Review on “Quantifying the Influence of Snow over Sea Ice Morphology on L-Band Microwave Satellite Observations in the Southern Ocean” by

Zhou et al.

Utilizing the limited but widely collected data for the sea ice and snow in the south ocean by buoys, ship bases, and aviation satellites, based on the snowpack and radiation transfer models, the authors simulated and analyzed the impact of snow cover and its stratification on the passive microwave brightness temperature observation results. The latter is crucial for satellite remote sensing inversion of sea ice concentration and snow/sea ice thickness. Overall, the authors conducted a very detailed analysis based on the complex data. The comprehensiveness of the data is the greatest contribution of the paper, and the complexity of the data also brings many uncertainties to the results of the paper. Therefore, I recommend that the paper needs to undergo major revisions before it can be considered for publication. My main concern is still to focus on the analysis of complex data. Here are my general comments:

• Does the snowpackmodel consider the effect of snowdrift? Is the thickness of snow simulated output or forced by observational data？
• How are the differences in snow and sea ice characteristics in different regions, especially between the regions ofdrifting ice and landfast ice, considered？

Special comments:

• Line 18 “facilitating key biological processes”is not belong to the influences on the polar climate system
• Line 57: Why the evolving climate conditions in Antarctica would lead tosnow melting and refreezing processes becoming more prevalent?
• Line 76 “vis-à-vis”it is not a commonly used word.
• Line 94 “We also use data collected from ice mass balance buoys (IMBs)”: I wonder if this is also SIMBA buoys, as it is also not equipped with sonars. Please verify.
• SIMBA-type buoy over Prydz Bay: the buoys deployed in the Prydz Bay included the SIMBA and CRREL-IMB.
• Line 127 “Vertical now salinity profiles”: It is snow profiles.
• Line 153 “ et al. (2015)”: typing error.
• ASPeCt snow and ice thickness data: This data actually has a significant error because it is visually judged by humans.
• Method by Cox and Weeks (1983) and Leppäranta and Manninen (1988)is used to calculate the volume fraction of brine inside sea ice. Whether it is also suitable for the slush layer?
• Snow evolution off Zhongshan and Davis Stations in Prydz Bay: According to observations, negative ice freeboards and slush layers are rare in this regions because of strong katabatic downwind. See also:
• Li N, Lei R, Heil P, Cheng B, Ding M, Tian Z, and Li B. 2023.Seasonal and interannual variability of the landfast ice mass balance between 2009 and 2018 in Prydz Bay, East Antarctica, The Cryosphere, 17, 917–937, https://doi.org/10.5194/tc-17-917-2023.
• Lei R, Li Z, Cheng B, Zhang Z, Heil P. 2010. Annual cycle of landfast sea ice in Prydz Bay,east Antarctica. Journal of Geophysical Research: Oceans, 115(C2), C02006, 1–

Thus, is the data of the ZS-2010 buoy representative？This buoy observed a negative ice freeboard because it was deployed near an iceberg. In addition, in the observation footprints of satellite remote sensing, the vicinity of this buoy should include signals from icebergs and land.

• Line 405 “Essentially, this means that the conditions observed at the buoy location are representative of the entire grid cell, ensuring that the satellite Tb data is a valid proxy for the conditions across the whole floe”

In fact, the surface brightness temperature at the floe scale depends not only on air temperature and lead distribution (or ice concentration), but also on the heterogeneity of snow and sea ice thickness, the latter of which undergoes significant changes at the floe scale of several tens of meters.

• Line 458 “we conduct Tbs simulations following the trajectory of the ZS-2010 buoy”

In fact, this buoy is deployed on landfast ice and would not drift, with the small distance from the shore.

• Figs. 4-6: Do you also consider using surface temperature observed by buoys to compare/verify brightness temperature?

Review of

Quantifying the Influence of Snow over Sea Ice Morphology on L-Band Microwave Satellite Observations in the Southern Ocean

by

Zhou, L., et al.

Summary

This very nice manuscript takes a look at the sensitivity of L-Band brightness temperatures as observed by satellite sensors such as SMAP and SMOS (and in the future CIMR) to snow properties on Antarctic sea ice. The focus lies on changes in the snow properties with respect to processes involved in sea-water entering the ice-snow interface, causing basal snow layers to become moist or wet, and saline, being a precursor for a slush present at the ice-snow interface and subsequent snow-ice formation. Since such L-Band radiometer measurements have been used for sea ice thickness retrieval of sea ice below about 70 cm to 100 cm thick in the Arctic, bearing also potential to estimate the thickness of the snow cover on sea ice, such work is really important. The authors used an impressive suite of various independent observations (in-situ and air-borne) to support their investigations based on numerical modeling of sea ice / snow properties using SNOWMACP and radiative transfer modeling using RADIS-L in two different versions. The manuscript is in general well written and provides a red line to follow through the comparably complex modeling tasks carried out. The presentation and interpretation of the results is well organized and bear sufficiently novel information to warrant publication.

