satellite products. We will add discussion regarding interannual variability of simulated coastal polynyas in the fully coupled model to add context for the results from the hindcast simulation.
Climatological sea ice concentrations (SIC) and extents (SIE) are very similar in the CESM-JRA and the CESM2-LE. Mean SIEs from the hindcast simulation fall within the variability of the fully coupled CESM2-Large Ensemble, with two regional exceptions. The first exception is found in the Weddell Sea in the summer (Jan-Feb) when mean SIE in the JRA-CESM lies below the range (and thus further away from the observations) in the CESM-LE. The second exception lies in eastern Antarctica during July-November, when mean JRA-CESM SIE is higher – and thus closer to the satellite-based SIE – than the CESM-LE (Figure 1 and Supplemental Figure 2). We will create figures of mean, monthly polynya areas over the historical time period for both the JRA-CESM and CESM-LE to compare when/if there are significant regional and seasonal differences, and add text to the paper describing any differences. We will include a figure showing this either in the manuscript or in the supplementary material.
Reviewer 1 suggestions for Figures will be addressed as follows:
- We will change colors to better distinguish in Figure 1
- We will change polynya marker colors in Figure 2
- We will change line colors in Figures 3 onwards to better distinguish the results
- We will increase font size for legends in Figures 4-6
Citation: https://doi.org/10.5194/egusphere-2024-3490-AC1
-
AC1: 'Reply on RC1', Laura Landrum, 13 Feb 2025
-
RC2: 'Comment on egusphere-2024-3490', Tarkan Bilge, 29 Dec 2024
The study under review presents an analysis of the sensitivity of polynya occurrence and area to polynya definition thresholds in satellite observations and climate model output. The authors subsequently use this analysis to construct methodological recommendations for future studies on polynyas. In my opinion, these two aspects represent the main two contributions of the study; methodological advice for polynya studies, and numerical evidence for the way in which resolution/data-type/season influences polynya statistics. This submission is original, being the first such study to quantitatively demonstrate the influence of resolution and data type (e.g. model/observational) on polynya number and area. The main significance of the study lies in the methodological recommendations and accompanying numerical justifications which can be used by future studies to consistently define polynyas without the requirement for authors to invent their own definitions (e.g. Mohrmann, 2021). In my opinion, the scientific content of the article is clearly articulated and well presented and only requires minor corrections prior to publication.
Below I have included a small number of specific issues which I would like to bring to the authors' attention.
The authors have explained that this study is focused on coastal polynyas, (line 165), but in the study abstract and in the the conclusions (e.g. conclusions 3, 4, 6) refer to "polynyas" more generally. Given that the mechanisms which cause coastal polynyas and open-water polynyas are different, it is not necessarily possible to extrapolate conclusions from the analysis of one to the other. For example, the authors explain that SIT may be a better metric in winter due to air temperatures causing almost immediate surface freezing - I would expect this effect to disproportionately impact coastal polynya identification, because open-water polynyas feature upwelling of warm water which might less readily freeze. Related to this comment, Mohrmann et al. (2021) found that the "CMIP6 CESM2 models show coastal polynyas, but never OWPs", so extending analysis to open-water polynyas using CESM2 might not be possible. Instead, I would recommend the authors to caveat these potential differences, and clearly state in the abstract, conclusions, and main text that results are relevant for *coastal* polynyas. This may sound like a pedantic point, but since this study aims to make recommendations on polynya definition and thresholding to future studies, it is important that future studies on open-water polynyas recognise that not all of the thresholding analysis has been carried out with open-water polynyas in mind, and therefore the conclusions may not be suitable for application in their studies.
In section 4.6, polynya number and area are analysed in the CESM2-LE, and are found to decrease towards the end of the century. I think the study would benefit from some analysis or informed speculation as to the mechanism which drives this decrease in future polynyas under emissions scenarios. The reduction could be a result of changing mean-wind fields, or more simply due to a reduction in overall Antarctic sea-ice. If the authors compare the ~2090 mean Antarctic July and November sea-ice field with the ~2020 fields, they might find that a mean field difference could explain this reduction. This might also explain why the November polynya statistics drop off more readily than the July ones.
On line 312, the authors say that April to October the polynya area is higher for daily mean than monthly mean data, often by more than one standard deviation. Should this mean that the dark blue line and light red shaded areas should be separate in Figures 3a, S2 (Apr-Oct)? Perhaps the lines are too thick or the plots are too small to appreciate this, could the authors check this statement. The authors go on to conclude than monthly mean and daily mean polynya areas are comparable on a hemispheric basis, so this point does not seem critical for the main argument.
On line 456, the authors mention that the SIC threshold modelled polynya areas are closer to the satellite areas, this seems to be only really true for the monthly data. On this note, there seems to be quite a large difference between daily and monthly averaged JRA-CESM SIC thresholds in November. Conclusion 5) states that polynya areas are comparable between monthly/daily thresholds on a hemispheric basis, but as mentioned in Line 596, and as is visible in Figure 4c, there is a seasonality to this, and while it is true on an annual-average basis, the CDR daily to monthly cross-correlation drops to ~0.65 in December. In general, Conclusion 5) is important result, and I think the authors should consider modifying the phrasing to be '...data are comparable on an annual averaged and hemispheric basis', or similar.
Further technical corrections:
Line 162: reference should be Mohrmann et al. (2021). Please check spelling of Mohrmann elsewhere in the text.
Line 243: could authors double-check that '≥' sign is correct. Will depend on journal precedent.
Line 293-297: repeated point about CDR vs NASA Team polynya comparison.
Line 454: unmatched bracket at end of line.
Line 547: Figure 8 caption; missing space before comma, should use 'a)' and 'b)' in caption text, formatting of following needs to be consistent "(red; 85% SIC)", "(85% SIC; black)", "(0.4m SIT, blue)".
Line 630: long space.
Line 715: Figure A2 caption '(REF)' left in.
Line 735: Figure B1 caption capitilisation of 'Sea Ice Concentration' inconsistent with other figures e.g. Figure B2.
Line 760: Figure C1 caption has no '(d)'.Many thanks to the authors for their submission, which I feel contains a valuable contribution to the study of Antarctic sea ice, and I hope that the comments above prove useful in finalising this work.
Citation: https://doi.org/10.5194/egusphere-2024-3490-RC2 -
AC2: 'Reply on RC2', Laura Landrum, 13 Feb 2025
Thank you to the reviewers for careful and considerate reading of our submission and thoughtful feedback that will make this a stronger paper. Thank you also for your general support of the manuscript and overall suitability for publishing.
After reading all three reviews we agree that we can do a better job of clearly communicating the intent and results of this paper. Our four primary intentions and results in this work are:
- An analysis of the sensitivity of polynya detection in gridded sea ice products (both satellite and model output) to sea ice product types, grids, and polynya definition metrics and thresholds. When comparing polynya areas from different data products it is important to use uniform grids, and optimal metric choices may also depend on season and region of interest.
- A comparison of Antarctic coastal polynyas detected in satellite observations and simulated in a climate model forced with atmospheric reanalysis data. A hindcast version of the CESM2 identifies similar integrated Antarctic coastal polynya areas, although the polynyas identified in the model are individually larger and less numerous than those in the satellite product.
- To assess the internal variability of coastal polynya areas in an ensemble of fully coupled climate model simulations for historical and future time periods. Coastal polynya areas identified in climate model output exhibit significant internal variability in historical and future time periods. The same metrics and thresholds tend to identify higher springtime coastal polynya areas in the hindcast simulation than in the fully coupled CESM2, which may be related to the slightly more rapid sea ice retreat in the hindcast simulation (e.g. Figure 1).
