Wavefield Decomposition of Ocean-Bottom Multicomponent Seismic Data with Composite Calibration Filters

Round 1

Reviewer 1 Report

Review of: Wavefield decomposition of ocean-bottom multicomponent seismic data
with composite calibration filters

This is a very ambitious paper that presents a procedure for comprehensive signal processing of ocean bottom reflections in marine seismic survey.  The authors present mathematical treatments of a multi-stage signal processing approach that proposes to suppress water layer multiples generated by sea surface reflections and also cross-contamination of S and P waves.  The approach is based upon wave equation deconvolution to separate up and down going components of P and S waves.  The approach accounts for large and small angles of incidence.  The performance of their signal processing scheme is demonstrated by application to OBS data from a deep water deployment.  I have a few questions that the authors should consider in making a final version of the paper, and I found many instances of minor corrections that should be made.  When these issues are adequately addressed, I recommend that the paper be published.  It makes a strong contribution to an already powerful signal processing technique.

There are far too many instances requiring corrections in English usage to list separately, so I have attached an annotated version of the article that contains all my comments.  The most serious instances and my critical remarks about some parts of the text are listed below.

1.  Page 1 line 38 (and several other instances (P5 line 157)): ‘summarizing’ should be replaced by ‘summing’.

2.  Page 3 line 106:  ‘Seabed bounce’ is an awkward and sloppy term to use in a scientific paper.  Instead, it would be better to use a more appropriate term.  'Water column multiples unavoidably affect....'

3. Page 5 line 172: ‘Point to point division’:  Is there any evidence of large magnitude events from very small values of the denominator?  I note the remark on line 183 that essentially confirms the problem.  Division in the frequency domain is a risky procedure that can cause serious problems in signal processing.  Some methods simply include a subjectively chosen ad hoc value in the denominator to mitigate creation of spurious large values after division. In my view, the authors should make a stronger case that they recognize the problem in equation 3, and then state more strongly that their approach introduces a way to mitigate the problem.

4.  Page 6 line 187: change ‘deserted’ to ‘rejected’  (this is probably a better term to use than ‘deserted’)

5. Page 7 line 238: change ‘letter’ to ‘parameter’.

6.  Page 7 line 237 and Equations 9 and 10: Is there an objective criterion that would support this choice of n  (n=1)?  I do not know that there is one, so the choice is generally subjective.  If this is a subjective choice, the same value of n may not work well with another data set.  (see also remark 7, the same question arises for Equation 20).

Regarding the inversion of the seabed properties, this aspect of the work is poorly described.  I do not see until later in the paper the way in which the method is actually applied to the data.  It would be useful to revise this section to explain in more detail what aspect of the data are used, or at least point the readers to the section later in the text where a revised description that provides more detail about the method can be found.

Finally, about the inversion of seabed properties.  I gather from equations 9 and 10 that the inversion uses the l-1 norm.  If this is so, what information is obtained about the errors of the seabed model parameters?  The estimates of these parameters, seabed density, P-wave and S-wave velocities, derived by inference from the data are important results that will have uncertainty in the estimated values.  They should be accompanied by a discussion of the errors involved.

7.  Page 9 line 277 and Equation 20: Is there an objective criterion to decide whether n=1 or n=2? It looks as though you use least squares optimization in this instance.  Least squares is likely a standard choice?    

8.  Page 9 line 298: Discussion of the noise in the data.  What was the frequency band of the raw data?  The sea bottom location eliminates ocean noise unless the sound speed profile in the water is bottom limited.  The very low frequency end is dominated by flow noise, but the high frequency end may be dominated by shipping of small (fishing ?) vessels?  Otherwise, the pass band of the data (10-90 Hz) is in the band of merchant shipping noise.  This noise source is can be mitigated by the location of the OBS on the seabed if the sound speed profile in the water is not strongly downward refracting.  Or perhaps the ship noise sources were at great distance?  Was the sound speed profile in the water measured during the experiment?  If not done, it should have been measured.

