Next Article in Journal
Mapping of the Successional Stage of a Secondary Forest Using Point Clouds Derived from UAV Photogrammetry
Previous Article in Journal
Classification of Conservation Tillage Using Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model
 
 
Article
Peer-Review Record

Real-Time and Continuous Tracking of Total Phosphorus Using a Ground-Based Hyperspectral Proximal Sensing System

Remote Sens. 2023, 15(2), 507; https://doi.org/10.3390/rs15020507
by Na Li 1,2, Yunlin Zhang 1,*, Kun Shi 1,3, Yibo Zhang 1, Xiao Sun 1,2, Weijia Wang 1,2, Haiming Qian 1,4, Huayin Yang 1,2 and Yongkang Niu 1
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Remote Sens. 2023, 15(2), 507; https://doi.org/10.3390/rs15020507
Submission received: 24 November 2022 / Revised: 28 December 2022 / Accepted: 13 January 2023 / Published: 14 January 2023

Round 1

Reviewer 1 Report

The manuscript Real-time and continuous tracking of total phosphorus using ground-based hyperspectral proximal sensing presents a new method for remote and continuous tracking of total phosphorus in rivers and lakes. According to the authors, the proposed method gives very good results and can be used to respond quickly in crisis situations.

The manuscript presents in an accessible and interesting way the proposed method of testing phosphorus in the aquatic environment.

I have two formal remarks:

- authors should correct formula numbers due to repeated numbers.

- the section "conclusions" should be called "summary" as it does not contain conclusions.

It was miss in the manuscript the impact of the time of operation of the ground equipment used for measurements on the quality of the obtained results, based on my own experience or literature. This additional information can be entered in the "discussion" section.

Author Response

December 28, 2022

 

Response to reviewers of remotesensing-2086088

Dear Editor and Reviewers:

We would like to express our gratitude to the three anonymous reviewers for their critical comments and constructive suggestions regarding our manuscript (remotesensing-2086088). We have revised the manuscript and made a detailed response for the comments of three reviewers point by point below.

To facilitate review of the revisions, we have highlighted the main changes in response to the reviewers’ comments in red text within the document. In addition, in our response letter, we use “….” at the start and end of the new manuscript text to show that it is a new version of the manuscript and thus distinguish between comments in responses and the actual new text in the revised manuscript.

We greatly appreciate your help and that of the three reviewers concerning improvements to this manuscript. We also acknowledge the three anonymous reviewers for their useful reviews of this manuscript. If the reviewers present further comments, we are willing to revise the manuscript again.

We hope that our responses are satisfactory for you and that this revised manuscript is scientifically acceptable to Remote Sensing. Thank you for your patience while we made our revisions.

We are looking forward to hearing from you. Best regards.

 

Dr. Yunlin Zhang

E-mail: [email protected]

Tel. (+86) 25-86882008; Fax (+86) 25-57714759

 

 

Reviewer 1

The manuscript Real-time and continuous tracking of total phosphorus using ground-based hyperspectral proximal sensing present a new method for remote and continuous tracking of total phosphorus in rivers and lakes. According to the authors, the proposed method gives very good results and can be used to respond quickly in crisis situations. The manuscript presents in an accessible and interesting way the proposed method of testing phosphorus in the aquatic environment.

Response: Thanks for your suggestions. The point-by-point response is addressed as follows.

 

Q1: authors should correct formula numbers due to repeated numbers.

Response: Thanks for your suggestions. We have corrected formula numbers (Page 6: lines 220)

 

Q2: the section "conclusions" should be called "summary" as it does not contain conclusions.

Response: We revised “conclusions” to “Summary” (Page 15: line 574) and rewrote the sentences as follows (Page 16: lines 584-603):

“In this study, real-time high-frequency and automatic GHPSs was first proposed for tracking TP dynamics, with characterized by a spectral resolution of 1 nm from 400 nm to 900 nm and a minimum interval of 20 s. A TP machine learning model was developed and validated with ideal accuracy based on XGBoost method using 377 pairs of synchronous samples (R2 = 0.97, RMSE = 0.017 mg·L-1, MAPE = 12.8%). Subsequently, compared with traditional TP monitoring equipment, GHPSs, with the advantages of simple operation, high-frequency, real-time, accurate and suitable for complex weather, supplement the defects of low monitoring frequency, poor timeliness and accuracy of existing monitoring equipment, complement the lack of TP data in overcast and cloudy weather. Short and rapid TP changes were observed within one day in LT and LR based on GHPSs minute scale monitoring, which highlighted the importance of high frequency observation of TP. In the future, GHPSs have great potential value and application prospects in raising our awareness of the dynamics and driving mechanisms of water quality for inland waters. Therefore, our findings will contribute to the deep understanding of short-term dynamics and long-term trends of TP concentration by using GHPSs, which greatly improve water environment management and prediction precision of algal bloom.

 

Q3: It was a miss in the manuscript the impact of the time of operation of the ground equipment used for measurements on the quality of the obtained results, based on my own experience or literature. This additional information can be entered in the "discussion" section.

Response: Thank you for your suggestion. As you said, time with different solar altitude angles will slightly affect the quality of GHPSs data. In this manuscript, due to the high consistency between Rrs(λ)and R(λ) collected from different times and weather after standardized at 574 nm, the impact of time on GHPSs data is considered a systematic error and has been removed. And We added an explanatory sentence as follows (Page 5-6: lines 216-219):

“In addition, upwelling and downwelling irradiance above the water surface were corrected by the ambient light sensor combining a convolutional neural network algorithm to minimize the influence of the skylight and the solar altitude angles (Fig. 1).”

Author Response File: Author Response.docx

Reviewer 2 Report

This manuscript presents a novel ground-based hyperspectral remote sensing system (GHPSs) for automatic, real-time, and continuous observation of TP dynamics under different weather conditions. It's a great innovation and contribution to water quality monitoring using hyperspectral remote sensing, which provides a new available monitoring method to make up for the deficiencies of existing monitoring. New monitoring techniques of the water environment will largely improve our understanding of the water environment key processes and driving mechanisms. However, a moderate revision is needed to improve the quality of this manuscript before this paper can be recommended for publication on RS.

1. Is the maximum observation frequency 20 s or 30 s? The descriptions in line 22 and line 199 are different.

2. Line 50-52: What are the “fluidity and environmental sensitivity” of water bodies? How do they contribute to the challenges behind TP monitoring?

3. Line 104: replace “an TP model” with “a TP model”

4. Line 112: there should be two spaces.

5. Line 125: check the spelling of " Xin’anjaing reservoir"

6. Line 25, 104,154 and 570: check the format of “in situ” for the whole text

7. Line 155: what is “surface water samples (0~0.5 m)” mean?

8. Line 200: A grammatical error should be revised in “that of LR and FR was set to 1 minute”.

9. Line 207: the number of formulates is the worry, and check for the whole text.

10. Line 208: Please give the unit of the parameters in the text, for example, Rrs(λ), Eu and Ed et al.

11. Line 203-209: the reflectance measurements acquired by the GHRS system do not comply with the above-water method, how does the author deal with the influence of skylight?

12. Line 217-220: the author used the NDVI threshold to remove the abnormal data of the water meter completely covered by cyanobacteria bloom. However, under the disturbance of wind and waves, cyanobacteria are sometimes suspended in the water body rather than gathered on the surface of the water body. How does the author deal with the spectrum of this mixed state of algae? Whether it is the abnormal spectrum

13. Line 228: why is the spatial matching standard 0.1 m? How to get it?

14. Line 280: replace “The mean value of TP, SPM……” with “The mean values of TP, SPM……”

15. Line 294: replace “the developed models” with “The developed models”

16. Line 322: the labels of 1:1 should be adjusted in figure 2.

17. Line 358: there should be two spaces at the beginning of the paragraph. and check for the whole text.

18. Line 376-385: GHPSs data are backscattered signals, the relationship between R(λ) and Rrs(λ) is based on the volume scattering function of the component and the external environmental conditions (i.e. sky conditions). So, I think that the Rrs(λ) - R(λ) relationship would vary by the type of optically-active constituents in the water and sky conditions.

19. Line 432-435: literature support needs to be added here.

20. Line 528: the research objects of the paper include lakes, rivers and reservoirs, and we know that the TP concentrations of rivers and lakes are often very different. What is the basis for the classification of water body types at the minute scale in the three study areas in Figure 6? Are the thresholds the same? What is the threshold for each water body? It is suggested to add a specific classification basis of water body type in the material and method part of the paper.

Are the threshold values for the classification of water body types at a minute scale in the three study areas in Figure 6 the same? What is the threshold for each water body? Please supplement the method.

21. What are the limitations of the study?

22. What is the spatial resolution of GHPSs? How much can it observe? Hyperspectral satellite sensors can observe a large area in the water body and that is significant to understanding water quality dynamics in the water, Is the data collected from GHPSs representative in large lakes and reservoirs that the water quality is heterogeneous?

23. What is the topic of this article? Is it to focus on the development of new observation equipment or the remote sensing estimation model of total phosphorus? The result part is the development TP models and dynamic change of TP, and the discussion part is the data stability, reliability, advantages and disadvantages and prospects of GHPSs.

 

24. After reading the manuscript, one question that confused me is the difference between proximal sensing and remote sensing. To my mind, the proximal sensing and remote sensing mentioned in the article are the same.

Author Response

December 28, 2022

 

Response to reviewers of remotesensing-2086088

Dear Editor and Reviewers:

We would like to express our gratitude to the three anonymous reviewers for their critical comments and constructive suggestions regarding our manuscript (remotesensing-2086088). We have revised the manuscript and made a detailed response for the comments of three reviewers point by point below.

To facilitate review of the revisions, we have highlighted the main changes in response to the reviewers’ comments in red text within the document. In addition, in our response letter, we use “….” at the start and end of the new manuscript text to show that it is a new version of the manuscript and thus distinguish between comments in responses and the actual new text in the revised manuscript.

We greatly appreciate your help and that of the three reviewers concerning improvements to this manuscript. We also acknowledge the three anonymous reviewers for their useful reviews of this manuscript. If the reviewers present further comments, we are willing to revise the manuscript again.

We hope that our responses are satisfactory for you and that this revised manuscript is scientifically acceptable to Remote Sensing. Thank you for your patience while we made our revisions.

We are looking forward to hearing from you. Best regards.

 

Dr. Yunlin Zhang

E-mail: [email protected]

Tel. (+86) 25-86882008; Fax (+86) 25-57714759

 

 

Reviewer 2

This manuscript presents a novel ground-based hyperspectral remote sensing system (GHPSs) for automatic, real-time, and continuous observation of TP dynamics under different weather conditions. It's a great innovation and contribution to water quality monitoring using hyperspectral remote sensing, which provides a newly available monitoring method to make up for the deficiencies of existing monitoring. New monitoring techniques of the water environment will largely improve our understanding of the water environment's key processes and driving mechanisms. However, a moderate revision is needed to improve the quality of this manuscript before this paper can be recommended for publication on RS.

Response: Thanks for your suggestions. The point-by-point response is addressed as follows.

