The Use of Spectral Indices to Recognize Waterlogged Agricultural Land in South Moravia, Czech Republic

Round 1

Reviewer 1 Report

This manuscript mainly studies the possibility of using a hyperspectral aerial survey for the diagnosis of waterlogged areas in the agricultural landscape, and uses the maximum entropy model (MAXENT) in the analysis of these indices. However, the authors should show more evidences to prove which indices relating to waterlogging are more useful in this study.

 

Major issues:

1. In the section of abstract, why NVI (New vegetation index) and CARI (Chlorophyll Absorption Ratio Index) are the most useful? Please list the specific results of the experiment to prove this conclusion.Moreover, please use other analysis methods to compare the results with the maximum entropy model. So the results would be more convincing.

2. In the section of study area, the periods of aerial imaging at different locations are not the same. Please try to control a single variable and collect more imaging.Please provide more detailed information of aerial imaging, such as shooting time.

3. In Fig. 8, the name of figure does not match the figure.

4. Lack of sufficient explanation of the results. Please explain your results in detail.

5. In the section of Discussion, the conclusion that the NVI index sensitive to foliage cover was the most sensitive in terms of waterlogged areas need more explanation.

6. The current manuscript needs to be polished by a native English speaker ora professional language editing service.

 

Minor issues：

1. In Fig. 3,the scale of the first figure is missing

2. In Fig. 4, the figure are messy. Please re-plot it.

3. In Fig. 5, the compass of the figure is missing.

Author Response

The use of spectral indices to recognise waterlogged agricultural land in South Moravia, Czech Republic – Responses to Opponent reviews

Thank you for all the comments and feedback; we have tried to incorporate them all in the updated version of the manuscript.

First reviewer:

In the section of abstract, why NVI (New vegetation index) and CARI (Chlorophyll Absorption Ratio Index) are the most useful?

This sentence has been reformulated to make it more understandable. The statement is based on the average percentage contribution to identifying (creating probability maps) water-logged areas. [lines 20-21]

Please list the specific results of the experiment to prove this conclusion.

The abstract has been revised accordingly. [10-29]

Moreover, please use other analysis methods to compare the results with the maximum entropy model. So the results would be more convincing.

Well, for sure, you can use some “combined” SDM models, e.g., glodep2 R library, which can incorporate more methods such as Random Forest, Boosted Trees, Support Vectors, Maxent, and others, but with the disadvantage of less-specific parametrization or inappropriateness for the investigated data, which could then raise problems with interpreting results, I can imagine that one unsuitable method can lower the overall results simply by averaging the results of different models.

Maybe if we know that there is an xxx index that is sensitive enough to identify waterlogging areas, we can use this approach to create the best probability maps.

But this does not apply in our case. We decided on another approach – selecting just one method (which, by the way, provides one of the best results, according to many reviews) and set it up correctly based on many steps carried out with the model actually running. Now we have a general idea of the possibility and accuracy of using Remote sensing for identifying these specific waterlogged areas, for which we could not find any relevant reference in remote sensing research.

In the section of study area, the periods of aerial imaging at different locations are not the same. Please try to control a single variable and collect more imaging. Please provide more detailed information of aerial imaging, such as shooting time.

The shooting time was present in the form of a date. We have added the time of shooting. The table has been updated. [lns. 155-156]

Well, as we write in the article, it would be nice to have much more hyperspectral imaging data. Unfortunately, such data are unavailable for the specific area, not to mention the extreme cost. This is the first attempt, as far as we know, to distinguish these particular waterlogged areas by means of hyperspectral indices. So, yes, it would be valuable, and we plan to extend the research in this way, but first, we need to know if this identification is at least possible. We tried to structure the analysis into two single variables –different seasons in one area, and different places at the same time. Unfortunately, more images were, and still are, unavailable for this study area, but the results of our research can support further investigation and more frequent shooting of hyperspectral images.

 

Lack of sufficient explanation of the results. Please explain your results in detail.

Well, we have added some paragraphs about the role of chlorophyll and LAI, about its role in the identification of waterlogged areas. Recently (2022), two studies (references and text have been updated) have explicitly described the effect of chlorophyll in comparison with LAI, but with no descriptive explanation in relation to our topic. We obtained some valuable results, and, in order to explain more thoroughly, further research should be done. This should focus on field surveys rather than remote sensing. [lns. 421-437]

 

In the section of Discussion, the conclusion that the NVI index sensitive to foliage cover was the most sensitive in terms of waterlogged areas need more explanation.