I have three general comments where I am convinced the authors need to work on before final publication, though. The authors will also find a suite of specific comments and quite a number of editoral comments and suggestions.

General comments:

GC1: A lot of your model simulations are based on sea ice with a thickness of 2.5 m. Also the sensitivity studies carried out are mainly based on thicker ice categories. I was wondering how representative your results are in the context of Antarctic sea ice usually being considerably thinner than this value. Typical values for most of the seasonal ice cover (e.g. at least 80% of the annual sea ice cover) are between 0.5 and 1.5 m, aren't they?

GC2: I am not sure I can follow well enough your motivation for the density and salinity values chosen for the slush layer. I find these not overly "slushy". I see slush as a very wet, more or less saturated layer of a mixture of ice crystals, liquid water and brine - with almost no air - if at all. Hence I would assume the density is as least as large as the one of first-year ice - simply because of the high content of liquid water. I (and the reader) shall be grateful to learn from you, why you considerably divert from this slush concept.

GC3: My impression is that quite a number of the causalities that are presented but also of the explanations given in the context of interpreting the results would benefit from revisiting the physical principles and taking another look at existing literature to sharpen the formulations and avoid misunderstandings.

Specific comments:

L1: "refrozen snow-ice" --> "snow ice", because this is per definitionem frozen, right?

L26: "thin ice" --> I suggest to drop the "thin" here as it can be mistaken as the WMO's classification where thin ice is sea ice of less than 30 cm thickness.

L28: Again you use the term "refrozen snow-ice". My understanding is that snow-ice is refrozen slush that once formed during a flooding process. Snow-ice is per se frozen - otherwise it would not be termed ice. I therefore suggest to drop "refrozen" whereever you write about "snow-ice"

L30-32: "this flooding .... on the top" --> I suggest to sharpen this statement. Is flooding just "accompanied" by the processes mentioned? Or are these processes perhaps rather a pre-requisite for flooding to occur?

- I suggest to add "of sea ice" behind "melting"

- I suggest to change "redistribution and precipitation of snow" into "precipitation of snow and its redistribution by winds"

L33: "roughly one-third" -- I am doubting that we already have a circum-Antarctic quantification of the amount of snow-ice based on observations. I therefore suggest to be a bit more careful here and write "up to one third".

L35: "will reassert the balance and produce a zero freeboard"  

i) Which balance? Please be more specific. 

ii) The freeboard you are refering to here is the "sea ice freeboard" and you should write it as such to distinguish it from the total (sea ice + snow) freebaord.

iii) I am not sure I buy the "zero" because slush might be spread on top of the sea ice both below and above the water line and the refreezing of it might turn the entire slush layer into snow-ice. Paired with brine drainage and the resulting decrease in density the sea ice with the new snow-ice layer on top is likely to be lifted out of the water a bit plus, as I wrote above, there might be several millimeters if not centimeters of slush that are above the water line that are refreezing as well (possibly these are the parts that freeze first). Therefore, the resulting sea ice freeboard might very likely not be zero but slightly positive.

L40: Isn't ice porosity rather related to its air content while here you are talking about the increased permeability due to the widening of brine inclusions and channels?

L42: "brine drainage channels" --> Have such channels been observed in snow on Antarctic sea ice? It would have thought this is a feature of sea ice only.

L43: What I am so far missing in this paragraph is i) the process of surface melting, percolation of melt water and formation of (fresh) snow-ice at the bottom of the snow pack and ii) rain-on-snow events that can cause any kind of snow property variations - including surface glaze. Most of these you mentioned already in the sentence ending here but you did not connect these well to the processes.

L57: I believe you could (and should) increase the list of sea-ice quantities which retrieval from satellite remote sensing data is influenced by these property changes by i) sea ice motion (e.g. Lavergne and Down, https://doi.org/10.5194/essd-15-5807-2023) and ii) sea ice type (e.g. Melsheimer et al., https://doi.org/10.5194/tc-17-105-2023)

L99: "above the snow/sea ice interface" --> For the initial installation of the snow buoys this is correct but later on the snow-ice interface migrates towards the ultrasonic sensors due to snow-ice formation. It might make sense to be more specific here in this regard therefore.

