Review of "Turbulent heat flux dynamics along the Dotson and Getz ice-shelf fronts (Amundsen Sea, Antarctica)" by B. Jacob et al.

The authors estimated the turbulent heat flux in the Amundsen Sea Polynya region during summer using in situ atmospheric data collected aboard the RV Nathaniel B. Palmer. They observed episodic heat loss events triggered by the outflow of cold, dry air from the Antarctic continent. A comparison with turbulent heat flux data from ERA5 revealed that ERA5, with its relatively coarse spatial resolution of 0.25 degrees, did not accurately reproduce the turbulent heat flux in the ocean along the ice shelf edge, leading to an underestimation. Heat flux estimates based on in situ observations in the Antarctic coastal areas, particularly in coastal polynya regions, are rare, making this study a valuable contribution to the polar science community. The data and analysis methods employed in this study appear to be reasonable. However, I have the following concerns and look forward to the authors' responses and revisions to the manuscript.

This study emphasizes the importance of estimating turbulent flux due to its impact on heat loss and sea-ice production in coastal polynyas (e.g., P. 1, L. 2–, P. 2, L. 37–). While this is undoubtedly true during the winter months, this study is based on summer observations. In winter, the dominant heat flux component is turbulent heat flux, whereas in summer, it is shortwave radiation, as shown in Fig. C2. This distinction should be clearly described in the manuscript. During summer, coastal polynyas act as "meltwater factories" due to solar heating of the upper ocean through open water with low albedo, contrasting with their role as "ice factories" in winter (Ohshima et al. ,1998 and Morales Maqueda et al., 2004). Therefore, I do not suggest removing the descriptions of coastal polynyas but rather believe they should be described with care. In recent years, the Antarctic sea-ice extent during summer has been unusually small (Purich and Doddridge, 2023). A prolonged open-ocean period in summer, resulting from anomalous sea-ice retreat, leads to increased solar heating and warming of the upper ocean, with this heat anomaly potentially influencing subsequent ice advance (Nihashi and Ohshima, 2001; Stammerjohn et al., 2012). The key factor here remains shortwave radiation, though heat loss to the atmosphere in autumn and winter is driven by turbulent heat flux. In the Amery Ice Shelf area, a reduction in summer sea-ice extent has been found to weaken the formation of Antarctic Bottom Water (Aoki et al., 2022). This is because anomalously small summer sea-ice extent leads to increased solar heating of the ocean, which accelerates the melting of the ice shelves and the supply of freshwater to the coastal polynya area, limiting the production of dense shelf water. Again, the primary heat flux component here is shortwave radiation, but turbulent flux also contributes to the total heat flux. Given the significant changes occurring in the Antarctic sea ice, I believe that incorporating these perspectives could be valuable.

• Ohshima, K. I., K. Yoshida, H. Shimoda, M. Wakatsuchi, T. Endoh, and M. Fukuchi (1998), Relationship between the upper ocean and sea ice during the Antarctic melting season, J. Geophys. Res., 103, 7601–7615, doi:10.1029/97JC02806.
• Morales Maqueda, M. A., A. J. Willmott, and N. R. T. Biggs (2004), Polynya dynamics: A review of observations and modeling. Rev. Geophys., 42, RG1004, doi:10.1029/2002RG000116.
• Purich, A. and E. W. Doddridge (2023), Record low Antarctic sea ice coverage indicates a new sea ice state. Commun. Earth Environ. 4, 314, doi:10.1038/s43247-023-00961-9.
• Nihashi, S. and K. I. Ohshima (2001), Relationship between the sea ice cover in the retreat and advance seasons in the Antarctic Ocean, Geophys. Res. Lett., 28, 3677–3689, doi: 1029/2001GL012842.
• Stammerjohn, S, R. Massom, D. Rind, and D. Martinson (2012), Regions of rapid sea ice change: An inter-hemispheric seasonal comparison. Geophys. Res. Lett. 39, L06501, doi:10.1029/2012GL050874
• Aoki, S., T. Takahashi, K. Yamazaki, D. Hirano, K. Ono, K. Kusahara, T. Tamura, and G. D. Williams (2022), Warm surface waters increase Antarctic ice shelf melt and delay dense water formation. Commun. Earth Environ. 3, 142, doi:10.1038/s43247-022-00456-z

Temperature and wind speed are crucial parameters for determining turbulent heat flux. A comparison of in situ observed wind speed with ERA5 data is shown in Fig. A1. How about including a similar comparison for temperature? As a reader, I believe such a comparison would provide valuable insights. In Fig. A1, the wind speed from in situ observations is higher than that from ERA5. Could this discrepancy be due to the difference in observation heights, with the ship's measurements taken at 34.4 m (Table 1) and ERA5's at 10 m? A similar consideration applies to temperature: the ship's observations are taken at 19.2 m, while ERA5's are at 2 m. I suspect there may also be a bias in the temperature data. I believe the impact of these biases in wind speed and temperature on the turbulent heat flux estimates should be quantitatively discussed.

