Publisher’s note: the content of this comment was removed on 24 January 2024 since the comment was posted by mistake.

This paper presents a nice technique for providing much-needed measurements.

Technically this is one of the cleaner papers I have reviewed. There are a number grammar and style issues, which comprise most of my comments. While largely optional, the paper’s readability would be improved by applying many of them. The authors may want to consider using something like Grammarly in the future, which would have fixed many of these minor issues.

A minor point is I would like to have seen more detail on the chip sizes used to do the speckle tracking, which may affect the biases (or differences in measurement techniques – see comment about GL location below).

The NISAR launch should happen in 2024. With the longer wavelength and finer resolution, there will be more glaciers where the phase can be resolved and unwrapped. In these cases, this technique could be applied using the phase in place of the offsets.  A sentence or two making this point would be useful.

As I commented below, it’s worth pointing out that any of these techniques only detect a signal that is the proxy for the grounding line. I don’t know of any work that has shown that the point of actual ungrounding is definitively given by any specific threshold for DiNSAR or this technique. What is most important is that a consistent proxy is used so that apples-to-apples comparisons are made when inferring grounding line retreat.  It would be good to make a point like this.

Minor Comments

Line 20. Since you don’t match the retreat numbers to specific glaciers, remove “, respectively”.

Line 25. In many cases the area above the grounding line is also dominated by longitudinal stresses and vertical shear has little effect. Lateral shear stress is often as if not more important than longitudinal stress in many cases on both sides of the grounding line. Maybe rephrase as the transition from the region influenced by basal shear stress to the region with no drag.

Line 33: Here an numerous other places hyphenate “ice-sheet” when its used as an adjective “ice-sheet mass loss” but not when used as noun, “Antarctic Ice Sheet”.

Line 37 Add “,” before “which”

Line 41 “and the extent” break the sentence in two here.

Line 43 “,” before “which”

Line 46 “this point is followed”

Line 55 “ice-flow”

Line 57 remote “state” the ice flow may be steady (the same over two periods), but not necessarily steady state (i.e, if the flow is causing thinning).

Line 58. Break start a new sentence at “, with the remaining” and add “differential” before “vertical”

Line 62. The data can be coherent but not suitable for mapping the GL. For example, the data can be well correlated but if the fringe rate exceeds the spatial sampling, the fringes are aliased even though they may remain coherent.  Maybe say something about the data should be adequately sampled so that the fringes can be resolved.

Line 71. A better reference than Joughin et al 2010b is Joughin et al 2016  https://doi.org/10.1002/2016GL070259

Line 78. Maybe say “slope-related shading effects”

Line 81. Strictly speaking Goldstein did not use DiNSAR since he only had a single interferogram. It happened to be one that with a relatively simple flow field so that GL was visible.

Line 103: I think you mean “formerly” not “formally”

Line 108: I think “sparse” is spelled the same way on both sides of the Atlantic (not sparce).

Line 110: Add a “,” after “methods”

Line 117. In general its good to have at least an introductory sentence between Level 1 and Level 2 heading (i.e., before 2.1 Physical Basis).

Line 121. Don’t start a sentence with an acronym.

Line 122 after “range direction” add “(line of sight)”

Line 122 “, and the speed” start a new sentence here.

Line 124 Don’t’ start with acronym. How about “There are several limitations with DROT;”

Line 125 add “,” before and after “, and hence, “

Line 126. Be careful with terms like “feature tracking”.  You are actually using a combination of speckle tracking and feature tracking here.

Line 126 remove “and”

Line 127 change “…motion,” to “…motion; “

Line 132 add a “,” before “we conclude”

Line 151 add “,” before “and” or better yet make it a new sentence.

Line 155 Consider breaking this long sentence in two.

Line 161 Sentence “Speed measurements are ….” Its not clear what is meant here. Rephrase.

