Water Turbidity Retrieval Based on UAV Hyperspectral Remote Sensing

Round 1

Reviewer 1 Report

The paper "Water Turbidity Retrieval based on UAV Hyperspectral Remote Sensing" describes a method for determining the degree of turbidity in inland water bodies based on hyperspectral UAV survey data and established regression relationships between the degree of turbidity and spectral reflective characteristics of a water body. The article contains original material and is of practical importance for specialists using remote spectral methods for diagnostics of natural environments and objects.

The article should be revised in accordance with the above comments.

Notes:

1. In the introduction, it is necessary to clearly formulate how the proposed method for restoring turbidity differs from the known ones and what is its novelty and advantages.
2. Line 110. "Therefore, this experiment cannot simulate the actual river environment."

Including: due to the limited amount of simulated water volume; inhomogeneity of turbidity (impurities) distribution via volume (depth); because the water in a natural body (in a river) is not perfectly smooth, waves are present.

These and other effects limit the ability to use the small black box regression equations for natural water bodies. Accuracy estimates are required.

3. Linу 177. "To eliminate the influence of the atmosphere, water vapor, and other factors when UAV flies to a certain height, we adopt black-and-white correction and atmospheric correction."

What is "black-and-white correction" here and how can it eliminate the influence of the atmosphere and, in particular, water vapor?

4. Formula (2). How was the reflectance correction done here, i.e. how are Ref_Hyp and Ref_gray obtained? Here, apparently, a gray board cannot be used (as in formula (1)), but instead a gray cloth is used, i.e. reflectance correction and atmospheric correction are performed simultaneously with the use of the gray cloth standard, therefore instead of (2) it is more correct to write Ref_fixed = (DN_Hyp*Ref_standard)/DN_gray
5. Lines 186-188. "The reflectance measured by the ground target can be converted into image reflectance, which can truly reflect the surface reflectance to achieve the requirements and purposes of the experiment." - Incomprehensible meaningless phrase!
6. Formula (3). Obviously, the wrong second wavelength, 83 nm?
7. Lines 216-217. "To solve the influence of the external environment in the process of collecting ultraviolet and NIR spectrum data." - What does this phrase mean? An unfinished phrase?
8. Lines 217-218. "To remove the characteristic spectral noise outside the band with high noise standard deviation, especially the spectral severe overlap problem of NIR spectrum ...". It is not clear here about what "characteristic spectral noise" and what "band" are you talking about? The overlap of the spectra of the second orders of diffraction?
9. Line 221. "... to weaken the positioning accuracy error."
10. Lines 263-264. "Normalization of the spectrum can improve the correlation between turbidity and remote sensing reflectance." What kind of spectrum normalization is used here? Give the formula! If you mean the normalized difference (the difference of the reflectances divided by their sum, see Table 3), then the effects of measurement distance, angle and weather conditions are not completely removed.
11. Lines 284-319: This part of the article is presented formally, without the necessary clarity, using general terms and designations of quantities without specifics. It does not allow any researcher to repeat this sequence of actions. Instead of using the general words "independent variables", "dependent variables" (and their designations), you should use the specific values that appear in your problem.

In particular:

what is “F is the normalized data of Y”? Why are the components (u₁ and subsequent) extracted from Y and not from F?

New undefined terms appear: variable group? input standard? the standardized index? standardized regression variable? original variable?

How many regression coefficients are there? (Step 4). What is the “coefficient of the equation” and “regression coefficient”?

12. Line 324-325. "After processing the spectral data of the standard solution, it was measured by the hyperspectral imager."

It was measured after processing the spectral data? Then, what was measured?

13. Line 325. "The independent variable matrix X dimension was 29 * 161"

How can spectral values act as independent variables, and what then are dependent variables?

14. Lines 380-381. “The regression coefficients of all bands in the hyperspectral band are the same, and change at about 0.1 before the 750 nm band” This statement is not supported by Figure 7a.
15. Lines 398-3399. "According to the establishment of the above model, we compare the multiple linear regression model and the PLS model."

What does “multiple linear regression model” mean here? According to your statement (see line 274) the multiple linear regression model and the PLS model are the same.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper is well written and organized. Results, Discussion, and Conclusions sections are acceptable for publication. I think that authors should improve Introduction and Methods sections. My comments are below.

Line 22 NTU – what does it mean? It should be explained as the abbreviation is used for the first time.

Line 25 there is lack of space between (PLS) and models.

