Crop Health Assessment from Predicted AGB and NPK Derived from UAV Spectral Indices and Machine Learning Techniques

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study developed a decision-making framework to categories crop health condition into eight classes and assessed the health condition of paddy crop across various growth stages of the crop from predicted AGB and NPK. However, the authors should revise the manuscript substantially.

1. Abstract: Please enhance the abstract by explicitly addressing the limitations of existing research and highlighting the novel contributions of this study. Finally, clearly expound the significance of this research.
2. In lines 39 and 54, please expand the abbreviations "AGB" (Aboveground Biomass) and "UAVs" (Unmanned Aerial Vehicles) upon their first appearance in the main text. Additionally, standardize the abbreviation "Vis" in line 58 and "VIs" in line 68 to ensure consistency throughout the manuscript.
3. It is recommended that the innovation of this study be clearly articulated in the penultimate paragraph of the introduction section. The final paragraph of the introduction should be restructured for improved coherence and logical flow. Furthermore, lines 109–114 should be condensed to eliminate redundancy and improve clarity.
4. Figures: The Data Analysis and Validation section in Figure 2 is currently disorganized and requires redesign to improve readability and visual clarity.

Figure 3 lacks aesthetic appeal; it is suggested to introduce spacing between subfigures and place the labels (a, b, c, etc.) beneath each subplot. For consistency across all figures in the manuscript containing labeled panels, please uniformly position the subplot identifiers either in the upper-left corner or directly below each subfigure. 

In Figure 5, consider adding reference lines to clearly demarcate results across different time periods. 

Enhance the resolution of Figure 7 and include indicators of statistical significance for the correlation coefficients. 

Clarify the meaning of the label “N” appearing in the upper-right corners of Figures 8–10. 

For Figures 12–15, ensure that subfigure (b) includes properly lines of the X-axis and Y-axis for completeness.

5. Standardize the date format throughout the entire manuscript, including both textual content and graphical elements. Currently, four distinct date formats are used, which undermines consistency.
6. It is advised to separate the Results and Discussion sections to enhance structural clarity. Additionally, compare and analyze this research with previous related work, and clearly discuss the limitations of the current study as well as potential strategies for future improvement.
7. The current list of references is insufficient. Please augment the literature cited at appropriate positions, such as lines 270–278.
8. The 4.2 section title “AGB and NPK Assessment” does not accurately reflect the content. It is recommended to revise it to “Temporal Variation of Measured Parameters” for greater precision and clarity.
9. In Section 3.4.1, please specify the total number of samples used in the analysis.
10. In the Conclusion section, begin with a concise summary of the overall research, followed by a numbered enumeration of the key findings.

Author Response

We would like to thank you for positive and constructive comments and suggestions. We revised the manuscript in accordance with your comments, and revised portion are marked in red in the paper. Below are our responses.

Point-by-point response to Comments and Suggestions for Authors

This study developed a decision-making framework to categories crop health condition into eight classes and assessed the health condition of paddy crop across various growth stages of the crop from predicted AGB and NPK. However, the authors should revise the manuscript substantially.

Response:       Thank you for your valuable feedback. We appreciate your suggestion and will revise the manuscript substantially to improve clarity, strengthen the methodology, and enhance the presentation of the decision-making framework and its application in classifying paddy crop health conditions across growth stages.

Comments 1: Abstract: Please enhance the abstract by explicitly addressing the limitations of existing research and highlighting the novel contributions of this study. Finally, clearly expound the significance of this research

Response 1: Thank you for your valuable suggestion. The abstract has been revised to explicitly address the limitations of existing research, particularly the constraints of conventional crop monitoring techniques such as low spatial resolution. The novel contributions of this study have been clearly highlighted. Additionally, the significance of this research in enabling precise, plant level crop health assessment and supporting sustainable agricultural practices through targeted nutrient management has been elaborated in the revised abstract.

Comments 2: In lines 39 and 54, please expand the abbreviations “AGB” (Aboveground Biomass) and "UAVs" (Unmanned Aerial Vehicles) upon their first appearance in the main text. Additionally, standardize the abbreviation “VIs” in line 58 and “Vis” in line 68 to ensure consistency throughout the manuscript.

