“I reviewed a previous version of this manuscript that was rejected by TC last year. The resubmitted paper has probably been revised based on the previous comments from the reviewers. However, unfortunately, the changes were minor. They are changes in wording and the deletion/increase of a few figures. At least for me, I felt that those were superficial modifications, and accordingly, I couldn’t satisfy with them. Therefore, I recommend rejecting this paper mainly for the following reasons.

The points at issue:

Too qualitative analysis; lack of error analysis and science.

The unclear main focus of this study.

Misinterpretation of inappropriate satellite data.”

Thank you for taking the time to review our paper, we respond to specific points below.

“This manuscript lacks quantitative analysis and also has no error analysis. This leads to a lack of science. Especially in SAR data analysis, this study only shows satellite images (without backscatter scale) and wind condition maps, and then they make stories. Fig. 8 of the previous manuscript was the only statistical analysis. However, the correlation was weak, and this figure has been deleted.”

We disagree with the suggestion that qualitative analysis and visual interpretation of imagery constitutes a lack of science. Visual interpretation by experts and qualitative analysis are important tools in remote sensing where they enable findings often not possible with existing quantitative methods (i.e. the dynamics and polynya events discussed in this paper).  Qualitative analysis and visual interpretation of SAR imagery have been used in previous polynya studies (e.g. Hollands and Dierking, 2016; Dai et al., 2020). Inclusion of a backscatter scale is not standard in the literature for such images and including absolute values would not add to the qualitative analysis. Furthermore, the polynya area and ice production sections are indeed quantitative, and the polynya areas identified by the SIC data are assessed against an image of the ‘open’ polynya in SAR (Fig 2).

“It can be read that the main focus of this study seems to be on the biological production and chemical processes in the summer coastal polynya (complete ice-free ocean) and the accompanying carbon dioxide absorption. However, the biochemical analyses were not conducted in this study.”

The main focus of the study is not biological production or chemical processes. We describe the aims of the study in the last paragraph of the introduction (lines 95-101). Biological/chemical processes are mentioned only as a motivation for better understanding polynyas.

“On the other hand, this manuscript estimates sea-ice production in a coastal polynya in winter (A small part may be an open water fraction, but it is mostly covered by frazil ice or thin solid ice). The purpose of estimating production is not clear; what is stated in L. 65 is insufficient. The dense water formed in winter coastal polynyas associated with the prominent ice production is an important source of AABW. This process of bottom water formation is thought to significantly impact the climate system through the transport of heat and substances such as carbon dioxide between the atmosphere and the deep ocean. However, this is not described at all in the manuscript.”

Mention of the role of ice production in polynyas in AABW formation was removed from this version of the manuscript, upon suggestion of another reviewer, due this region not being an important contributor to AABW formation (e.g. Gordon, 2009). We could re-include mention of the role of ice production in AABW in general, while being clear it is not important in this specific case.  Nevertheless, we emphasize that the goal of this paper is not to address bottom water formation and rather the study of the ice formation in polynyas.  The importance of polynya ice production is discussed on lines 56-66, but in a revision we will also include mention of the importance of quantifying ice production for understanding the overall sea ice mass balance in the region.

“The use of SAR data is a challenging point in this study. However, as is clear from video S1, it has many temporal and spatial discontinuities. This suggests that it is not suitable for monitoring a coastal polynya whose variability is large.”

We acknowledge that there are spatial and temporal discontinuities and this is a limitation (e.g. leads to gaps in qualitative analysis and is one of the barriers to an attempt at quantitative analysis) but disagree that this makes it not suitable for qualitative analysis. Processes such as polynya events and back-flow are observable despite these gaps, and indeed are only observable in winter with SAR.

“The authors defined surface conditions in SAR images as follows.

Open ocean: a low backscatter and appears dark

Older icepack: relatively high backscatter and appears bright and more granular

Recently-formed polynya produced ice: an intermediate backscatter

Frazil ice: distinct bands of varying brightness

However, this is very qualitative as it is affected by the SAR’s incident angle. For example, in video S1, the open ocean may also appear white (e.g., 21-23 November 2016; 15-17 December 2016; 8-10 January 2017). These examples indicate the difficulty of conducting “quantitative” discussions of sea-ice and open water areas from SAR images.”