 

Q1. Is the maximum observation frequency 20 s or 30 s? The descriptions in line 22 and line 199 are different.

Response: The maximum observation frequency is 20 s, and we have revised it on Page 5 Line 209.

“The spectrometer has a field of view of 3° with a spectral resolution of 1 nm from 400 nm to 1000 nm at the highest frequency of 20 s.”

 

Q2. Line 50-52: What are the “fluidity and environmental sensitivity” of water bodies? How do they contribute to the challenges behind TP monitoring?

Response: Under the effect of wind waves and topography, natural water bodies will have fluidity, and TP concentration will also change frequently with time, especially in some rivers with large flow velocities, lakes with large wind wave disturbances and strong lake currents. In addition, a lake is a gathering of water and erosive substances in the basin. Therefore, the water quality in lakes and rivers is sensitive to the changes in the natural environment and human activities in the basin or a certain distance around them, which puts forward higher requirements for monitoring the dynamics of TP, especially in extreme episodic or unpredictable pollution. And we rewrote the sentence (Page 2: lines 58-61)

“However, spatiotemporal heterogeneity and environmental sensitivity of water quality in lakes and rivers bring great challenges to obtain effective and high-frequency TP data in short-lived, extreme episodic, or unpredictable pollution [14,16].”

 

Q3. Line 104: replace “an TP model” with “a TP model”

Response: We have revised it (Page 3: Line 112)

 

Q4. Line 112: there should be two spaces.

Response: We have indented 2 characters at the beginning of the paragraph (Page 3: Line 121)

 

Q5. Line 125: check the spelling of "Xin’anjaing reservoir"

Response: We replaced “Xin’anjaing reservoir” with “Xin’anjiang reservoir” (Page 3: Line 134)

 

Q6. Line 25, 104,154 and 570: check the format of “in situ” for the whole text

We have set “in situ” to skew and checked the full text (Page 1: Line 25, Page 3: Line 112, Page 4: Line 163)

 

Q7. Line 155: what is “surface water samples (0~0.5 m)” mean?

Response: Surface water samples refer to the mixed water column collected within the range of 0 m to 0.5 m below the water surface. They were all collected by an acid-cleaned plastic water collector with a height of 0.5 m and a volume of 2.5 L. And we have revised the corresponding sentence as follows (Page 4-5: lines 163-167)

“The in situ dataset contained 172, 96 and 109 water samples, which were sampled at LT, LR and FR every 10 or 15 minutes from October 28 to November 3, November 7 to 9 and November 10 to 13, 2020, respectively (Fig. 1a). The mixed water column was collected by a 2.5 L acid-cleaned plastic water collector within range of 0 m to 0.5 m below the water surface.”

 

 

Q8. Line 200: A grammatical error should be revised in “that of LR and FR was set to 1 minute”.

Response: We have revised it on Page 5 line 210.

 

Q9. Line 207: the number of formulates is the worry, and check for the whole text.

Response: Thanks for your suggestions. We have corrected formula numbers (Page 6: line 220)

 

 

Q10. Line 208: Please give the unit of the parameters in the text, for example, Rrs(λ), Eu and Ed et al.

Response: we added the unit of Rrs(λ) Ed, Eu, ,  and  on Page 5 lines 183, 197-198, and 200-201 and Page 6 lines 222.

 

Q11. Line 203-209: the reflectance measurements acquired by the GHPS system do not comply with the above-water method, how does the author deal with the influence of skylight?

Response: Although the accurate Rrs(λ) was obtained using the above-water method which eliminates the influence of skylight through measuring water bodies and sky respectively (section 2.3), the measurement process is cumbersome. In this study, R(λ) collected by GHPSs was the ratio of upwelling and downwelling irradiance above the water surface, which were corrected by the ambient light sensor combining convolutional neural network algorithm to eliminate the influence of skylight. And We added an explanatory sentence as follows (Page 5-6: lines 216-219):

“In addition, upwelling and downwelling irradiance above the water surface were corrected by the ambient light sensor combining a convolutional neural network algorithm to minimize the influence of the skylight and the solar altitude angles (Fig. 1).”

 

Q12. Line 217-220: the author used the NDVI threshold to remove the abnormal data of the water meter completely covered by cyanobacteria bloom. However, under the disturbance of wind and waves, cyanobacteria are sometimes suspended in the water body rather than gathered on the surface of the water body. How does the author deal with the spectrum of this mixed state of algae? Whether it is the abnormal spectrum

Response: Given the spectral characteristic of cyanobacteria bloom is similar to that of vegetation, with high absorption in the red band and high reflection at the near-infrared band, NDVI is used to detect the cyanobacteria bloom in this study. In addition, a significant positive relationship between NDVI and vegetation pigments exists. Therefore, no matter whether the cyanobacteria floats on the water surface or suspends in the water, once the NDVI exceeds 0.4, it means that there are enough cyanobacteria in the water. In this case, TP is mainly determined by cyanobacteria, which is different from the true concentration of TP in the water body. So, this spectrum is eliminated as an abnormal spectrum. And we revised the manuscript for easy understanding as follows (Page 6: Line 229-234):

“Moreover, given the spectral characteristics of cyanobacteria pigments with high absorption in the red band and high reflection in the near-infrared band, Normalized Difference Vegetation Index (NDVI) was introduced to detect and eliminate the abnormal spectral of cyanobacteria blooms completely covering the water surface and obstructing water quality information.”

 

Q13. Line 228: why is the spatial matching standard 0.1 m? How to get it?