- To test if coastal polynyas identify marine productivity “hot spots” and how both polynyas and sea ice zone NPP may change in the future. Coastal polynyas identify “hot spots” in marine productivity, relative to the rest of the sea ice zone, in the CESM. Average NPP within coastal polynyas does not change significantly in future scenarios, although the contribution of NPP within polynyas relative to the total sea ice zone productivity decreases as sea ice zone NPP increases and total polynya areas decrease. This is particularly true in the spring.
If given the opportunity to rewrite this submission based on reviewer feedback, we will restructure the paper based on reviewer comments regarding lack of clarity and highlight and distinguish between these four primary results. A revised abstract reads:
Antarctic polynyas are key components of Antarctic marine ecosystems, influencing light and nutrient availability and open water access for marine predators. Thus, changes in the physical characteristics of polynyas can influence how these ecosystems respond to a changing climate. Here, we explore how to identify polynyas using satellite and Earth System Model data, and we assess the impacts of using different polynya-identification metrics and thresholds (sea ice concentration or thickness). Our results show optimal metrics for coastal polynya definition will depend on both the data resolution, as well as the season and region of interest. Temporal resolution (daily vs monthly) has an impact on polynya identification in specific regions and seasons but has no significant impact on integrated hemispheric coastal polynya area. Our results highlight the importance of identifying polynyas on grids of the same type and resolution when comparing polynyas from different data products. We find that sea ice thickness is more suitable for identifying polynyas in model data in winter months in contrast to spring months when both sea ice thickness and concentration may be suitable metrics. The Community Earth System Model Version 2 (CESM2) identifies similar integrated Antarctic coastal polynya areas and locations as satellite-based sea ice data, although polynyas identified in the CESM2 tend to be larger and fewer in number than those identified in the satellite products particularly in the winter. We then use the CESM2 to investigate ecosystem function within polynyas and find that there is enhanced phytoplankton productivity in modeled polynya features in both hindcast and fully coupled simulations. Springtime polynyas remain an important control on Antarctic productivity under future climate change, although the relative role of polynyas decreases as polynyas diminish and phytoplankton productivity outside of polynyas increases as sea ice cover decreases.
We will also provide information on modifications that are made within the text to address the reviewer concerns.
Reviewer 2
Reviewer 2 points out that although the anlayis in this study is focused on coastal polynyas and yet this is not clearly stated particularly in the abstract and the discussion.
This point is well taken – thank you for bringing this to our attention. We will rewrite the abstract and the discussion to make this clear. In addition, we will add text specifically aimed at clarification and caveats with respect to open water polynyas, which result from mechanisms different than coastal polynyas. We will explicitly state that because of this, our conclusions may not apply to open water polynyas.
We do not delve into analysis and description of why polynyas decline at the end of the 21st Century in the CESM-LE and agree that some analysis and discussion would benefit the paper. DuVivier et al, 2023, found that polynyas in the Ross Sea during winter decrease and sea ice concentrations and thicknesses increase in the 21st century in the CESM2-LE, likely due to both large scale circulation changes and local feedbacks. We will investigate regional changes in coastal polynyas for a broader understanding of some of the causes of the projected decreases in both winter and spring polynya areas, although an in-depth analysis of the mechanisms is beyond the purview of this paper.
Reviewer 2 found a mistake on line 312 in the submission – thank you!
We replace:
“In general, polynyas identified using daily data lead to larger total polynya area April through October across the SIC thresholds than those based on monthly data, and the mean integrated polynya areas based on the daily data often lie more than one standard deviation higher than those based on the monthly mean data. As spring turns to summer, the sensitivity of polynya area to threshold value becomes larger than any differences in the means calculated from different grid and temporal resolutions (Fig. 3b and Supplemental Fig. 2).”
“When moving beyond an example day and looking across all years, there are significant differences in integrated total SH area of coastal polynyas in the satellite-based data products as identified by SIC threshold value, grid resolution, temporal resolution (daily vs monthly) and season (Fig. 3 a-b and Supplemental Fig. 2). The total area of coastal polynyas is larger for a given threshold in the CDR on the original higher-resolution EASE grid than when regridded to the 1° CESM grid, and these differences are most pronounced from late fall through early spring (May-Oct; Supplemental Fig. 2). In general, polynyas identified using daily data lead to larger total polynya area April through October across the SIC thresholds than those based on monthly data, and the mean integrated polynya areas based on the daily data often lie more than one standard deviation higher than those based on the monthly mean data. As spring turns to summer, the sensitivity of polynya area to threshold value becomes larger than any differences in the means calculated from different grid and temporal resolutions (Fig. 3b and Supplemental Fig. 2). Throughout most of the year, differences due to temporal resolution are much smaller at a given SIC threshold than the differences due to grid-size. That is, more grid cells are identified as polynyas, and the total SH polynya area is larger on the original, finer-grid EASE grid than on the regridded data for a given threshold value.”
With:
“When moving beyond an example day and looking across all years, the primary differences in integrated total SH area of coastal polynyas in the satellite-based data products as identified by SIC threshold values arise from grid resolution during the late fall through early spring (May-October; Fig. 3 a-b and Supplemental Fig. 2). Integrated southern hemisphere (SH) coastal polynya areas are larger for a given threshold in the CDR on the original higher-resolution EASE grid than when regridded to the 1° CESM grid. Differences in integrated SH coastal polynya areas identified using daily data lie within one standard deviation of polynyas identified in the monthly data. As spring turns to summer, the sensitivity of polynya area to threshold value becomes larger than any differences in the means calculated from different grid and temporal resolutions (Fig. 3b and Supplemental Fig. 2). Throughout most of the year, differences due to temporal resolution are much smaller at a given SIC threshold than the differences due to grid-size. That is, more grid cells are identified as polynyas, and the total SH polynya area is larger on the original, finer-grid EASE grid than on the regridded data for a given threshold value.”
Reviewer 2 finds that our summary on regional polynya areas (Section 4.4.2) and our conclusion regarding monthly vs daily data for polynya identification needs clarification.
We will replace lines 459-461:
“In general, the spatial correlations between polynya area as a function of longitude identified in the satellite product and the JRA-CESM polynya area (Fig. 4c) are higher than they are for the SH timeseries (Fig. 460 5c).”
With:
“In general, the spatial correlations between polynya area as a function of longitude identified in the satellite product and the JRA-CESM polynya area (Fig. 4c) are higher than the temporal correlations for the integrated SH timeseries (Fig. 5c)”
In addition, in the primary recommendations (in our conclusions) we will replace:
“polynya areas identified from monthly and daily data are comparable on hemispheric basis”
with:
“polynya areas identified from monthly and daily data are comparable on an annually averaged, integrated hemispheric basis”
If given the opportunity to revise our submission, we will check and change accordingly the 9 specific technical corrections found by reviewer 2.
References
DuVivier, A.K., Molina, M.J., Deppenmeier, AL. et al. Projections of winter polynyas and their biophysical impacts in the Ross Sea Antarctica. Clim Dyn 62, 989–1012 (2024). https://doi.org/10.1007/s00382-023-06951-z
Ivanova, N., Pedersen, L. T., Tonboe, R. T., Kern, S., Heygster, G., Lavergne, T., Sorensen, A., Saldo, R., Dybjkaer, G., Brucker, L. and Shokr, M.: Inter-comparison and evaluation of sea ice algorithms: towards further identification of challenges and optimal approach using passive microwave observations, The Cryo., 9, 1797-1817, https://doi.org/10.5194/tc-9-1797-2015, 2015.