L102: "from ... 2017" --> Really? You only used data from two specific days? Or did you want to refer to the period April 30 2016 to January 1 2017?

L127/128: "Vertical .... (2017b)" --> Compared to other publications the degree of detail provided here is relatively low. My questions are: 

i) were the samples used for the density also used for the salinity measurements or were additional samples taken? If so, how close to the density samples and at which vertical spacing?

ii) How large was the snow volumne that was melted? At which temperature did the melting occur? At which ambient temperature where the salinometer measurements carried out?

iii) What type of a salinometer was used and what is its measurement accuracy and sensitivity?

iv) As you described yourself, the snow cover can be very heterogeneous. How did you treat visually discernible strong gradients in density such as caused, e.g. by an ice lens buried in the snow?

v) How did you treat slush at the snow/ice interface if there was any?

Weren't there any temperature measurements?

There is also a typo: "now" --> "snow"

L130-135:

What is the motivation to take samples from earlier cruises (2004, 2006) into account?

How were the layers defined? Visually or my means of the density observations described further up?

Layer hardness was recorded using standard measures, I assume?

L143/144: "Snow density and salinity ..." --> Were the devices used here of the same type as for the measurements carried out during the Polarstern expeditions? How about the measurement accuracy and sensitivities? How about the means by which snow stratigraphy and grain size were observed. Were these also the same as described above?

L147-151: Please review the instrumentation that was actually used during the OIB flights in the Antarctic over sea ice. I am not convinced that LVIS (or other lidar / laser devices) was used. I would think that the main products derived by Kwok et al. (2012) and available over sea ice from these campaigns rely on observations of the ATM and of the snow radar. Giving a bit more information about these instruments would be good - also, for instance, what the flight altitude has been for the Antarctic OIB flights.

L152-162: I was wondering whether it would make sense to write a bit more about the accuracy and the limitations of these observations here. After all, OIB measurements are not the truth. There is a snow radar resolution based bias for thin snow covers which are under-represented in the measurements and there are substantial difficulties encountered of sea ice with complex surface topography (i.e. ridged ice).

L168: Where did you get the L3B SMOS TBs from and which processing version has been used for these?

L173: Ok, now you have introduced two SMOS TB products but which of these did you actually use?

L184/185: What is the projection used by the ASI SIC product? Please add this information.

L191/192: "missing snow value due to leads and melting snow" 

i) It is not clear where the "melting snow" information is coming from. Does the NSIDC product contain a flag? Is this also a 5-day running mean? 

ii) It appears a bit strange that you use a coarser resolution product (12.5 km) to detect the influence of leads on "missing snow (depth) values" when you are using a much finer resolved product for the SIC. Please consider re-phrasing your wording here.

L194: "To intuitively ..." --> I strongly recommend to back up this "intuition" by citing respective published literature. Since I am not sure what you mean by "small-scale ice surface variability" I cannot assist you further here. But it seems important that you inform the reader in more detail about the scale you are refering to here (millimeter, cms, meters?) and why you want to see which feature in the SAR images. Also, why is this intuitive? Wouldn't OIB ATM measurements be much more intuitive as they provide high-resolution information about the actual surface topography?

In short: It is not entirely clear what the aim of including ALOS PALSAR imagery is.

Besides this: This is data from the first ALOS, right? Did you consider to use ALOS-2 PALSAR data? They might overlap with many more of the other data sources you are actually using. e.g. SMOS, SMAP, the OIB flights, the buoy observations and so forth.

L221: "calibrated negative freeboards" --> not clear what is meant by "calibrated" in this context. How were negative freeboards calibrated? And against which source are these calibrated?

L236-237: What about the ocean heatflux? Is this set to zero?

You introduced SNOWPACK as allowing to handle several snow AND sea ice layers. You did not comment in your description about the ice layer(s) you used in SNOWPACK. Did you use a one-layer ice slab?

L245: Maass et al. applied this model to Arctic sea ice only, didn't they? How compatible is this for the Antarctic?

L284: "only valid ... -3degC" --> Where does this threshold come from? Is this also from Geldsetzer et al. 2009?

L329: I note that taking a thickness value for sea ice of 2.5 m is a rather high value for typical Antarctic sea ice (see GC1).

L332-334: I am a bit surprised about the choice of the air content of the slush - and also about you choice of slush density. Isn't slush basically a high-density mixture of freshwater ice crystals, brine solution, and liquid water with an even higher density than, e.g. first-year ice? Are there really no slush density estimates in the literature from which you could have taken an actual value rather than assuming a density of snow composed of ice layers? This kind of snow potentially contains some air pores but no liquid water or liquid brine as slush does (see GC2).

