| 13 Jan 2026
Improving the representation of coastal polynyas in high resolution climate models with satellite datasets
Abstract. Coastal polynyas are primarily wind driven openings in the sea ice close to shore. They are important drivers of sea ice production, ocean circulation and have important consequences for polar ecology. Accurate understanding and modelling of the Earth system in the polar regions requires a thorough representation of polar climate processes including polynyas and sea ice. This paper investigates the evolution of coastal polynyas and their atmospheric impacts in Greenland and Antarctica by analysing two satellite products alongside simulations from a regional climate model (RCM). We examine how the representation of polynyas differs between the satellite datasets and how these differences affect the RCM when used as boundary conditions. We further explore how the modelled polynyas influence surface atmospheric conditions, boundary layer and cloud properties. Our analysis is based on the Harmonie-Climate (HCLIM) regional climate model, which employs spectral nudging and a surface scheme optimised for polar regions. The model provides a flexible framework for process studies, including with direct use of Earth Observation data such as a new higher spatial resolution satellite product, developed by the ESA climate change initiative (CCI) for sea ice. We run the model over both polar regions, to examine the impact of coastal polynyas in northern Greenland in 2018 and Terra Nova Bay in Antarctica in 2010. Comparison between runs shows an improved representation of climate with the high resolution sea ice product, particularly in boundary layer properties and near the coast which standard resolution sea ice data often has difficulty capturing. Our results show that using the high resolution CCI sea ice data directly in regional climate models can improve characterisation of near-surface weather and climate, supplementing the use of fully coupled atmosphere -ocean- sea ice modelling for process understanding and argue for the inclusion of this data in assimilation systems used for reanalysis.
Competing interests: At least one of the (co-)authors is a member of the editorial board of The Cryosphere.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Review of
Improving the representation of coastal polynyas in high resolution climate models with satellite datasets
by Torres-Alavez, et al.
Summary: This manuscript provides an example how data sets of the sea-ice concentration derived from satellite observations in both hemispheres could be used to improve the representation of coastal polynyas in high spatial resolution (2.5 km to 4 km) climate models. The preferred method to ingest the information provided by the sea-ice concentration data sets is nudging. Quantification of the improvement in the representation of the polynyas is realized by investigating the potential impact two sea-ice concentration products with different grid resolutions have on a selected set of meteorological parameters that are modeled and output by the high-resolution climate model. The manuscript seems to show some evidence that a finer grid resolution of such products results in more realistic latent heat fluxes at the surface, in elevated low-level cloud formation and subsequent elevated downward longwave radiation. One of the main conclusions seems to be that this improvement is more pronounced for particularly small-scale polynyas.
General Comments:
GC0: Overall the manuscript is difficult to read. Usage of terms and definitions is not well thought-through and formulations are often not specific enough. Examples of this can be found throughout the manuscript. Examples include things like "sea-ice value", the usage of sea ice extent versus sea ice area, the naming of polynyas and regions. You may note that the review contains a large number of editoral comments / typos.
GC1: The manuscript would benefit from a more thorough description of datasets and methodologies used. This comprises more information about 1) the meteorological / AWS stations used, 2) the motivation for the time periods chosen, 3) the motivation for the initialization data sets chosen, 4) the motivation to use CARRA as an evaluation source for the Arctic while not having an evaluation source for the Antarctic, 5) the motivation of the meteorological parameters chosen to investigate both the sources and the impact of polynyas in the RCM used, 6) the choice and the size of the regions used to compute sea-ice area and why sea-ice area is preferred over other definitions of the polynya size, and possibly a few more aspects.
GC2: The manuscript contains errors.
GC3: The critical discussion of the obtained results is rather light. In some aspects, the manuscript does not go beyond an investigation of sea-ice concentration variations and their (potential) impact on selected meteorological quantities. The linkage to polynyas is not overly well made in this regard - possibly because of the way you tried to "define" where polynyas were located in your two case studies. The difference in the temporal sampling (daily vs. hourly) and its potential impact on the results is not discussed sufficiently well. It is not sufficiently clear whether and to which degree the sea ice concentration data products chosen are suitable to resolve small-scale polynyas such as those in the Nares Strait or the Terra Nova Bay polynya and to which degree the amount of thin ice within the regions of interest did bias the sea-ice concentration estimates used for the nudging; this applies in particular to the polynya North of Greenland and the Ross Ice Shelf polynya.