9.  Page 12 figure 6 and Page 16 figure 14:  It is very difficult to see any features in these figures.  Is it possible to revise the figure to show only the areas where the differences are significant, and choose a different figure color scheme?

10.  Page 13 line 370: One example doesn't prove anything.  It does demonstrate the successful application in this case and only this case.  ‘Prove’ should be changed to ‘demonstrate’ here and in other instances of its use in the text (see line 393).

11.  Page 13 line 383: The estimates of these parameters are important, and it is further important to know whether the estimated values are realistic.  Is there any comparison available with ground truth information about the sea floor sediment in this area?

12.  Page 14 line 407: ‘vertical velocity component’ should be ‘horizontal velocity component’.

13.  Page 15 line 423: ‘vertical velocity component’ should be ‘horizontal velocity component’.

14. Page 17 line 448: What do you mean by seabed bounces?  I think you mean water multiples.  'Seabed bounces' should not be used in the text.

15. Page 21 Figure 19: Explain in the revised text the apparent reflections from the grid boundaries in panels b and d.  Are these artifacts of the signal processing?

 

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Review of paper by Chu and Yu.

The paper presents a new technique to analyse the data from Ocean-Bottom Multicompoennt geophones and hydrophones instruments. The authors improved the present methods by proposing a different calibration technique. The paper is well written. The methods are illustrated in detail as well as the results of applications in OBS located in the South China sea at a depth of about 1km, that support the conclusions.

Reading the paper, I just have some questions and very minor suggestions that I suggest to the authors to check.

 

·        Line 10. I think this sentence can be improved. You can try to shift the sentence from “Extended” to “decomposition method” after “velocity components”, changing “extended” in “extending” and removing “from”; so, 1758471 it would sound “We proposed [...] velocity components, extending the wave equation-based adaptive decomposition method.”

·        Line 18. I would suggest “compensated” in the past tense.

·        Line 32. I think estimations” is more proper than “estimates”.

·        Line 69. I think it’s better to say “in the optimization procedures”.

·        Line 107 (and Figure 1). You used the term “particle velocity” that I think it’s not proper for this case as I think you measure the wave velocity (or ground velocity by geophone) not a single particle of the water. Please, check.

·        Line 140. Maybe is it better “contribution” instead of “participation”?

·        Line 182. You can eventually replace “thus” with “therefore”

·        Line 186-187. Sorry, I don’t understand this sentence. DO you mean that “must be designed to make null the modulus of c_pz^raw(w,x)” ?

·        Line 250. I would suggest adding similar to what? I think you mean: “similarly to stage 1”.

·        Line 264. I think it’s “is” not “are”

·        Line 300. Please by 9-10-90-100 Hz, do you mean that 9Hz is low-stop, 10Hz low-pass, 90Hz high-pass, and 100 Hz high-stop? If so okay, but I would like you precise in some way (maybe in brackets specifying the limits are cut-off and full-pass frequencies).

·        Line 347. I think it’s better to say “no calibrated” or “not-yet calibrated” instead of “yet-to-be calibrated”

·        Figure 3, 6, 7, 8, 10, 12, 13, 14, 15, 16, 19 and 21 missed the indication of Unit of measurements of acoustic pressure on the colour (or grey) band, please add.

·        Figure 9, 17 and 20 also missed indication of unit of measurement. AN amplitude normally has a unit of measure except is relative amplitude like a transfer function that is the ratio between two amplitudes. In any case clarify this please.

·        Line 472. I would suggest considering to replace “obvious” with “evident” but it’s not fundamental

·        Line 516. I think the sentence is correct, but maybe it could be a little more readable if you add a comma between “in” and “when”.

·        Line 542. I suggest checking this sentence. I wonder if you would say: “the optimization necessities without introducing any additional dimension of calibration.”

 

·        Line 552. Is it “accelerated” or maybe “emphasised”, please?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf