A review of “A new dataset of Southern Ocean sea-ice leads: First insights into regional lead patterns, seasonality and trends, 2003–2023” by Dubey and coauthors.

This is a well-written, detailed and generally well-supported characterization of leads in Antarctic sea ice, and is an aggregation and subsequent quantitative analysis of the underlying dataset of daily leads from Reiser et al, 2020. I find that the paper is well-structured, and that the content fits well within The Cryosphere (as opposed to ESSD). Furthermore, the content is of great interest to those of us trying to disentangle the rollercoaster ride of Antarctic sea ice over the past decade, and places many other works in new context. 

I have only one major comment, detailed below, regarding Section 3.6. All others can be regarded as quite minor, and I think the author team will deal with these without too much trouble.

Major:

The whole of Section 3.6 is a little perplexing. If my understanding is correct, based on the description presented, the following procedure is followed to determine regional trend magnitude/significance:

1) Each grid cell, in each month, has 21 data points (one point per month over the 21 year dataset)

2) In each grid cell, the linear trend is calculated from these 21 points. 

3) For each grid cell, the trend is either determined to be “significant” or “insignificant” - based on what criteria? A “5% level” is mentioned but this isn’t really sufficient information to replicate your method. Is a linear regression calculated How many degrees of freedom used? Two tailed or one tailed? Etc.

4) Then these grid cells are aggregated into regions - but only significant (positive or negative) pixels are aggregated? It’s a little baffling why perfectly valid (but insignificant) pixels are left out of this calculation.

5) Then in Table 1, it’s not clear what a value of “-” means. Is this 0? Or are you taking 0 to also mean insignificant, and also reporting that as “-”? In that case, it’s very suspicious that a value of 0.1 just happens to coincide with your threshold of significance.

Basically I’m not convinced that leaving out insignificant pixels is the right thing to do, and not convinced that all values in Table 1 are significant trends. 

Minor, referenced by line number where possible:

Abstract/Intro: 

7 - recommend change “can be inferred” to “are shown”

8 - lots of “rather” in this sentence

9 - I think Reiser et al 2020 could probably claim “first results on the spatial….” - reword to make the claim accurate. 

18 - an inconsistency in the use of a hyphen between “sea” and “ice” where it precedes a noun. E.g. if you use “sea-ice leads”, as you do, then you should use “sea-ice drift” and “sea-ice deformation”. This is inconsistent in many places.

21 - regarding the citations for ice-albedo feedback, I recommend to cite an Antarctic-specific example - perhaps https://doi.org/10.1029/2005JC003447

32 - the use of “winter” is a little too unspecific in many places of this manuscript. April isn’t exactly a winter month.  Also line 108, 123, etc...

35 - deduct isn’t quite the right word - is “deduce” better?

35 - this sentence should make clear that it’s an aggregation of the Reiser dataset, and also give the months in which the technique is valid.

Introduction in general - it’s not actually clear, especially for someone unfamiliar with the Reiser dataset, what you’re doing. I think the introduction would benefit from more description of the Reiser dataset. E.g. what instrument? Aqua only? Or Terra too? What are the drawbacks of just using the Reiser dataset (without the aggregation that you present here)?

Data and methods:

49-50 - are these anomalies in space, time or both?

52 - this is the first mention of the April - Sept data validity - I think it needs to be done in the intro.

57 - please give some ideas of other datasets to combine.

61 - “annual” is only April - September, right? 

67 - I think “averaging” is more correct than “integrating”

75 - there may be nothing that can be done within the style guide, but that subscript r is so small!

76 - You take anomalies from the long-term mean? Does this mean all months from April to September as your baseline? Wouldn’t it make more sense to just choose the same month as your baseline?

78 - sentence beginning “This approach” is hard to interpret. Do you mean “not to avoid…”? 

89 - This section would benefit from first stating your motivation for looking at surface currents. E.g. “Surface currents are the main determinant of leads, so….” 

94 - This CPE technique isn’t quite straightforward, and also no reference is provided. 

96 - “study is” to “study are” (data is a plural)

97 - DBM is never used again, so don’t define the acronym

Results

107 - “most dominant” is subjective - reword

116 - a publication showing the fast-ice long-term mean distribution would be important to reference here - https://doi.org/10.5194/tc-15-5061-2021

121 - MIZ is defined here but would be better to define 2 lines earlier.