GC4: I recommend to carefully re-think the main conclusions of the manuscript and ensure that these are sufficiently well backed up by the results presented and discussed in the manuscript.
Specific Comments:
L19-22: "Coastal ... 2025)" --> I find this is a VERY condensed description of the role polynyas play. I am also not sure that everything you write and combined in one sentence here is correct. I am questioning, for instance, that ocean dynamics play a key role. Since you are dealing with both, polynyas near Greenland and in the Antarctic, the "Antarctic bottom water formation" does not apply in general. I would also put the role of polynyas for biology / biogechemistry in a different sentence.
L33: "and ocean-atmosphere ... drive them." --> I am content with the high surface wind speeds but I would say that elevated ocean-atmosphere heat and moisture fluxes are rather a result of polynyas than their cause. An exception would be if you are referring to sensible heat polynyas where the ocean heat flux is contributing to the opening of the polynya. Latent heat polynyas are kept open due to the interaction of the wind pushing the sea ice away from a barrier (coast, ice shelf, sea ice arc) and advecting the newly formed ice downstream, and the latent heat of fusion that provides heat to the water, contributing to keeping the polynya open. I suggest to change the writing accordingly.
L34/35: "They also have a tendency to drift ..." --> I am not sure what you want to say here with the notion of the drift of the model. I would say that a drifting model is not suitable to use it to investigate processes on climate relevant time scales - but whether it "performs well" against in-situ observations is perhaps not that relevant here?
L35/36: "An intermediate ... features." --> This sentence needs to be backed up with more detail about the status of the numerical models' landscape and the way how polynyas are treated. A reader might want to know whether (and how successfully) polynyas are parameterized, what the spatial resolution of such models is, what the typical spatial resolution of available data sets to be used for this solution are, and so forth. Please change the writing accordingly.
L47/48: "Coastal polynya ... occurrence." --> This statement needs to be backed up by references.
L59: The polygons in Fig. 1: Are they (at all) related to the sea ice area that is computed later on? Isn't the polygon chosen for the North Greenland polynya - and also the two polynyas in the Ross Sea - far too small?
Figure 1: What is "NG"?
What are the red dots and the names next to them representing?
In the caption you write RS and TN for Ross Sea and Terra Nova ... but the polynya names that are typically used (and should be used in this paper as well) are "Ross Ice Shelf polynya" and "Terra Nova Bay polynya".
L65-67: "generating ... early March" --> This is not correct as written. I invite the authors to take a closer look especially at the paper they cite: Ludwig et al (2019). In that paper several figures clearly demonstrate that i) air temperatures increased to near 0degC AFTER Feb. 10 and that there was considerable new ice formation, creating quite some spatiotemporal dynamics in both the sea ice concentration as well as in the polynya area.
L81: Both references focus on the physical properties and mechanisms. Please add an additional reference for these "Several studies ..." in relation to the seal population.
What I have been missing so far is the typical size of these 4 polynyas - both in terms of the open water area where the largest turbulent heat fluxes are observed and in terms of the thin ice area downstream of the principal wind direction. Learning about the typical size is required to understand your motivation to use RCMs at the specific grid resolution you are detailing in the next section.
L133/134: I can understand that the main motivation for the time period chosen to investigate polynyas near Greenland comes from the large polynya event north of Greenland. But what was the motivation to investigate the polynyas in the Ross Sea for the time period chosen, i.e. Oct. 10 through Nov. 9? What is the scientific rationale behind that choice?
L144: Later on you show that there is a bias in the pressure between the observations and the HCLIM runs. Would it make sense to estimate the altitude-induced bias in the surface pressure to better understand the bias that is shown later?
L144: There is no Table 1.
L146/section 3.2: Your model experiments are using a grid resolution finer than 5 km. Why did you not look for data sets other than those you are using here? The data set which has been used by Ludwig et al. (2019) for instance. Or the ARTIST Algorithm SIC based on the AMSR2 near-90 GHz channels providing a grid resolution of 3.125 km. Or high-resolution SAR - AMSR2 based SIC products? I strongly recommend to come up with a discussion that justifies the usage of the two products used. The length of the time series could be one argument - but since you are investigating specific cases and/or events this argument seems not to be holding.