Response: Thanks for your comments. The spectrometer of GHPSs with a field of view of 3° placed at a height of approximately 4 m above the surface of the water. According to the following formula, the observation radius value of 0.1 m can be calculated:

tan( )=                             (1)

where w is the field of view, r and h represent the observation radius and the height of the spectrometer of GHPSs (m)

 

Q14. Line 280: replace “The mean value of TP, SPM……” with “The mean values of TP, SPM……”

Response: We have revised it (Page 7: lines 301).

 

Q15. Line 294: replace “the developed models” with “The developed models”

Response: We have replaced “the developed models” with “The developed models” (Page 8: line 314).

 

Q16. Line 322: the labels of 1:1 should be adjusted in figure 2.

Response: We have modified and placed the 1:1 line in the appropriate position (Page 9: lines 336).

 

Q17. Line 358: there should be two spaces at the beginning of the paragraph. and check for the whole text.

Response: We added two characters at the beginning of the paragraph (Page 10: lines 386).

 

 

Q18. Line 376-385: GHPSs data are backscattered signals, the relationship between R(λ) and Rrs(λ) is based on the volume scattering function of the component and the external environmental conditions (i.e. sky conditions). So, I think that the Rrs(λ) - R(λ) relationship would vary by the type of optically-active constituents in the water and sky conditions.

Response: Thanks for your advice. Although the measurement methods of Rrs(λ) and R(λ) are slightly different, they are all the backscattered signals which were related to the volume scattering function of the component and concentration of optically active substances. The light received by the equipment is mainly divided into two parts, one part is the real water-leaving radiation and the error caused by sky light and reflection. Theoretically, no matter what the optically-active constituents in the water is, the water-leaving radiation received by Rrs(λ) and synchronous R(λ) should be the same. Therefore, the different slopes may be related to sky conditions. In addition, the different optically-active constituents showed different spectral characteristics and spectral shapes. Hence, the type of optically-active constituents mainly affects the determination coefficient between Rrs(λ) and R(λ). In general, I thought the slop of Rrs(λ) and synchronous R(λ) was mainly affected by the sky conditions, and the determination coefficient was mainly affected by the composition of matter.

 

Q19. Line 432-435: literature support needs to be added here.

Response: we added three pieces of literature on Page 13 line 473.

 

Q20. Line 528: the research objects of the paper include lakes, rivers and reservoirs, and we know that the TP concentrations of rivers and lakes are often very different. What is the basis for the classification of water body types at the minute scale in the three study areas in Figure 6? Are the thresholds the same? What is the threshold for each water body? It is suggested to add a specific classification basis of water body type in the material and method part of the paper.

Are the threshold values for the classification of water body types at a minute scale in the three study areas in Figure 6 the same? What is the threshold for each water body? Please supplement the method.

Response: There are different thresholds of water type between lakes (including reservoir) and rivers according to TP in Environmental Quality Standard for Surface Water (https://www.mee.gov.cn/ywgz/fgbz/bz/bzwb/shjbh/shjzlbz/200206/W020061027509896672057.pdf). The classification thresholds of 6 grades of water bodies are shown in the following table.

Water types

Lakes and reservoirs

River

Class â… 

≤ 0.01

≤ 0.02

Class â…¡

0.01<TP ≤ 0.025

0.02<TP ≤ 0.1

Class â…¢

0.025<TP ≤ 0.05

0.1<TP ≤ 0.2

Class â…¥

0.05<TP ≤ 0.1

0.2<TP ≤ 0.3

Class â…¤

0.1<TP ≤ 0.2

0.3<TP ≤ 0.4

Inferior Class â…¤

0.2<TP

0.4<TP

Therefore, we added a new section “Classification of lake and river” in the manuscript on Page 7: lines 269-275

2.7. Classification of lake and river

The water of lakes (including reservoirs) and rivers can be divided into 6 types according to TP pollution in Environmental Quality Standard for Surface Water (https://www.mee.gov.cn/ywgz/fgbz/bz/bzwb/shjbh/shjzlbz/200206/W020061027509896672057.pdf). Specific classification thresholds are given in Table 1.

Table 1. The classification thresholds of 6 water types according to TP

Water types

Lakes and reservoirs

River

Class â… 

≤ 0.01

≤ 0.02

Class â…¡

0.01<TP ≤ 0.025

0.02<TP ≤ 0.1

Class â…¢

0.025<TP ≤ 0.05

0.1<TP ≤ 0.2

Class â…¥

0.05<TP ≤ 0.1

0.2<TP ≤ 0.3

Class â…¤

0.1<TP ≤ 0.2

0.3<TP ≤ 0.4

Inferior Class â…¤

0.2<TP

0.4<TP

 

Q21. What are the limitations of the study?

Response: There are three limitations of this study described in 4.2 section. (1) The single-point hyperspectral imager needs to upgrade for obtaining a wide field and more information. (2) To build a more robust and reliable water quality estimation algorithm, more water samples with complex spectral characteristics need to be collected from rivers, lakes, reservoirs and ponds. (3) More diverse algorithms need to be introduced for establishing stable and general estimation models, including deep learning, semi-analysis and analysis algorithms (Page 14: lines 524-533).

 

Q22. What is the spatial resolution of GHPSs? How much can it observe? Hyperspectral satellite sensors can observe a large area in the water body and that is significant to understanding water quality dynamics in the water, Is the data collected from GHPSs representative in large lakes and reservoirs that the water quality is heterogeneous?

Response: The GHPSs is a single-point spectrometer, which can only observe one point below the spectrometer. Just as you said, a large observation is the advantage of the hyperspectral satellite sensors. For large lakes and reservoirs where the water quality is heterogeneous, the data collected from GHPSs only represents the temporal variations of nearby water bodies and cannot display the spatial variations of the whole lake or reservoir. It is vital to monitor TP dynamics with high temporal resolution for key water quality section in water environment management. In addition, the spatial distribution of TP can be obtained by deploying a large number of equipment and spatial interpolation methods.

