Improved Spectral Detection of Nitrogen Deficiency and Yellow Mosaic Disease Stresses in Wheat Using a Soil Effect Removal Algorithm and Machine Learning

Round 1

Reviewer 1 Report

Improved spectral detection of nitrogen deficiency and yellow mosaic disease stresses in wheat through a soil effect removal algorithm and machine learning

A summary 

This study proposed a method to detect spectral responses to nitrogen deficiency and yellow mosaic disease stresses in wheat. A pre-process spectral approach and a ML spectral analysis were investigated.

General concept comments

The work reports an interesting approach which try to solve a specific problem with a ML approach. I suggest the authors consider the extension of Discussion part, providing more details about methodology improvement and integration at the farm level.

Review

A minor revision is required in order to mind the gap between the research findings and the farm application.

Specific comments 

Abstract

·       Line 21: try to avoid using the first plural person “we” in the manuscript.

·       The abstract is well made but I suggest summarising results giving only the most important information. Further details will be provided inside the work.

Materials and methods

·       Lines 209-2011: In order to make the next sentence clearer, I suggest authors to add the acronym at the end of each pre-processing methods.

·       Lines 277-280: Does the research team think that precision and recall metrics could improve the results interpretation and comparison with other works?

·       Line 278: what does “PE” mean?

Discussion

·       The discussion section is well written but no considerations about detection performances have been made and how this approach could be applied on a larger scale using UAS and satellite techniques.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

  Thank you very much for the review. The detailed explanation of what we have changed in response to your concerns is given point by point in the response reports to each reviewer. Upon the request, a “tracking change” version is also provided. Please see the details in the attachment.

Wei Feng, on behalf of all authors,

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript titled "Improved spectral detection of nitrogen deficiency and yellow mosaic disease stresses in wheat through a soil effect removal algorithm and machine learning" addresses an important problem in the field of agriculture. The authors investigate the wheat nitrogen deficiency and yellow leaf disease fields, conduct physiological and biochemical experiments, and collect spectral data over four years to develop a crop stress detection method. The study evaluates different preprocessing methods and spectral indices, applies feature selection techniques, and tests various machine learning models to determine the most effective approach. The manuscript appears well-structured and provides a comprehensive analysis of the problem. However, a few areas could be improved to strengthen the paper further.

The introduction provides a comprehensive background on the importance of wheat, the effects of biotic and abiotic stresses, and the challenges in differentiating nitrogen deficiency from WYMV infection. However, some sentences are quite long and complex, which may make it difficult for readers to follow the main points (lines 82-88, lines 110-115). Moreover, it would be helpful to provide more context on the global significance of wheat production and the potential impact of WYMV and nitrogen deficiency on food security. Finally, the authors could elaborate on the limitations of previous studies (there is some information in line 110, but it should be extended) and later discuss how their proposed method aims to address these gaps in the literature.

In the material and methods section is not clear the rationale behind choosing these specific spectral preprocessing methods and how they contribute to the study's objectives. Therefore, the authors should elaborate further on this.

The results are fine, but the discussion is the weakest part of the manuscript. The authors did not address how the findings of this study compare to those of similar studies in other regions or crops or if there are there any differences in the spectral response characteristics that can be attributed to regional or crop-specific factors (there is some information in line 479, but it should be expanded). Moreover, they should explain how robust the separability thresholds for spectral indices are when applied to different datasets or under varying environmental conditions and they should add a specific section with the limitations of the 3SV-TVIF-SVM model, and how might they be addressed in future research (or, at least, the authors should include this in line 459 “4.3. Advantages…”)

Finally, the conclusions are good, the authors have presented a comprehensive summary of the findings, highlighting the most sensitive regions for detecting nitrogen deficiency and yellow mosaic disease stresses in wheat.

Therefore, I recommend a minor revision of this manuscript prior to its publication.

English needs a comprehensive revision all over the manuscript. In general, the document does not have critical errors, but some incorrect word choices and subject-verb concordance problems make it sometimes difficult to follow. Some examples: Line 52 "The high confusion of these plants is detrimental to disease prevention and control and crop field management" should be "The high similarity of these plants is detrimental to disease prevention and control and crop field management". Line 80 "The responses of vegetation leaf chemicals, the canopy structure and soil to canopy spectra is mixed [14], and different stresses may exacerbate this effect, which would be detrimental to the remote sensing detection of different stresses.". there is an error in the verb. It should be “The responses of vegetation leaf chemicals, the canopy structure, and soil to canopy spectra are mixed [14], and different stresses may exacerbate this effect, which would be detrimental to the remote sensing detection of different stresses."

Author Response

Dear Reviewer,

  Thank you very much for the review. The detailed explanation of what we have changed in response to your concerns is given point by point in the response reports to each reviewer. Upon the request, a “tracking change” version is also provided. Please see the details in the attachment.

Wei Feng, on behalf of all authors,

Author Response File: Author Response.pdf

Reviewer 3 Report

This study presents a good work for identifying crop stress types (nitrogen deficiency and yellow mosaic disease stresses) using hyperspectral data. In particular, this study investigated wheat nitrogen deficiency and yellow leaf disease fields and conducted wheat physiological and biochemical experiments to collect agronomic indicators, collected four years of reflectance spectral data at green-up and jointing. 

Overall, the topic is interesting that can be beneficial to the scientific community. However, this manuscript could benefit from minor editing before publication. 

Specific comments:

Line 203 (Section 2.3): The reviewer suggests adding more introduction sentences for the flowchart. 

Line 207 (Section 2.3.1): The reviewer suggests adding continuous wavelet transform (CWT) in the spectral preprocessing methods. CWT has been demonstrated as a promising method for detecting stress in numerous studies. 

Line 237 (Section 2.3.2): Please give specific calculation equations and reference papers for VIF. 

Line 325 (Section 3.3): The reviewer is confused about how can you calculate the OA and Kappa values when using spectral indices? Based on which machine learning method? 

Line 353 (Section 3.4): The reviewer is confused about whether the OA and Kappa values are calculated for separating two classes (nitrogen deficiency and yellow mosaic disease stresses), or separating six classes (N0, N1, N2, Level1, Level2, Level3)? 

Minor editing of English language required.

Author Response

Dear Reviewer,

  Thank you very much for the review. The detailed explanation of what we have changed in response to your concerns is given point by point in the response reports to each reviewer. Upon the request, a “tracking change” version is also provided. Please see the details in the attachment.

Wei Feng, on behalf of all authors,

Author Response File: Author Response.pdf