In addition to SIC data sets there exist other data sets of polynya area and/or frequency of occurrence and/or ice production that were derived using different methodologies and data, e.g. SAR imagery, MODIS imagery, or resolution-enhanced passive microwave imagery (Polynya Signature Simulation Method - PSSM and others). Some of these methods and data sets do not only offer a finer spatial resolution but also a finer temporal resolution - i.e. sub-daily. Particularly in a highly dynamic area such as a coastal polynya and in combination with the high-resolution RCM modelling it seems a bit odd that you do not take into account - at least to list and discuss - the value and/or degree of being fit-for-purpose of these other data sets for your specific study aim.
L162/163: "SICCI-HR-SIC is ..." --> It is not really clear from this sentence where the improvement is with respect to the 25 km grid data set. Please spend one more sentence to describe this with sufficient detail.
L164: "resampled" --> So you do not apply any averaging or spatial interpolation. All 2.5 km grid cells that fall into a 25 km grid cell did receive the same SIC value?!?
L173-181: It is not sufficiently clear why CARRA can be used as a means of evaluation of your experiments if CARRA and your method are based on the same SIC input data source. How is the SIC treated differently in CARRA as compared to your study and why is this difference in the treatments sufficient to use CARRA data as evaluation data source?
How about the Antarctic? It is not sufficiently clear how you are going to evaluate your results for the Terra Nova Bay polynya or the Ross Ice Shelf polynya.
L182-185 / section 3.4: This section cannot be complete. Please add the information that is required to understand what your aim is in this regard. There are several AWS stations located near your two focus areas in the Antarctic and there are certainly also more stations than the one you mentioned in the context of the North Open Water polynya and the other polynyas in Greenland. Their names and locations, periods of measurements, parameters used and several more details should be provided.
L204: "hourly time series" --> Both data sets have daily temporal resolution (or sampling). How can you come up with an hourly resolution?
"annual minimum" --> The time series does not extend over an entire annual cycle. How can you then speak of an "annual minimum"?
L228/229: "Compared ... certain periods" --> When I look at Figure 4, left hand side, I would say that CARRA shows higher LLC fractions for 90% - 95% of the time. Only for the polynya north of Greenland there are a few short periods when HCLIM exhibits higher LLC fractions. And these occur at times when Fig. 3 does not really indicate a particularly low sea ice area or a particularly high latent heat flux. This does not go together very well. The comparison of the LLC fractions between CARRA and HCLIM does not seem to be overly conclusive.
L236/237: I agree that enhanced formation of low-level clouds can lead to enhanced downward longwave radiation which in turn can contribute to elevated near surface air temperatures. I am wondering, however, how large the contribution of the sensible heat flux due to the increased open water area and the increase in ocean-sea ice-atmosphere heat flux due to more thinner / snow-free sea ice is. Wouldn't this result in a larger and also more direct contribution to the near surface air temperature?
L238/239: "After ... polynyas." --> Will you get back to this in your discussion section? No ...
L246-254: I am a bit sceptical about this paragraph and I am not sure it is really needed. The primary influence of more open water / more thinner sea ice is enhanced moisture and heat input into the atmosphere and consequently there could be more low level cloud formation - as you tried to demonstrate already. But whether and from which point onwards (downstream of the polynya) these clouds would be thick enough to produce precipitation is certainly much more difficult to capture and I doubt that the case study presented here is mature enough to dwell into this. Only the polynya north of Greenland seems to have been large enough to promote development of enhanced precipitation.
L255: Please specify a bit more clearly what you mean by "stronger impact". Also: Is this "stronger impact" leading really to an improvement in the representation of polynas or polynya-like features?
Figure 6: Clearly, after the polynya event the two HCLIM runs begin to diverge in temperature and also in pressure - not so much in the relative humidity. Please discuss potential causes for that.
For the bias in the pressure see also my comment to Line 144.
Neither in Fig. 6 nor in Fig. 7 the text labeling the x-axis is located at the respective tick mark; the text has to be moved to the left.
L282-: Looking back at the comparison between HCLIM runs forced with two different SIC products and the in-situ observations I am missing an inter-comparison of the near surface wind speed. This parameter is essential for the formation and maintenance of a polynya (besides the temperature). I can understand that from the viewpoint of a numerical modeller it is important to illustrate that surface pressure and humidity are correctly represented - and I am not advocating for not showing these parameters but I am asking in addition for a comparison of the near surface wind vector.
L304/305/306: "comparing sea ice areas in the HCLIM simulations" --> Where can I see these sea ice areas visualized?