L347/348: "... depth hoar ... began forming due to rain and higher air temperatures." --> I am not convinced that higher air-temperatures contribute to the formation of depth hoar while at the same time snow wetness increases. Isn't depth hoar formation rather related to strong temperature gradients within the ice-snow system / large humidity gradients within the snow, triggered by particularly cold air-temperatures? Please check the available literature. (GC3)

L386/387: Isn't wind slab not also a form of the snow cover that is simply formed by the action of the wind - without precipitation events, i.e. from wind-induced snow redistribution?

L393: Aren't brine drainage channels features developing during the formation of sea ice - mainly during columnar sea ice growth? I would then rather change the formulation towards that brine drainage channels are widened.

I was also wondering whether the flooding through such channels is really the main mechanism through which sea water enters the ice-snow interface for Antarctic sea ice. Isn't the role of lateral flooding, i.e. from the floe boundaries, and flooding through cracks in the ice cover the more prominent mechanism, given the fact that there is considerably less columnar sea ice growth in the Antarctic than in the Arctic?

L394: The capillary wicking of moisture into the basal snow layer mentioned here requires that there is some liquid water already available at the ice-snow interface; it appears hence to be a consequence of whatever flooding process that might have happened before?

L451/452: How large is the land influence on the SMAP / SMOS TBs measured in Prydz Bay in comparably close proximity to the coast?

L469-471: Just a comment: These results using the ASPeCt data are kind of surprising because the ASPEeCt observations are just estimates and are certainly much less realiable than the buoy data with respect to the sea ice thickness and snow depth values.

L507-511: These are lines of a more generic, summarizing character that might better be placed into the discussion of even into the conclusions. Try to avoid repeating the same message several times.

Instead, what might perhaps be more interesting to learn - in the context of Fig. 7 - is why the agreement between RADIS-L v1.1 modeled TBs and observed TBs is much better for Ice Stations 4 and 6 than for the other three stations? If you have written this somewhere else, then I am sorry for my oversight.

L522: "The overestimation of sea ice concentration is directly observed ..." 

I suggest to re-phrase this statement, because the ALOS PALSAR image does not provide credible enough information about the sea ice concentration itself. The fact that you see a mode in the backscatter values at lower values is probably an indication of leads but whether these are open water or covered with thin ice remains unknown at this stage. Hence the sea ice concentration in all PALSAR images shown might be pretty close to 100% within the sea ice cover. You might also bear in mind please, that comparably high backscatter values could be caused by wind roughened open water. While you might not find that in the scenes shown, I recommend to be less blunt with statements such as "averaged -13.1 dB, indicating sea ice" in Line 528. There are other issues you could comment on, that are much more clearly visible in the PALSAR images. Basically all these images show a very clear transition between brighter and darker features in the backscatter. I recommend that you check the literature whether the contrast in L-Band backscatter observed here could have to do with i) surface melting, ii) subsurface melting (you might actually see a wet ice/snow interface at L-Band because it penetrates the snow cover), and iii) ice type (especially the images of Oct. 30 show nicely a bright backscatter area kinf of hugging the coast of the Antarctic peninsula which could well be the signature of perennial ice, while further east you find seasonal ice).

L529-535: You might need to re-phrase this paragraph after having worked on the previous one.

L539-541: In view of Fig. 9 I was wondering whether you also tried to separate the influence of theta_a and theta_w? How do TB values change if you only vary theta_w? How do they change it you vary theta_a? How would you be able to explain the changes modeled?

L542-545: I have to admit that I am surprised to see that the overall reduction in TB between dry snow and 80% of the 0.5m thick snow layer consisting of brine wetted snow is only about 6K. Does this make sense in view of the theory? I am also surprised to see that an increase in the fraction of brine wetted snow actually results in a decrease in the modeled TB. Intuitively, since an increase in brine-wetted snow is associated with more liquid water in the snow (the snow is at least moist, right?), I would have expected an increase in the TB values. Can you explain why an increase in the liquid water content in this case results in a decrease in the TBs? Is this the effect of the salinity? If so, why? GC3

L545-547: "The inclusion ... more akin to ice than to snow ..." --> I get the impression that the TBs change into contrasting directions.  On the one hand, without a slush layer, TBs decrease with an increase in brine-wetted snow layer thickness (and hence net total amount of water in the snow). On the other hand, with a slush layer, a larger water fraction Theta_w results in a TB increase - hence exactly the opposite development. Why? 