Lines 40-42 Please rewrite this sentence and avoid repetition (water body) in a sentence

Line 42 should be “monitoring OF spatial”

Lines 50-62 and 63-78. Two below papers should be at least shortly mentioned in this and the next paragraph, and further in the paper, as the topic of the paper is not novel as state by the authors.

Investigation of Sediment-Rich Glacial Meltwater Plumes Using a High-Resolution Multispectral Sensor Mounted on an Unmanned Aerial Vehicle

KA Wójcik, RJ Bialik, M Osińska, M Figielski

Water 11 (11), 2405

Application of drone technologies in surface water resources monitoring and assessment: a systematic review of progress, challenges, and opportunities in the global south

M Sibanda, O Mutanga, VGP Chimonyo, AD Clulow, C Shoko, ...

Drones 5 (3), 84

Line 123. I would suggest authors to avoid such strong statements, pointing out only one company (DJI). As presented in the above mentioned publications also others drones were used for that purpose.

Please read the paper again and look for all editorial mistakes. Improve text.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The article starts by highlighting the results of the work instead of providing a clear, recognized knowledge gap in the scientific field. It is unclear what is the unsolved problem, hence what is the principal contribution of the study to the field. The article follows rather a case study form.

The article proposes and verifies turbidity retrieval models. While interesting, the article does not provide a clear comparison between the proposed models and the other state-of-the-art solutions, therefore being difficult to assess positively the performance.

There are minor modifications done to paragraphs and the whole paragraphs are marked as improved since the last submitted version. The authors should only highlight key aspects of the article that were improved.

In the 3.2. section there are four proposed models presented, but the authors’ contribution is not clear. It should be clear what is already available and what is the contribution of the article.

Most importantly, there is no mathematical model to present and support the contributions, making the reproduction of the solution very difficult for other researchers.

Another important point is that there is no comparison with other solutions/processes. While the proposed solution may be scientifically sound, in order to be valuable, it must be better in some aspects to other available solutions, or it must be a pioneer in a research direction.

The recommended cost estimation and the presentation of the solution from the implementation point of view (i.e. argue that it is or it is not more efficient than others) was not provided.

With unclear knowledge gap, unclear contributions, no mathematical model or clear details for ease of identical reproduction by other researchers and lacking comparisons with other solutions and implementation details like costs, performance requirements etc., the article is not in a form suitable for being published.

Author Response

Dear academic editor,

Thank you for your comment, and please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper "Water Turbidity Retrieval based on UAV Hyperspectral Remote Sensing" describes the method for determining the degree of turbidity of inland water bodies based on data from UAV hyperspectral imaging and established regression relationships between the degree of turbidity and spectral reflective characteristics of a water body. The article contains original materials and is of practical importance for specialists using remote spectral methods for diagnostics of natural environments and objects.

 

The article can be recommended for publication after revision in accordance with the given comments.

 

Comments:

1. Lines 78-80: «The UAV remote sensing system has flexibility, rapid response, simple structure, and good real-time performance. It can realize the acquisition of ‘three high’ remote sensing image data at low altitude and effectively make up for traditional satellite remote sensing technology [13].»

What does “three high remote sensing image data” mean in this context? Why three?

2. Section 2.1.1:

It is not described how the turbidity of the prepared standard solutions was determined in terms of NTU? It seems logical that it should be determined by the same method (instrument) as the turbidity of subsequently taken samples from water bodies (as in line 166: HACH 2100q portable turbidimeter).

In addition, the choice of the size of the container (black box with a volume of 13L) is not justified. It is not clear how the light coming out of the container relates to the light coming from a deep body of water? It would be necessary to estimate the depth of light penetration into the water column of different turbidity. And then the size of the container should be such that it corresponds to the depth and width (taking into account the angular aperture of the device) of that effective volume in an open reservoir, which forms the main contribution (for example, more than 95%) to the backscattered (outgoing the water) light signal. It is easy to estimate it by calculation or by means of appropriate experiments. Absorbed paper can only distort the signal.

3. Figure 1 is not informative and can be omitted.
4. Line 134: What does lens correction and black-and-white correction mean? Shooting of all objects should be carried out with unchanged parameters (settings) of the equipment, including the focal length of the lens and black-and-white level.
5. Line 144: Where did the standard reflectance of the gray board come from?
6. Lines 158-159:” Before the UAV took off, a standard gray cloth with reflectivity of 20% and an area of 1m * 1m had been placed in the aerial photography area".

It is not specified how many pixels in the image of the hyperspectrometer from a height of 200 m correspond to the image of the gray cloth? The accuracy of measuring the spectrum of gray cloth depends on this. In particular, adjacency effect can introduce a significant error.