You write that "ESA CCI consistently represented smaller polynya extents than OSI SAF" --> it is not clear whether this is a statement based on model results. Also, it is not clear how you define "polynya extent" in this context.
L306/307: The numbers for the differences are far too high. Please also check my comment to Fig. 9.
L310: "2nd-9th" --> The episode of high latent heat flux ended on Nov. 7, not on Nov. 9. Please correct.
For the second polynya event, which resulted - seemingly - in an even larger polynya area (the sea ice area is even lower than during the first episode) the HCLIM simulations do not show elevated latent heat fluxes. Why?
L311-314: "Differences ... system." --> This statement is not backed up by your Figure 9. While there is indication that the ESA CCI forced HCLIM simulations provide higher fluxes for the Ross Ice Shelf polynya, differences do not reach to 20 W/m². And for the Terra Nova Bay polynya example I don't see any improvement between ESA CCI and OSI SAF forcing. On the contrary, the differences seem to be rather arbitrary. Also: the decrease of 40 W/m² you are reporting about in the previous sentence ... where do I find that and why (and how) is this related to any change in sea ice area in the region of the Terra Nova Bay polynya? I have the impression that the HCLIM simulations do not follow the forcings of either of the two sea ice concentration products particularly well - perhaps because the actual fluctuations in the polynya size or area happen at spatio-temporal scales which are not resolved by any of the two products.
L315-318: I don't find that the results for the low-level cloud cover align well with either the observed sea-ice area or the temporal development of the latent heat fluxes.
Figure 9: Please check the exponent of the sea ice area. It cannot be correct that the sea ice area in the Terra Nova Bay polynya region is as large as 1.3 x 10^7 km²; the same applies to the Ross Ice Shelf polynya. The numbers are far too large.
Please also check my comment to Fig. 1 where you show polygons that you may or may not have used to compute these sea ice area values. It is not clear from your writing.
The panels on the right hand side are supposed to also show the results of the CARRA reanalysis. But there are none for the Antarctic.
The ESA CCI sea ice area in the Terra Nova Bay polynya region is constantly smaller than the OSI SAF sea ice area but this is not visible in the latent heat flux time series. A smaller sea ice area should lead to a higher latent heat flux - which is only sporadically the case.
L323-324: "over the Ross Sea ... on 31 October" --> While this statement seems to be true, the timing of this largest difference between the two forcings seems not to be related at all to any polynya activity because it falls into an episode of rather high sea ice area in the Ross Ice Shelf polynya region.
L325-327: Neither your figures nor your text backs up this statement sufficiently well. Unlike in the Arctic there is little one can learn from the results.
L333: Please explain in more detail why you think that the consistent cold bias stems from a difference in the land-sea mask.
L337: I cannot see from your results that either of the two polynyas, the Terra Nova Bay polynya or the Ross Ice Shelf polynya, reached this maximum extent. This statement is not well connected to the rest of the paper.
L340: "Fig. 7" --> wrong figure. Also, there is no figure showing sensible heat flux values.
L341-344: Not clear where these lines are refering to.
Section 5: --> see also GC3
Missing elements:
- critical reflection on the model results and whether / how the observations can be used as they were used.
- reflection about the potential of other data sets is missing
- daily input but hourly results
- Noise and uncertainties created
- The role of leads
- Influence of sea ice thickness on model results of the larger polynyas.
L349: No, you did not investigate an impact on the "sea ice extent".
Please be more specific: You investigated LATENT surface fluxes, LOW LEVEL cloud COVER, and LONGWAVE DOWNWELLING radiation AT THE SURFACE.
L357/358: The results obtained for the Ross Sea are not sufficiently mature to back up the statement "A similar pattern ... Sea Polynya."
L360-362: Please avoid speaking about "polynya extent" in the context of your work. You are not dealing with the extent of the polynya; you did not work with a clear definition for it. What you are working with are time series of the sea ice area (i.e. the integral surface area of the ocean covered by sea ice) within some more or less rigidly defined regions within which the polynyas of interest are located. You take the sea ice area as an approximate measure of the polynya size - no matter whether there are other factors influencing the sea ice concentration and with that the sea ice area computed.
I am fine with the conclusion about the elevated latent heat flux but the investigations (and results) about the low level cloud cover are more speculative - and those about longwave heatflux, near surface air temperature and precipitation even more. I recommend to be more critical with the interpretation of your results.
L371-374: I suggest to remove the part about precipitation completely. I don't think that the results shown are credible in that regard.
Editoral Comments / Typos:
L10: "including with direct use"? --> "including direct use"?
L13: "an improved representation of climate" --> This reads a bit odd. Perhaps it would be sufficient if you would directly relate to the processes that you mention later in this sentence.
L14: "near the ... difficulty capturing" --> not clear what you want to say here. Please reformulate.
L22/23: What are the common features you are refering to? What exactly do you mean by "at the surface interface with the atmosphere"?
L25: "the high resolution needed to resolve them" --> this sentence should be backed up earlier with a clear statement about the typical size of the polynyas studies - which, by the way, can be quite variable if you compare the size of the Terra Nova Bay polynya with the Ross Ice Shelf polynya.
L27/28: "including for failing ... 2012)." --> This paper is about 13 years old. I recommend to update this statement with more recent information - e.g. from the CMIP6 team.
L30: "is not well represented (Rinke et al., 2006)" --> Also this reference is quite old. You could consider updating this reference with more recent work.
L31: "suffer similar" --> "similar from similar"
L38-30: "satellite ... the OSI-SAF" --> This sentence needs to be re-written. What is "from the ESA CCI"? What is "by the OSI-SAF"? The part: "data set of high resolution sea ice data ... sea ice concentration" should possibly be condensed.
L38: "also use" --> why "also"?
L49: "in the pack ice" --> should be deleted because you are not dealing with open ocean polynyas such as the one in the Weddell Sea or the Cosmonaut Sea. Coastal polynyas are formed close to the coast, often along a landfast ice border or along an ice shelf.
L50: "off shore away from the coast" ... either "away from the coast" or "off shore" is enough.
L51: What you should perhaps add to this sentence is the role of the latent heat of fusion that supports the polynya to stay open - as described in the paper of Morales-Maqueda you are citing further down.
L57 / L71-72: The North Water polynya is not solely a latent heat polynya. This polynya is a mixture of a latent heat and a sensible heat polynya. Please check available literature about additionally factors such as tides and upwelling of warmer water masses and correct your writing accordingly.
L63: "in" --> "by"
L63: "2010" --> "2018"
L71: "Vincent, 2019" --> This is a publication focusing on the aspect of the ice arch formation and how ice conditions have been monitored using satellite remote sensing. It is not a valid reference for the various topics you touched upon in this sentence. Please find 1-2 references that are fit-for-purpose.
L77: "Ross Sea" --> I strongly recommend to not use the term "Ross Sea polynyas" - simply because there are several: The Terra Nova Bay polynya is one of them. The more accurate name would be "Ross Ice Shelf polynya". See also GC0
L87/88: "permafrost ... on regional climate." Please make sure that you provide references for these two statements as well.
L94/95: "inherit ... models" --> I don't understand what you want to say here. What are "bosses" in this context?
L111: "Terra Nova Antarctica" --> Not sure what you are referring to here - only the Terra Nova Bay polynya or also the Ross Ice Shelf polynya?
L112: "and time step of 45 second" --> "and a time step of 45 seconds."
L114-116: "Notably ... Von Storch et al., 2000)" --> Would it make sense to move these sentences a bit further down to where you actually begin writing about the nudging (i.e. L124)?
L158-160: Please provide the temporal resolution as well. I guess it is daily? Did you eventually take into account to contact the data providers (after you have read the paper by Lavergne et al. (2019), to ask for the swath data of the SIC data?
L167-169: The SIC uncertainty sources listed are, no doubt, correct. The question a reader may have is how relevant these uncertainty sources are for your specific aims.
L187: "polynya over Greenland" --> Not sure what you mean: "North of Greenland"? "Near Greenland"? Please correct. A polynya cannot develop on the Greenland ice sheet. Please check usage also in your following text. An example of GC0.
L189/190: I suggest to refer to the two polynyas you are aiming to investigate. In addition: using "intense" in the context of a minimum seems a bit odd. Please try to provide a more specific description of what you observe. You might want to say: "ESA-CCI sea ice concentration values are lower than OSI SAF sea ice concentration values in these two regions." or "ESA-CCI sea ice concentrations exhibit more details than OSI SAF sea ice concentrations."?
L191: "This polynya" --> Which one? I guess you used a name for that polynya before and it makes a lot of sense to stick to your own notation throughout the manuscript.
"Narest Strait" --> Do you mean "Nares Strait"? You need to correct this also in following instances (e.g. L195).
What is a "sea-ice value"? Please use correct language and be more specific. There are also several other instances of this.
L195: "over" --> "at" or "along"
L196/197: "with OSI-SAF ... of Peary Land." --> This half sentence seems to be out of context.
L199/L201: "Wadel Sea" --> please use the correct name.
Figure 2: Is it correct that you show the two SIC products at their native grid resolution?
L203: "Sea ice area" Please see my comment to Figure 1 related to the polygons.
L215: What are "sea-ice differences"? Please be more specific.
"latent heat flux" --> Please state clearly that you are (since when) talking about results you obtained using your model. This is not clear here.
L216/217: "when sea ice reached its minimum" --> what "sea ice" parameter? thickness? drift speed? coverage? concentration? Please be more specific.
L218: "differences between ESA CCI and OSI SAF remained below" --> You must make the connection to the model and the model results.
L221: "minimum sea ice" what? area? thickness?
L222/223: "analyzed " ... using which tool or method? Again you need to connect to your model.
L226/227: "ESA CCI indicated ..." --> Again the connection to the model is missing. ... "Using ESA CCI sea ice concentration data as sea ice cover boundary condition (or forcing) the HCLIM simulations result in higher low-level cloud fractions than using OSI SAF sea ice concentrations."
L260+: Whenever you write about a temperature difference it would be good if you'd consider to use the unit K. It avoids confusion with actual temperature values expressed as degC.
Figure 5: Please add in the caption of the figure what the "Difference" is.
Please consider plotting a horizontal line denoting a difference of zero degrees. This applies to all other figures containing panels of the same sort in which you are combining actual values with a difference.
L270: Make sure that you are not changing the name of the experiment / or run. So far it was "HCLIM-ESA CCI" and "HCLIM-OSI SAF". There are more instances of this later in the text.
L284+ The entire paragraph needs to be re-written, checking carefully the language used, the terms used, and other elements. Please consider shortening the paragraph substantially because a reader can take a look at the maps and figure our most of the spatio-temporal development of the sea ice concentration by themself. Formulations should be improved.
L276-277: "However, ..." --> Please check this sentence.
Why did you switch to the precipitation as an additional parameter here and did not also compare surface pressure and relative humidity?
L283: "of the polynya over Terra Nova" --> "of the Terra Nova Bay polynya"
L303: Why now "5th" of October? I thought so far the period of investigation begins Oct. 10.
L303: "reached its minimum sea-ice extent" --> I know what you want to express here but: i) Sea-ice extent has a fixed definition that should not be used in the context of describing the properties of a polynya. ii) Fig. 8 and 9 show sea-ice area. I suggest to write something along the lines: "The Ross Ice Shelf polynya reached its maximum size - illustrated by means of the minimum sea ice area on days y and z"
Figure 8: The panels are too small to adequately see the development of the polynas in the sea ice concentration maps. Their size must be increased.
Figure 10 & 11: These figure cannot contain CARRA results. Please update the caption.
L330/331: Please connect Figure 12 and its description better to the work done about polynyas - I assume that you are doing this comparison to better understand the results obtained for the Terra Nova Bay polynya?
L331-332: You need to switch the colors; the blue is for the finer (higher) resolution boundary conditions; In addition - why do you here not again refer to ESA CCI and OSI SAF?
L350: "sea ice products" --> "sea-ice concentration products"
L355: "in narrow" --> better: "for small"
L356: "such as the" --> "such as occurring in the"
L358: "sea ice products" --> which?
L363/364: I don't think "smaller ice concentration" is the correct expression here. You possibly should write something that expresses the larger degree of detail. Speaking globally of a "smaller" ice concentration gives the wrong impression.
As stated above, I don't find the results for the Terra Nova Bay polynya that convincing.
L366-369: "In Greenland ... exceeded 1.5degC" --> not clear what you want to state here. Please re-phrase.
L366: "resulted in" --> "contributed to"
L382/383: "and our results show ... impact" --> I don't understand what you want to say here. Why is this?
L392: I don't understand. Why do your results show that an Antarctic (atmospheric) reanalysis is plausible?
L502: Please check this reference. The name is not correct as written here. I guess this needs to be the CMIP community.
L542/543: The doi is not resolved. I could not access the paper.