Lavergne, T., Sørensen, A. M., Kern, S., Tonboe, R., Notz, D., Aaboe, S., Bell, L., Dybkjær, G., Eastwood, S., Gabarro, C., Heygster, G., Killie, M. A., Brandt Kreiner, M., Lavelle, J., Saldo, R., Sandven, S., and Pedersen, L. T.: Version 2 of the EUMETSAT OSI SAF and ESA CCI sea-ice concentration climate data records, The Cryo., 13, 49–78, https://doi.org/10.5194/tc-13-49-2019, 2019.
Citation: https://doi.org/10.5194/egusphere-2024-3490-AC2
-
AC2: 'Reply on RC2', Laura Landrum, 13 Feb 2025
-
RC3: 'Comment on egusphere-2024-3490', Anonymous Referee #3, 17 Jan 2025
Review of “Defining Antarctic polynyas in satellite observations and climate model output to support ecological climate change research” by Laura Landrum et al.
The MS is comparing the polynya size and location from satellite SMMR, SSM/I – SSMIS microwave radiometer sea ice concentrations (SIC) and climate model simulations and then use the model simulations to assess the polynya ecosystems and long-term trends over the period where satellite data are available (since 1979). The topic is very welcome and the MS is well written and structured. My primary concern is the relatively poor match between polynyas detected from the satellite observations and the model simulations. This does not give high confidence in the results. Part of that mismatch is due to real differences between observed and model simulated polynyas and part of that is due to the definition of polynyas in the two datatypes, spatial resolution etc. as discussed in the MS. The satellite microwave radiometer SIC data are relatively consistent over the study period (because the SIC algorithm is using the 19 and 37 GHz channels) and it covers from 1979 until today. However, as also mentioned in some of the references (Markus and Burns), this channel combination is not ideal for mapping polynyas. For example, due to coarse resolution and sensitivity to new and mature ice types and also the way the data are processed excludes low SIC from being detected (weather filters). The NASA Team SIC is included for comparison, but it is using the same 19 and 37 GHz channels as the “CDR” and it has approximately the same spatial resolution and new and mature ice sensitivity. It also uses weather filters which excludes the detection of low SIC. The microwave radiometer swath data are resampled to daily 25 km x 25 km grids while the spatial resolution of the SIC is approximately 45 km. Other satellite radiometer datasets for detection of polynyas mentioned in the MS (Arrigo & van Dijken using the Markus & Burns algorithm and detections based on OSISAF SIC in Mohrman et al.) are not used for comparison which contributes to limited confidence in the detected location and size of the polynyas.
I would suggest to compare the detection of polynyas in the CDR dataset with other (already published) algorithms for polynya detection. Hopefully this will lead to a better match with the model simulations. I think that this is needed to give confidence in the model simulations.
Specific comments
L51 and 53: I don’t think that “Passive microwave images” is a descriptive name for the CDR sea ice concentrations. Please find a better term for it and change it throughout the MS.
L53: “Passive microwave images tend to underestimate SIC’s… “ This is not accurate as a general statement about “passive microwave” data. The sensitivity of the microwave radiometer data to sea ice concentration and the inherent noise depends on the selection of channels and the algorithm used for processing the SIC as described in Ivanova et al. I would suggest to rewrite this entire paragraph L47-L87 with reference to the specific data which is used in this study (CDR). The level of detail about the CDR data is not adequate.
L200: Polynyas are formed in a dynamic environment and it is unlikely that nilas grows to a thickness of 5 cm on scales of the satellite radiometer footprint or resolution of the model. The type of new ice formed in a dynamic environment is pancake ice and pancake ice does not reach 100% SIC before consolidation is reached at a thickness higher than 5 cm. In addition, different SIC algorithms have different sensitivity to new and mature ice. Please specify which algorithm that you are referring to.
L258: 10-14% is that approximately 2 million km2? Please provide the difference in percent and in square kilometers.
L425: Are the thin ice formation processes in the model realistic considering the dynamic environment where polynyas appear? Please, include a discussion of ice formation processes in polynyas.
L691: Please specify the source and date of the “satellite image” in figure AA2 (MODIS?). L705 (fig. AA2) the title is truncated and there is a reference missing in the fig-text.
L735: in the figure title, I think that it should be “25 x 25 km2”
L740: If polynya detection depends on grid resolution a lat-lon grid is not optimal. It never is in polar regions.
Citation: https://doi.org/10.5194/egusphere-2024-3490-RC3 -
AC3: 'Reply on RC3', Laura Landrum, 13 Feb 2025
Thank you to the reviewers for careful and considerate reading of our submission and thoughtful feedback that will make this a stronger paper. Thank you also for your general support of the manuscript and overall suitability for publishing.
After reading all three reviews we agree that we can do a better job of clearly communicating the intent and results of this paper. Our four primary intentions and results in this work are:
- An analysis of the sensitivity of polynya detection in gridded sea ice products (both satellite and model output) to sea ice product types, grids, and polynya definition metrics and thresholds. When comparing polynya areas from different data products it is important to use uniform grids, and optimal metric choices may also depend on season and region of interest.
- A comparison of Antarctic coastal polynyas detected in satellite observations and simulated in a climate model forced with atmospheric reanalysis data. A hindcast version of the CESM2 identifies similar integrated Antarctic coastal polynya areas, although the polynyas identified in the model are individually larger and less numerous than those in the satellite product.
- To assess the internal variability of coastal polynya areas in an ensemble of fully coupled climate model simulations for historical and future time periods. Coastal polynya areas identified in climate model output exhibit significant internal variability in historical and future time periods. The same metrics and thresholds tend to identify higher springtime coastal polynya areas in the hindcast simulation than in the fully coupled CESM2, which may be related to the slightly more rapid sea ice retreat in the hindcast simulation (e.g. Figure 1).
- To test if coastal polynyas identify marine productivity “hot spots” and how both polynyas and sea ice zone NPP may change in the future. Coastal polynyas identify “hot spots” in marine productivity, relative to the rest of the sea ice zone, in the CESM. Average NPP within coastal polynyas does not change significantly in future scenarios, although the contribution of NPP within polynyas relative to the total sea ice zone productivity decreases as sea ice zone NPP increases and total polynya areas decrease. This is particularly true in the spring.
If given the opportunity to rewrite this submission based on reviewer feedback, we will restructure the paper based on reviewer comments regarding lack of clarity and highlight and distinguish between these four primary results. A revised abstract reads:
Antarctic polynyas are key components of Antarctic marine ecosystems, influencing light and nutrient availability and open water access for marine predators. Thus, changes in the physical characteristics of polynyas can influence how these ecosystems respond to a changing climate. Here, we explore how to identify polynyas using satellite and Earth System Model data, and we assess the impacts of using different polynya-identification metrics and thresholds (sea ice concentration or thickness). Our results show optimal metrics for coastal polynya definition will depend on both the data resolution, as well as the season and region of interest. Temporal resolution (daily vs monthly) has an impact on polynya identification in specific regions and seasons but has no significant impact on integrated hemispheric coastal polynya area. Our results highlight the importance of identifying polynyas on grids of the same type and resolution when comparing polynyas from different data products. We find that sea ice thickness is more suitable for identifying polynyas in model data in winter months in contrast to spring months when both sea ice thickness and concentration may be suitable metrics. The Community Earth System Model Version 2 (CESM2) identifies similar integrated Antarctic coastal polynya areas and locations as satellite-based sea ice data, although polynyas identified in the CESM2 tend to be larger and fewer in number than those identified in the satellite products particularly in the winter. We then use the CESM2 to investigate ecosystem function within polynyas and find that there is enhanced phytoplankton productivity in modeled polynya features in both hindcast and fully coupled simulations. Springtime polynyas remain an important control on Antarctic productivity under future climate change, although the relative role of polynyas decreases as polynyas diminish and phytoplankton productivity outside of polynyas increases as sea ice cover decreases.