7. It's not entirely clear what the demonstration of Figure 3 tells us?
8. In formula (3), in addition to wavelengths, one should indicate what values are substituted for these wavelengths (DN or Ref)?
9. Lines 259-260: It is not clear why The band selection range of the ratio model is narrow? The search for pairs of wavelengths must be carried out in the entire range of 400-1000 nm.
10. Line 278: "He used the near-infrared sensors to obtain spectral information ..." Who is he?
11. Lines 285-318: This part of the text should be rewritten as it is poorly written to understand.
12. Line 300: "Step 1: Data standardization". What does standardization mean here?
13. Line 303: What does set a standard mean?
14. Table 2 does not indicate what the summation index i means and what is the value of n?
15. Why is the x-axis shown in Figure 5 for channels rather than wavelengths (as in Figure 6)? Should be consistent for ease of comparison and analysis.
16. In fig. 5 there does not indicated the units for the standard deviation value. Is the noise shown for the original image or for the reflectance?
17. Lines 365-368: It is also advisable to give the values of the correlation coefficient of turbidity with band reflectance at individual wavelengths, as is done for the ratio and normalized ratio.
18. Lines 371-372: “This result is also consistent with the previous research results”. What is the previous result? You must provide a link.
19. Lines 372-373: “Figure 7 (b) shows a similar result to the band ratio correlation”. Here is an error: there should be a "normalized ratio correlation".
20. Lines 380-381: "The regression coefficient of each band in the hyperspectral band 450 ~ 640nm is negative ...". In fig. 8a this is not entirely true.
21. Lines 380-381: "The value of each band's regression coefficient in the hyperspectral band is almost identical and changes around ± 0.05." This statement is again not supported by Figure 8a.
22. It is necessary to sign the horizontal coordinate axes of the graphs on Fig. 9.
23. Line 417: “The correlation coefficient R2 and…”. This is not a correlation coefficient.

Author Response

Dear academic editor,

Thank you for your comment, and please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The present paper:  Water Turbidity Retrieval based on UAV Hyperspectral Re- mote Sensing, establishes  a set of turbidity retrieval models through the artificial control experiment, and verifies the model's  accuracy through UAV flight and water sample data in the same period. Using turbidity retrieval model established in this paper, the water quality parameters of small and medium-sized waters can be quickly monitored through UAV hyperspectral technology, and the retrieval results are in line with the actual situation.

The paper have an appropriate research design, the methods and design are  adequately  described.

Author Response

Dear academic editor,

Thank you for your comment, and please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Dear Authors,
I have reviewed the paper entitled “Water Turbidity Retrieval based on UAV Hyperspectral Remote Sensing”. In my opinion the paper is very interesting. I would only have a few additional comments that the Authors may introduce to improve the quality of the article.
Kind regards

In the Abstract, I would like the Authors to add information on how their technology can be used in other water bodies.
In the introduction, I miss information relating to the possibility of using the technology. In what research can it be used? The goal of popularization must be to encourage readers to read. Therefore, this element should be included with the appropriate citations. In addition, have the authors made comparisons to other methods? Sometimes the UAV is quite limited in area, so airborne or satellite technology can be used for similar assessments. I think there should be a reference to this in Introduction.

*Figure 2 is illegible.

Overall, the experiment looks like an interesting study. However, the use of methods and showing the solution is in my opinion quite debatable. Appropriate comparisons to other methods, e.g. those related to coastal zone research, may be the key to better reception of the article. I also suggest that the authors pay special attention to it.

 

Author Response

Dear academic editor,

Thank you for your careful review, and please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Unfortunately, the authors did not provide the requested information. A mathematical model and comparisons with other solutions or a clear and detailed description about how this solution is completely new and cannot be compared with others are essential elements of a highly ranked journal papers.

Some of the other aspects mentioned in the last review were also not integrated in the paper.  There is not concordance between the cover letter and the provided revised manuscript. The cover letter should help the reviewer easily identify the requested modifications.

Reviewer 2 Report

Unfortunately, I did not receive a proper answer to some of my comments, you reacted formally, not on the merits of the issue. This applies to comments: 2, 4, 6, 7, 15, 17.
In addition, the line numbers you referenced in the coverletter, where the changes were made do not match the line numbers in the new version of the article.

Reviewer 4 Report

Dear Authors, 

The instructions for authors include information relating to presenting research in a broad aspect, defining controversial. My main point was that it was neither included in the discussion nor in the introduction.