Estimation of Arctic Sea Ice Thickness Using HY-2B Altimeter Data
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper presents a new methodology for the estimation of sea ice thickness (SIT) using HY-2B altimeter data. Considering the scientific importance of sea ice monitoring in the Arctic Ocean, I agree that HY-2B data can be a good data source for SIT estimation. However, the authors fail to present (i) a clear motivation in the introduction section, (ii) a detailed description of datasets in the data section, (iii) methodological details in the method section, and (iv) the significance of their product compared to the other altimeter-based SIT products in the result section. The method and result sections also lack logical soundness, and some figures have poor quality. Therefore, I would not recommend this manuscript for publication. Please see the comments below for details.
- Title: I’m not sure the terminology “inversion” is appropriate. I encourage the authors to replace “inversion” with “retrieval” or “estimation” in the title and throughout the manuscript (e.g., L42, L71, L75).
- L85-86: The introduction misses a clear explanation of the motivation and objective of this work. Please include (1) Why did you use HY-2B data? I know some details of HY-2B data are mentioned in section 2.1, but you need to clarify the characteristics, advantages, and disadvantages of HY-2B compared to other sensors. (2) What methodology did you use? Please describe why you chose that methodology. You don’t need to provide every detail, but at least you need to provide a clear background and motivation based on the limitations of previous studies and characteristics of HY-2B.
- L89: HaiYang-2B (HY-2B) should have been specified in L86, where it has been first mentioned.
- L94: Figure 0 -> Figure 1
- L96: We downloaded
- L96: What does “SGDR” stand for?
- Section 2.1: What are the spatial resolution (i.e., the radar footprint size) and temporal resolution (i.e., revisiting period)?
- Section 2.2: Please provide any detail about how this sea ice type is produced (e.g., what types of sensors are used, what methodology is used for the sea ice classification).
- L110: Why does the data coverage period start in Nov 2021? I believe all the CS2, CPOM, CS-SMOS, and PIOMAS data are available in 2019 and 2020, covering the same period as HY-2B (Nov 2019 - Apr 2022).
- Section 2.3: This section lacks a lot of explanation about each dataset. The authors should have provided details of sensor features (e.g., frequency, resolution of CryoSat2) and the algorithm each product employs (e.g., what waveform conversion is used, how PIOMAS sea ice modeling works). Understanding the different characteristics between each data product and the HY-2B is the key to sea ice thickness comparison, but unfortunately, the authors have missed a lot of important descriptions in this section. In the same context, please provide details of the OIB and ICEBird field observations.
- L118-120: I don’t understand this sentence. You project the data into a 25 km grid “due to the high resolution”? What do you mean?
- L150-153: I’m not sure why it is necessary to calculate hr’ instead of directly using hr in Equation 3. If the SSHA is calculated within 25 km, doesn’t it remove the residual? I’m not convinced that the processing of Equation 2 is really required between Equation 1 and 3.
- Equation 4: The text in this equation seems to be stretched horizontally. Please fix it.
- L182-183: Where did you obtain these “in-situ” measurements? Are they from OIB and ICEBird? Can this parameter be applied to the HY-2B Ku band in the same way?
- L250-254: How about snow depth (hs)? How did you determine snow depth? It seems that you used two snow depth products (L126-130), but you did not describe any details. How did you handle these two different products to derive snow depth?
- L258-273: This part is really tough to understand:
(1) There should be a bunch of parameters that affect ice thickness besides lead fractions, e.g., uncertainties in ice/snow density and snow depth, sea ice drift, time discrepancy between HY-2B and in-situ data, and penetration rate of Ku-band radar in snow layer, etc. I’m not convinced that lead is the only and the most significant parameter you should use to “correct” ice thickness.
(2) The linear fitting in Fig. 8(c) does not really look statistically significant. I believe there is no significant linear trend between lead fraction and SIT differences. If you want to argue the statistical significance of this linear fitting, please provide the correlation coefficient and p-value of this linear fitting.
(3) There is no detailed description of the 107 matching dataset and the remaining dataset. I would like to see how these datasets are selected and what part of the Arctic Ocean these datasets come from.
(4) I’m surprised that you obtained a perfect linear line in Fig. 8e with the noisy linear fitting in Fig. 8c. I would doubt that the result in Fig. 8e is overfitted or that only a good result is intentionally picked by the authors.
(5) Since you used the lead data only from April 2019, I doubt the general applicability of this lead-based correction for the other months.
- Figure 9: What does the x-axis mean? Without an explanation of the x-axis, I could not understand what this figure means.
- Figure 10: (b) I don’t think the y-axis is the frequency. What is the unit of the y-axis?
- L303-322: I don’t understand the logical flow of this part.
(1) In section 3.2, you have already analyzed the best combination of features using the KS test and mutual information. You already mentioned that “From the results, the correlation between the eight feature values is not very high, with most values below 0.5. This indicates that the selected feature values are representative to a certain extent and can independently represent waveform data without redundancy.” (L237-240). According to this description, it is natural that combination 16 shows the best result, as you also mentioned in L315: “consistent with the calculation results of KS value and mutual information”. Is the additional analysis of Table 1 and Fig. 11 really necessary?
(2) In L317-318, you said: “However, due to the small number of effective points for Sigma0, combination 14 (PP, LEW, LTPP, and AGC), which has the second highest classification accuracy, is selected.” What do you mean “small number of effective points”? If you wanted to ignore Sigma0 because it is NOT an effective variable, why didn’t you exclude this variable in the first feature analysis in section 3.2? If this Sigma0 is not an effective variable, you should have excluded this variable at first. The logic that you select combination 14 is the final combination, even though this is NOT the highest-accuracy combination according to Fig. 11, does not sound reasonable.
- Figure 13: I believe (b) is not really thickness, but it looks like the freeboard result. Please check it.
- Section 4.2.3: The SIT comparison between different products is too superficial. The authors just mentioned the numbers, such as differences, RMSD, and R, but they did not discuss the sources of errors at all. If the authors would like to highlight the significance of their study as the FIRST study to use HY-2B altimeter data for SIT retrieval, they should have highlighted the inherent methodological differences between them and provided any insights on the sources of their differences. This is essential to highlight the significance of their work.
- Section 5.2: I don’t understand what you want to present in this section. How does this section show the “impact of sea ice type on ice thickness inversion”?
- Figure 22: It is tough to understand what (a) and (b) show respectively. What is the difference between (a) and (b)? The caption is even incomplete.
- L474-476: How can you say “the accuracy of inversion was improved by using enhanced data for sea ice density, seawater density, snow depth, and snow density.” What improvement are you talking about? I don’t think you have presented any “improvement” in the manuscript by this data.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper tested the application of HY-2B altimeter data for Arctic sea ice thickness estimation. This paper is well written and organized. The proposed methodology is correct and can provide relevant information about sea ice thickness estimation from radar altimeter data. I only have a few minor comments.
1. Line 2: In this study, the word “Inversion” is used, which means that there exists a forward model for sea ice thickness. In fact, no forward model is described in this study. Therefore, “estimation” is suggested instead of “inversion”.
2. Line 43: What does sea ice discharge height mean?
3. Line 92: Dual-band? Please specify.
4. Line 94: Figure 0? Please correct.
5. Line 121: In this study, snow depth products from Bremen University and Ocean University of China were utilized for sea ice thickness estimation. However, snow depth over MYI is missing in the snow depth product of Bremen University. Please clarify. Meanwhile, uncertainty in snow depth will lead to uncertainty in sea ice thickness estimation. Please discuss the uncertainty in sea ice thickness induced by snow depth.
6. Line 273: The Equation (9) is used for sea ice thickness over the whole Arctic Ocean, while this relationship is derived from data over Beaufort. Please discuss the uncertainty induced by Equation (9).
7. Line 285: This paper derived penetration coefficients for FYI and MYI, and used them for the correction of sea ice freeboard. A discussion on the effect of penetration coefficient on the resulted sea ice thickness is necessary.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Comments and Suggestions for AuthorsReview of “Inversion of Arctic Sea ice thickness using
HY-2B altimeter data” By Pang et al.
The manuscript presents a study of using the HY-2B radar
altimeter for estimating sea ice thickness. The penetration
coefficient of HY-2B is calculated to correct radar freeboard.
Also, it demonstrates that correcting the proportion of leads
benefits the accuracy of SIT. I suggest the author use the
conventional algorithms for sea ice thickness retrieval
recognized internationally to avoid introducing controversial
processing methods, ensuring that the HY-2B sea ice thickness
product aligns with international products. This would make
comparisons more reasonable.
Also, it does not discuss in detail the impact of Ku-band
penetration through snow on SIT, which significantly reduces
the readability of this manuscript. The study could benefit from
these additional analysis. Overall, I believe a manuscript that
shows the retrieval ability of HY-2B would be beneficial, as it
could provide yet another means of estimating thickness from
altimeters.
General comments
I think the discussion of region selection and impact of sea ice
type on ice thickness inversion are not really necessary. These
two steps are not so essential for retrieving SIT from the results.
I recommend the author to discuss the impact of Ku-band
penetration through snow on SIT.
The author use this equation to correct SIT.
SITcor = 0.0032x - 0.2516 + SIT, x is lead%
0.0032/0.2516≈12.4%
So the radio of lead would be very little influence on SIT. It just
have a constand error, around 0.25 m. I suggest the author revise
the relevant statements and avoid writing about the correction
effect of proportion of lead within 25 km on SIT.
Specific comments
Abstract
Line 24-26: Please delete this sentence.
Introduction
Line 45: Please rephrase this sentence.
Line 46-47: Sea ice thickness is obtained indirectly through altimetry,
which directly measures radar freeboard or sea ice freeboard. It is
recommended to replace sea ice thickness with radar freeboard.
Line 61-62: Please rephrase this sentence.
Line 48-86: The Introduction need to be rewritten to highlight the
necessity and innovativeness of this research, rather than simply
compiling others' studies.
Data
Line 96-97: Is such a limited dataset convincing? Is it possible
to obtain more data to enhance the credibility of the manuscript?
Line 97, 119: How is the gridding performed? What's your
projection grid?
Line 112: Is it fusion SIT products provided by CPOM? Please
provide all of product links of SIT.
Line 102, 121: Please provide all of auxiliary data links.
Line 129: Why did the author use two kinds of snow depth
products? Please provide data links.
Method
Line 176-178: Currently, it's difficult to take into account
change in the scattering horizon within the snowpack, so AWI
have assumed that the radar pulses penetrate through any snow
cover on ice floes and scatter from the snow–ice interface. For
AWI sea ice thickness prodcut, it just applied a geometric
correction that accounts for the slower wave propagation speed
of the radar signal in the snow layer. Are all the other processing
steps and auxiliary data used in the processing identical to the
AWI processing so that a proper comparison can be made?
Line 263: Obviously, it isn't a significant linear trend between
the proportion of leads and the diffenence of SIT. I have to
doubt its reliability. Please give the coefficient of determination
of the fitting line.
Result
Line 275: Please delete ‘first’.
Line 323: I'm glad the author classify FYI and MYI with
waveform features using sea ice type products as true values.
Did you compare the results of SIT with classification and
without classification? Is't improved?
Line 345: Please give the mean error and RMSE. I suggest to
combine Table 2 and Figure 14.
Line 381: Please give the mean error and RMSE. I suggest to
combine Table3 and Figure 17.
Discussion
Line 345: As the author said, ‘The results show that the two ice
thickness comparisons with CS-SMOS are very similar’. How
come not directly use sea ice type to retrieve sea ice thickness? I
suggest the author to delete the discussion of sea ice type. It’s
still not very clearly the mechanism of partial snow penetration
for Ku-band radar altimeter due to lack of measurement data. I
would recommend the author to discuss the impact of snow
penetration for retrieving sea ice thickness.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsI would like to thank the authors to handle all of my comments to improve the manuscript. The authors addressed most of my major concerns well in the data and method section. However, I still see so many typos and space errors throughout the manuscript, so I encourage the authors to review the manuscript very carefully. Please see the minor comments below.
L89: Therefore
L90: Please remove “undoubtedly”. It is a too strong word.
L94: HaiYang-2B (HY-2B) was first mentioned in L71, so it should be noted there.
L92-96: Please consider moving these sentences to somewhere earlier in the introduction section. I don’t think these sentences are in the appropriate context.
L99-101: Can you provide any reference for this information about HY-2B?
L103-108: Please remove this paragraph. It is totally duplicated with the next paragraph.
L112: Please remove one period on the middle of the sentence.
L124: CryoSat-2 is already specified in L49.
L131: Please add a space between [7] ; and CPOM. Actually, I found a bunch of this spacing error throughout the manuscript (e.g., L133, 134, 136, 137, 146 and more). Please correct them.
L146: therefore
L148: projection. During this process,
L161-163: Please provide any reference that (i) DTU snow depth is overestimated and (ii) Ocean University of China snow depth is more accurate than that for MYI.
L194: Please fix this reference source error.
L199: Please add a period after the sentence.
L201: Please fix this reference source error.
L211: Please fix this reference source error.
Figure 8: On the X-axis, what is the “number of matching point”? Is it physically meaningful metrics?
L356: They are not uniform distributions.
Table 2 and Table 3: Please change the date format into November 2021, December 2021, etc. Please avoid using a date format such as 2021.11, 2021.12, etc. This is not appropriate for scientific journals.
Figure 19a: It is hard to see the OIB and ICEBird points. Please consider enlarging their sizes.
L457 and 459: Please correct the spacing. I saw so many space errors throughout the manuscript.
L466: Please remove comma.
L469: was discussed -> is discussed.
L481: was discussed -> is discussed.
Author Response
Please see the attachment.
Author Response File: Author Response.docx