Regarding the estimation of turbulent flux (Eqs. 1 and 2), the influence of atmospheric stability on the heat transfer coefficient should be mentioned in this manuscript, even though it is discussed in the cited paper. Furthermore, since this study primarily focuses on the open ocean area of the summer coastal polynya region, I believe the influence is minimal. However, turbulent flux is also estimated in the sea-ice area (Fig. 3). In regions where sea ice and open water coexist, the estimation of turbulent flux is complicated by the significant thermal contrast between the sea ice, which acts as an insulator, and the open water. Additionally, considering atmospheric stability in such areas is challenging. How was the insulating effect of sea ice accounted for in the estimation of turbulent flux in this study?

Minor comment:

1. 1, L. 3–: "The Amundsen Sea Polynya is the fourth largest coastal polynya ..." Is this referring to the size of the polynya or the sea-ice production?

1. 1, L. 5: "NBP22/02" This expression makes sense to readers familiar with the observations by RV Nathaniel B. Palmer but is confusing to those who are not. It would be better to be more specific. Additionally, the description of the ship observation data begins on P. 2, L. 54, but the specific ship name does not appear until P. 3, L. 63. The ship name should be described earlier.

1. 6, L. 98: "... a heat loss (gain) for the ocean surface." would be appropriate.

1. 7, L. 118: "... fresh water flux ..." During summer, the freshwater supply from melting sea ice is significant. Does this study consider that, or only precipitation?

1. 8, Fig. 2c: It would be helpful if you could show the freezing point. Furthermore, since temperatures below 0°C are important, I would appreciate it if you could display them in a taller figure.

Review of Turbulent heat flux dynamics along the Dotson and Getz ice-shelf fronts (Amundsen Sea, Antarctica), by Jacob et al. (2024).

Summary

The authors make use of in situ observations taken from a research vessel located in the Amundsen Sea Polynya, Antarctica, to investigate ocean-atmosphere turbulent heat fluxes at this location. The authors examine the performance of the ERA5 reanalysis turbulent flux product in this region by comparing with the fluxes derived from observations. A 1D model is then used to evaluate the impact of differing turbulent heat fluxes on the sea surface temperature and heat content of the ocean mixed layer.

The paper is generally well-written, clear, and easy to read. It is interesting to see measurements of turbulent heat fluxes derived from in situ observations at the Amundsen Sea Polynya. Comparing these to the ERA5 reanalysis product provides a useful validation, and showing the impact of turbulent heat fluxes on SST and ocean heat content is also of value.

However, the paper is missing some detail on background, methods and implications of these results, which would make it more informative.

Questions asked below are intended as pointers to aid improvement of the paper, rather than solely as questions to answer in response to this review.

Major comments

Background

The introduction should make clear what work in the paper is novel, and how it fits in with the existing literature. Literature should be referenced in the introduction to provide background and to put this research into context. What does this work add to the existing literature?

More general background on coastal polynyas should be included, for example could use Morales Maqueda et al. (2004) as a starting point: Morales Maqueda, M. A., A. J. Willmott, and N. R. T. Biggs (2004), Polynya dynamics: A review of observations and modeling, Rev. Geophys., 42, RG1004, doi:10.1029/2002RG000116.

More information on previous in situ observations at the ASP should be included as well as satellite observations and available reanalyses.

More background information on the ASP itself would be helpful. What is the polynya size? Frequency of occurrence? Does it change on the timescale of the field campaign?

Methods

What are the sensible and latent heat transfer coefficients (C_t , C_q) used for the COARE 3.5 algorithm? Are these valid for open ocean or sea ice? Do they change depending on the surface? How were they calculated? Is neutral stability being assumed? Is this a reasonable assumption?

How are ERA5 fluxes calculated? What heat transfer coefficients are used here? Is this different when the reanalysis thinks the surface is an ice shelf? Where does the ice shelf location dataset come from? What dataset does ERA5 use to define sea ice or open water? How is surface temperature determined? What validation has previously been performed on this product? What is the spatial and temporal resolution of this product? A spatial map of the data in the polynya region would be helpful.