We will also provide information on modifications that are made within the text to address the reviewer concerns.
Reviewer 3
Reviewer 3 suggests that the paper will be much stronger by including other satellite-based data that have used algorithms different that that used for the sea ice concentrations of the NSIDC-CDR.
This is an excellent recommendation, and we appreciate the thoughtfulness and insight of this recommendation from a clear satellite data expert. Upon a closer read of Ivanova et al. (2015), we believe it would be informative to include results using the OSIASF SIC (Lavergne et al., 2019) in our analysis. The algorithms used by the NASA Team product and the OSIASF product result in particularly large differences in SICs in thin and low sea ice areas and thus represents an important comparative tool for this study. We anticipate that this additional analysis with a new product will be a major modification to this work.
Additionally, we also noticed a mistake in the labels in Figure 8. We will correct – orange lines are for JRA-CESM using SIT 0.4m threshold and navy lines are for JRA-CESM SIC 85% threshold.
References
DuVivier, A.K., Molina, M.J., Deppenmeier, AL. et al. Projections of winter polynyas and their biophysical impacts in the Ross Sea Antarctica. Clim Dyn 62, 989–1012 (2024). https://doi.org/10.1007/s00382-023-06951-z
Ivanova, N., Pedersen, L. T., Tonboe, R. T., Kern, S., Heygster, G., Lavergne, T., Sorensen, A., Saldo, R., Dybjkaer, G., Brucker, L. and Shokr, M.: Inter-comparison and evaluation of sea ice algorithms: towards further identification of challenges and optimal approach using passive microwave observations, The Cryo., 9, 1797-1817, https://doi.org/10.5194/tc-9-1797-2015, 2015.
Lavergne, T., Sørensen, A. M., Kern, S., Tonboe, R., Notz, D., Aaboe, S., Bell, L., Dybkjær, G., Eastwood, S., Gabarro, C., Heygster, G., Killie, M. A., Brandt Kreiner, M., Lavelle, J., Saldo, R., Sandven, S., and Pedersen, L. T.: Version 2 of the EUMETSAT OSI SAF and ESA CCI sea-ice concentration climate data records, The Cryo., 13, 49–78, https://doi.org/10.5194/tc-13-49-2019, 2019.
Citation: https://doi.org/10.5194/egusphere-2024-3490-AC3
-
AC3: 'Reply on RC3', Laura Landrum, 13 Feb 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3490', Anonymous Referee #1, 16 Dec 2024
I see four main results in this paper:
- Sensitivity analysis of the various choices involved in polynya detection using observations (thresholds, concentration vs thickness, grid type, daily vs monthly, CDR vs NASA Team);
- Assessment of the representation of polynyas in CESM2-LE and the effect of internal variability;
- Comparison of the representation of polynyas between the fully-coupled CESM2 and the forced JRA-CESM, i.e. detangling the contributions of ocean/ice and atmosphere;
- Polynyas as a hotspot for ecosystem productivity.
All these four are interesting and important areas of research.
Unfortunately, the first three are mixed together in a story that I struggled to follow. In particular, the representation of polynyas in the models is presented as a detection issue; the fact that ”there may be model biases” is only finally acknowledged line 618. Yes, one needs to know how to detect polynyas in models before being able to evaluate them, but that’s why you are doing all these tests on the CDR where you degraded the resolution and/or changed the low values to reproduce known model/observation differences: You have a reference to compare your tests to. I would recommend you set the detection method, using the one most adapted to the models' detection based on the many tests you did on the observations, and only then look at the models, using this on method only. The model comparison of the two temporal resolutions would remain useful though, since different communities use the monthly and daily output.
It is also confusing that for most of the analyses you present only the results of JRA-CESM, when 1. Figure 1 suggests that CESM-LE has a more accurate sea ice and 2. The NPP analysis at the end is done on CESM-LE. That is, the case for using JRA-CESM is not well motivated. As I said above, the comparison between JRA-CESM and CESM-LE would allow you to discuss the effect of forcing vs full-coupling, but you do not discuss this for now. Removing JRA-CESM from this paper could allow you to keep this discussion for a dedicated study. You could even consider using only the subset of ensemble members that are the most accurate for your NPP analysis; according to Figure 8b, bottom panels, some are ok-enough.
Despite how this may sound, I suspect that all the work is there already, and that it is only the text that needs re-arranging for your argument to be convincing. The rewriting would be substantial though, so I do not provide minor comments that could become irrelevant this time.
The figures need adjusting as well to increase readability:
- Figure 1, black and dark blue are hard to distinguish (grey instead of black?);
- Figure 2, the dark blue asterisks are hard to see against the dark blue low ice concentration (have them e.g. orange, and switch red to magenta?);
- Figure 3 onwards, the four different shades of blue are barely distinguishable;
- Figures 4-6, please have bigger fonts in the legend
Citation: https://doi.org/10.5194/egusphere-2024-3490-RC1 -
AC1: 'Reply on RC1', Laura Landrum, 13 Feb 2025
Thank you to the reviewers for careful and considerate reading of our submission and thoughtful feedback that will make this a stronger paper. Thank you also for your general support of the manuscript and overall suitability for publishing.
After reading all three reviews we agree that we can do a better job of clearly communicating the intent and results of this paper. Our four primary intentions and results in this work are:
- An analysis of the sensitivity of polynya detection in gridded sea ice products (both satellite and model output) to sea ice product types, grids, and polynya definition metrics and thresholds. When comparing polynya areas from different data products it is important to use uniform grids, and optimal metric choices may also depend on season and region of interest.
- A comparison of Antarctic coastal polynyas detected in satellite observations and simulated in a climate model forced with atmospheric reanalysis data. A hindcast version of the CESM2 identifies similar integrated Antarctic coastal polynya areas, although the polynyas identified in the model are individually larger and less numerous than those in the satellite product.
- To assess the internal variability of coastal polynya areas in an ensemble of fully coupled climate model simulations for historical and future time periods. Coastal polynya areas identified in climate model output exhibit significant internal variability in historical and future time periods. The same metrics and thresholds tend to identify higher springtime coastal polynya areas in the hindcast simulation than in the fully coupled CESM2, which may be related to the slightly more rapid sea ice retreat in the hindcast simulation (e.g. Figure 1).
- To test if coastal polynyas identify marine productivity “hot spots” and how both polynyas and sea ice zone NPP may change in the future. Coastal polynyas identify “hot spots” in marine productivity, relative to the rest of the sea ice zone, in the CESM. Average NPP within coastal polynyas does not change significantly in future scenarios, although the contribution of NPP within polynyas relative to the total sea ice zone productivity decreases as sea ice zone NPP increases and total polynya areas decrease. This is particularly true in the spring.