Line 162-164. Remove “We select ….melt season” And start a new sentence first explaining melt effects. “Surface melt between….” Then a new sentence “Thus, we select only SAR images from winter periods” or something like that.

Line 176 add a “,” before “while”

Line 177. I don’t like the sentence “In the absence of noise…”  Whenever you estimate a correlation, you get a random variable, not something that’s precisely 0. Moreover, with no noise, you might pick up subtle speed variations that correlate with the tide. I am more surprised that the threshold is as low as 0.1. So you don’t have to justify not using 0. Remove the sentences referencing 0. And just say something like “We found a threshold of 0.1 give a good balance between false and missed detections of grounded ice”.

Line 186: Add introductory text before 3.1

Line 188: Read better if you say what “this” is. “This data set covers…”

Line 190: “These data” not “this data”

Line 203: Nothing wrong with “map” but how about “detect”

Line 209: Do you have a reference to backup this belief (i.e., most termini are grounded)?

Line 244: Add a “,” before “we”

Line 251 “,” before “which”

Line 263 Change “which” to “that”

Line 270. How about “These ambiguous regions..” rather than just “This…”

Line 274: Not sure why there is an initial in the Rignot citation.

Line 274: Change “This might be expected, as fast…” to “This might be caused because fast…” The issue is cause and effect concerning the glaciers, not what people might expect.

Paragraph with Line 290. It might be good to say with either method a proxy for the grounding line is being used. No one can really say exactly where the grounding line is based on the fringes. Instead, what we have are educated guesses used to define conventions. The bias is not with respect to the true grounding line, but instead to the inland limit of fringes as used by MEASURES.

Line 300: Spell out Grounding line when staring a sentence.

Line 312 “Glaciers’” remove the “’”. It is not needed as written.

Line 312 “respectively” should be “, respectively, “

Line 316 “,” before “we”

Line 340 “which” to “that”

Line 345 Break sentence “… by TMOC. Following a comparison with …. DInSAR, we were…”

Line 360: Add a sentence or two before 5.1

Line 370: With short enough baselines so the phase could be unwrapped, you could apply your technique, which should be an improvement over single DInSAR.

Line 371: Rewrite to move acronym from starting the sentence.

Line 530. Another overly long sentence break into 2 or even 3 pieces.

The proposed research approach is sound, interesting, and well-organized.

However, I am afraid the authors may be, in some ways, overstating the results and the capabilities of the proposed method. For instance when they claim: “the method performs well compared to highly precise DInSAR GL measurements, with a mean offset between these data of 185 m and a standard deviation of 295 m” they asses the presence of a bias in their data, but they do not address the actual accuracy of the technique itself.

For instance, Single DInSAR Inteferogram grounding line measurements have an accuracy of about +/- 300 m when locating grounding lines.

The proposed technique could most likely map grounding lines within +/- 1-2 km accuracy. This is because of the Sentinel-1 pixel size, the accuracy of pixel tracking technique itself, and the inaccuracies due to the non-accurate CATS model near the grounding line together with the coarse resolution of IBE corrections. Finally, this technique assumes no horizontal velocity changes during the observation period which can also affect the correlation with tidal levels.

I would like these numbers to be specified in the text in order to better characterize the applicability of the presented technique.

Most importantly, all the grounding line retreat rates of the same magnitude of the grounding line mapping accuracy (i.e. anything slower than 1 to 2  km/yr retreat)  would lose statistical significance. This should be specified in the discussion and the conclusions.

Further comments below:

Line 43-45 No reference to a figure where point F or G are located.

Lines 72-74 The sensitivity of the pixel offset mapping method also depends on the Radar pixels size. Achievement displacement accuracies are usually 1/8 of the radar pixel size (De Zan 2014)

De Zan, F. (2013). Accuracy of incoherent speckle tracking for circular Gaussian signals. IEEE Geoscience and Remote Sensing Letters, 11(1), 264-267.