The co-location method to match up THF observations and the ERA5 reanalysis would be improved by using interpolation (e.g. nearest neighbour interpolation) of the reanalysis to the observation location, rather than a simple selection of the nearest neighbouring point. It would also be usual to perform a quality check on the data beforehand, which should remove points classified as the ice shelf in the reanalysis dataset.

How was time co-location of the matchups of the THF observations with the ERA5 reanalysis performed? What is the temporal resolution of the ERA5 product?

Even if using the existing co-location method, why not simply choose the nearest ERA5 THF that is an ocean point, rather than recalculating the THF using ocean temperature and the COARE 3.5 algorithm?

Should include the method used to calculate the air humidity and saturated humidity for the latent heat flux from the relative humidity recorded by the measurements (as given by Table 1).

The method of calculation of the ocean heat content is also missing.

Were there no ship-board observations of sea ice concentration? What is the accuracy of the ASI product?

Equation 1 refers to skin temperature of the ocean. However, the SST observations here are measured at 5 m depth, which is not the same thing. How will this impact the sensible heat flux calculation? A good description of the different SST definitions is here, if needed: https://www.ghrsst.org/wp-content/uploads/2021/04/SSTDefinitionsDiscussion.pdf

CTD and TSG measurements should also be discussed in the observations section. How accurate are the observations? Are uncertainty estimates available?

What is the uncertainty on the results? How does this magnitude compare to the differences seen between results from using different methods?

Implications

What are the wider implications of the results? E.g. for biological production in the ASP or melt of the nearby glaciers? Dense water formation? Although the paper mentions that some of this would be affected, it is not stated how.

What is the impact on heat loss over the whole polynya from the error in ERA5 ice shelf location vs observations? Is it just a small surface area missing from the turbulent flux? Does the fact that the area closest to the ice shelf experiences the highest fluxes change things? Can this impact be quantified? What is the impact over a year? On longer timescales? It is stated in the paper that the impact on sea ice formation, ice shelf melt and primary production is not negligible, but this needs to be expanded on and quantified in some way if possible.

Minor comments

Cold air outbreaks are mentioned in the literature review section, but the first half of the paper refers to this phenomenon as heat loss events. Terminology for these events should be made consistent throughout the paper.

The validation of the ERA5 reanalysis product highlighted a mismatch between the ice shelf edge in the product and in reality. While it is useful to point this out, this is not unusual, due both to the resolution of the product, and the fact that ice shelf edges are not static over time. It looks from Figure 7 like the ship is not located at any point designated 100% land. This implies that the issue is with the spatial resolution of the product, rather than the ice shelf mask used for the ERA5 product being out-of-date.

It would be useful to show the RMS difference between the products in Table 4, as this would aid in the interpretation of the differences between Fig. 8b,d and Fig. 6b,d (line 211). Note also that these difference might reduce if using interpolation in the co-location method (see above).

The distributions of the different results shown on Figure 10 should be discussed further.

The discussion in Section 4.1.1 should be worked in to the relevant parts of the paper rather than separated out, e.g. see comment about line 97 below.

Much of the discussion of relevant literature in Section 4.1.2 should instead form part of the introduction. This would provide background for the research and then context for the discussion of results in this section.

Specific (minor) comments

Line 4: Differentiate between in situ and satellite observations – ASP is only poorly observed through in situ observations.

Line 4: Models and reanalyses are not wholly unvalidated, clarify that what is meant here is against in situ observations only

Line 8: “along the ice shelves” should say “along the ocean in front of the ice shelves” or similar, for clarity, as it’s not the ice shelves themselves losing heat

Line 12: underestimation of 28 W m^-2 perhaps? (double negative if -28 W m^-2). This is similar on lines 271-2 (heat loss of -230 W m^-2), but you may prefer to leave it as it is.

Line 31: Reference Figure 1 in this sentence, e.g. “The Amundsen Sea, West Antarctica (Figure 1)…”

Line 37: More description of coastal/wind-driven polynyas and “ice-factory” mechanism is needed here, including the continual off-shore transport of ice by winds.

Line 39: Clarify here whether the observations in this study are the only in situ turbulent heat flux observations available for the ASP

Line 54: Clarify that THF includes sensible and latent heat fluxes (it’s mentioned later in section 2.2.1 but the question first arises here)

Line 55: What is an “underway system”?

Line 56: Clarify that it is the “bulk” turbulent heat flux being calculated, rather than using an eddy covariance method using high-frequency observations.

Line 56: “Punta Arena” should be “Punta Arenas”

Line 59: What is the sampling frequency of the observations from which the hourly averages (would be better to state “means” here, rather than “averages”) are made?