We agree and acknowledge that our analysis is qualitative here.  We disagree, however, that this discounts the merit of our study. The differences between these key features is generally clear in the context of the images. It was an oversight for us not to mention that open ocean may also have high backscatter during windy conditions – as it the case in the example dates given here. We will mention this in a revision. However, it is clear from the context of the images that these bright areas are open ocean and not sea ice, and therefore it did not lead to mis-interpretations.

“The use of such sparse and unquantifiable data leads to misinterpretation. The authors state that “approximately all of the ice produced between 30 April and 4 November by the main polynya is contained within the red outline on 4 November in Fig. 4” from the SAR images in Video S1 and Fig. 4 (L. 403-405). This is a lack of science to tell this from SAR data alone. It is more natural to assume that the sea ice will grow both thermodynamically and dynamically during this long period of 6-months, resulting in a backscatter similar to that of one-year ice. In any case, this cannot be suggested solely from the SAR images.”

We do not agree this is a mis-interpretation. This was determined by carefully qualitatively analyzing the video and imagery. It is clear from the image presented in Fig. 4 that the highlighted area of ice has a distinct backscatter from the pack ice in the region, and it is clear from carefully analyzing the video and imagery that this is ice produced by the polynya during that season. We agree that the ice would continue to grow dynamically and thermodynamically during this period, but that does not contradict that it was initially formed in the polynya.

This result is not central to the study and can be removed if the editor agrees that it is problematic.

“The definition of a winter coastal polynya area based on AMSR2 sea-ice concentration (SIC) is questionable. Firstly, this study ignores heat loss and sea-ice production in thin ice areas, the dominant type of sea-ice in winter coastal polynyas. Secondly, an area with SIC <70% was defined as a polynya area, but SIC by the ASI algorithm underestimates SIC in thin ice areas. A similar analysis had to be performed using SIC with other AMSR2 algorithms, and error analysis also had to be performed.

In a previous review, I pointed out the definition of a coastal polynya area. As in the previous manuscript, this manuscript also treats the area where the SIC by AMSR2 is <70% as the polynya (open water) area. The previous manuscript was based on studies of the comparison between a coastal polynya area from the polynya signature simulation method (PSSM) by Markus and Burns (1995) and SIC (Parmiggiani, 2006; Morelli & Parmiggiani, 2013; Preußer et al., 2015). The new manuscript added a comparison with SAR images (Figs. 2b-g). The black low backscatter area (open ocean) in Fig. 2b corresponds to the low SIC area (SIC <70%) in Fig. 2e. Since this is not a SIC map, I do not know the details, but the correspondences between the open ocean and the low SIC (not open ocean) areas seem strange. On the other hand, Figs 2c-d shows a bright band-like feature, which is considered to be covered with frazil ice. Since no dark areas can be seen, it is assumed that this area is mostly covered with sea ice. In other words, the SIC must be close to 100%. However, the AMSR2 map shows a low SIC of <70%. The two areas certainly coincide, but they are not consistent in terms of SIC”

On lines 170-173, during winter, we state “Frazil ice, that may form when a polynya opens up and the open ocean begins to freeze, forms in distinct bands of varying brightness (Fig. 2c-d). Note that what we refer to as ‘open’ polynya area during the winter will typically be filled with thin, newly-forming frazil ice.”

On lines 187-191 we state “Following other studies (e.g. Dai et al., 2020) during the winter we also use the term ‘open polynya’ for areas that we include in the polynya, where an opening has been created and new ice production is taking place. However, during the winter we expect thin ice to immediately begin forming when an opening is created, and thus we note the area is not truly ‘open’ ocean.”