124 - this sentence structure (sentence beginning “Areas, ..“) sounds unnatural. This structure is used in other places too (e.g., figure captions). 

137 and Fig 4 - Somehow it’s a little strange to me for this figure to be presented as an anomaly. First of all, what’s it an anomaly from? I guess Fig 2a, right? Secondly, I kind of prefer to see this figure presented as a mean for each month, rather than anomaly from the mean. Without seeing the means, it occurred to me that the authors maybe already explored this and decided the anomalies are more meaningful or easy to interpret. That’s fine, if that’s the case, but for baseline characterisation I find that means can sometimes be more important. Just a thought. I have no problem with the maps in Fig 3, and somehow in my mind it’s clear that Fig 3 should be presented as anomalies, but somehow it doesn’t seem as natural for the baseline monthly figures in fig 4.

156 - unnatural sentence around “months, where”

Fig 5 caption - can you state the colours used for months in the caption? It’s a bit cramped and I think these would help.

178 - Fraser et al 2009 (https://doi.org/10.1109/TGRS.2009.2019726) showed that leads often coincide with an erroneous “cloud” determination from the MODIS cloud mask (see their Fig 6). Is this perhaps what you’re seeing here? Does the Reiser dataset rely on the MODIS cloud mask? 

208 - a good idea to reference the map of long-term mean fast-ice persistence from Fraser et al., 2021 (https://doi.org/10.5194/tc-15-5061-2021)

210 - “inverted fast-ice frequencies” - I see what you’re saying but this might be overstating it. E.g. your plot in Fig 8b, right column shows the climatology of lead frequency increasing throughout the winter, but fast ice also increases throughout the winter in this region. 

222 - missing an accent in “Adelie”. NB also typo in Fig 7, box 4 “Adleie”.

243 - I feel like Maud Rise is commonly-enough known to stand without a lat/lon, but Gunnerus Ridge probably isn’t.

253 - “These anomalies are superimposed…”  not really - Fig 4 is *also* anomalies, not “general seasonal patterns”.

Leads is an important indicator of Antarctic sea ice deformation, and their changes reflect the vulnerability of Antarctic sea ice. However, it is a pity that the relevant research has not attracted enough attention, partly because such a small-scale phenomenon is difficult to detect or simulate. This paper using the latest high-resolution leads detection data push the research in this field to make significant progress, making me exciting. For the first time, it reported regional and seasonal trends in the Antarctic leads. I think it would be important circumstantial evidence that Antarctic sea ice is weakening under the dramatic climate change of the last two decades. I also sincerely commend the authors for their willingness to share leads dataset.

I believe the conclusion of the article is attractive enough, but a few changes are needed.

1. This paper claims to have developed a new dataset, but the improvement over Reiser's work is very limited: it seems to have only filled in missing values with multi-year means and reduced the resolution to monthly. Therefore, I think it is overstated to emphasize this point in the title and abstract.
2. Figure 6 shows that there are many misses after 2020, but this period contributes to most of the increasing trend of East Antarctica leads, which weakens the reliability of the trend and would shake the foundation of this article. It is necessary to check the code and recalculate the leads frequency during this time. There are missing AMSR-based sea ice concentrations in 2011-2012, and the authors do not explain how they dealt with this period.

Some small comments and presentation issues are also worth noting.

1. Section 3 needs to be better organized, especially Section 3.5 and the following paragraphs. Analysis that is redundant and jumpy with the previous description is confusing. The authors need to find a clear logical thread to rearrange how to present these results.
2. Line 80: What’s the meaning of “absolute” pIE?
3. Line 126: For color blindness, please don't put red and green together. You can replace green with cyan. The same goes for the other figure.
4. Line 160: Is the trend significant? It seems that many of the trends in this paper lack p-values.
5. Line 211: Please explain "pronounced regional differences" in detail and explain how leads and fast ice are physically related.
6. Line 233: Given the large uncertainties in ocean current simulations, I recommend that you validate your results using multiple analyses/reanalysis.
7. Figure 7. This figure does not seem to have any difference from Figure 2a. This one looks a bit prettier, so you can keep just this one.