Validation of the Ocean Wave Spectrum from the Remote Sensing Data of the Chinese–French Oceanography Satellite

Round 1

Reviewer 1 Report

The authors evaluate the accuracy of the remote sensing data of the Chinese-French Oceanography Satellite, which was launched on 29 October 2018. There are quite a lot of figures and tables in this article. However, the description in English is not clear, and some explanations for the results are inconsistent with the theory. This draft needs major revision. The review comments and some typos on the draft are shown as follows

Comments

1.    Some tables and figures show the difference between the results of three incidence angles. However, the results and their differences are rarely discussed in the text.

2.    Total SWH in Table 3 has a bias close to zero. However, negative biases for wind waves and positive biases for swell are shown in table 4 and 5, respectively. The authors explain that the reason results from swell noise (line 283-287). As I know that the vertical acceleration of a swell is small, so it is difficult to accurately measure the swell with a buoy.  Therefore, the height of the swell by a buoy will be underestimated. The positive bias for the swell should be due to underestimation of buoy measurements rather than low frequency noise. The authors should search the relevant literature to clarify the cause.

3.    More generally, a swell consists of wind-generated waves that are not greatly affected by the local wind at that time. It seems redundant to compare the relationship between the SWH of a swell and corresponding local wind speed in Fig. (10).

4.    Low wind speed will only cause small waves, so the measurement results with different instruments will be quite different. Therefore, there will be negative correlation coefficients at low wind speeds in Figure 8 and Figure 9. Considering the condition of high wind speed, the correlation coefficient should be more meaningful.

5.    The author uses the revised results to obtain the relationship with SWH in Eq. (11). The reviewer do not know what is the purpose of this regression?

Suggestion

1.    Subscript M can be used for the result with mask as Eq. (8).

2.    Original figure caption in Fig. 8 is “Distribution of the wind speed and Rs of wave. The red line is the mean R for per 3 m/s width wind speed window”. The plot is a scatter plot instead of “distribution”. The red line is the mean Rs, not R. The use of Rs and R should be identified.

3.    Line 506: It is recommended to delete “and prospects”.

4.    Line 322 and later: “strong dispersion” is recommended to be “extremely scattered”.

5.    English in the whole draft should be modified for clear description.

 

Typos

1.    Water depth, H or d, in Eq. (2) and Eq. (3).

2.    RMSE in Eq. (5) or RMS in some tables.

3.    Figure caption of Fig. 9 and Fig. 10 should be for wind and swell.

Author Response

Dear reviewer,

      Thank you for your comments, please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

 

I have read the paper by Li et al., submitted for publication in Remote Sensing. This work attempts to validate the wave spectra of CFOSAT against NDBC buoys for a three-year-long period. The quality metrics presented in the manuscript suggest that higher errors are obtained for the swell part of the spectrum. The spectra are of good quality for open-sea regions and wind-sea-dominated sea states. The paper will interest to the engineering/numerical modelling community, especially for numerical modelling data assimilation purposes in wave forecasting and numerical modelling validation. Therefore I suggest its publication after minor corrections are applied. 

Please see below a list of suggestions:

 