Response 2: Thank you for the valuable suggestion. The abbreviations “AGB” (Aboveground Biomass) and “UAVs” (Unmanned Aerial Vehicles) have now been expanded upon their first appearance in the main text (line no. 45 and 59), in addition to their mention in the abstract. Furthermore, the abbreviation for Vegetation Indices has been standardized as “VIs” throughout the manuscript to ensure consistency.

Comments 3: It is recommended that the innovation of this study be clearly articulated in the penultimate paragraph of the introduction section. The final paragraph of the introduction should be restructured for improved coherence and logical flow. Furthermore, lines 109–114 should be condensed to eliminate redundancy and improve clarity.

Response 3:     Thank you for the valuable suggestion. The novelty of the study has been clearly articulated in the penultimate paragraph of the introduction section in revised manuscript. Additionally, the final paragraph has been revised to enhance coherence and logical flow. Lines 109–114 (now 108-113) have also been condensed to eliminate redundancy and improve clarity.

Comments 4: Figures:

• The Data Analysis and Validation section in Figure 2is currently disorganized and requires redesign to improve readability and visual clarity.

Response: Thank you for the valuable suggestion. Figure 2 has been revised and redesigned to enhance its organization, readability, and visual clarity.

• Figure 3 lacks aesthetic appeal; it is suggested to introduce spacing between subfigures and place the labels (a, b, c, etc.) beneath each subplot. For consistency across all figures in the manuscript containing labeled panels, please uniformly position the subplot identifiers either in the upper-left corner or directly below each subfigure. 

Response: As per the reviewer’s suggestion, we have introduced spacing between the subfigures in Figure 3 and repositioned the labels beneath each subplot. Additionally, we have ensured consistency in the placement of subplot identifiers across all figures in the manuscript.

• In Figure 5, consider adding reference lines to clearly demarcate results across different time periods. 

Response: Thank you for your valuable suggestion. To enhance clarity and improve the interpretation of temporal variations, we have replaced the combined vegetation indices map with individual maps for every indices, allowing for better visibility and understanding.

• Enhance the resolution of Figure 7 and include indicators of statistical significance for the correlation

Response:  Thank you for your valuable suggestion. In accordance with the reviewer’s recommendation, the resolution of Figure 7 has been enhanced, and indicators of statistical significance for the correlation coefficients have been incorporated into the revised manuscript.

• Clarify the meaning of the label “N” appearing in the upper-right corners of Figures 8–10. 

Response: Thank you for pointing out the inconsistency. The label “N” in the upper-right corners of Figures 8–11 was intended to represent the North direction; however, the North arrow was not properly aligned in the original figures. As per your suggestion, we have now corrected this by clearly indicating and aligning the north arrow in Figures 8–11.

• For Figures 12–15, ensure that subfigure (b) includes properly lines of the X-axis and Y-axis for completeness.

Response: Thank you for the valuable suggestion. As advised, the X-axis and Y-axis lines have been appropriately added to subfigure (b) in Figures 12–15 to ensure completeness and clarity.

Comments 5: Standardize the date format throughout the entire manuscript, including both textual content and graphical elements. Currently, four distinct date formats are used, which undermines consistency.

Response 5:     As per the reviewer’s suggestion, the standardized date format used in the throughout the manuscript text for consistency. Additionally, a short date format has been used in the graphical elements to enhance clarity and visualization.

Comments 6: It is advised to separate the Results and Discussion sections to enhance structural clarity. Additionally, compare and analyze this research with previous related work, and clearly discuss the limitations of the current study as well as potential strategies for future improvement.

Response 6:     Thank you for the insightful suggestion. The Results and Discussion sections have been separated to improve the structural clarity of the manuscript. The Discussion section now includes a detailed comparison and analysis with relevant previous studies. Additionally, the limitations of the current study have been explicitly discussed, along with potential directions for future research and improvement.

Comments 7: The current list of references is insufficient. Please augment the literature cited at appropriate positions, such as lines 270–278.

Response 7:     Thank you for your suggestion. We have reviewed all the sections and added relevant references to strengthen the literature and discussion including the lines 270–278 (now 270-279), ensuring appropriate citations are included to enhance the context and credibility.