 

The discussion part has been slightly extended and modified. Also, NVI sensitivity is now mentioned more in the Abstract.[421-437][25-29]

 

The current manuscript needs to be polished by a native English speaker or a professional language editing service

The manuscript has been checked by a native-English reader with extensive experience of translation in this field. We include the confirmation.

 

Minor issues：

1. In Fig. 3,the scale of the first figure is missing

Done.

2. In Fig. 4, the figure are messy. Please re-plot it.

We have tried to replot it. For better interpretation, we have changed the symbol type and also added jittering.

3. In Fig. 5, the compass of the figure is missing.

Added.

 

 

 

Author Response File: Author Response.docx

Reviewer 2 Report

This manuscript by Marek is generally well written. This study takes South Moravia, one of the warmest and driest regions in the Czech Republic, as an example, and uses the maximum entropy model (MAXENT) to analyze and evaluate 33 spectral indexes related to soil waterlogging in previous studies, describing the possibility of using hyperspectral aerial surveys to diagnose waterlogged areas in agricultural landscapes. It is concluded that the spectral index based on chlorophyll usually produces the best results regardless of the season. Of all the other indices only one can be used, namely the NVI, especially in springtime. The lower overall sensitivity allows the use of Chlorophyll-based and NVI indices for the initial identification of waterlogged sites, which should then be confirmed by field survey. This paper is very meaningful but there are several issues which must be solved and clarified before it is considered for publication.

 

General comments:

1)      It is noted that your manuscript needs careful editing by someone with expertise in technical English editing paying particular attention to English grammar, spelling, and sentence structure so that the goals and results of the study are clear to the reader.

2)      The proposed methods are too heuristic. A strong theoretical basis is necessary.

3)      The size and format of the pictures and tables in the manuscript need to be further adjusted and standardized.

4)      The discussion part did not describe the future research plan, and did not describe where the next research of this article is directed.

5)      In the conclusion, it is recommended to summarize by points

 

Specific minor comments:

1)      The article begins to say that the Czech Republic faces the problem of drought, but the research is to identify the farmland of waterlogging disaster, which is a bit contradictory.

2)      In the keyword section, Maxent should be followed by ";" instead of ",".

3)      In line 103 of the full text, it is said that the research uses statistical methods based on logical regression, GLM, or more complex nonparametric methods in the form of random forest methods, but the following method research uses the maximum entropy model (MAXENT), please explain it clearly.

4)      The blue boundary in Figure 5a is not obvious in the figure and is difficult to find. Please mark the area with a bright color.

5)      The last paragraph on page 5 does not explain what the red areas in Figures 5b, 5c and 5d represent

6)      In the "This section is not mandatory but can be added to the manuscript if the discussion is unusually long or complex" part of the conclusion, do not use descriptive statements.

Author Response

The use of spectral indices to recognise waterlogged agricultural land in South Moravia, Czech Republic – Responses to Opponent reviews

Thank you for all the comments and feedback; we have tried to incorporate them all in the updated version of the manuscript.

It is noted that your manuscript needs careful editing by someone with expertise in technical English editing paying particular attention to English grammar, spelling, and sentence structure so that the goals and results of the study are clear to the reader.

Done. The manuscript has been checked by a native-English reader with  extensive experience of translation in this in this field. We include the confirmation.

The proposed methods are too heuristic. A strong theoretical basis is necessary.

We agree that heuristics are, to some extent, applied in the overall processing of Maxent models but not in the methodology of preparing and running the Maxent model itself (in this case, we followed the recommended procedure mentioned in the Bantalab laboratory). Of course, there are other preprocessing procedures (such as variable selection through PCA analysis) or the absence of variable selection. Still, even here, there are specific heuristics, and it is not clear which approach is the best. As for subsequent processing, it is essentially just averaging the results (success) of individual indices. If the indices were replaceable (not selected on the basis of correlation with a significant variable), their contribution was linearly dependent on  correlation with  a variable  corresponding to logic. Information in line with this explanation has been added in the discussion section. [lns. 369-375]

The size and format of the pictures and tables in the manuscript need to be further adjusted and standardized.

The magazine's editorial team determines the size and format of the image. We have fulfilled these conditions. Within the template of the final document, there were only two options – to include the image (table) in one column or two columns. Also, the final layout seems strange to us, but, in our opinion, that is a question of graphic preparation of the article in the editorial team.

The discussion part did not describe the future research plan and did not describe where the next research of this article is directed.

In the discussion and conclusion, we have added a plan for future research, which also depends on the results achieved in this research. [lns. 435-437, 513-515]

In the conclusion, it is recommended to summarize by points

We have revised the conclusion as per recommendations.[493-519]

The article begins to say that the Czech Republic faces the problem of drought, but the research is to identify the farmland of waterlogging disaster, which is a bit contradictory.