1. I suggest the authors rewrite the abstract. In the first part of the abstract, you refer to CFOSAT and its validation against buoys, yet on line 18, you already refer to the discrepancy between satellite and buoy data. I suggest you present indicative metrics on the performance of satellite data and subsequently provide the source of discrepancy, but also the strengths of the data. 
2. Please try to minimise using acronyms in the abstract as much as possible. 
3. Lines 43 - 45: please consider explicitly referring to the benefits of accurate spectral representation for ocean engineering applications. Since altimeters already provide trustworthy SWH data, having access to accurate spectral representation can help us improve data assimilation, validate numerical models (such as partitions of SWH, wave periods, spectral bandwidth), detect the occurrence of cross seas, etc. Although you provide a thorough literature review on the instruments and applications later within the text, it is still vital to stress the importance of the spectral representation earlier within the manuscript. 
4. Line 46: revise "insitu" to "in-situ" throughout the text.
5. Lines 110-115, I find these sentences hard to follow; consider rewriting them. I understand you mean the following: Grigorieva et al. analysed the difference between SWIM spectral partition data and VOS data, examining SWH, wind-sea wavelength, swell-sea wavelength(?). Their results indicated that none of the spectral partitions can be confidently detected. Refine lines 114-116. 
6. Figure 1(b): please include a legend of the symbols on the map. 
7. Caption of Figure 2: it is not clear to me what is meant by "incomplete statistics".
8. Line 162: A sample size equal to 2918 for the satellite data of three years seems a bit small to me. Could you clarify this within the text?
9. Line 239 - 246 includes very important information on processing such data. I wonder, do you also observe this low-frequency noisy peak when dealing with wave number-dependent data? Is this something occurring after the conversion to frequency-dependent data?  
10. Figure 4: a legend in the figure for (b), (d) and (f) would be very helpful (although you provide a good description in the caption). 
11. Line 264, please explain why this separation method was used to detect wind-sea and swell components. Have you inspected the shape of spectra obtained from buoys? Unlike pure single-peak spectra, spectral partitioning can be quite challenging when dealing with mixed sea states. It can also be the source of discrepancies when validating the partitions. 
12. As a reader, I prefer the methods and results to be presented separately. I suggest you first introduce the methods used for validation and filtering, and then in the results and discussion section, you present and discuss the quality metrics and the shortcomings associated with the processing. 
13. Figures 5,6,7: Please include a legend with the colours corresponding to the reference and trend line. Units for standard deviation are missing. 
14. Including a mask tends to increase RMS (poorer quality) for some instances (e.g. figure 5), yet it yields a higher correlation coefficient (higher quality indicator). Please explain more.  
15. Tables 3,4,5: include units for standard deviation. 
16. I could not find within the text where Table 5 is discussed. 
17. It would be helpful for the reader to add a legend in Figures 9-16.
18. Line 467: "We folding the sea wave spectrum", please reformulate. 
19. Line 531: "But in the regions with complex terrain, such as Southeast Asia, the estimated spectra are bad." I think this type of statement kind of undermines the data/processing undertaken. I would rewrite as: "In regions with complex terrains, such as Southeast Asia, the wave spectra are characterised by lower quality; hence they should be used cautiously." Revise similar statements within the text, avoid using "bad" (e.g. lines 278, 284), instead use, lower quality representation, poorer metrics, etc

Overall use of English language is good. Carefully read the text before submitting the revised version.  

Author Response

Dear reviewer,

      Thank you for your comments, please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper describes the evaluation of an important new satellite wave measuring instrument.  The results are important and the paper should be published, but the methods used need to be clarified.

The description of the spectral partitioning and use of the resulting masks is confusing.  My understanding is that a watershed algorithm is used to partition the directional spectra.  Are all spectra divided into three partitions regardless of spectral shape?  The partitions are used to mask out portions of the spectra.  The masks could be applied in a lot of ways.  Each of the three masks could be used separately or two could be used at once.  The evaluations are done with and without “a mask”.  But which mask?  How is the mask selected?  Are they marked by frequency or described as swell and sea?

Then another kind of partitioning based simply on frequency is used to estimate which parts of the spectrum are likely to be swell or sea.  How are those partitions related to the watershed partitions?

The abstract gave me the impression that the accuracy of the spectra measured by the satellite was not very good, but the results in the body of the paper show rather good accuracy.  It would be a good idea to revise the abstract to say that.  The first sentence in the last paragraph of the paper does that well.

The zero degree beam is said to be similar to a radar altimeter.  Since satellite altimeters have been shown to give accurate measurements of significant wave height, why are the measurements from the zero degree beam not used to adjust the total energy of the measured directional spectra?

It would be helpful to show a few example plots comparing satellite and buoy frequency spectra for different values of Rs so that the reader can get a visual impression of the accuracy measure. 

The 180 degree ambiguity in the wave direction can often be eliminated by reference to the wind direction.  Does the scatterometer provide that data?  It is a shame that it does not seem possible to validate the directional spectra because of a lack of sea truth.

 

Minor editing would help

Author Response

Dear reviewer,

      Thank you for your comments, please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

The manuscript is about evaluating of CFOSAT spectral data in terms of quality. The error analysis depicts the accuracy of retrieving oceanographic parameters using this satellite. Here are my comments about this manuscript:

1-     The novelty of this research is not quite clear in the paper, particularly in Abstract. Is it only about accuracy evaluation of the satellite data? Clarify this.

2-     Barrick [4] did not propose electromagnetic scattering in Microwave frequency band, as you indicated in line 53. The radio frequency scattering (in the HF frequency band) was proposed by him.

3-     I believe, Eqs. 1-3 are from some other references. Please cite them accordingly.

4-     The Materials and Methods section lacks a clear and thorough explanation of the method employed.

Author Response

Dear reviewer,

      Thank you for your comments, please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors followed the reviewer's comments to correct the draft.

The revision is accpetable.

Reviewer 4 Report

I appreciate the response from the authors, and I am satisfied with the provided answers.