I also note that the increment by which the TBs increase, is substantially larger in case of a slush layer present than without a slush layer present (particularly in case of the slush layer with the highest air content and the lowest water content). How can this be explained? GC3

L548/549: "A larger slush depth ..." --> Here you are referring to the non-linear decrease in TBs with increase slush layer depth, right? Do we understand why this decrease is non-linear? What is the physics behind this observation?

L568-570: So the TB decrease is about twice as large for an increase in snow salinity than for an increase in snow density. This is interesting. I was wondering, however, why you begin with a snow salinity of 2 g/kg? I was also wondering to which degree the ranges you used are representing typical conditions. Finally, I was wondering, how much the snow salinity is decoupled from the snow moisture. Are we talking about (completely) dry cold snow? Or could it be that a certain (unknown) fraction of the decrease in TBs is due to snow wetness / moisture? One possible way to answer this last question would be to look into results where the snow salinity is zero but the wetness / moisture is not. GC3

L575/576: I have a problem with understanding the explanation given here. My understanding of the insulating capabilities of snow so far was that the less dense and the drier the snow is the better it insulates. Hence a very fluffy, 5 cm thick snow cover might insulate better than a 20 cm thick, coarse grained, high density (but stull dry) snow cover. However, once the snow cover is wet and/or saline, shouldn't it insulate much less well? Please clarify. GC3

L612/613: "basal snow ... ice-surface flooding" --> Please check the causality here. Isn't it the other way round? Flooding of the ice-snow interface leads to basal snow layers having a high salinity and/or moisture content?

L616: "Interestingly ... land-fast ice regions" --> Really? In which of the land-fast ice regions in Antarctica did you observe ice-snow interface flooding?

L650-659: Undoubtly it is necessary to point out the limitations of the ASPeCt data set. But I have difficulties to understand why you mention in this context "like the algorithm use, satellite sensor, observation technique" in Line 651. These are all visual, manual ship-borne observations from the ship's bridge that should follow the ASPeCt protocol. The largest uncertainties are the observers themselves and the fact that the ships tend to follow sea ice that is easily navigable, hence avoiding thick, compact and/or ridged sea ice as much as possible, ending up in a negative biases in both hi and hs with respect to the "general" conditions.

I am also sure that nobody would use "ship-based measurements for validating Tbs" (Lines 657/658), and I also doubt you did this. I understood your usage of ASPeCt observations as a means to provide more observations you can feed into your radiative transfer model, hence you are simply broadening the data basis. I suggest to condense this paragraph according to your specific usage of the ASPeCt data and the specific limitations that results from that and leave it with that.

L626-634: "In particular ... and retrievals." --> I was wondering whether you could not substantially shorten these lines because it has been well known since the early days of sea ice thickness retrieval using SMOS that small variations in sea ice concentration play a crucial role. Hence you could simply write that the large sensitivity of L-Band Tbs to the presence of open water requires to work with sea-ice concentration data sets of an as fine as possible spatial resolution - such as the one suggested by Ludwig et al (2019) or based on SAR. --> one sentence is enough here, I guess.

L629: How do "the turbulent flux exchanges" influence the "surface Tb values at L-Band?

L641-649: I am not so sure your work points into this direction and I have to admit that these lines are very generic. Of course we need more measurements and they need to be more detailed and we need to do both, field and laboratory studies. But which parameters do we need to observe in a contemporary manner over which spatial and temporal scales with which accuracy to make progress?

L645: "most notably in regions with thinner ice" --> this part of the statement is not backed up sufficiently well by your manuscript since the majority of your results are based on modeling using 2.5 m thick ice.

Fig. 10 a) and b): I am curious how the curves shown continue towards zero snow salinity and a snow density of, say, 100 kg/m3.

I was wondering, whether the snow density shown in Fig. 10b) refers to the snow above the brine wetted layer or to the entire snow layer? If the latter, how realistic is it to assume the same snow density for fresh and saline snow if you are considering brine-WETTED snow? Shouldn't the snow densities be considerably larger for the latter case?

Fig. 10 c) and d): You seem to have chosen a constant proportional relationship between sea ice thickness hi and snow thickness hs. hs is always 10% of hi. Why? Does this reflect actually encountered conditions? How would the violin for hi=2m look like if you would have used 0.6 m snow thickness? How would violin plots for more realistic Antarctic sea ice thickness values of 0.5 m to 1.5 m look like for the same range of snow thickness values?

Typos / editoral comments:

L20: "underscores" --> "underscore"

L37: please check: "capillary action" or "capillary suction". I learned it is the latter.