 

Q23. What is the topic of this article? Is it to focus on the development of new observation equipment or the remote sensing estimation model of total phosphorus? The result part is the development of TP models and dynamic change of TP, and the discussion part is the data stability, reliability, advantages and disadvantages and prospects of GHPSs.

Response: We are grateful for your patience. The paper is about a new high-frequency TP monitoring instrument called ground-based hyperspectral proximal sensing system (GHPSs), in which the TP estimation model is a part of the GHPSs. High-frequency observation and applicability of complex weather conditions are two of the main advantages, which overcome the problem of monitoring frequency in remote sensing, limited observation time of the unmanned aerial vehicle, and the sample residue of in-situ underwater monitoring equipment. Due to the robustness and accuracy of GHPSs data and the TP model being the core part, we developed the TP model and applied it to the time-series of GHPSs data in the result part, and discussed the robustness and accuracy of GHPSs data as well as the advantages and potential implications of GHPSs in monitoring TP.

 

Q24. After reading the manuscript, one question that confused me is the difference between proximal sensing and remote sensing. To my mind, the proximal sensing and remote sensing mentioned in the article are the same.

Response: Proximal sensing and remote sensing are not new concepts. Remote sensing means detecting objects from a distance, whose applicable scene is satellite or UAV remote sensing. Proximal sensing proximal sensing refers to close detection of targets. In my manuscript, they are all non-contact sensing based on the spectral characteristics of the target. The basic principle is that separating the characteristic bands from the satellite sensor containing water body information jointly determined by the absorption and scattering of each component concentration in the water body, then the material component and concentration of the water body were estimated according to the relationship between characteristic bands and the absorption and scattering characteristics of the water body. Since GHPSs are only 4 m away from the water surface which is much closer than satellite remote sensing, “proximal sensing” is selected in this manuscript.

Author Response File: Author Response.docx

Reviewer 3 Report

- The abstract should state briefly the purpose of the research, the principal results and major conclusions. An abstract is often presented separately from the article, so it must be able to stand alone.

- The major defect of this study is the debate or Argument is not clear stated in the introduction session. Hence, the contribution is weak in this manuscript. I would suggest the author to enhance your theoretical discussion and arrives your debate or argument.

-  It is mentioned in p.4 that “After the samples were decomposed by alkaline potassium persulfate obeying the “Standard Methods for the Examination of Water and Wastewater”, the TN and TP were measured by a Shimadzu UV-2550 PC UV-Vis spectrophotometer [22,38].” What are other feasible alternatives? What are the advantages of adopting this particular approach over others in this case? How will this affect the results? The authors should provide more details on this.

-  More explanations should be presented regarding figure 5. It is unclear.

- Methods section determines the results. Kindly focus on three basic elements of the methods section.

a. How the study was designed?

b. How the study was carried out?

c. How the data were analyzed?

- Much more explanations and interpretations must be added for the Results, which are not enough.

- More suitable title should be selected for the figure 3 instead of “Short-term series of TP variations in Lake Taihu (a), Liangxi River (b) a…...”.

- It is suggested to add articles entitled Nkansah et al. Preliminary Studies on the Use of Sawdust and Peanut Shell Powder as Adsorbents for Phosphorus Removal from Water” and “Hussain & Al-Fatlawi. Remove Chemical Contaminants from Potable Water by Household Water Treatment System to the literature review.

- Page 7: the following paragraph is unclear, so please reorganize that:

“In addition, the retrieved TP was in good agreement with the measured TP, which is evenly distributed around the 1:1 line. Therefore, to reduce the estimation error as much as possible, the XGBoost model for TP estimation with the highest determination coefficient and low errors was selected, which indicated that the model could quantify the dynamics of TP with the satisfactory performance and good applicabilit.

- Please make sure your conclusions' section underscore the scientific value added of your paper, and/or the applicability of your findings/results, as indicated previously. Please revise your conclusion part into more details. Basically, you should enhance your contributions, limitations, underscore the scientific value added of your paper, and/or the applicability of your findings/results and future study in this session.

Author Response

December 28, 2022

 

Response to reviewers of remotesensing-2086088

Dear Editor and Reviewers:

We would like to express our gratitude to the three anonymous reviewers for their critical comments and constructive suggestions regarding our manuscript (remotesensing-2086088). We have revised the manuscript and made a detailed response for the comments of three reviewers point by point below.

To facilitate review of the revisions, we have highlighted the main changes in response to the reviewers’ comments in red text within the document. In addition, in our response letter, we use “….” at the start and end of the new manuscript text to show that it is a new version of the manuscript and thus distinguish between comments in responses and the actual new text in the revised manuscript.

We greatly appreciate your help and that of the three reviewers concerning improvements to this manuscript. We also acknowledge the three anonymous reviewers for their useful reviews of this manuscript. If the reviewers present further comments, we are willing to revise the manuscript again.

We hope that our responses are satisfactory for you and that this revised manuscript is scientifically acceptable to Remote Sensing. Thank you for your patience while we made our revisions.

We are looking forward to hearing from you. Best regards.

 

Dr. Yunlin Zhang

E-mail: [email protected]

Tel. (+86) 25-86882008; Fax (+86) 25-57714759

 

 

Reviewer 3

Q1: The abstract should state briefly the purpose of the research, the principal results and major conclusions. An abstract is often presented separately from the article, so it must be able to stand alone.

Response: Thanks for your suggestion and we rewrote the abstract part as follows (Page 1: Line 17-43):

“Total phosphorus (TP) is the main limiting factor of eutrophication for global most inland waters. However, the traditional low temporal-spatial manual sampling, poor spectral resolutions and weather-vulnerable satellite observations yielded great data gaps of TP dynamics in short-lived, extreme episodic, or unpredictable pollution. Hence, a novel ground-based hyperspectral proximal sensing system (GHPSs) with the maximum observation frequency of 20 s and a spectral resolution of 1 nm from 400 to 900 nm was firstly developed for automatic, real-time and continuous observation of TP. Focusing on GHRSs, the TP machine learning model was developed and validated with ideal accuracy (R2 = 0.97, RMSE = 0.017 mg·L-1, MAPE = 12.8%) using 377 pairs of in situ TP measurements collected from Fuchunjiang Reservoir (FR), Liangxi River (LR) and Lake Taihu (LT). Second-scale TP results showed a low-value stable period and a sharp change period in LT from October 29~31 and November 1~3, respectively. The exponential increase (R2 = 0.65, p < 0.05) on November 1 and the two complete variations with peak values of 0.32 mg·L-1 and 0.42 mg·L-1 were recorded in LT on November 2 and 3 respectively. Simultaneously, a significant decrease (R2 = 0.57, p < 0.05) with the observation days was obtained in LR and no obvious change were observed in FR, respectively. High consistency between GHPSs spectral standardized at 574 nm and the measured reflectance in different weather demonstrated the accuracy of GHPSs spectral (R2 > 0.99, slop = 0.98). Short and rapid TP changes were observed within one day in LT and LR based on GHPSs minute scale monitoring, which highlighted the importance of high frequency observation of TP. Several advantages of real-time, high accuracy and wide applicability to complex weather of GHPSs for TP monitoring were highlighted comparing the traditional equipment. Therefore, the potential application of GHPSs in the integrated space-air-ground TP monitoring, as well as emergency monitoring and early-warning system in the future, can raise our awareness of the dynamics and driving mechanisms of water quality for inland waters.”

 

Q2: The major defect of this study is the debate or Argument is not clearly stated in the introduction session. Hence, the contribution is weak in this manuscript. I would suggest the author enhance the theoretical discussion and arrives at your debate or argument.

Response: Thank you for your suggestion. The paper is about a new high-frequency TP monitoring instrument called ground-based hyperspectral proximal sensing system (GHPSs). High-frequency observation and applicability of complex weather conditions are two of the main advantages, which overcome the problem of monitoring frequency in remote sensing, limited observation time of the unmanned aerial vehicle, and the sample residue of in-situ underwater monitoring equipment. Due to the robustness and accuracy of GHPSs data and the TP model being the core part, we developed the TP model and applied it to the time-series of GHPSs data in the result part, and discussed the robustness and accuracy of GHPSs data as well as the advantages and potential implications of GHPSs in monitoring TP.

Therefore, in the introduction, we described the ecological significance of TP and the serious environmental problems caused by TP to highlight the importance and necessity of high-frequency TP monitoring. In the second paragraph, we reviewed the existing TP monitoring equipment with the shortcomings of low monitoring frequency, data missing and inaccurate data, etc., to show the necessity of high-frequency equipment development. In this condition, we introduced the novel GHPSs for TP monitoring and its design concept in the third paragraph. Finally, we gave the purposes of this manuscript. Hence, to enhance our theoretical discussion, we put forward the argument at the beginning of each paragraph (Page 2: lines 65-69, 95-97 and Page 3: lines 109-111).

“However, the main existing monitoring methods and measurement equipment of TP have lagged behind the needs of water environment management and decision-making departments in terms of data collection frequency, timeliness and representativeness, especially some sudden TP pollution events.”

“Therefore, developing real-time, high-frequency, continuous and reliable TP monitoring equipment under complex weather conditions is the primary challenge to be solved.”

“In this paper, we first proposed a high-frequency, real-time automated equipment namely GHPSs and elucidate the feasibility, advantages as well as potential applications for tracking TP dynamics.”

 

Q3: It is mentioned on p.4 that “After the samples were decomposed by alkaline potassium persulfate obeying the “Standard Methods for the Examination of Water and Wastewater”, the TN and TP were measured by a Shimadzu UV-2550 PC UV-Vis spectrophotometer [22,38].” What are other feasible alternatives? What are the advantages of adopting this particular approach over others in this case? How will this affect the results? The authors should provide more details on this.

Response: Measuring principle of total phosphorus: polyphosphates and other phosphorus-containing compounds in water are hydrolyzed under high temperature and high pressure under acidic conditions to generate phosphate radical; For other phosphorus compounds that are difficult to oxidize, they are oxidized to phosphate by strong oxidant sodium persulfate. There are two methods to measure TP as follows: (1) Molybdenum blue method: phosphate ions react in a strong acid solution containing molybdate to form a yellow molybdophosphate complex, which is reduced to blue molybdophosphate by ascorbic acid. Measure the absorbance of phosphomolybdate at the wavelength of 850nm, and compare it with the standard to obtain the total phosphorus content of the sample. (2) Molybdenum alum method: phosphate ion reacts in a strong acid solution containing molybdenum salt to generate yellow phosphomolybdate complex, and alum reacts with this compound to generate yellow alum phosphomolybdic acid. Measure the absorbance of aluminophosphomolybdic acid at the wavelength of 430nm, and compare it with the standard to obtain the total phosphorus content of the sample.

In this manuscript, the TP measurement method we adopt is not special compared with other methods, and it is often used to measure the TP concentration in surface water with a large range of 0.01 mg/L to 0.6 mg/L. Considering the water samples collected from the highly eutrophic Lake Taihu and the poor eutrophic Fuchunjiang Reservoir, we choose the spectrophotometric analysis after the potassium persulfate decomposed. Accurate in situ TP data, directly and indirectly, influence the accuracy and robustness of the TP estimation model. Hence, the accuracy of high and low TP values is very important. Given that the TP measurement method has been mature and described in many literatures in detail, we will not repeat it but quote the literature.

 

Q4: More explanations should be presented regarding figure 5. It is unclear.

Response: In this manuscript, figure 5a displayed the time and cloud conditions of the two satellites namely MODIS (square) and GOCI (triangle) crossing Lake Taihu from October 29 to November 3, 2020. The green and blue represent cloudless while the black represents the cloud covering Lake Taihu. Figure 5b is the distribution of the monthly cloudless MODIS image covering LT during 2003~2020. In figure 5b, 70.8% of the months had less than 10 effective images, and 96.8% of the months were less than 15 effective images. Therefore, the lack of enormous image data caused by cloud and fixed revisit time was an important reason that hinders the continuous recording of the water environment by satellites. And we rewrote the sentences as follows (Page 13: Line 489-499):

“Figure 5a displayed the time and cloud conditions of the Moderate Resolution Imaging Spectroradiometer (MODIS, square) and Geostationary Ocean Color Imager (GOCI, triangle) crossing LT from October 29 to November 3, 2020. The green and blue represent cloudless while the black represents the cloud covering LT. The effective cloudless images of MODIS and GOCI were less than 50% during observation (Fig.5a). Moreover, figure 5b was the distribution of the monthly cloudless MODIS image covering LT during 2003~2020. 70.8% of the months were less than 10 effective images and 96.8% of the months were less than 15 effective images. Therefore, the lack of enormous image data caused by cloud and fixed revisit time was an important reason that hinders the continuous recording of the water environment by satellites.”

 

Q5:Methods section determines the results. Kindly focus on three basic elements of the methods section.

  1. How the study was designed?
  2. How the study was carried out?
  3. How the data were analyzed?

Response: Thanks for your comments. Just as you said, the method section determines the results. In this manuscript, the reasons for organizing the methods section are as follows:

In the 2.1 study area section, we described the information of the three study areas and high nutrient loading promoted eutrophication which posed threat to the water supply.

In the 2.2 GHPSs framework section, we proposed the compositions and data flow of the novel GHPSs, so that readers can understand how GHPSs work.

In the 2.3 field data collection and measurement section, we described the collection, pre-processing and measurement process of in situ data including water quality and the remote-sensing reflectance.

In the 2.4 GHPSs dataset and preprocessing section, we described the specific equipment parameters and time of GHPS data acquisition, and how to obtain and preprocess GHPSs data.

In 2.5 matchups between GHPS and field data section, to obtain the TP estimation model, it is necessary to acquire the dataset matched by the measured TP and GHPSs data. Therefore, in this section, matching criteria are specified here.

In the 2.6 section, we introduced the empirical model and machine learning mothed (XGBoost) to develop the TP model.

In the 2.7 section, we gave the statistics analysis and accuracy assessment indexes, and how the data were analyzed.

In the 2.8 section, we gave the specific classification thresholds of 6 water types according to TP in lakes and rivers.

Therefore, we have organized the methods section according to the three basic elements.

 

Q5: Much more explanations and interpretations must be added for the Results, which are not enough.

Response: Thanks for your comments. In my opinion, the result section is the result of data analysis without too much explanation. In the “Materials and Methods” section, we gave a detailed description of how to get and deal with the data as well as the modeling process. So, we will not repeat but directly present the water quality conditions and modeling results of the empirical and machine learning method, as well as the variations of TP derived from the optimal model and GHPSs data. And we added three tables and additional information as explanatory material (Page 8: line 310, Page 9: line 335 and Page 9: line 349-361)

“Table 2. Statistical analyses of in situ measurements were collected from three study areas.

 

Parameters

Max.

Min.

Mean

S.D.

LT

2020.10.28~11.3

N = 172

SPM mg·L-1

127.83

25.37

43.05

17.88

TN mg·L-1

6.73

0.98

1.66

1.01

TP mg·L-1

0.62

0.08

0.14

0.10

Chla μg·L-1

442.94

3.08

52.26

75.89

LR

2020.11.7~11.9

N = 96

SPM mg·L-1

61.90

19.86

40.66

9.37

TN mg·L-1

2.73

0.93

1.54

0.44

TP mg·L-1

0.21

0.06

0.12

0.04

Chla μg·L-1

123.60

11.34

45.72

29.09

FR

2020.11.10~11.13

N = 109

SPM mg·L-1

16.88

6.92

11.79

2.11

TN mg·L-1

2.17

0.93

1.76

0.21

TP mg·L-1

0.10

0.04

0.05

0.01

Chla μg·L-1

1.72

0.70

1.11

0.19

Overall data

2020.10.28~11.13

N = 377

SPM mg·L-1

127.83

6.92

33.49

18.96

TN mg·L-1

6.73

0.93

1.66

0.73

TP mg·L-1

0.62

0.04

0.11

0.08

Chla μg·L-1

442.94

0.70

35.76

57.65

“Table 3. The R2 of four fitting models for TP with optimal spectral indexes

Spectral index

Linear

Exponential

Logarithmic

Power formulations

R(800)

0.75

0.73

0.54

0.72

R(750)+R(800)

0.75

0.74

0.53

0.72

R(690)-R(710)

0.83

0.71

/

/

R(740)/R(670)

0.85

0.64

0.68

0.85

(R(510)-R(520))/(R(510)+R(520))

0.76

0.76

/

/

“To further verify the superiority of our model, we introduced three published TP models for comparison (Table 4). In the 6 models, our XGBoost TP model outperformed better than the YH-model proposed by Xiong et al (R2 = 0.64, RMSE= 0.06 mg·L-1 and MAPE = 34.13%), the semi-analytical TP model proposed by Du et al. based on the absorption of 675 nm (R2 = 0.87, RMSE= 0.04 mg·L-1 and MAPE = 16.8%), and the empirical TP model proposed by Xiong et al for Lake Hongze ( R2= 0.75, RMSE = 0.03 and MAPE = 37.6%).