Comments 8: The 4.2 section title “AGB and NPK Assessment” does not accurately reflect the content. It is recommended to revise it to “Temporal Variation of Measured Parameters” for greater precision and clarity.

Response 8:     Thank you for the valuable suggestion. As recommended, the title of Section 4.2 (now updated as Section 3.2) has been revised to “Temporal Variation of Measured Crop Parameters” to enhance clarity and better reflect the content of the section.

Comments 9: In Section 3.4.1, please specify the total number of samples used in the analysis.

Response 9:     Thank you for the valuable suggestion. In response, the total number of samples used in the analysis has been specified and incorporated in Section 2.2.4.1 (now Section 3.4.1in the revised manuscript).

Comments 10: In the Conclusion section, begin with a concise summary of the overall research, followed by a numbered enumeration of the key findings.

Response 10:   Thank you for the valuable suggestion. As recommended, the Conclusion section has been revised to begin with a concise summary of the overall research, followed by a numbered list highlighting the key findings for better clarity and readability.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors use UAV (Unmanned Aerial Vehicle) remote sensing and spectral indices to assess paddy crop health across different growth stages. By analyzing vegetation indices, the researchers predicted Above Ground Biomass (AGB) and nutrient levels (Nitrogen, Phosphorus, and Potassium) using four machine learning models. Random Forest and XGBoost outperformed others, providing accurate predictions and generating detailed crop health maps. These maps visualize spatial variability in crop conditions, enabling targeted interventions, efficient resource use, and sustainable yield optimization.

 

1. Does Figure 1 include sufficient cartographic elements (e.g., scale bar, north arrow, legend) to interpret spatial extent and orientation? If not, what additional map features would improve its usability? What are the key geographical features or agricultural practices specific to the Dharmasagar village region that make it suitable for UAV-based paddy crop health assessment?
2. How do the selected vegetation indices (e.g., NDVI, GNDVI, NDRE) contribute differently to the estimation of AGB and nutrient levels (NPK), and which indices showed the strongest predictive performance at various crop stages?
3. What are the strengths and limitations of using Random Forest (RF) and XGBoost compared to MLR and PLSR for predicting AGB and NPK in this study?
4. How effective was the decision-rule framework developed in the study for classifying crop health into eight categories, and how can it be practically applied in real-time precision agriculture?

Author Response

We would like to thank you for positive and constructive comments and suggestions. We revised the manuscript in accordance with your comments, and revised portion are marked in red in the paper. Below are our responses.

Point-by-point response to Comments and Suggestions for Authors

The authors use UAV (Unmanned Aerial Vehicle) remote sensing and spectral indices to assess paddy crop health across different growth stages. By analyzing vegetation indices, the researchers predicted Above Ground Biomass (AGB) and nutrient levels (Nitrogen, Phosphorus, and Potassium) using four machine learning models. Random Forest and XGBoost outperformed others, providing accurate predictions and generating detailed crop health maps. These maps visualize spatial variability in crop conditions, enabling targeted interventions, efficient resource use, and sustainable yield optimization.

Response:       Thank you for your valuable feedback.

Comments 1: Does Figure 1 include sufficient cartographic elements (e.g., scale bar, north arrow, legend) to interpret spatial extent and orientation? If not, what additional map features would improve its usability?

Response 1: Thank you for your valuable suggestion. Yes, Figure 1 includes essential cartographic elements such as a scale bar, north arrow, legend, spatial extent, and map body to aid in proper interpretation and orientation.

Comments 2: What are the key geographical features or agricultural practices specific to the Dharmasagar village region that make it suitable for UAV-based paddy crop health assessment?

Response 2: Thank you for the suggestion. The key geographical features and agricultural practices that make Dharmasagar village suitable for UAV based paddy crop health assessment have been addressed in the revised manuscript under Section 2.1 (lines 115–122).

Comments 3: How do the selected vegetation indices (e.g., NDVI, GNDVI, NDRE) contribute differently to the estimation of AGB and nutrient levels (NPK), and which indices showed the strongest predictive performance at various crop stages?

Response 3: Thank you for the suggestion. The differential contribution of VIs to the estimation of AGB and nutrient levels has been addressed in Section 3.4. Additionally, the temporal variation in feature importance across different crop stages is illustrated in Fig. S1 (supplementary file), highlighting which indices were most predictive at each stage.

Comments 4: What are the strengths and limitations of using Random Forest (RF) and XGBoost compared to MLR and PLSR for predicting AGB and NPK in this study?

Response 4:     Thank you for the suggestion. As per the reviewer suggestion, the strengths and limitations of using RF and XGBoost compared to MLR and PLSR for predicting AGB and NPK is included in the discussion section.

Comments 5: How effective was the decision-rule framework developed in the study for classifying crop health into eight categories, and how can it be practically applied in real-time precision agriculture?

Response 5:     Thank you for the comment. The decision-rule framework effectively classified crop health into eight distinct categories by integrating AGB and NPK thresholds derived from UAV based observations. It showed strong agreement with ground truth data, demonstrating its robustness and it was presented in Section 3.6 and 4.2. For practical application, the framework can be integrated into real-time precision agriculture systems through automated UAV data processing and classification pipelines, enabling timely interventions and optimized input management.

Reviewer 3 Report

Comments and Suggestions for Authors

• The topic needs thorough improvement because it does not fully reflect the stated objective. The results explicitly mention prediction and the use of machine learning methods. Unless the authors wish to remove this thread from the paper.
• ABSTRACT: Line 23: No specific information on which model and with which values obtained the best result. No presentation of meaningful numerical data in the abstract. Correct it.
• INTRODUCTION: The introduction is to be improved, too long. I would focus attention on key information related to the topic of the paper. Please make it shorter.
• Lines 116-123 - remove this paragraph and put the information in methods in one sentence.
• DISCUSSION: Another major problem that out of 25 references 18 of them are included in the Introduction, which translates into a lack of discussion in the paper. No comparison to current literature, this needs to be improved.
• Figures not clear.
• Another serious point, since each journal sets a certain canon of what the structure of the article should look like the authors should also take this into account when submitting the paper, which unfortunately they did not do. This solution presented by the authors is unreadable. Please correct it (https://www.mdpi.com/journal/agronomy/instructions).
• The purpose of the work also needs improvement.
• When analysing the results at this stage, I would question whether it is worth giving and describing all the metrics we know. I would suggest focusing on metrics only in terms of prediction, regression.

In light of the key issues and without looking at references to important information in the text of the article, the article is not suitable for publication at this stage.

Author Response

We would like to thank you for positive and constructive comments and suggestions. We revised the manuscript in accordance with your comments, and revised portion are marked in red in the paper. Below are our responses.

Point-by-point response to Comments and Suggestions for Authors

Comments 1: The topic needs thorough improvement because it does not fully reflect the stated objective. The results explicitly mention prediction and the use of machine learning methods. Unless the authors wish to remove this thread from the paper.

Response 1: Thank you for the insightful comment. The title of the manuscript has been revised from “Geospatial Analysis of Paddy Crop Health Assessment Using UAV Derived Spectral Indices” to “Crop Health Assessment from Predicted AGB and NPK derived from UAV Spectral Indices and Machine Learning Techniques”, to clearly reflect the study's focus on prediction and the application of machine learning methods, ensuring alignment with the stated objectives and presented results.

Comments 2: ABSTRACT: Line 23: No specific information on which model and with which values obtained the best result. No presentation of meaningful numerical data in the abstract. Correct it

Response 2: Thank you for the valuable suggestion. The abstract has been revised to include specific numerical results of the best performing models, which consistently yielded superior performance across most temporal dates. This addition aims to enhance clarity and provide a more meaningful summary of the study’s key outcomes.

Comments 3: INTRODUCTION: The introduction is to be improved, too long. I would focus attention on key information related to the topic of the paper. Please make it shorter.

Response 3: Thank you for your valuable suggestion. The introduction section has been revised and shortened to focus on the key information relevant to the topic of the paper, ensuring better clarity and improved understanding for the readers.

Comments 4:   Lines 116-123 - remove this paragraph and put the information in methods in one sentence

Response 4: Thank you for the valuable suggestion. As per the reviewer’s recommendation, the content from lines 116–122 describing the study area has been integrated into the methodology section (now Section 2.1) and presented concisely.