L46/47: "affecting ... parameters" --> perhaps better: "affecting the retrieval of various sea ice quantities." ?

L53: "Comiso et al., 1997" is focussing on sea ice concentration algorithm intercomparison. While it might mention these processes I was wondering whether there isn't a more specific publication you could cite here in which these processes are detailed from the viewpoint of in-situ observations or radiative transfer modeling results.

L95: "sea ice temperature" --> perhaps better: "the temperature profile in sea ice and its snow cover" ?

L111+ You might want to include the years during which these buoys operated.

L113: ",identified" --> ", identified" (blank missing)

L116: You distinguish between an "acoustic sounder to track the distance to the snow surface" and an "underwater sonar altimeter" to track the distance to the ice bottom. Both sensors work with acoustics and both are operated such that one derives a distance. But whether it is warranted to call one "altimeter" I don't know and seems not common to me.

L117/118: "temperature string" ... so these buoys indeed do not also use a thermistor string?

L138: Since you have described other snow pit measurements already further up, you might consider to begin this paragraph with "Additional" instead of "The"

L169: You might want to add "at 70degrees Southern or Northern latitude" since the actual size of the NSIDC grid cells changes with latitude away from the tangential plane used for the projection.

L190: ".This" --> ". This" (blank missing)

L252: "over the Arctic" --> perhaps better "in the Arctic" or "for Arctic sea ice"

L305: "brine" --> "saline"

L316: "determined by" --> "determined following"

L331: "i, a, w are" --> I guess the epsilon is missing here?

L343: "consistent" --> Did you mean "constant"?

L369-372: "a central tendency defined by a mean value of" --> perhaps better "a mean value of"?

Is the mean value you mention here in fact the "column-averaged value" mentioned in L372? Please clarify.

Is the density value of 396.7 kg/m3 the value that has been computed for the rounded crystals / snow-ice / slush cases? This is not entirely clear from your writing.

L382: "frequencies" --> I suggest to try to find a different expression here because what you are describing in this subsection is the vertical distribution or variability of the snow stratigraphy.

L385: "region(Massom" --> "region (Massom"

L395: "Thus, .." --> Why "thus"? Did you mean "subsequently"?

L397: "ICE" --> Why do you use capital letters here?

L398: "However ..." --> perhaps better: "In addition, ..."

L411: "declining trend in Tbs" --> I suggest to write either "negative trend in Tbs" or "decline in Tbs" because a "declining trend" suggests that the trend value itself is decreasing.

L415/416: "nearly constant ice concentration approximating 100%" contradicts "increase in open water" --> please check and correct your writing.

L500: "contrasted" --> Where is the contrast here? If this was the maximum median snow depth then you might write so.

L502: "more substantial"? --> perhaps better: "larger"

L503: "with and devoid" --> "with and without"

L504/505: "When compared to ... consistently demonstrated an overestimation bias of 8.8K" --> perhaps more simple: "Compared to ... are biased high by 8.8K."

L518: "incorrect sea ice concentration value" --> I suggest that you quantify this better by stating whether the sea ice concentration was too high or too low.

L546: "and lower air content" --> you could add "and hence higher density"

L572-574: Just for my clarification: By this percentage you mean a larger fraction of the snow cover that is composed of brine-wetted snow, right? You are not referring to a higher brine volume fraction.

L581-586: Please check these lines. Something seems not to fit well in the context of the "However, Further, ..."

In this context: Did you think about that the thinner the brine-wetted snow layer is, the higher is the likelihood to receive a signal from the sea ice underneath?

L598: "ice thickness deepens" --> I guess a snow layer can deepen (even though I like to talk about snow thickness as it is simply the vertical extent between the snow surface and the underside of the snow and hence similar in definition to that of the sea ice thickness) but not an "ice thickness". So maybe rather write "increase"

L602: Why "dramatic"? While we know a lot about the Arctic sea ice thickness from submarine and moored sensors in addition to the satellite observations a thickness decrease is present, yes, but I would not call it "dramatic" - especially during the past 5-10 years when, for instance, the PIOMAS time series of the Arctic sea ice volume shows a plateau of stagnating values rather than a continuation of a decrease. And for the Antarctic, we know much less about the past sea ice thickness distribution and perhaps should not come up with adjectives like "dramatic". What we do know is that the Antarctic sea ice cover is substantially more variable than the Arctic one.

L614: One ")" can be deleted.

L618: By "changes in the depth" I would understand changes of at which point, when measured from the snow surface, the region of brine-wetted snow begins. But what I guess you want to say here is "changes in the thickness and/or vertical extent"

L623: By "reanalysis-driven" you refer to atmospheric re-analyses? Not entirely clear.

