Turbidity Estimation from GOCI Satellite Data in the Turbid Estuaries of China’s Coast
and Zhongfeng Qiu 3,*
Round 1
Reviewer 1 Report
The study is well presented. However, addressing the following comments will enhance your paper.
Introduction
Line 64-66: Grammatical error.
Introduction is relatively short.
Discuss advantages and disadvantages of Remote sensing and NN.
Please discuss hydrology of the study area. Also discuss reasons of why the study area was selected.
More discussion regarding quality control of in situ turbidity datasets (Line 126-129).
Satellite Data: discuss reasons for choosing GOCI.
Line 148: provide justifications of why using spatial windows of 3×3 pixels (e.g. why not 5*5 pixels).
Line 214: ROMS was only implemented for the year 2018, please justify.
Include a table for statistical information shown in figure 3 and delete from figure 3 for more clarity.
Figure 5: Add a legend to the plot on the left.
In sub-section 3.3.1 Yellow River Estuary, try to shorten sentences to minimize errors. Also discuss more clearly of why the Yellow River Estuary shown in figure 6 is more turbid on 04/7/2018 (middle panel) compared to the other dates.
Figures 8 and 9: delete dates on sub-plots and only keep dates in captions.
Add advantages of NN to your discussion.
Figures 10 and 11: Describe subplots a through i in the caption.
Conclusion: Overall, well written. However, more discussion of the importance of NN and its applicability is needed.
Author Response
Dear Reviewer:
We thank you for your comments to help improve the manuscript.
Attached is our reply of how to revise and improve the new version.
Yours sincerely,
Yuanzhi Zhang
On behalf of all co-authors
Author Response File: 
Author Response.pdf
Reviewer 2 Report
A very interesting paper. Very well organized, with good English, easy to read, adequated support and background, and adequate references.
The only suggestion I propose is to include legends in figures 9 and 10 in order to identify the turbiditic zone and figures 10 and 11, the experimental (in red) data, and estimated values (in black).
Author Response
Dear Reviewer:
We thank you for your comments to help improve the manuscript.
Attached is our reply of how to revise and improve the new version.
Yours sincerely,
Yuanzhi Zhang
On behalf of all co-authors
Author Response File: Author Response.pdf
Reviewer 3 Report
This is a well written paper that is based on substantial work by the authors. The paper proposes the use of a Neural Network algorithm to replace standard remote sensing empirical turbidity relationships used in remote sensing form satellites. The results presented will therefore be of current interest to the community of scientists concerned with estimating seawater turbidity. For this reasons I recommend publication of this paper. However there are in my opinion several issues that must be addressed before publication can proceed.
- The authors need to specify completely the source of the Neural Network algorithm builder they use. Does it come from a standard commercial package? If so which (give references).
- The authors use 70% of a data set for training and 30% for validation. This is reasonable but what is not clear is if the subsequent tests were from areas outside those of the training set. If not their results on the applicability of their NN approach are obviously favorably biased. This should be made clear.
- There is no explicit comparison with any of the standard empirical algorithms performance on the same data set as NN. To support their claim they must at a minimum give a table with comparative statistical parameters to those they give in figure 5
- It’s true that any empirical algorithms are based on results from specific areas but in a very real sense so is any NN model based on the location of its training set. The capability of the NN model to outperform standard empirical algos must be backed up by its application in areas and waters completely outside and far removed from the training set.
Section 3.3 the authors don’t mention the turbidity was calculated using the NN algo??
Some minor typos noted. A more complete review should in any case be done:
Line 162 to 173 need to be removed as they seem to have been replaced by lines 174-183
Line 56 “fewer studies have been used” should be replaced by ”some rarer studies have used”
Line 65 “may not useful for” should be replaced by ”may not be useful for”
Line 78 “for investigating the response of turbidity pattern to tidal action” should be replaced by ”and investigated the response of turbidity patterns to tidal action”
Line 65 “which were used” should be replaced by ”were used for”
Line 149 “were defined the final” should be replaced by “defined the final”
Line 214 “We run the ROMS” should be replaced by “we ran the ROMS”
Line 230”were shown” should be replaced by “are shown”
Author Response
Dear Reviewer:
We thank you for your comments to help improve the manuscript.
Attached is our reply of how to revise and improve the new version.
Yours sincerely,
Yuanzhi Zhang
On behalf of all co-authors
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The paper has been improved. Thank you for addressing my comments in your revised version.
Author Response
Dear Reviewer,
We have revised and updated the English checking as suggested.
Yours sincerely,
Yuanzhi Zhang
On behalf of all co-authors
Author Response File: Author Response.pdf