Therefore, Fig 2 shows an expanding ‘open’ polynya during a polynya event and the corresponding identified ‘open’ polynya using 70% SIC as a threshold. The figure shows good agreement (as does comparing Video S1 and S2) for other such events. This definition of ‘open’ polynya during winter (i.e. when a polynya event takes place, opening the ocean, where thin/frazil ice ~immediately starts forming due to cold temperatures) is consistent with that in the literature (e.g. WMO, 1970; Nihashi and Ohshima, 2015; Cheng et al., 2017; 2019; Dai et al., 2020; Nakata et al., 2021). In fact Nakata et al, 2021 recently mapped major polynyas around Antarctica during winter specifically by identifying frazil ice areas (We will add reference to this paper in a revision). And the WMO (1970), quoted within Nihashi and Ohshima (2015), states “Polynyas are defined as ‘‘any non-linear shaped opening enclosed within sea ice, and may contain brash ice and/or may be covered with new ice, nilas, or young ice’’ (WMO 1970).” Dai et al., 2020 measure polynya area by manually mapping open/frazil-covered areas in Sentinel-1 akin to that shown in Fig. 2. In summary: we already state that ‘open’ polynyas include newly-forming ice, this definition is consistent with previous studies, and comparison of SAR and our SIC data show that the SIC data (with 70% threshold) effectively identifies ‘open’ winter polynya.

If it would be clearer to use the term ‘active’ polynya rather than ‘open’, we can make that change. We could also draw on a boundary of the interpreted ‘open/active’ polynya on the SAR images in Fig 2 if it is not clear enough from the image alone. We will also review the text in general for clarity regarding the definition of an open/active polynya during winter.

We suggest that comparing the SIC 70% threshold results to an actual SAR image from the same day is a better form of validation than comparing to another SIC product.

“Furthermore, even though the area is considered to be covered by ~100% frazil (thin) ice, the authors did not consider the presence of sea ice at all in their estimate of sea-ice production and assumed an ice-free (open) ocean (eq. 1-10).”

The area will be open ocean at the initial point of ice production when the old ice moves away as part of the polynya event, exposing the ocean, where the new ice forms. We acknowledge this is an imperfect modelled estimate.

“There is an area of intermediate backscatter below (west of) the areas shown as low backscatter area (open ocean, Fig 2b) and the area shown as the band-like feature (frazil ice, Figs. 2c-d). These areas are considered to be “the recently-formed polynya produced ice” as described by the authors. This area is considered to be covered by thin sea ice with ~100% SIC. The sea-ice production in this area is expected to be significant. Sea ice acts as a heat insulator between the atmosphere and the ocean, but its effect decreases rapidly as the ice thickness decreases. The authors had ignored the sea-ice production in thin ice areas.”

We agree that ice growth will take place in these areas but the study focuses on ice production by the ‘open’ polynya i.e. during polynya events. We will review the text for clarity on this.

“SIC estimated from brightness temperatures observed by satellite passive microwave radiometer is underestimated in areas covered with new and thin ice areas, such as coastal polynya (Cavalieri et al., 1994). The authors stated an error in the ASI SIC algorithm but ignored the effects of thin ice (L. 204). As described in the previous review comment, a comparison of SIC using the ASI algorithm in the Ross Ice Shelf polynya and the Mertz Glacier polynya in Antarctica with the PSSM polynya map clearly shows that coastal polynyas are covered by thin ice, not open water, in winter (Kern et al. 2007).”

We acknowledge on lines 170-173 and 187-191 that newly-forming thin ice exists in an ‘open polynya’ (i.e. in a polynya event) in winter. We show in Fig. 2 that the ASI SIC data with a 70% threshold effectively captures these ‘open’ areas, and this approach was also followed by Cheng et al., 2017; 2019) in the Ross Sea. As mentioned above, although ‘open polynya’ has been used elsewhere in the literature we could change this term to ‘active polynya’ if it would be clearer.

“Moreover, the SIC is underestimated in these regions. The authors should show an error analysis for this. In addition to ASI, NT2 and Bootstrap are other major algorithms for estimating SIC. The analysis had to be done using these SICs as well. Their spatial resolution (about 12 km) is coarser than that by ASI, but it is sufficient to resolve the Amundsen polynya.”

As mentioned above, we suggest that comparing the SIC result to actual image of the polynya on the same day is a better form of checking and validating the method than comparing to another, low-resolution SIC product.

Regarding the point in the opening paragraph about the study’s focus being unclear, we will review the introduction for clarity. Presently we do attempt to explicitly lay out the focus on lines 90-101, including specific objectives on lines 95-101.

Best regards,

Dr Grant Macdonald, on behalf of all authors.

Additional References

Gordon, A. L. Bottom water formation. Ocean Currents, 263, 269, 2009.