L625: "radiative properties of ice surfaces." --> Suggest to add: "at L-Band frequencies."

L626: You are referring to the modeled Tb values here? Then I suggest to add "modeled".

"surface Tbs values" --> "surface Tb values"

L637: "properties" --> please mention which properties you refer to here.

"depth" --> "thickness"

Fig. 3 caption: What is a "white dost"?

Why do you write "ICE crust" instead of "Ice crust" at the x-axis annotation of panels a/b) and d)?

"Decomposing and Fragmented" in d) is missing in a) and b)? Also, did you mean "Decomposed"?

Fig. 6: If panels b) and c) are heat maps (please correct the caption) then you need to provide a legend which translates to color into counts.

Fig. 6 caption: I suggest to write "ASPeCt observations" instead of "ASPeCt measurements".

Fig. 7 left y-axis: "freeboard thickness" does not make sense. Please correct accordingly into "Sea ice/snow thickness & sea ice freeboard" as this is what you want to refer to. 

Also the first line of the caption needs "thickness" to be added behind "sea ice" and "snow" plus "sea ice" to be added in front of "freeboard".

Fig. 8: It might be a matter of taste but I would prefer to have the images of the earlier date to the left (Oct. 29) and those of the later date to the right (Oct. 30).

Fig. 9: The legend within the figure lacks the line for 80% percentage of brine-wetted layer.

I suggest to make clear in the caption that the overall snow depth used is 0.5 m and that the "slush layer depth" is a "relative slush layer depth" or perhaps even better a "relative slush layer thickness".

Fig. 10 caption: 

(a) and (c) is salinity and (b) and (d) is density - contrary to what you write in the caption.

There are no "e" and "f"

Fig. B.3: The y-axis is slightly misleading. This is not the "snow height".

Fig. B.5: How many data points are shown here? Would it make sense to turn this into a heat map?

Figure B.6: Which AMSR-E snow depth product is shown here?

In the caption you write "SMOS Tbs" but actually you seem to show both SMOS and AMSR-E Tbs; hence in the first line of the caption it needs to read "...between simulated and observed Tbs".

Fig. B.7: What are the tracks in blue-white-red denoting?

Since you write the units of the parameters shown below the columns of panels you might not need to repeat the units in the caption. How can a "net precipitation" be negative? Is this perhaps E-P? I note that the scale of the legend of this quantity is not well chosen because large regions are shown with a saturated color. You might consider to change this.

L907/908: Reference needs to be Studinger, N. K., ....

Summary

Flooding and slush over Antarctic sea ice have been a major reason to cause errors and uncertainty for snow depth and/or sea ice thickness retrieval from Satellite remote sensing for southern oceans. The purpose of the paper hopes to improve snow depth and ice thickness retrieval from passive microwave (1.4 GHz) brightness temperature via satellites such as SMOS and SMAP, by improving the existing RADIS-L v1.0 to v1.1, by introducing brine-wetted snow and lush layers over Antarctic sea ice.  The modeling results are validated based on three types of datasets: 4 buoys, some ASPeCT data, and OIB data. Very surprisingly, the ASPeCT data actually gave the best results, only 1 out of 4 buoys gave reasonable results, validation from OIB seems not as good as stated. They attribute the biases between modeled and observed Tbs mostly to sub-grid scale ice surface variability and snow stratigraphy.  One of the results is that the adding of a slush layer into the model decrease the modeled Tbs. This is obviously true. Anyway, I agree this is very important topic of study, but the paper needs to prove that the v1.1 is indeed effectively modeling the wet and slush layers for Tbs, and in my opinion, it is not the case in its current format.

General comments

The paper in somehow confuses me on the brine-wetted and lush layers. I hope they can make corrections throughout the paper, to brine-wetted snow layer and slush layer, since slush layer is not snow layer any more. Slush layer is a mixed snow, ice, and water, more close to water in my opinion. And the slush layer does not exist too long (not sure how long, may be a few hours?) before refreezing to become the ice (or snow-ice).

Also need to make sure the improvement of the modeling from v1.0 to 1.1 is only adding the “slush layer”, since the v1.0 already has the brine-wetted snow layer. Right?

The paper in somehow confuses me about active and passive microwave sensor. From the title, I thought it uses radar data, but clearly the paper did not use radar data. see my comment on L166, I suggest to change “L-band” to either “passive” or “1.4 GHz”. If this is agreed, you should do it throughout the entire paper.

