the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Jul 2026
High-resolution Antarctic sea-ice thickness and snow depth from drift-aware multi-mission altimetry
Abstract. Since 2016, Antarctic sea-ice extent has undergone an abrupt regime shift, with a sequence of record low years not observed in the preceding satellite record. These recent anomalies further underline the need for improved characterization of sea‑ice and snow thickness. We primarily use radar (CryoSat-2 and Sentinel-3) and laser (ICESat-2) satellite altimetry to estimate radar and total freeboard. The difference between the two freeboard retrievals provides an estimate of snow depth. However, snow on Antarctic sea ice is very complex, affecting elevation distribution of radar backscattering targets, leading to higher uncertainties than in the Arctic. Monthly gridded datasets are produced by averaging along‑track altimetry data, but sea-ice drift can affect spatial consistency, especially when combining measurements from different satellites. To address this, we combine altimetry with satellite-derived sea-ice drift retrievals to generate drift‑aware daily freeboard, thickness, and snow-depth estimates. Within the ESA “Sea Ice Mass Balance Assessment: Southern Ocean” (SO-SIMBA) project, we advect along‑track CryoSat‑2, Sentinel‑3, and ICESat‑2 measurements over a ± 15‑day window to preserve spatial structures, reduce temporal smearing, and improve the co‑location of radar and laser freeboards. Merging several altimeters increases the sampling density and enables an effective spatial resolution of 12.5 km for the sea-ice thickness dataset. We also track uncertainties from the raw satellite measurements to the final gridded products, combining measurement uncertainties and drift‑related propagation errors. Here we present a nearly 7-year long dataset (October 2018 to August 2025) of year-round daily updated, monthly sea‑ice thickness and snow depth along with uncertainties. Initial results show coherent and realistic spatial patterns, consistent with features visible in SAR imagery. Comparisons with independent airborne measurements, and available in-situ observations, indicate improved spatial fidelity and internal consistency compared to standard monthly gridded approaches, especially if only one sensor is used.
- Preprint
(18499 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 16 Sep 2026)
- RC1: 'Comment on egusphere-2026-4063', Anonymous Referee #1, 07 Sep 2026 reply
-
RC2: 'Comment on egusphere-2026-4063', Anonymous Referee #2, 07 Sep 2026
reply
General comments
This paper proposes a sea ice thickness (SIT) and snow depth (SD) product for the Antarctic derived from multi-mission altimetry. Corrections are applied to account for the sea ice drift over the month-long windows in which parcels of sea ice may appear in co-located satellite tracks.
The paper is clear and easy to follow, and the research is relevant and timely. The inputs and methodology are stated in good detail. There are significant gaps in the literature review and with the treatment of assumptions, particularly around Ku-band scattering within the snowpack in Antarctica (which are stated but subsequently ignored).
A broad evaluation of the “drift-aware” product is attempted, including exploring pan-Antarctic spatial patterns, intercomparisons with in situ data, and considering differences with non-drift-corrected products. However, quantitative assessment to justify the “drift-aware” product is missing. Some supporting interpretations of the figures do not accurately represent the findings and require revision. Further validation is needed.
The discussion needs a complete revision. Inclusion of references which place the study within the current literature is missing and limitations of radar scattering within the snowpack needs addressing. Some inaccurate interpretations from the findings propagate into the discussion.
Generally, there is work required throughout the paper to correct typographical inconsistencies, sentence clarity and brevity, and precision of language.
Most figures require format alterations including larger text and plots, clearer colour scales and more revised panel positioning.
Scientific/specific comments
Section 1 – introduction:
A reasonable introduction to the approaches but there is no comparison with other approaches, e.g. Lawrence et al. (2024) vs Garnier et al. (2021) vs Bocquet et al. (2024). What are the uncertainties from each? How well validated are they? There is no justification for your approach included in the paper, this is fundamental.
The uncertainties of radar scattering and penetration must be addressed with a section in the introduction.