Comments 5: DISCUSSION: Another major problem that out of 25 references 18 of them are included in the Introduction, which translates into a lack of discussion in the paper. No comparison to current literature, this needs to be improved.

Response 5:     Thank you for the valuable feedback. We have revised the manuscript to enhance the discussion section by incorporating comparisons with relevant recent studies, ensuring a more balanced distribution of references and a stronger connection to existing literature.

Comments 6: Figures not clear.

Response 6:     Thank you for your valuable comment. The UAV processing layout images have been developed at a high dpi in JPEG format to ensure high clarity. For improved visualization, all figures have been shared separately with the journal editorial team. Please let us know if any further modifications are required.

Comments 7: Another serious point, since each journal sets a certain canon of what the structure of the article should look like the authors should also take this into account when submitting the paper, which unfortunately they did not do. This solution presented by the authors is unreadable. Please correct it (https://www.mdpi.com/journal/agronomy/instructions).

Response 7:     Thank you for highlighting this important point. We have thoroughly revised the manuscript to ensure it adheres to the journal’s prescribed structure, improving both readability and consistency.

Comments 8: The purpose of the work also needs improvement.

Response 8:     Thank you for your valuable suggestion. The main purpose of the study has been rephrased and clearly articulated in the last paragraph of introduction section (lines 102–113) to enhance clarity and better reflect the objectives of the work.

Comments 9: When analysing the results at this stage, I would question whether it is worth giving and describing all the metrics we know. I would suggest focusing on metrics only in terms of prediction, regression.

Response 9:     Thank you for your insightful comment. We have streamlined the analysis by focusing specifically on regression and prediction metrics that directly reflect the model performance. This targeted approach ensures the results remain relevant and concise, avoiding unnecessary inclusion of general metrics that do not contribute significantly to the interpretation of predictive accuracy. We appreciate your suggestion in helping to improve the clarity and focus of the results section.

Comments 10: In light of the key issues and without looking at references to important information in the text of the article, the article is not suitable for publication at this stage.

Response 10:   Thank you for your critical feedback. We understand your concerns regarding the current structure and clarity of the manuscript. In response, we have thoroughly revised the text to ensure that all key findings and important information are clearly presented and well-supported with appropriate references throughout the manuscript. We have also improved the organization and coherence of the content to enhance overall readability and scientific rigor. We hope that these comprehensive revisions address your concerns and make the article more suitable for publication.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

I am returning an article for review that previously needed critical revisions before publication. In re-evaluating, I see that the authors have largely addressed the earlier comments and made many changes, showing their dedication to improving the work. The work still requires further improvement in style and language. My comments are below:

• The summary is too lengthy; corerect it.
• Line 180 - The authors used incorrect terminology, e.g., change ‘traits’ to ‘characteristics’.
• If the authors used 'acronym' for machine learning, i.e., ML, then they should apply this change throughout the article.
• Line 86 - Use a different word than “Studies” to correct this sentence.
• Line 98 - Correct this sentence stylistically by removing ‘However...’
• Line 133 - Improve the sentence stylistically; do not start with „For...”.
• Line 141 - The unit is only given once; after the value and the range of error, correct it.
• Line 153 - Provide a reference with a link to the literature, including page number and access date.
• Line 197 - Clarify why ‘(Eq.(1))’ is used when continuing with a formula numbered 1; correct according to guidelines.
• Remove the separator ‘,’ at the end of every equation.
• Correct the units in the equations to improve clarity and transparency.
• Line 336 - Reconsider if starting sentences or paragraphs with ‘Where’ is stylistically appropriate; correct it.
• Line 384 - Fix the illegible reference to the figure, e.g., ‘(fig. 6(a))’.
• Line 648 - Clarify the meaning of “minimal prediction errors ... its robustness”.
• Review and improve the references according to journal guidelines, once again.

Author Response

We would like to thank you for positive and constructive comments and suggestions. We revised the manuscript in accordance with your comments, and revised portion are marked in red in the paper. Below are our responses.

Point-by-point response to Comments and Suggestions for Authors

I am returning an article for review that previously needed critical revisions before publication. In re-evaluating, I see that the authors have largely addressed the earlier comments and made many changes, showing their dedication to improving the work. The work still requires further improvement in style and language.