Line 61: It should be mentioned that there will be a positional bias in these observations (in a different way to e.g. airborne data) as the ship presumably won’t travel into regions of higher concentration sea ice. What is the maximum SIC for this dataset? (Note that polynyas can be covered in high concentration, but thin ice, with large heat fluxes still associated with them)

Line 63: Define RV acronym

Line 66: What depths did the glider cover? How close to the surface did it take measurements? Why 40 m for the HC calculations – does this provide good coverage of the mixed layer?

Figure 1: Suggest making the inset map a bit bigger. What is the grey in Figure 1(b)? CTD should be mentioned in the observation section too, and the acronym defined.

Line 80: “ASMR2” should be “AMSR2”

Line 87: Suggest rewording “the evaporation” to “sea surface evaporation” or similar, for clarity

Line 95: Does the logarithmic wind speed adjustment assume neutral atmospheric stability?

Line 97: What impact on the overall results would neglecting ocean surface velocity have? Suggest move lines 291-293 here.

Line 109: Suggest changing “/=” notation as this implies “does not equal”

Line 114: Suggest changing section title to include “the heat content of the mixed layer”

Line 116: Add “produced” before “using observations and ERA5” for clarity.

Line 119: “blue rectangle” is described as a square on figure caption, should be consistent

Figure 2: Were there any sea ice covered points which were also along the ice shelf front? The colour coding method doesn’t allow for this, so it should be clarified if this was the case for any points. Clarify also that the “main sea surface and atmospheric variables” are from the ship observations. Suggest choosing a different colour for the blue area in (f) as it looks similar to the grey, and to the Southern Ocean classification too.

Line 131: How are the uncertainties calculated? These are very large compared to the magnitude – this should be commented on.

Figure 3: caption “latter” is not needed here

Line 135: What is the reason for the differing domination of LHF and SHF in different regions?

Figure 4: Suggest replace “on top of the black rectangles” with “outside of the black rectangles” as some are shown underneath. Suggest could show empty (-130, -90] bin for LHF so that the two figures line up.

Line 148: “each time step” – suggest rewording this as it implies a model. Calculated for hourly data?

Table 3: “indice” should be “index”

Line 166: Mention that it’s the grey shaded area shown on Figure 3.

Line 169: add “visible on the map (Fig. 5d)”

Line 172-3: Should this be Fig. 3a,c not Fig. 3a,b? Fig. 3e might be better to refer to though.

Line 179: Suggest reword “comparing them to the COARE 3.5 fluxes from the observations” to “comparing them to the observed fluxes, calculated using the COARE 3.5 algorithm” for clarity.

Figure 6 caption: Clarify that where SST is NaN it is classed as ice shelf. What happens to the SST if the surface is covered in sea ice?

Line 184: What is the ERA5 product temporal frequency? Would it be expected to capture hourly events?

Table 4: Suggest swapping LHF and SHF columns as this is the order in which they are discussed in the text. Suggest changing column header to e.g. “LHF max diff to obs” for clarity. Which way round is the difference calculated, e.g. obs minus product? Define “ds”.

Line 210: The mean LHF difference to observations in Table 4 is actually higher for the hybrid version (though very small) – this should be addressed in the text. What is the uncertainty range of these results?

Figure 9: The colour scale is quite hard to see on the plots when comparing the three simulations, particularly the palest colour. Is it possible to use more contrasting colours?

Line 239: What is the wider implication of this? (See comments above).

Lines 245, 258, 340: Should replace “seasonal” with “over a month” or similar wording, as 35 days is not a full season.

Line 245-6: “not negligible” – this needs to be expanded upon. What impact would this have on these processes? Can it be quantified? (See comments above).

Line 253: Should replace “less important” with “smaller”

Figure 10: SST depth is given as 6 m here (and in Table 5), whereas in Table 1 it is quoted as ~5 m. Which is correct?

Line 275: Is 12 W m^-2 within the range of uncertainty?

Line 283: “any pattern of wind intensity” – this wording is a bit confusing, perhaps “any pattern of increasing wind intensity” or similar

Line 330: Add something along the lines of “as the air was cold and dry enough to enhance the air-sea temperature and humidity gradients” despite it being summertime.

Line 333: Is ERA5 likely to be the same as ERA-Interim in this respect?

Line 349: Define Chla.

Figure A1: How are points outside of the range defined as outliers? (Meaning, why are they not included within the range?)

Line 384: Typo in LHF equation: I believe the overbar should not extend over the delta_q’