Section 2 – Methods:
Line 127: Is there any issue with only using coverage where SIC > 40% vs IS-2 with the drift correction? What happens between SIC 40% and 15%?
Line 149: Why was this important and is it the best approach for this newer SAR data?
Line 162: Justify whether these uncertainties are actually independent and can be treated in this way?
Line 165: This method was developed for the Arctic. Do you make any modification for it’s use in the Antarctic? Justify its applicability.
Line 180-188: There is information missing here. Look at references by Giles et al. (2008), Fredensborg Hansen et al. (2025) x2, Willatt et al. (2010), Willatt et al. (2025) and discuss how you account for the Ku-band scattering occurring above the ice surface.
Line 195: Are there any more recent seasonal densities of snow and sea ice published available for use?
The drift correction seems like a good idea, however how much difference does it make to use this approach? Can this be quantified? The drift patterns seem to converge on similar tracks in Figure 1 (particularly for S-3A). Is this explained by physical behaviour or a systematic bias in the method? Further, understandably the terminology “drift-aware” has been inherited from Ricker et al. (2025) but this terminology could be clearer. “Drift-corrected”, “drift-subtracted” or similar is more intuitive to describe the method.
Line 263-267: This need to be discussed fully. Why does it occur and why are weightings required? How do you account for it elsewhere?
Have you considering applying weights to days within the +-15 day windows? Weighting retrievals on days closer to the target day higher address some of the product’s uncertainty.
Line 271-272: Justify this treatment of uncertainty. The Ku-band is used for all radar altimeters over similar surfaces, is this independent? Clarify what the uncertainty represents.
Equation 5: Defines the snow depth uncertainty and states an assumption that the three terms are independent. However, the laser and Ku-band freeboard are used in the calculation of the snow density term, i.e. not independent. This seems flawed and requires justification or correction.
The methodology summary in Section 2.6 may be elevated with a flow-chart style schematic to provide visual explanation of the processing chain.
Section 3 – Results:
Figure 3 analysis: As most uncertainty is due to drift, is it actually worthwhile? What happens if is not done?
Figure 4 analysis: Some interpretation provided does not accurately reflect what is displayed in the figure.
- A statement should be made here that positive relationships shown in Figure 4 are possibly unsurprising, as this is comparing the output (i.e. the product) with its inputs.
- Line 344: Statement here does not really seem true, there is large differences occurring in the Weddell, Amundsen Seas and East Antarctic coast in Figure 4 which is not discussed. Mean differences of 0.03 m is quite large when snow depth is only ~10 cm. A non-zero mean difference may not directly prove that there is no-systematic bias.
- Line 347: Why is sampling differences a reason here, please explain?
- Line 348: It cannot be concluded that residual differences are random based from standard deviation values alone – The requires more evidence.
Figure 6 analysis: Some interpretation provided does not accurately reflect what is displayed in the figure.
- It is not clear what is being investigated here and why. Please explain.
- Line 370: An additional panel in figure 6 to support this statement on the April-November period is required as evidence of this statement.
- Line 372 to 377: Not true. It seems fundamentally incorrect to state that the close relationship displayed in Figure 6 is not a causal relationship given that the MMS snow depth product is directly derived from IS2 freeboard. The result of Kacimi and Kwok (2020) is framed incorrectly here. The correct interpretation is that IS-2 freeboard explains most of the variance in MMS (IS2 – CS2) snow depth. In the case of Kacimi and Kwok (2020), it seems they suggest it points to the potentially for using this relationship to skip the IS2-CS2 comparison step where CS2 observations are not available. This entire paragraph needs re-consideration.
The motives for the analysis conducted in section 3 needs to be made clearer. Additionally, bring the analysis from Section 5 earlier into Section 3 may build a stronger argument for justifying benefits of this of this product, as this should be a core aim of the conducted analysis.
Section 4 – Intercomparison and validation:
How far the MMS product aligns with the Fons 2023 results is overstated throughout this section and requires reframing.