Therefore, to reduce the estimation error as much as possible, the XGBoost model for TP estimation with the highest determination coefficient and low errors was selected to quantify the dynamics of TP with satisfactory performance and good applicability.

 

Table 4 The R2, RMSE and MAPE of different TP estimation models

Models

R2

RMSE (mg·L-1)

MAPE

Source

Linear model

0.84

0.033

15.5%

This study

Power model

0.83

0.037

15.0%

This study

XGBoost model

0.97

0.017

11.8%

This study

YH-TP model

0.64

0.06

34.13%

[14]

Semi-analytical model

0.87

0.04

16.8%

[17]

Empirical model

0.75

0.03

37.6%

[40]

 

Q6: More suitable title should be selected for figure 3 instead of “Short-term series of TP variations in Lake Taihu (a), Liangxi River (b) a…...”.

Response: we rewrote the title of figure 3 as follows (Page 11: lines 400-404):

“Time series of TP derived from GHPSs and XGboost TP estimation model in Lake Taihu (a) from October 29 to November 3, 2020; Liangxi River (b) from November 7-9, 2020 and Fuchunjiang Reservoir (c) from November 11-13, 2020. Note the data of the grey represent the abnormal data with NDVI values above 0.4, which are not used for statistics and calculation.”

 

Q7: It is suggested to add articles entitled “Nkansah et al. Preliminary Studies on the Use of Sawdust and Peanut Shell Powder as Adsorbents for Phosphorus Removal from Water” and “Hussain & Al-Fatlawi. Remove Chemical Contaminants from Potable Water by Household Water Treatment System” to the literature review.

Response: Thanks for your suggestion. We have added the two papers in the manuscript on Page 2 lines 52 and 54.

 

Q8: Page 7: the following paragraph is unclear, so please reorganize that:

“In addition, the retrieved TP was in good agreement with the measured TP, which is evenly distributed around the 1:1 line. Therefore, to reduce the estimation error as much as possible, the XGBoost model for TP estimation with the highest determination coefficient and low errors was selected, which indicated that the model could quantify the dynamics of TP with satisfactory performance and good applicability.

Response: we rewrote the sentences as follows (Page 9-10: lines 346-360):

“In addition, compared with the models based on the ratio of 740 nm and 670 nm, the retrieved TP based on XGBoost estimation model and the in situ TP are evenly distributed around the 1:1 line. The result implied that the developed XGBoost models performed well in estimating TP from GHPSs data. To further verify the superiority of our model, we introduced three published TP models for comparison (Table 4). In the 6 models, our XGBoost TP model outperformed better than the YH-model proposed by Xiong et al (R2 = 0.64, RMSE= 0.06 mg·L-1 and MAPE = 34.13%), the semi-analytical TP model proposed by Du et al. based on the absorption of 675 nm (R2 = 0.87, RMSE= 0.04 mg·L-1 and MAPE = 16.8%), and the empirical TP model proposed by Xiong et al for Lake Hongze ( R2= 0.75, RMSE = 0.03 and MAPE = 37.6%).

Therefore, to reduce the estimation error as much as possible, the XGBoost model for TP estimation with the highest determination coefficient and low errors was selected to quantify the dynamics of TP with satisfactory performance and good applicability.

 

Q9; Please make sure your conclusions' section underscore the scientific value added of your paper, and/or the applicability of your findings/results, as indicated previously. Please revise your conclusion part into more details. You should enhance your contributions, and limitations, underscore the scientific value added to your paper, and/or the applicability of your findings/results and future study in this session.

Response: Thank you for your advice. We have revised the conclusion section as follows: (Page 16: lines 586-603)

“In this study, real-time high-frequency and automatic GHPSs was first proposed for tracking TP dynamics, with characterized by a spectral resolution of 1 nm from 400 nm to 900 nm and a minimum interval of 20 s. A TP machine learning model was developed and validated with ideal accuracy based on XGBoost method using 377 pairs of synchronous samples (R2 = 0.97, RMSE = 0.017 mg·L-1, MAPE = 12.8%). Subsequently, compared with traditional TP monitoring equipment, GHPSs, with the advantages of simple operation, high-frequency, real-time, accurate and suitable for complex weather, supplement the defects of low monitoring frequency, poor timeliness and accuracy of existing monitoring equipment, complement the lack of TP data in overcast and cloudy weather. Short and rapid TP changes were observed within one day in LT and LR based on GHPSs minute scale monitoring, which highlighted the importance of high frequency observation of TP. In the future, GHPSs have great potential value and application prospects in raising our awareness of the dynamics and driving mechanisms of water quality for inland waters. Therefore, our findings will contribute to the deep understanding of short-term dynamics and long-term trends of TP concentration by using GHPSs, which greatly improve water environment management and prediction precision of algal bloom.”

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

After carefully read the revised MS, I think all issues I raised have been addressed. My concerns are closed.

Reviewer 3 Report

The article has been revised very well, so I would suggest to accept in its present form.

Back to TopTop