If given the opportunity to rewrite this submission based on reviewer feedback, we will restructure the paper based on reviewer comments regarding lack of clarity and highlight and distinguish between these four primary results. A revised abstract reads:
Antarctic polynyas are key components of Antarctic marine ecosystems, influencing light and nutrient availability and open water access for marine predators. Thus, changes in the physical characteristics of polynyas can influence how these ecosystems respond to a changing climate. Here, we explore how to identify polynyas using satellite and Earth System Model data, and we assess the impacts of using different polynya-identification metrics and thresholds (sea ice concentration or thickness). Our results show optimal metrics for coastal polynya definition will depend on both the data resolution, as well as the season and region of interest. Temporal resolution (daily vs monthly) has an impact on polynya identification in specific regions and seasons but has no significant impact on integrated hemispheric coastal polynya area. Our results highlight the importance of identifying polynyas on grids of the same type and resolution when comparing polynyas from different data products. We find that sea ice thickness is more suitable for identifying polynyas in model data in winter months in contrast to spring months when both sea ice thickness and concentration may be suitable metrics. The Community Earth System Model Version 2 (CESM2) identifies similar integrated Antarctic coastal polynya areas and locations as satellite-based sea ice data, although polynyas identified in the CESM2 tend to be larger and fewer in number than those identified in the satellite products particularly in the winter. We then use the CESM2 to investigate ecosystem function within polynyas and find that there is enhanced phytoplankton productivity in modeled polynya features in both hindcast and fully coupled simulations. Springtime polynyas remain an important control on Antarctic productivity under future climate change, although the relative role of polynyas decreases as polynyas diminish and phytoplankton productivity outside of polynyas increases as sea ice cover decreases.
We will also provide information on modifications that are made within the text to address the reviewer concerns.
Responses to specific reviewer comments:
Reviewer 1
Reviewer 1 points out that did not adequately address potential model biases early on in the paper.
We will add specific caveats in the introduction highlighting that many climate models have large biases in the Antarctic sea ice and that our work is focusing on polynya identification in gridded products and an analysis and application of polynya identification on one climate model. We will add broad discussion (w.r.t. climate models in general) and model-specific (e.g. the CESM) biases in sea ice and how this may also impact polynya detection beyond the identification methodologies analyzed in this submission.
Comments from reviewer 1 highlight that our motivation for using the hindcast simulation (JRA-CESM) was not clear.
Our comparisons between coastal polynyas identified in satellite data and model output are presented primarily using the hindcast version of CESM (JRA-CESM) as this provides the closest analogue to the observations. The hindcast (which is an ocean-sea ice only simulation) is forced by atmospheric reanalysis products. Using the hindcast thereby removes differences between satellite observations and the model that would be due to atmospheric forcing or internal variability in the fully coupled CESM simulations. Thus, much of the simulated ice conditions and variability should be more directly comparable to satellite products. We will add discussion regarding interannual variability of simulated coastal polynyas in the fully coupled model to add context for the results from the hindcast simulation.
Climatological sea ice concentrations (SIC) and extents (SIE) are very similar in the CESM-JRA and the CESM2-LE. Mean SIEs from the hindcast simulation fall within the variability of the fully coupled CESM2-Large Ensemble, with two regional exceptions. The first exception is found in the Weddell Sea in the summer (Jan-Feb) when mean SIE in the JRA-CESM lies below the range (and thus further away from the observations) in the CESM-LE. The second exception lies in eastern Antarctica during July-November, when mean JRA-CESM SIE is higher – and thus closer to the satellite-based SIE – than the CESM-LE (Figure 1 and Supplemental Figure 2). We will create figures of mean, monthly polynya areas over the historical time period for both the JRA-CESM and CESM-LE to compare when/if there are significant regional and seasonal differences, and add text to the paper describing any differences. We will include a figure showing this either in the manuscript or in the supplementary material.
Reviewer 1 suggestions for Figures will be addressed as follows:
- We will change colors to better distinguish in Figure 1
- We will change polynya marker colors in Figure 2
- We will change line colors in Figures 3 onwards to better distinguish the results
- We will increase font size for legends in Figures 4-6
Citation: https://doi.org/10.5194/egusphere-2024-3490-AC1
-
RC2: 'Comment on egusphere-2024-3490', Tarkan Bilge, 29 Dec 2024
The study under review presents an analysis of the sensitivity of polynya occurrence and area to polynya definition thresholds in satellite observations and climate model output. The authors subsequently use this analysis to construct methodological recommendations for future studies on polynyas. In my opinion, these two aspects represent the main two contributions of the study; methodological advice for polynya studies, and numerical evidence for the way in which resolution/data-type/season influences polynya statistics. This submission is original, being the first such study to quantitatively demonstrate the influence of resolution and data type (e.g. model/observational) on polynya number and area. The main significance of the study lies in the methodological recommendations and accompanying numerical justifications which can be used by future studies to consistently define polynyas without the requirement for authors to invent their own definitions (e.g. Mohrmann, 2021). In my opinion, the scientific content of the article is clearly articulated and well presented and only requires minor corrections prior to publication.
Below I have included a small number of specific issues which I would like to bring to the authors' attention.
The authors have explained that this study is focused on coastal polynyas, (line 165), but in the study abstract and in the the conclusions (e.g. conclusions 3, 4, 6) refer to "polynyas" more generally. Given that the mechanisms which cause coastal polynyas and open-water polynyas are different, it is not necessarily possible to extrapolate conclusions from the analysis of one to the other. For example, the authors explain that SIT may be a better metric in winter due to air temperatures causing almost immediate surface freezing - I would expect this effect to disproportionately impact coastal polynya identification, because open-water polynyas feature upwelling of warm water which might less readily freeze. Related to this comment, Mohrmann et al. (2021) found that the "CMIP6 CESM2 models show coastal polynyas, but never OWPs", so extending analysis to open-water polynyas using CESM2 might not be possible. Instead, I would recommend the authors to caveat these potential differences, and clearly state in the abstract, conclusions, and main text that results are relevant for *coastal* polynyas. This may sound like a pedantic point, but since this study aims to make recommendations on polynya definition and thresholding to future studies, it is important that future studies on open-water polynyas recognise that not all of the thresholding analysis has been carried out with open-water polynyas in mind, and therefore the conclusions may not be suitable for application in their studies.
In section 4.6, polynya number and area are analysed in the CESM2-LE, and are found to decrease towards the end of the century. I think the study would benefit from some analysis or informed speculation as to the mechanism which drives this decrease in future polynyas under emissions scenarios. The reduction could be a result of changing mean-wind fields, or more simply due to a reduction in overall Antarctic sea-ice. If the authors compare the ~2090 mean Antarctic July and November sea-ice field with the ~2020 fields, they might find that a mean field difference could explain this reduction. This might also explain why the November polynya statistics drop off more readily than the July ones.
On line 312, the authors say that April to October the polynya area is higher for daily mean than monthly mean data, often by more than one standard deviation. Should this mean that the dark blue line and light red shaded areas should be separate in Figures 3a, S2 (Apr-Oct)? Perhaps the lines are too thick or the plots are too small to appreciate this, could the authors check this statement. The authors go on to conclude than monthly mean and daily mean polynya areas are comparable on a hemispheric basis, so this point does not seem critical for the main argument.
On line 456, the authors mention that the SIC threshold modelled polynya areas are closer to the satellite areas, this seems to be only really true for the monthly data. On this note, there seems to be quite a large difference between daily and monthly averaged JRA-CESM SIC thresholds in November. Conclusion 5) states that polynya areas are comparable between monthly/daily thresholds on a hemispheric basis, but as mentioned in Line 596, and as is visible in Figure 4c, there is a seasonality to this, and while it is true on an annual-average basis, the CDR daily to monthly cross-correlation drops to ~0.65 in December. In general, Conclusion 5) is important result, and I think the authors should consider modifying the phrasing to be '...data are comparable on an annual averaged and hemispheric basis', or similar.