Figure 7 analysis: Some interpretation provided does not accurately reflect what is displayed in the figure.
- Line 398-400: Commentary on the SIT from Fons 2023 being aligned to the MMS seems overstated as there are clear absolute differences in the bias maps including Weddell (~2m) and Ross (~1m) seas, as well systematically around the ice edge perimeter. There is a 20% difference between the stated means of MMS and Fons 2023.
- Line 403: The histograms distributions are not aligned.
- Line 405: The CSAO bias with MMS do not look that spatially different from Fons 2023, albeit of larger values, so this comment is puzzling.
- Line 410: The Fons SIT seasonal variability does not look weaker than CSAO, it looks a similar amplitude.
- Line 414: A similar comment is missing on Fons 2023 for snow depth here.
Figure 8 analysis: Some interpretation provided does not accurately reflect what is displayed in the figure.
- Line 447: Panel b differences may be proportionally large which is not discussed. It is hard to see due to shallow snow depth on this transect.
- Line 450: Not convinced of this from the figure. Using this figure you can make comments on “following similar trends”, however there is no quantification provided to prove that MMS is within the range of the two OIB observations. For example, providing root-mean squared values for quantifying the differences with respect to the OIB range.
- Line 451-452: Not really, this only looks like it in panel c.
- Line 453: MMS distribution does not really reproduce the central part of the observed distributions in figure 7.
- Line 455: Not demonstrated, requires quantification and re-writing.
Figure 9 analysis: Some interpretation provided does not accurately reflect what is displayed in the figure.
- Line 473: It is not reasonable to make the statement. Figure 9 SIT and snow depth values are completely different.
- Line 474-477: Would ASPECT-BIO underestimate? This requires expanding and references to justify inclusion.
Including “conventional gridding” (as in Section 5) in this analysis in Section 4, along with quantification will help indicate how much improved the “drift-aware” product is with respect to these observational datasets.
Section 5 – Impact analysis:
Line 492: Couldn’t it instead mean that the drift correction introduces the errors?
Figure 10 analysis: Some interpretation provided does not accurately reflect what is displayed in the figure.
- Line 488: This is not the case in the figure, drift-aware variability looks like it occurs more when the SIT change slows down, i.e. at the peaks.
- Line 494: Figure 10 does not seem to directly indicate an improvement using MMS, moreover MMS captures more variability. There is no evidence provided here to trust this temporal variability and therefore this figure alone cannot be used claim an improvement. This needs re-writing. This also propagates into Line 531 in Section 6.
Section 5.2 must be expanded to a full quantitative treatment, e.g. passive microwave snow depth. This is critical for this paper as it justifies the entire concept. After doing so, this should be promoted into results Section 3. [major corrections]
Discussion and conclusions:
There are no references to the literature in this discussion. Needs to be completely re-written with references and discussion of quantified analysis. [Major corrections]
Line 545: Why did you not try other re-tracking?
Technical corrections
Abstract
2: “sea-ice” and “sea ice” is used inconsistently (throughout paper).
2: Remove “further”.
3: Use of “we”, not clear if it refers to authors or sea ice community?
5: Quality – Change “provides” to “is assumed to provide”.
6: Quality - Remove “very” from ”very complex”.
12: Merging “retrievals from” several altimeters, not merging the altimeters themselves.
14: Change “measurements” to “observations” when talking about retrievals from satellites (throughout paper).
15: Description “daily updated, monthly sea ice…” needs rewording to clarify - “daily updated, monthly-window averaged”?
18: “especially if only one sensor is used.” – Clarify. Do you mean that whilst multiple sensors result in some improvement (line 12) the combined results do not compare as well to airborne sensors?
Introduction
21-25: Re-write – Insufficient references and not up to date.
29: Revise sentence for readability and brevity.
32: Quality - Remove “uniquely” from “considering uniquely”. Improve this sentence’s readability.
33: Choose more appropriate word than “reflects”.
33-35: Add a comment on sea ice volume here.
35: Does Massonnet et al. (2013) actually say this? Is this paraphrased correctly?
39: Add “which TOGETHER enabled…”
41: Define snow, ice and total freeboard here.
43-45: Quality – Split up sentence for brevity and readability.
45: What is meant by “such as negligible ice freeboard”? – Clarify wording
54: Remove “strongly” and change “depends” to “depend”.
60: Re-write for quality and readability.
61-64: Provide typical and max velocities to quantify ice drift.
Methods
87: Add “merging RETREIVALS of the”. What does “dedicated” refer to here? Change “delayed” to “delay”.
94: Ensure all types of units all have consistent font and spacing (i.e. 13 GHz not 13GHz) (throughout entire paper).
95: Change “providing” to “which provides”.
97: Add “same TYPE OF radar…” and remove “called the”. Revise sentence for brevity and readability.
100: “similar enough” – poor quality, please quantify and clarify.
102: Remove “using satellite altimetry alone”.
109: Specify the quality flag used here.
131: Correct format to “Lavergne and Down (2023) and Lavergne et al. (2010)”.
139: Correct format of Paul et al. citation.
142-144: Re-write for quality and readability.
145: Need to supply another reference here as Laxon et al. (2003) did not do this.
151-152: Re-write for quality and readability.
151-155: Requires references.
159-160: Why the need to mention this? Also, notation rFB is not defined.
164: Why is this value set? This conflicts with the statement above. Please clarify.
168: Ulaby F.T and R.K (1987) citation seems wrong format (throughout paper).
169: Centre all equations (throughout paper).
175-178: Needs explaining further.
182 Change “affects significantly” to “significantly affects” or remove “significantly”.
203: Add “AS uncertainties In radar…”
216: Typically drifted 100 km from where to where? Please explain further. Is “has typically drifted” to “can typically drift” more appropriate?
223: The relevance of this statement is unclear, please explain further if it is important.
236: Some headings in the paper are missing their number, change to “2.3.1 Uncertainties” throughout. Also ensure capitalisation of words in all headings is consistent (throughout paper).
Figure 2: Plots and text too small (All figures throughout the paper). Histogram y-axis label is confusing. Distributions are not transparent so they block each other.
246: Change “25-km” to “25 km” (in line with the line 94 correction above).
257 & 262: Anonyms for satellite missions are used inconsistently throughout paper. Define in the introduction and subsequently use acronyms throughout (i.e. CS-2, IS-2, S-3A etc.)
275: Justify this statement and reference. Define “total freeboard”.
276-277: This statement requires reference over Antarctic sea ice and justification, it is not in line with current literature.
275: See correction 14 above.
307: Change “Ku freeboards” to Ku-band freeboards. Clarify what is meant hear co-located? Grid pixels or fixed distance?
308: Change “using” to “by applying the principal of”.
310: Is step 5 a repeat of step 2?
316: Specify sea ice thickness, as this reads as snow thickness.
Results
319: Define SIT in introduction and do not repeatedly define (also line 354).
320: See correction 94 above (1.24 m not 1.24m). Change SIT from italics to normal. Specify pan-Antarctic.
322: Revise for brevity and readability.
332-335: No evidence provided yet, this is a concluding statement of this section and lacks quantification.
Figure 3: Too small. Colour scales need revising, particularly for uncertainties and snow depth as smaller values are lost (throughout paper). 10 cm of snow is impossible to identify in plot G (similar in later figures). Colour bar has four ticks across 30 cm intervals, change to three.
Figure 3 caption: Re-write for quality and readability. Remove capital letters.
350: Choose more appropriate word for “coherent”.
Figure 5: impossible to sea snow depth detail between 0 and 10 cm. Plots too small.
354-355: This should be in figure caption instead.
358: Also the east Antarctic coast in sept?
362-364: This statement needs evidence to justify and reference. Are these uncertainties isolated to the months stated?