Nakata, K., Ohshima, K. I., & Nihashi, S., Mapping of active frazil for Antarctic coastal polynyas, with an estimation of sea-ice production. Geophysical Research Letters, 48, e2020GL091353, 2021.

WMO: WMO sea-ice nomenclature, terminology, codes and illustrated glossary. WMO/OMM/BMO Rep. 259, Tech. Paper 145, World Meteorological Organization, 147 pp, 1970.

We add another reference that supports our definition of ‘polynya area’ in winter, and the ability to visually interpret this in SAR imagery. Aulicino et al. (2018) manually measure polynya area (‘extent’) in Terra Nova Bay (Antarctica) in ENVISAT ASAR images, based on visual interpretation, as a form of validation for an approach they develop using MODIS ice surface temperature. They do this during winter, when as in our winter examples, the polynya includes thin ice formation (e.g. frazil ice streaks).

“In the SAR imagery polynyas are clearly visible and well defined thanks to an exceptional spatial resolution (up to 5 m). This resolution decreases to 150 m for the ASAR WS images reported in this study. Nevertheless, they allow us to define the shape of the TNBP  [Terra Nova Bay Polynya] and to estimate its area with a remarkable precision as for ASAR images reported in Figures 6c and 7c.”

As mentioned in the above quote from the paper, examples of the polynya in winter (August) identified in SAR can be seen in Figures 6c and 7c, with similar characteristics to those observed in our study in the Amundsen Sea using Sentinel-1 SAR (40 m resolution).

We will add this paper as a reference in a revised version.

Aulicino, G., Sansiviero, M., Paul, S., Cesarano, C., Fusco, G., Wadhams, P., and Budillon, G.: A new approach for monitoring the Terra Nova Bay polynya through MODIS Ice Surface Temperature Imagery and its validation during 2010 and 2011 winter seasons, Remote Sens., 10, 366, https://doi.org/10.3390/rs10030366, 2018.

Thank you again for taking the time to review our paper, and for the suggestions that have improved its quality and value. We include responses below.

GENERAL COMMENTS

Since The Cryosphere is meant to be an "open" journal, I'll point out that this manuscript is a resubmission (previously submitted to the same journal), that I've reviewed the manuscript before, and that my earlier review is available at:



https://doi.org/10.5194/tc-2021-250-RC2

I would also like to point out that remote sensing isn't my expertise. Therefore, I cannot comment on aspects related to e.g. SAR imagery or AMSR2 algorithms. I rely on the other reviewers (and on the editors) to evaluate these specific aspects.

My "general comments" from the earlier review still hold, so I'm repeating them here:



"I think the manuscript's topic fits nicely within the scope of the journal. As far as I know, this manuscript is unique (i.e. original) in its detailed description of Amundsen Polynya, providing far more information than earlier publications. The study is also scientifically significant given the regional importance of the Amundsen Polynya, either in terms of sea ice production (4th in Antarctica) or for its role in phytoplankton production and atmospheric CO2 uptake."

RECOMMENDATION: The authors have addressed all but one of my earlier comments, so unless my co-reviewers identify major flaws that I've missed, I think the manuscript will be acceptable for publication after minor revisions (see "Specific Comments", below).

SPECIFIC COMMENTS: MINOR

(1) About Section 4.4, "Wind and polynya area": I think the existing analysis of Sec.4.4 is very basic, and that a lot more could be done to establish the relation between the winds and the polynya area. In my earlier review, I was making a number of suggestions: (a) Rather than limiting your analysis to wind velocity, try to incorporate information about wind direction; (b) see if the *time-derivative* of the polynya area is better correlated to winds than the area itself, (c) see if the *cumulated* wind correlates better with the polynya area than the wind itself (this metric works well in the context of coastal upwelling; see https://doi.org/10.1029/2006GL027149 for an application.)

Such additional analyses are not necessary for the paper to be worth publishing, but I do believe the authors are in a unique position to discover something interesting about winds and the area of the Amundsen Polynya, and it would worth giving it one more try.

Thank you for the suggestions – we agree this is worth revisiting. We will start with investigating the relationship between the time-derivative of the polynya area and winds, and will also consider suggestions (a) and (c).