Perhaps a better expression than "drought" would be "agricultural drought," where crops do not have enough water in a given period. The total amount of precipitation may decrease due to climate change, but the timing of rainfall plays a more prominent role. This is why identifying waterlogged areas is essential, as appropriate measures could be implemented to retain water and make it available during dry periods. This explanation has been added to the article. [33-39]

In the keyword section, Maxent should be followed by ";" instead of ","

Done.

 

In line 103 of the full text, it is said that the research uses statistical methods based on logical regression, GLM, or more complex nonparametric methods in the form of random forest methods, but the following method research uses the maximum entropy model (MAXENT), please explain it clearly.

We have corrected the relevant text. The sentence was meant differently.

The blue boundary in Figure 5a is not obvious in the figure and is difficult to find. Please mark the area with a bright color.

Done.

The last paragraph on page 5 does not explain what the red areas in Figures 5b, 5c and represent.

This is just the border of the waterlogged area. Information has been added to the Figure description.

 

In the "This section is not mandatory but can be added to the manuscript if the discussion is unusually long or complex" part of the conclusion, do not use descriptive statements.

Sorry, this part has been removed. It is a remnant of the Word template.

 

Author Response File: Author Response.docx

Reviewer 3 Report

1.      Title: revise the title to “The use of spectral indices to recognise waterlogged agricultural land in South Moravian, Czech Republic”

2.      Please thoroughly review the manuscript for its grammar.

3.      Abstract: Appropriate

4.      Introduction: The content of the section needs revising. The author should focus on adding reviews and descriptions of other related works. What is state-of-the-art? Their findings and challenges, etc.

5.      Study Area: add one paragraph on the Socio-economic and Physio-Environmental setup of South Moravian, Czech Republic.

6.      Methodology: Make the section more concise. The flow chart must be placed at the beginning of the methodology section. Describe this in the manuscript accordingly.

7.      Results: Must be improved

8.      Discussion:  Appropriate

9.      Conclusion: "Must be improved"

The section needs to be reconstructed. The conclusion is not based on the findings that were presented in the main body of the manuscript, since the validation process was not performed to show the accuracy of the estimation, and there are no data that shows the accuracy assessment. Add some key recommendations for decision and policymakers

 

Author Response

The use of spectral indices to recognise waterlogged agricultural land in South Moravia, Czech Republic – Responses to Opponent reviews

Thank you for all the comments and feedback; we have tried to incorporate them all in the updated version of the manuscript.

Title: revise the title to “The use of spectral indices to recognise waterlogged agricultural land in South Moravian, Czech Republic”

Done.

Please thoroughly review the manuscript for its grammar.

Done. The manuscript has been checked by a native-English reader with  extensive experience of translation in this  field. We include the confirmation.

 

The content of the section needs revising. The author should focus on adding reviews and descriptions of other related works. What is state-of-the-art? Their findings and challenges, etc.

We would like to include the results of other professional studies on identifying waterlogged areas. However, we have yet to find any relevant references in literature on waterlogged areas as we understand them in the context of agricultural land. To the best of our knowledge, specific research on this topic has yet to be conducted, which we find surprising. We would be happy to expand the introduction if you know of similar studies.

 

Add one paragraph on the Socio-economic and Physio-Environmental setup of South Moravian, Czech Republic

Done. [lns. 129-143]

 

Make the section more concise. The flow chart must be placed at the beginning of the methodology section. Describe this in the manuscript accordingly

A flow chart has been added. The text has been updated with references to the flow chart.

Results: Must be improved

The complete results are too extensive to be included in the article. Therefore, they have been stored at: Figshare Digital Repository (https://doi.org/10.6084/m9.figshare.21710309.v1), where they are available for viewing. We believe that the results presented are sufficient; any greater amount would only lead to poorer understanding of the text.

 

Conclusion: "Must be improved"

The section needs to be reconstructed. The conclusion is not based on the findings that were presented in the main body of the manuscript, since the validation process was not performed to show the accuracy of the estimation, and there are no data that shows the accuracy assessment. Add some key recommendations for decision and policymakers.

 

The Conclusion section has been expanded and supplemented, including recommendations for policymakers. Validation is part of the Maxent model; in our case, it uses the leave-one-out method for testing and validating the model. The results of sensitivity, i.e. the accuracy of the resulting probability map, are present – in the form of AUC in Table 3. [490-528]

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Accept in present form