Figure 6: Poor colour scale for scatter points. Paler points are hard to see. No colour bar provided. Figure 6 is not explicitly cited anywhere in the text.
372: Specify what variance, i.e. add “94% of the MMS SNOW DEPTH variance”.
370: Check if capital R-squared notation is correct here.
379: Change “no systemic offset” to “less systemic offset” or similar.
Intercomparison and validation
Title: change “validation” to “evaluation”
403: Reword “aligned with MMS” to accurately reflect results in figure 7.
Figure 7: Again colour scale is bad for looking at snow depth from 0 to 10 cm.
Figure 7 caption: Change “histograms shown on right” to “left”. “Relative to MMS for MMS” is confusing.
423: Does “CSAO” ASD need stating here for clarity?
432: Define OIB here (not line 439)
Figure 8: Revise the panel configuration here (Possibly one column of transect plots, one column of histograms, one column with a larger map). Histograms too small and cluttered. CSAO and Fons 2023’ line colours switch between transect plots and histograms. Change labels from “rolling average” to “rolling mean”.
Figure 8 caption: 25 km rolling mean applies to OIB Peakiness and Wavelet - revise the description.
457: ASPECT-BIO section has not been introduction and requires further detail to explain this comparison.
Figure 9 caption: Requires more detail including the explanation of error-bars.
466: Change to “x- and y- directions”
470: Clarify, error or standard deviation? What are the bars in the figure?
471: “essentially zero” is not great accuracy - revise.
Impact analysis
Figure 11 plots are too small. Contour lines require defining in caption.
Discussion and conclusion
Re-write with references to literature and discussion of quantified analysis.
258L Correct “static”.
579: Change to “Antarctic applications”.
Appendix
594: Define DTU21.
609-610: Justify 5 degree, 350 m and 700 m values.
614 and 620: Hyphens too big.
A3 requires references throughout.
647: Add “retrieve THE same elevations”.
659: Change “Wavelet at can” to “and”.
Supplements
Line 681: add “doi:” to start of link to video supplement.
Citation: https://doi.org/10.5194/egusphere-2026-4063-RC2
Video abstract
Sea-Ice Thickness animation for 2018-2025 Marion Bocquet and Robert Ricker https://doi.org/10.5281/zenodo.21257446
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 241 | 109 | 21 | 371 | 20 | 21 |
- HTML: 241
- PDF: 109
- XML: 21
- Total: 371
- BibTeX: 20
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript implements a drift-correction method to align freeboard measurements from multiple satellite altimetry missions and generate high-resolution Antarctic snow depth (SD) and sea-ice thickness (SIT) maps for 2018–2025. The manuscript is generally well written and logically presented. However, I have several comments that should be addressed before the manuscript can be recommended for publication.
My main concern is the literature review and the justification for the selection of datasets used for validation and comparison. Retrieving Antarctic sea-ice thickness remains an important challenge for the community, particularly because of the heterogeneous snow cover and the complexity of radar interactions within the snowpack. In the Introduction, the authors mention several existing Antarctic sea-ice thickness products and studies, including Kurtz and Markus (2012), Schwegmann et al. (2016), Kacimi and Kwok (2020), and Fons et al. (2023). Some of these references (e.g., Kacimi and Kwok (2020) and Fons et al. (2023).) were use for comparisons of the derived MMS results. While some datasets used in Sections 4 and 5 (Paul et al. (2024) and CSAO) were not introduced or discussed in the Introduction.
Therefore, I suggest that the authors provide a clearer overview of the currently available Antarctic SIT and SD products and datasets. For example, a table could summarize the existing products/datasets, the satellite missions or measurement platforms from which they are derived, their spatial and temporal coverage, and indicate which datasets are used for validation or comparison in this manuscript. This would help readers understand how the new MMS product relates to existing datasets and why particular datasets were selected.
Related to the above point, I also suggest that the authors reconsider or better explain the organization of Sections 4 and 5.
Different datasets are introduced at different stages of the validation and comparison:
-In Section 4.1, Fons et al. (2023) and CSAO are used for intercomparison of SIT and SD.