Specifical comments

Line 7, change to “brine-wetted snow and slush layers”

L35, should be “zero ice freeboard”

L127, do you mean “vertical snow salinity”?

L166, “this fully polarized sensor….” is wrong. I do not think it is fully polarized. Since SMOS is passive and it is either H or V, not HH, HV, VH, or HH. Also my comment about the title. L-band is mostly for active sensor, for passive sensor, we usually use frequency.

L199, should be “horizontal transmit and receive polarization”, not the other way as here. Remember if it is active, it transmits first, then receives.

L204, I do not know why you want to use 35 degree, not the real incidence angle for each pixel? I think SNAP can do it by directly use the real incidence angle of each pixel.

Section 3.1 SNOWPACK. Is this RADIS-L v1.0? if yes, should you change the section to it? If not, what it is the difference between them?

L243, should “X- and L-band” be the relevant frequency of them?

L282, equation 4, the surface Temperature Tsurf, I believe it is more controlled by the air temperature, rather than as provided by the equation.

L305, “Therefore, we treat the snow slush …..”, please check this sentence and make sure it is the right statement? I am little confused.

Section 3.3, wondering why you want to use 30-50km, not use the 9km for SMAP and 12.5 km for SMOS?

L353-354, when there was heavy snowfall, it will general flooding, slush, then slush freezing to snow-ice; this entire process should increase ice thickness not decrease ice thickness. I do not understand your statement here.

L402-406, this is hard to align the satellite pixel with the buoy location, given the buoy collecting continuous data while the satellite only gets one data per day. Also the buoy (ice floe) is moving. Not sure how you process your buoy data to match the satellite data? You average all data for the day or you only use the data collected when the satellite passing? Maybe averaging these data a few hours before and after the satellite passing?

L407-408, this is a good statement but it is only true for 1 out of 4 buoys. See figure 4 (a). this makes me to wonder something is wrong. This would be due to my comment on the line 402-406, or other reasons we do not know. I believe you need far more example than these 4 and may be also expand to other sections of the Antarctic, currently 3 in Weddell sea. The one not in Weddell sea is a fast ice which could be very different from the pack ice situation. And I am not sure if the flooding and slush layer would happen at all for fast ice case. I take it back, from figure 2d, it shows negative freeboard. But please check if this is indeed.

L415-416. It is true “Tbs reduction from 243.8k ….to 226.1K”,  but it also involves with a few increases and decreases in between. Can you explain that.

L458, if ZS-2010 buoy was on the fastice, why you talk about trajectory? Please explain here.

L462-467 and figure 4d, your statements seem be suspicious. I see v1.0 and 1.1 have the same patterns but different magnitudes. Compare with satellite Tbs, however, there are a few cases, with inverse change. R2 of 0.35 does not mean a good relation, in my opinion.

Section 4.2.2, validation with ASPeCT shows better results than the buoys. In line 488-489, please explain to me what you mean small-scale? What kind of parameters in small-scale would increase your modeling results? In term of SAR imagery, it will not give Tbs.

L492-493, not sure what you mean “This analysis is …. Near Wilkes Land”?

L495-497, not sure where you mentioned ice stations 2,3,4,6,7? They are in the figure 1?

L565-569, mentioned about 2.5m ice and 0.5 m snow for the experiment, I am not sure if this snow depth would cause flooding and slush over ice? It may be but I am not sure. but figure 10 shows 0.2m snow with 2m ice, 0.3m snow with 3m ice, etc.., I am pretty sure these cases would not cause flooding and slush…; please explain. Also as you know and talked about that most of the Antarctic sea ice is thin ice. Your ice here are 2-6m, which is very unnormal. I would suggest to model ice thickness from 0.5 to 3m, the maximum.

L657-658, “Given these considerations…”, this statement is very interesting, since from your results, the ASPeCt data seems gave the best validate of simulation. Please explain.

Conclusions section includes a lot of contents that are not really from this paper, many of them could be included in the Discussion section, some of others are very suspicious statements. For example, line 642-643, I do not see where you discussed this and how you come to this:  “The scenarios we have discussed …in various settings”.

Figure 1. why you only show three layers, not the fourth layers that you claimed to improve in the v1.1?

To me this is for the v1.0, right?

Figure 4, very similar pattern between v1.0  and 1.1, but really not that match with the observations, except for the case in (a), i.e., the buy 2016S31.

Figure 8, I do not see “the differences between simulated and SMOS Tbs …”, where are those Tbs in this figure?

Figure 10. I hope to see the same value range of 242-256 for all y-axis of a, b, c, d.