Thank you for your detailed feedback and for acknowledging the efforts made to address the earlier comments. We appreciate your recognition of the improvements so far. We will carefully review the manuscript again to further enhance the style and language, ensuring it meets the highest standards for publication. Your guidance is invaluable in this process.

Comments 1: The summary is too lengthy; correct it.

Response 1:     Thank you for the suggestion. While we acknowledge that the summary is lengthy, other reviewers recommended expanding it to include key findings. However, we have revised and streamlined several statements to improve clarity and reduce length where possible.

Comments 2:   Line 180 - The authors used incorrect terminology, e.g., change ‘traits’ to ‘characteristics’.

Response 2:     Thank you for pointing out the mistake. The terminology changed from “traits” to “characteristics” in line no. 64 and 80.

Comments 3:  If the authors used 'acronym' for machine learning, i.e., ML, then they should apply this change throughout the article.

Response 3:     Thank you for your suggestion. We have now standardized the use of the acronym “ML” for machine learning consistently throughout the manuscript.

Comments 4:    Line 86 - Use a different word than “Studies” to correct this sentence.

Response 4:     Thank you for your suggestion. The sentence has been revised using alternative wording as recommended.

Comments 5:  Line 98 - Correct this sentence stylistically by removing ‘However...’.

Response 5:     Thank you for your suggestion. The statement is revised as per the reviewer suggestion.

Comments 6:  Line 133 - Improve the sentence stylistically; do not start with „For...”.

Response 6:     Thank you for your suggestion. The sentence has been revised for improved style, avoiding the use of “For” at the beginning, as recommended (lines 132–135).

Comments 7:  Line 141 - The unit is only given once; after the value and the range of error, correct it.

Response 7:     Thank you for the suggestion. The units have been corrected and are now consistently included after both the central value and the range of error for each spectral band.

Comments 8:  Line 153 - Provide a reference with a link to the literature, including page number and access date.

Response 8:     Thank you for your comment. A reference has been added as requested, including the link to the source, and the date it was accessed to ensure transparency and traceability of the cited information.

Comments 9: Line 197 - Clarify why ‘(Eq.(1))’ is used when continuing with a formula numbered 1; correct according to guidelines.

Response 9:     Thank you for the feedback. The use of “(Eq.(1))” was an oversight. According to standard formatting guidelines, the correct reference should be “Eq. (1)” without the extra parentheses. This has been corrected throughout the document for clarity and consistency.

Comments 10: Remove the separator ‘,’ at the end of every equation.

Response 10:   Thank you for pointing that out. The commas at the end of the equations have been removed to comply with the formatting guidelines.

Comments 11: Correct the units in the equations to improve clarity and transparency.

Response 11:   Thank you for your valuable feedback. The units in all equations have been reviewed and corrected to improve clarity and ensure consistency and transparency across the manuscript.

Comments 12: Line 336 - Reconsider if starting sentences or paragraphs with ‘Where’ is stylistically appropriate; correct it.

Response 12:   Thank you for the suggestion. I have reviewed the use of "Where" at the beginning of sentences and revised the text to ensure it is stylistically appropriate.

Comments 13: Line 384 - Fix the illegible reference to the figure, e.g., ‘(fig. 6(a))’.

Response 13:   Thank you for noticing that. The reference has been corrected to follow proper formatting and now reads as “fig. 6” for clarity and consistency.

Comments 14: Line 648 - Clarify the meaning of “minimal prediction errors ... its robustness”.

Response 14:   Thank you for your comment. By “minimal prediction errors,” I mean that the model’s predictions closely matched the observed values, as indicated by the low RMSE. This low error rate reflects the model’s robustness, meaning it consistently provides accurate and reliable estimates of above-ground biomass across different conditions. I have revised the statement to clarify this connection.

Comments 15: Review and improve the references according to journal guidelines, once again.

Response 15:   Thank you for your comment. The references have been carefully reviewed and updated in accordance with the journal’s formatting guidelines, using the Zotero reference management tool. We have ensured consistency, accuracy, and completeness throughout the reference list.

Author Response File: Author Response.pdf