-In Sec 4.2, OIB data are used to validate regional MMS snow depth.
-In Sec 4.3, ASPECT-Bio in-situ measurements are used to validate MMS SD and SIT.
-In Sec 5.2, SIT and SD estimates from Fons et al. (2023) and Paul et al. (2024) are compared with MMS.
Is the intention to distinguish between pan-Antarctic and regional-scale comparisons? If so, this distinction should be stated more clearly. At present, the rationale for separating the different validation and comparison analyses is not super apparent.
An alternative structure might be to organize the analyses according to the retrieved variables - for example, one section focusing on the validation/intercomparison of MMS snow depth and another focusing on sea-ice thickness. However, the current structure could also work if the authors provide a clearer explanation for why the analyses are organized in this way.
For the OIB snow-depth measurements, ASPECT-Bio in-situ observations, and Sentinel-1 imagery, could the authors provide more justification for the specific regions and dates selected for validation/comparison? Since the MMS products cover the pan-Antarctic region from 2018 to 2025, I would expect that several OIB campaigns may overlap with the study period and provide snow-depth measurements. There are also ample in-situ Antarctic SIT and SD measurements from various expeditions, for example through AWI or the AAPP data portal. Sentinel-1 also provides extensive spatial and temporal coverage of Antarctic sea ice.
The authors should therefore explain why these particular dates, regions, and datasets were selected. Were the choices determined by temporal overlap, spatial coverage, data quality, availability, or other criteria? A clearer explanation would help readers assess how representative the validation results are for the overall pan-Antarctic MMS product.
minor points:
Lines 160–165: There are many mathematical symbols introduced in this section. Please ensure that each symbol is clearly defined when it first appears.
Lines 227–228: Could the authors provide some additional explanation of how the change correction factor is applied in this study?
Line 287: Why was a three-day rolling window selected? Are the results sensitive to the selected window length.
Lines 322–325: The authors state that the thickest ice is observed in certain regions and that similar spatial distributions are found across the study period. Could the authors be more quantitative and specific here? For example, during which years are similar spatial patterns observed? Are the maximum thickness values also comparable between years? What are the main differences or biases between the annual results?
Figure 4: I may have misunderstood the purpose of this comparison, but I did not fully follow why comparing MMS with each of these three datasets separately, given that MMS itself is generated by merging observations from these missions. If the purpose is to assess consistency in freeboard measurements among the different missions, would it be more appropriate to use spatially and temporally overlapping observations and directly compare CS2 with S1-A/S1-B? Please clarify the purpose of this analysis and how it should be interpreted.
Figure 5: Is this figure showing the monthly averages calculated over the entire 2018–2025 period? Could the authors also provide monthly SIT and SD results for each individual year in the Supplementary Material? This would allow readers to assess the interannual variability that may be hidden in the multi-year monthly averages.
Line 364 (bottom of Page 16): The authors mention that processes such as flooding can increase the uncertainty. Is this conclusion based on observations in the present study, or is it supported by previous studies? It would also be useful to show the spatial distribution of uncertainty, potentially alongside the SIT and SD results in Figure 5.
Figure 6: Section 3.3 discusses the fitting relationship between freeboard and snow depth for each month. Could the authors provide the monthly fitting results in the Supplementary Material? It would be interesting to examine the seasonal variation in the relationship between freeboard and snow depth.
Lines 487–489: The authors state that the drift-aware approach captures distinct temporal anomalies. Could the authors identify specific examples in the figures where this improvement can be clearly observed?
Figure 10: For segments where the differences exceed the estimated uncertainty, could the authors discuss the possible reasons for these discrepancies and support the interpretation with relevant references where appropriate?
Code and Data Availability: Do the authors plan to make the derived MMS SIT and SD products publicly available? Given the pan-Antarctic coverage and the 2018–2025 time series, public access to these products would potentially be valuable to the broader sea-ice community.