Figure 1C is confusing, A straight line does not give the idea of when images have been acquired.

The authors propose a new method for mapping grounding line, evaluate its precision, and apply to the Antarctic Peninsula to detect changes since the 1990s. They report O(10km) retreat in the Peninsula since 1996. The paper is well written, the results are significant, the analysis of error is good, I would however recommend reducing the discussion section a bit to remove many paragraphs of literature review, perhaps trim by 30%. Overall, I would recommend publication after minor revision. I am only curious why the Mohajerani et al. 2021 dataset from year 2018, Antarctic wide, was not used for comparison.

Detailed comments.

Line 36: Not sure I agree with this definition of the grounding zone. The grounding zone defines the range of migration of the grounding line, whereas what is described here is the flexure zone, which is the zone over which ice adjusts to flotation. It would be good to agree on that nomenclature, and I would recommend that the authors adopts this one.

Line 52. One of the earlier demonstration of GL mapping with DInSAR was Rignot, J. Glaciol., 1996 and Rignot et al. Science, 1997; Rignot, Science, 1998. The earliest I know of that used differential interferometry (which Goldstein et al., 1993 did not) was: Hartl, P., K.-H. Thiel, X. Wu, C. Doake & J. Sievers 1994 Application of SAR interferometry with ERS-1 in the Antarctic — Earth Observ. Quarterly 43: 1-4, but this is not a peer review article .. Just FYI.

Line 60. I actually do not agree with this. See Rignot et al. GRL 2011. The inland limit of fringes, the limits where you see ice lift up from the bed is the grounding line. What you mention here at the hinge limit is a mathematical model fitting to the elastic deformation which relates to the point of hinging and is typically displaced 1 km upstream of the GL. I sort of abandoned this point F over time, because the only point that matters is G, and Rignot et al. GRL 2011 describes how it is deduced from Differential interferometry.

Line 61. SAR coherence can ALWAYS be maintained. The true limitation of the repeat pass of the satellite.

Line 68. DInSAR is not used to delineate F, but G. See Rignot et al. GRL 2011. Also see above comments.

Line 71. Did Joughin really map “F”?

Line 95. If you call it “new” method, I need to be sure. Unfortunately, Joughin et al. Geophys. Res. Lett. (2016) used feature tracking on Pine Island Glacier to delineate the grounding line. Please refer to that earlier article by Ian J. and explain what is truly new in your TMOC method, which I think is the time series concept. Correct?

Line 116. Yes.

Line 125. Speckle tracking is 10 times less sensitive to vertical than phase. See documentation of that in Mouginot et al. Geophys. Res. Lett. 2019 (phase map of Antarctica). A factor 10 would speak more than “significantly less”.

Line 134. This applies only if Delta S is in ground range. In slant range, this formula is wrong. Please correct text accordingly to say “ground range”.

Line 159. What reference pressure do you use?

Line 182. You expect .. is this evaluated later on?

Line 208. Who says they are only tidewater?

Line 219. Ref. needed.

Line 225. Please note that Mohajerani et al. Nature Sci. Rep. 2021 produced an Antarctic wide data set for year 2018 that includes the East coast of the Antarctica Peninsula. Why did not you use that dataset, also distributed at NSIDC, for comparison? Or could you add it?

Line 256. I am not sure this is true for the Mohajerani et al 2021 dataset. They include ALL of these areas. Can you please include?

Line 310. Please add error bars on these retreat estimates.

Line 405-444. This is more of a review that a new contribution. Is this necessary? You already covered this in the intro. The Same comment may apply to 5.3. I would recommend cutting a bit into this, remove lit. review and focus on new elements.

Line 540. This should be done PRIOR to acceptance. I am surprised that this journal does not impose this to be done while the paper is submitted. I have therefore, as a reviewer, no access at this information for evaluation. Disappointing.

Figures are good. Figure 5 should explain what year is used for the MEaSURes grounding lines.