(2) Line 258: The sentence on this line needs to be flipped for consistency with the revised Eqs.2,3. It should be:

"The temperature of the water surface (T_S, in K), was assumed to be at the freezing point of seawater (T_0, in K) which was calculated following..."

Thank you for catching this, we will revise as suggested.

(3) Another reviewer commented that the manuscript was "too qualitative". I do agree that the present manuscript is, to a large extent, descriptive. On the other hand, I believe that scientific studies should first and foremost be *relevant* to the community, regardless of where they fall on the descriptive-to-quantitative scale. The present manuscript is definitely relevant to my own research projects, and I would cite it in a heartbeat if it were to be accepted by the journal.

We agree and acknowledge that a large portion of the paper is descriptive, and agree that it is nevertheless relevant and informative for the community. We also suggest that a large part of the descriptive portion cannot be well-captured by quantitative analysis.

Best regards,

Dr Grant Macdonald, on behalf of all authors.

Thank you for taking the time to review our paper. We include responses to each point below.

Evolution of the dynamics, area and ice production of the Amundsen Sea Polynya, Antarctica, 2016-2021 by Macdonald and others.

This study looks at the dynamics of the Amundsen Sea Polynya (ASP) using a combination of active and passive microwave observations. While the results present potentially useful information about the dynamics of the ASP, I agree with Reviewer #1 and Reviewer #3 (first draft) that additional scientific quality assurances need to be taken.  The major points raised by those Reviewers would only serve to improve the quality of the manuscript so I was surprised they are rebutted and not implemented. The major problem with only using video is the statements about processes/characteristics cannot be supported clearly with evidence and therefore subject to miss-interpretation.  This paper is full of casual statements that maybe true but lack quantitative support which is not scientific. The threshold approach for polynya area is also problematic.

Regarding the qualitative analysis of the video: of course with observations, whether of optical imagery, SAR or from the field, there is the chance of mis-interpretation. However, this does not mean expert interpretation of observations is not valuable without quantification. Visual interpretation of various forms of imagery and field observations has long had a place in physical and geographic sciences. Furthermore, it is not without evidence, the imagery is the data and evidence. They are not ‘casual statements’ but conclusions from hours of analysis of the imagery, and examples are provided in the main paper of processes (e.g. polynya events, back-flow, secondary polynya formation), as well as in the time-lapse video. It is not clear that the key claims of the video analysis (such as that polynya events occur throughout winter, that ice ‘backflows’, secondary polynyas occur, the role of topography in their location) require some form of quantitative result to be supported when they are observable. Visual interpretation of SAR imagery has formed an important part of numerous sea ice studies, some of which are referenced within.

Quantifying these observations would be a huge task, if indeed possible – and in some cases it is not clear exactly what about the process would be quantified in in order to provide further evidence for them, within the scope of this study. Automated quantification of polynya states (e.g. area, frazil ice cover) using SAR is an active area of challenging research that groups are working on, typically involving machine learning techniques. To disregard what can be seen in the imagery because it cannot currently be quantified would mean missing the opportunity to take advantage of an excellent observational tool.  We are clear that the SAR analysis is qualitative and descriptive, but we believe, as stated by Reviewer 2, that importantly this contributes relevant information to the community and therefore merits publication.

(Note Reviewer 1 of this draft is the same reviewer as one of the reviewers on previous iteration)

I would encourage the author’s to actually implement the suggestions of the previous Reviewers also taking into consideration my comments:

A very large part of the critiques (and associated suggestions) were based on a different understanding of what we mean by ‘open polynya’ in the winter. We have attempted to clarify that in the response and pointed out that our definition of an ‘open’ polynya in winter is already established in the literature, but pledged to make it clearer in the paper by instead referring to ‘active’ polynya. The other primary criticism was of the qualitative nature/visual interpretation element of the analysis. We have outlined that we dispute the idea that qualitative analysis and visual interpretation lacks scientific value and highlighted the unfeasibility of quantifying most of what we qualitatively describe in the SAR. We have implemented numerous other changes in response to all the previous reviewers, and suggested others in response to Reviewer 1’s comments on the current iteration. Regarding suggestions we have not implemented, we have already explained in our responses why we think they are unnecessary or are based on a misunderstanding of something we have made/will make clearer when revising.

Reviewer 1 of this iteration makes only one actual suggestion – to compare our quantitative (SIC-derived) polynya-area estimates to results from lower-resolution SIC products. We suggested comparing to actual observations of the polynya from SAR is a better form of checking and validating.

1. The problems with passive microwave data underestimating thin ice (and ponded ice) are well-known so a threshold approach is not ideal. I think a better approach would have been to construct the time series of open water area from PM rather than polynya area based on a threshold. This is done nicely in Moore et al. (2021; 10.1029/2021GL095099).

For readers not familiar with the above-mentioned Moore paper, it uses the same SIC product as in our paper to present/analyse the opening of a polynya in the Arctic in May 2020. It also includes a plot of ‘open water’ from this data in May 2020 and the mean of each May 2003-2021. They do not detail how they define ‘open water’ but they seemingly classify it as when SIC = 0.

Their use of this data supports our use of the same dataset for polynya identification, at least in summer. We would say that their approach also effectively uses a threshold to define ‘open water’, just that the threshold is 0 rather than ours of 70% to identify ‘open/active polynya’. Based on checking the SIC data against direct observations of the polynya in the SAR imagery, we found a threshold of 70% to be better at estimating the ‘open/active polynya’ than 0 for our case. A threshold of 70% has also been used in several previous studies referenced in our paper (e.g. Parmiggiani, 2006; Morelli & Parmiggia 2010, 2013; Preußer et al., 2015; Cheng et al., 2017; 2019). In a revision we could include a figure in the supplementary material showing the SIC values (i.e. with no threshold applied) for an open/active polynya, alongside a SAR image.

We will also clearly acknowledge that passive microwave data underestimates SIC where there is thin ice. (But as noted in the paper and in responses, an open/active winter polynya is expected to include thin ice that forms almost immediately upon opening, and therefore classification of these areas as open/active due to this issue and associated low SIC values is not likely to lead to mis-classification – as shown by our comparison with SAR)

Additionally, we will reference Moore et al. in a revision.

1. Polynya or open water area can indeed be better estimated by SAR but some attempt to do this quantitatively and consistently needs to be made to illustrate variability and ensure reproducibility. There is no problem manually extracting polynya or open water area from SAR imagery but this should be done carefully and ideally in time series image analysis format. Further, the imagery must be pre-processed (i.e. calibrated and corrected for incidence angle).  Not doing these basic things comes across as scientifically lazy and also introduces errors. Moreover, inclusion of the backscatter scale ISthe standard in the literature with respect to ice monitoring.    

We suggest that we could manually delineate and calculate the polynya area in sample SAR images for comparison with our automated approach, for the purpose of validating our automated approach using SIC data. For example, compare the areas from SAR and SIC for the case study in Fig. 2. Manually delineating all the SAR images would be an unfeasibly laborious task for this paper, and as mentioned above, an automated identification approach with SAR is currently beyond our capabilities. I am not sure I understand what exactly is meant by ‘should be done…in a time series image analysis format’.

SAR imagery was pre-processed. All Sentinel-1 imagery Google Earth Engine is pre-processed (i.e. those in the time-lapse), and images presented in figures and downloaded for further analysis were pre-processed in SNAP. Although we do state images were pre-processed in SNAP (lines 160-163) were not clear about the images that were processed in Google Earth Engine – we will amend this in a revision.

We will include a backscatter scale showing the decibels of SAR images in a revision.

1. Studies that use passive microwave to identify polynya area (or open water) together with SAR need to at least quantitatively compare the two estimates. In the current version of the paper this is not very rigorous and at the very least should also be placed on Figures 5 and 6. Timing of formation/closure should also be considered.

Regarding a quantitative comparison we repeat from our response to point 2:

“We suggest that we could manually delineate and calculate the polynya area in sample SAR images for comparison with our automated approach for the purpose of validating our automated approach using SIC data. For example, compare the areas from SAR and SIC for the example in Fig. 2. Manually delineating all the SAR images would be an unfeasibly laborious task for this paper, and as mentioned above, an automated identification approach with SAR is currently beyond our capabilities.”

Additionally, we could add an error estimate to Figures 5 and 6 based on this comparison.

Best regards,

 Dr Grant Macdonald, on behalf of all authors.