Further technical corrections:
Line 162: reference should be Mohrmann et al. (2021). Please check spelling of Mohrmann elsewhere in the text.
Line 243: could authors double-check that '≥' sign is correct. Will depend on journal precedent.
Line 293-297: repeated point about CDR vs NASA Team polynya comparison.
Line 454: unmatched bracket at end of line.
Line 547: Figure 8 caption; missing space before comma, should use 'a)' and 'b)' in caption text, formatting of following needs to be consistent "(red; 85% SIC)", "(85% SIC; black)", "(0.4m SIT, blue)".
Line 630: long space.
Line 715: Figure A2 caption '(REF)' left in.
Line 735: Figure B1 caption capitilisation of 'Sea Ice Concentration' inconsistent with other figures e.g. Figure B2.
Line 760: Figure C1 caption has no '(d)'.Many thanks to the authors for their submission, which I feel contains a valuable contribution to the study of Antarctic sea ice, and I hope that the comments above prove useful in finalising this work.
Citation: https://doi.org/10.5194/egusphere-2024-3490-RC2 -
AC2: 'Reply on RC2', Laura Landrum, 13 Feb 2025
Thank you to the reviewers for careful and considerate reading of our submission and thoughtful feedback that will make this a stronger paper. Thank you also for your general support of the manuscript and overall suitability for publishing.
After reading all three reviews we agree that we can do a better job of clearly communicating the intent and results of this paper. Our four primary intentions and results in this work are:
- An analysis of the sensitivity of polynya detection in gridded sea ice products (both satellite and model output) to sea ice product types, grids, and polynya definition metrics and thresholds. When comparing polynya areas from different data products it is important to use uniform grids, and optimal metric choices may also depend on season and region of interest.
- A comparison of Antarctic coastal polynyas detected in satellite observations and simulated in a climate model forced with atmospheric reanalysis data. A hindcast version of the CESM2 identifies similar integrated Antarctic coastal polynya areas, although the polynyas identified in the model are individually larger and less numerous than those in the satellite product.
- To assess the internal variability of coastal polynya areas in an ensemble of fully coupled climate model simulations for historical and future time periods. Coastal polynya areas identified in climate model output exhibit significant internal variability in historical and future time periods. The same metrics and thresholds tend to identify higher springtime coastal polynya areas in the hindcast simulation than in the fully coupled CESM2, which may be related to the slightly more rapid sea ice retreat in the hindcast simulation (e.g. Figure 1).
- To test if coastal polynyas identify marine productivity “hot spots” and how both polynyas and sea ice zone NPP may change in the future. Coastal polynyas identify “hot spots” in marine productivity, relative to the rest of the sea ice zone, in the CESM. Average NPP within coastal polynyas does not change significantly in future scenarios, although the contribution of NPP within polynyas relative to the total sea ice zone productivity decreases as sea ice zone NPP increases and total polynya areas decrease. This is particularly true in the spring.
If given the opportunity to rewrite this submission based on reviewer feedback, we will restructure the paper based on reviewer comments regarding lack of clarity and highlight and distinguish between these four primary results. A revised abstract reads:
Antarctic polynyas are key components of Antarctic marine ecosystems, influencing light and nutrient availability and open water access for marine predators. Thus, changes in the physical characteristics of polynyas can influence how these ecosystems respond to a changing climate. Here, we explore how to identify polynyas using satellite and Earth System Model data, and we assess the impacts of using different polynya-identification metrics and thresholds (sea ice concentration or thickness). Our results show optimal metrics for coastal polynya definition will depend on both the data resolution, as well as the season and region of interest. Temporal resolution (daily vs monthly) has an impact on polynya identification in specific regions and seasons but has no significant impact on integrated hemispheric coastal polynya area. Our results highlight the importance of identifying polynyas on grids of the same type and resolution when comparing polynyas from different data products. We find that sea ice thickness is more suitable for identifying polynyas in model data in winter months in contrast to spring months when both sea ice thickness and concentration may be suitable metrics. The Community Earth System Model Version 2 (CESM2) identifies similar integrated Antarctic coastal polynya areas and locations as satellite-based sea ice data, although polynyas identified in the CESM2 tend to be larger and fewer in number than those identified in the satellite products particularly in the winter. We then use the CESM2 to investigate ecosystem function within polynyas and find that there is enhanced phytoplankton productivity in modeled polynya features in both hindcast and fully coupled simulations. Springtime polynyas remain an important control on Antarctic productivity under future climate change, although the relative role of polynyas decreases as polynyas diminish and phytoplankton productivity outside of polynyas increases as sea ice cover decreases.
We will also provide information on modifications that are made within the text to address the reviewer concerns.
Reviewer 2
Reviewer 2 points out that although the anlayis in this study is focused on coastal polynyas and yet this is not clearly stated particularly in the abstract and the discussion.
This point is well taken – thank you for bringing this to our attention. We will rewrite the abstract and the discussion to make this clear. In addition, we will add text specifically aimed at clarification and caveats with respect to open water polynyas, which result from mechanisms different than coastal polynyas. We will explicitly state that because of this, our conclusions may not apply to open water polynyas.
We do not delve into analysis and description of why polynyas decline at the end of the 21st Century in the CESM-LE and agree that some analysis and discussion would benefit the paper. DuVivier et al, 2023, found that polynyas in the Ross Sea during winter decrease and sea ice concentrations and thicknesses increase in the 21st century in the CESM2-LE, likely due to both large scale circulation changes and local feedbacks. We will investigate regional changes in coastal polynyas for a broader understanding of some of the causes of the projected decreases in both winter and spring polynya areas, although an in-depth analysis of the mechanisms is beyond the purview of this paper.
Reviewer 2 found a mistake on line 312 in the submission – thank you!
We replace:
“In general, polynyas identified using daily data lead to larger total polynya area April through October across the SIC thresholds than those based on monthly data, and the mean integrated polynya areas based on the daily data often lie more than one standard deviation higher than those based on the monthly mean data. As spring turns to summer, the sensitivity of polynya area to threshold value becomes larger than any differences in the means calculated from different grid and temporal resolutions (Fig. 3b and Supplemental Fig. 2).”
“When moving beyond an example day and looking across all years, there are significant differences in integrated total SH area of coastal polynyas in the satellite-based data products as identified by SIC threshold value, grid resolution, temporal resolution (daily vs monthly) and season (Fig. 3 a-b and Supplemental Fig. 2). The total area of coastal polynyas is larger for a given threshold in the CDR on the original higher-resolution EASE grid than when regridded to the 1° CESM grid, and these differences are most pronounced from late fall through early spring (May-Oct; Supplemental Fig. 2). In general, polynyas identified using daily data lead to larger total polynya area April through October across the SIC thresholds than those based on monthly data, and the mean integrated polynya areas based on the daily data often lie more than one standard deviation higher than those based on the monthly mean data. As spring turns to summer, the sensitivity of polynya area to threshold value becomes larger than any differences in the means calculated from different grid and temporal resolutions (Fig. 3b and Supplemental Fig. 2). Throughout most of the year, differences due to temporal resolution are much smaller at a given SIC threshold than the differences due to grid-size. That is, more grid cells are identified as polynyas, and the total SH polynya area is larger on the original, finer-grid EASE grid than on the regridded data for a given threshold value.”
