Satellite-Derived Bathymetry with Sediment Classification Using ICESat-2 and Multispectral Imagery: Case Studies in the South China Sea and Australia
Round 1
Reviewer 1 Report
Thank you. Very good work. A few important questions. (See comments through out PDF. Significant notes also repeated here.)
Equation 2: even though you have cited a source, the variables in the equation should be defined (theta, L, R, h, and subscripts). (Or, possibly, leave out the equation, assuming the citation is sufficient.)
Multiple Figures: In scatter plots (estimated vs true): Using different (color) symbols for sediment type would be helpful/insightful.
Most important clarification: Can/should one use different model regressions/coefficients for the different sediments? It appears like you used sand to train and then apply to all types. If I misunderstood, then perhaps some additional wording could help to clarify. Put another way: Can the sediment types be used to generate a complete map using separate regressions/trained models for each type? i.e. use a sand regression for only sand points, use rubble regression for only rubble points, ...
Note that I may have misunderstood how you assembled your final statistics and figures, so if you can clarify what I might have missed, please let me know, and I will re-read with that additional information.
Comments for author File: 
Comments.pdf
Author Response
Dear Reviewer,
Thank you very much for all comments and suggestions of this manuscript. In the attached response letter, we have answered the questions item-by-item and revised the contents accordingly in the main text (the black font corresponds to the reviewers’ comments and the blue font corresponds to our response). Please see the response letter and revised manuscript in detail.
Best regards.
Author Response File: 
Author Response.doc
Reviewer 2 Report
This manuscript proposes a classification method that can improve the accuracy of SDB in the study area, and the author has obtained good results. In general, it may bring some value to the research of detecting large-scale water depths in remote and sensitive shallow-water areas.
1. The serial number of the discussion part in the manuscript should be 5.
2. The first two paragraphs in the conclusion describe that different reasons may cause certain errors in the research. It is suggested to write this part in the discussion section and conduct a more detailed discussion.
3. The author uses a large number of words such as "may be" and "may" in the manuscript, which will affect the rigorousness of the study.
4. The author needs to optimize the English expression of the manuscript.
5. Although the introduction is well written, it is suggested to select some references that are relatively new in time.
Author Response
Dear Reviewer,
Thank you very much for all comments and suggestions of this manuscript. In the attached response letter, we have answered the questions item-by-item and revised the contents accordingly in the main text (the black font corresponds to the reviewers’ comments and the blue font corresponds to our response). Please see the response letter and revised manuscript in detail.
Best regards.
Author Response File: Author Response.doc
Reviewer 3 Report
This study examines the benefits of using the seabed classification to augment or correct the reflectance data in satellite derived bathymetry of shallow water environments. The results show a mixed improvement, primarily in the classes where adequate training data is available, and may be an interesting way to produce a more precise result. Unfortunately, the study sites chosen appear to lack in situ ship or lidar bathymetry so no improvement in "accuracy" per se can be claimed. The result is interesting based on the precision improvements documented in Tables 2 and 3, thus worthwhile for publication in my opinion.
A few minor points to consider revising by line number, below:
Line 62: the use of word "replaced" is very strong and could be better phrased as ICESat-2 bathymetry is in common usage in areas where in situ is unavailable.
Line 83: the notion that these parameters were "corrected" implies that now it is perfectly correct. What the authors mean is that the data were "processed to correct for..."
Line 86: did the authors perform sediment classifications as this line suggests? This does not seem to be the case.
Line 91: Study Materials is an odd way to describe data sources. How about "Study Sites and Data Sources"?
Line 163: I don't find where the definitions of the symbols given in this equation are defined in this paper. Please define for readers
Line 224: what are the characteristics of "gross error points" and how do they differ from non-gross error points, or even regular points?
Figures 4 and 5: it would be easier for readers to compare these models if both models were on the same graph, perhaps linear band in orange and band-ratio in blue?
Line 340-348: the number of training points for sand and rubble are large enough (n ~ 1000) to result in robust models. This argument about sand being uniform is very likely incorrect since rubble is not uniform reflectance or backscatter, but did produce a precision improvement.
Line 355: I believe the word "viability" is not quite right. Perhaps variability?
The first two paragraphs of the Conclusions are not what the authors conclude, as that beings on line 394 and is quite well written. Lines 376-393 are somewhat redundant with the Discussion, very speculative, and should not be included in the Conclusion section.
Author Response
Dear Reviewer,
Thank you very much for all comments and suggestions of this manuscript. In the attached response letter, we have answered the questions item-by-item and revised the contents accordingly in the main text (the black font corresponds to the reviewers’ comments and the blue font corresponds to our response). Please see the response letter and revised manuscript in detail.
Best regards.
Author Response File: Author Response.doc