With:
“When moving beyond an example day and looking across all years, the primary differences in integrated total SH area of coastal polynyas in the satellite-based data products as identified by SIC threshold values arise from grid resolution during the late fall through early spring (May-October; Fig. 3 a-b and Supplemental Fig. 2). Integrated southern hemisphere (SH) coastal polynya areas are larger for a given threshold in the CDR on the original higher-resolution EASE grid than when regridded to the 1° CESM grid. Differences in integrated SH coastal polynya areas identified using daily data lie within one standard deviation of polynyas identified in the monthly data. As spring turns to summer, the sensitivity of polynya area to threshold value becomes larger than any differences in the means calculated from different grid and temporal resolutions (Fig. 3b and Supplemental Fig. 2). Throughout most of the year, differences due to temporal resolution are much smaller at a given SIC threshold than the differences due to grid-size. That is, more grid cells are identified as polynyas, and the total SH polynya area is larger on the original, finer-grid EASE grid than on the regridded data for a given threshold value.”
Reviewer 2 finds that our summary on regional polynya areas (Section 4.4.2) and our conclusion regarding monthly vs daily data for polynya identification needs clarification.
We will replace lines 459-461:
“In general, the spatial correlations between polynya area as a function of longitude identified in the satellite product and the JRA-CESM polynya area (Fig. 4c) are higher than they are for the SH timeseries (Fig. 460 5c).”
With:
“In general, the spatial correlations between polynya area as a function of longitude identified in the satellite product and the JRA-CESM polynya area (Fig. 4c) are higher than the temporal correlations for the integrated SH timeseries (Fig. 5c)”
In addition, in the primary recommendations (in our conclusions) we will replace:
“polynya areas identified from monthly and daily data are comparable on hemispheric basis”
with:
“polynya areas identified from monthly and daily data are comparable on an annually averaged, integrated hemispheric basis”
If given the opportunity to revise our submission, we will check and change accordingly the 9 specific technical corrections found by reviewer 2.
References
DuVivier, A.K., Molina, M.J., Deppenmeier, AL. et al. Projections of winter polynyas and their biophysical impacts in the Ross Sea Antarctica. Clim Dyn 62, 989–1012 (2024). https://doi.org/10.1007/s00382-023-06951-z
Ivanova, N., Pedersen, L. T., Tonboe, R. T., Kern, S., Heygster, G., Lavergne, T., Sorensen, A., Saldo, R., Dybjkaer, G., Brucker, L. and Shokr, M.: Inter-comparison and evaluation of sea ice algorithms: towards further identification of challenges and optimal approach using passive microwave observations, The Cryo., 9, 1797-1817, https://doi.org/10.5194/tc-9-1797-2015, 2015.
Lavergne, T., Sørensen, A. M., Kern, S., Tonboe, R., Notz, D., Aaboe, S., Bell, L., Dybkjær, G., Eastwood, S., Gabarro, C., Heygster, G., Killie, M. A., Brandt Kreiner, M., Lavelle, J., Saldo, R., Sandven, S., and Pedersen, L. T.: Version 2 of the EUMETSAT OSI SAF and ESA CCI sea-ice concentration climate data records, The Cryo., 13, 49–78, https://doi.org/10.5194/tc-13-49-2019, 2019.
Citation: https://doi.org/10.5194/egusphere-2024-3490-AC2
-
AC2: 'Reply on RC2', Laura Landrum, 13 Feb 2025
-
RC3: 'Comment on egusphere-2024-3490', Anonymous Referee #3, 17 Jan 2025
Review of “Defining Antarctic polynyas in satellite observations and climate model output to support ecological climate change research” by Laura Landrum et al.
The MS is comparing the polynya size and location from satellite SMMR, SSM/I – SSMIS microwave radiometer sea ice concentrations (SIC) and climate model simulations and then use the model simulations to assess the polynya ecosystems and long-term trends over the period where satellite data are available (since 1979). The topic is very welcome and the MS is well written and structured. My primary concern is the relatively poor match between polynyas detected from the satellite observations and the model simulations. This does not give high confidence in the results. Part of that mismatch is due to real differences between observed and model simulated polynyas and part of that is due to the definition of polynyas in the two datatypes, spatial resolution etc. as discussed in the MS. The satellite microwave radiometer SIC data are relatively consistent over the study period (because the SIC algorithm is using the 19 and 37 GHz channels) and it covers from 1979 until today. However, as also mentioned in some of the references (Markus and Burns), this channel combination is not ideal for mapping polynyas. For example, due to coarse resolution and sensitivity to new and mature ice types and also the way the data are processed excludes low SIC from being detected (weather filters). The NASA Team SIC is included for comparison, but it is using the same 19 and 37 GHz channels as the “CDR” and it has approximately the same spatial resolution and new and mature ice sensitivity. It also uses weather filters which excludes the detection of low SIC. The microwave radiometer swath data are resampled to daily 25 km x 25 km grids while the spatial resolution of the SIC is approximately 45 km. Other satellite radiometer datasets for detection of polynyas mentioned in the MS (Arrigo & van Dijken using the Markus & Burns algorithm and detections based on OSISAF SIC in Mohrman et al.) are not used for comparison which contributes to limited confidence in the detected location and size of the polynyas.
I would suggest to compare the detection of polynyas in the CDR dataset with other (already published) algorithms for polynya detection. Hopefully this will lead to a better match with the model simulations. I think that this is needed to give confidence in the model simulations.
Specific comments
L51 and 53: I don’t think that “Passive microwave images” is a descriptive name for the CDR sea ice concentrations. Please find a better term for it and change it throughout the MS.
L53: “Passive microwave images tend to underestimate SIC’s… “ This is not accurate as a general statement about “passive microwave” data. The sensitivity of the microwave radiometer data to sea ice concentration and the inherent noise depends on the selection of channels and the algorithm used for processing the SIC as described in Ivanova et al. I would suggest to rewrite this entire paragraph L47-L87 with reference to the specific data which is used in this study (CDR). The level of detail about the CDR data is not adequate.
L200: Polynyas are formed in a dynamic environment and it is unlikely that nilas grows to a thickness of 5 cm on scales of the satellite radiometer footprint or resolution of the model. The type of new ice formed in a dynamic environment is pancake ice and pancake ice does not reach 100% SIC before consolidation is reached at a thickness higher than 5 cm. In addition, different SIC algorithms have different sensitivity to new and mature ice. Please specify which algorithm that you are referring to.
L258: 10-14% is that approximately 2 million km2? Please provide the difference in percent and in square kilometers.
L425: Are the thin ice formation processes in the model realistic considering the dynamic environment where polynyas appear? Please, include a discussion of ice formation processes in polynyas.
L691: Please specify the source and date of the “satellite image” in figure AA2 (MODIS?). L705 (fig. AA2) the title is truncated and there is a reference missing in the fig-text.
L735: in the figure title, I think that it should be “25 x 25 km2”
L740: If polynya detection depends on grid resolution a lat-lon grid is not optimal. It never is in polar regions.
Citation: https://doi.org/10.5194/egusphere-2024-3490-RC3 -
AC3: 'Reply on RC3', Laura Landrum, 13 Feb 2025
Thank you to the reviewers for careful and considerate reading of our submission and thoughtful feedback that will make this a stronger paper. Thank you also for your general support of the manuscript and overall suitability for publishing.
After reading all three reviews we agree that we can do a better job of clearly communicating the intent and results of this paper. Our four primary intentions and results in this work are:
- An analysis of the sensitivity of polynya detection in gridded sea ice products (both satellite and model output) to sea ice product types, grids, and polynya definition metrics and thresholds. When comparing polynya areas from different data products it is important to use uniform grids, and optimal metric choices may also depend on season and region of interest.
- A comparison of Antarctic coastal polynyas detected in satellite observations and simulated in a climate model forced with atmospheric reanalysis data. A hindcast version of the CESM2 identifies similar integrated Antarctic coastal polynya areas, although the polynyas identified in the model are individually larger and less numerous than those in the satellite product.
- To assess the internal variability of coastal polynya areas in an ensemble of fully coupled climate model simulations for historical and future time periods. Coastal polynya areas identified in climate model output exhibit significant internal variability in historical and future time periods. The same metrics and thresholds tend to identify higher springtime coastal polynya areas in the hindcast simulation than in the fully coupled CESM2, which may be related to the slightly more rapid sea ice retreat in the hindcast simulation (e.g. Figure 1).
- To test if coastal polynyas identify marine productivity “hot spots” and how both polynyas and sea ice zone NPP may change in the future. Coastal polynyas identify “hot spots” in marine productivity, relative to the rest of the sea ice zone, in the CESM. Average NPP within coastal polynyas does not change significantly in future scenarios, although the contribution of NPP within polynyas relative to the total sea ice zone productivity decreases as sea ice zone NPP increases and total polynya areas decrease. This is particularly true in the spring.
If given the opportunity to rewrite this submission based on reviewer feedback, we will restructure the paper based on reviewer comments regarding lack of clarity and highlight and distinguish between these four primary results. A revised abstract reads:
Antarctic polynyas are key components of Antarctic marine ecosystems, influencing light and nutrient availability and open water access for marine predators. Thus, changes in the physical characteristics of polynyas can influence how these ecosystems respond to a changing climate. Here, we explore how to identify polynyas using satellite and Earth System Model data, and we assess the impacts of using different polynya-identification metrics and thresholds (sea ice concentration or thickness). Our results show optimal metrics for coastal polynya definition will depend on both the data resolution, as well as the season and region of interest. Temporal resolution (daily vs monthly) has an impact on polynya identification in specific regions and seasons but has no significant impact on integrated hemispheric coastal polynya area. Our results highlight the importance of identifying polynyas on grids of the same type and resolution when comparing polynyas from different data products. We find that sea ice thickness is more suitable for identifying polynyas in model data in winter months in contrast to spring months when both sea ice thickness and concentration may be suitable metrics. The Community Earth System Model Version 2 (CESM2) identifies similar integrated Antarctic coastal polynya areas and locations as satellite-based sea ice data, although polynyas identified in the CESM2 tend to be larger and fewer in number than those identified in the satellite products particularly in the winter. We then use the CESM2 to investigate ecosystem function within polynyas and find that there is enhanced phytoplankton productivity in modeled polynya features in both hindcast and fully coupled simulations. Springtime polynyas remain an important control on Antarctic productivity under future climate change, although the relative role of polynyas decreases as polynyas diminish and phytoplankton productivity outside of polynyas increases as sea ice cover decreases.
We will also provide information on modifications that are made within the text to address the reviewer concerns.
Reviewer 3
Reviewer 3 suggests that the paper will be much stronger by including other satellite-based data that have used algorithms different that that used for the sea ice concentrations of the NSIDC-CDR.
This is an excellent recommendation, and we appreciate the thoughtfulness and insight of this recommendation from a clear satellite data expert. Upon a closer read of Ivanova et al. (2015), we believe it would be informative to include results using the OSIASF SIC (Lavergne et al., 2019) in our analysis. The algorithms used by the NASA Team product and the OSIASF product result in particularly large differences in SICs in thin and low sea ice areas and thus represents an important comparative tool for this study. We anticipate that this additional analysis with a new product will be a major modification to this work.
Additionally, we also noticed a mistake in the labels in Figure 8. We will correct – orange lines are for JRA-CESM using SIT 0.4m threshold and navy lines are for JRA-CESM SIC 85% threshold.
References
DuVivier, A.K., Molina, M.J., Deppenmeier, AL. et al. Projections of winter polynyas and their biophysical impacts in the Ross Sea Antarctica. Clim Dyn 62, 989–1012 (2024). https://doi.org/10.1007/s00382-023-06951-z
Ivanova, N., Pedersen, L. T., Tonboe, R. T., Kern, S., Heygster, G., Lavergne, T., Sorensen, A., Saldo, R., Dybjkaer, G., Brucker, L. and Shokr, M.: Inter-comparison and evaluation of sea ice algorithms: towards further identification of challenges and optimal approach using passive microwave observations, The Cryo., 9, 1797-1817, https://doi.org/10.5194/tc-9-1797-2015, 2015.
Lavergne, T., Sørensen, A. M., Kern, S., Tonboe, R., Notz, D., Aaboe, S., Bell, L., Dybkjær, G., Eastwood, S., Gabarro, C., Heygster, G., Killie, M. A., Brandt Kreiner, M., Lavelle, J., Saldo, R., Sandven, S., and Pedersen, L. T.: Version 2 of the EUMETSAT OSI SAF and ESA CCI sea-ice concentration climate data records, The Cryo., 13, 49–78, https://doi.org/10.5194/tc-13-49-2019, 2019.
Citation: https://doi.org/10.5194/egusphere-2024-3490-AC3
-
AC3: 'Reply on RC3', Laura Landrum, 13 Feb 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,149 | 297 | 40 | 1,486 | 121 | 48 | 54 |
- HTML: 1,149
- PDF: 297
- XML: 40
- Total: 1,486
- Supplement: 121
- BibTeX: 48
- EndNote: 54
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Alice K. DuVivier
Marika M. Holland
Kristen Krumhardt
Zephyr Sylvester
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(6046 KB) - Metadata XML
-
Supplement
(2420 KB) - BibTeX
- EndNote
- Final revised paper
I see four main results in this paper:
All these four are interesting and important areas of research.
Unfortunately, the first three are mixed together in a story that I struggled to follow. In particular, the representation of polynyas in the models is presented as a detection issue; the fact that ”there may be model biases” is only finally acknowledged line 618. Yes, one needs to know how to detect polynyas in models before being able to evaluate them, but that’s why you are doing all these tests on the CDR where you degraded the resolution and/or changed the low values to reproduce known model/observation differences: You have a reference to compare your tests to. I would recommend you set the detection method, using the one most adapted to the models' detection based on the many tests you did on the observations, and only then look at the models, using this on method only. The model comparison of the two temporal resolutions would remain useful though, since different communities use the monthly and daily output.
It is also confusing that for most of the analyses you present only the results of JRA-CESM, when 1. Figure 1 suggests that CESM-LE has a more accurate sea ice and 2. The NPP analysis at the end is done on CESM-LE. That is, the case for using JRA-CESM is not well motivated. As I said above, the comparison between JRA-CESM and CESM-LE would allow you to discuss the effect of forcing vs full-coupling, but you do not discuss this for now. Removing JRA-CESM from this paper could allow you to keep this discussion for a dedicated study. You could even consider using only the subset of ensemble members that are the most accurate for your NPP analysis; according to Figure 8b, bottom panels, some are ok-enough.
Despite how this may sound, I suspect that all the work is there already, and that it is only the text that needs re-arranging for your argument to be convincing. The rewriting would be substantial though, so I do not provide minor comments that could become irrelevant this time.
The figures need adjusting as well to increase readability: