Predicting Nitrogen Flavanol Index (NFI) in Mentha arvensis Using UAV Imaging and Machine Learning Techniques for Sustainable Agriculture
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsAre references 12, 14, 21 from co-authors?
Avoid using the same keywords from title.
line 14: is it better to use yield instead of productivity? review for all text.
line 16: should be term instead of turn?
Use italic for Mentha arvensis (line 18- review for all text).
Make clear the objective of the paper on abstract.
line 28: to improve or improved?
Was the study done with one area and one season? The concern is about its validation and overfitting of the results on modeling.
line 39: use point after finishing the sentence.
line 45: every crop?
line 50: should be practices?
line 62: use the format of citation from periodic, is it missing the last name of author? review for all text.
Study area: why that region and period were selected for study? Specify crop cultivar, size of the area, type of soil. Make clear the importance of the crop for the selected region.
Figure 1: use (a), (b), (c) to facilitate understanding to the readers. The same for other Figures.
line 127: why those settings were chosen?
line 141: planting or plantation?
line 156: Table 3 instead of table.
line 160: specify the version of arcgis.
Ground truth: are 10 plants representative?
Model development: specify the parameters used for modeling. Add information about the programs/softwares and packages used to implement it. Explain how split train and test, its number of points, etc.
Figure 2: make sense NFI > 1?
Feature selection: make clear the features selected for each model. Why spectral bands weren't considered individually?
line 265: what procedures did you adopt to avoid overfitting?
Table 4: explain better what is "Combined".
Figures 4: include 1:1 regression line.
Figure 5: the same, and adjust the same x-y scale between the graphs, use (a), (b), (c).
line 295: Figure 5.
line 349-363: I am not sure about the stated, mainly because you had just one area and season for development and application of the model.
Conclusion: focus to answer the objective of the paper.
Comments on the Quality of English Language
Improve.
Author Response
Comments 1: Are references 12, 14, 21 from co-authors?
Response: Thank you for pointing this out, the references cited at number 12 and 21 are authored by one of the corresponding author/co-authors of this manuscript. However, the reference cited at number 14 is not from any of the co-authors.
Comments 2: line 14: is it better to use yield instead of productivity? review for all text.
Response: Agreed. We have revised “productivity” to “yield” in line 15 and reviewed the entire manuscript to ensure consistent usage.
Comments 3: line 16: should be term instead of turn?
Response: Yes, it should be “term”. We have corrected this in line 16.
Comment 4: Use italic for Mentha arvensis (line 18- review for all text).
Response: Have revised the manuscript and ensured that Mentha arvensis is italicized consistently throughout.
Comment 5: Make clear the objective of the paper on abstract.
Response: Abstract have been revised as suggested from line 17 to 19
Comment 6: line 28: to improve or improved?
Response: Rectified as ‘to improve’. We have corrected this in line 27.
Comment 7: Was the study done with one area and one season? The concern is about its validation and overfitting of the results on modeling
Response: We have acknowledged this as a limitation in the Conclusion and recommended multi-location, multi-season validation for future work (lines 414-419).
Comment 8: line 39: use point after finishing the sentence.
Response: Corrected as suggested.
Comment 9: line 45: every crop?
Response: Studies have used these techniques for major crops like paddy, wheat, etc and hence we have added the word ‘major’.
Comment 10: line 50: should be practices?
Response: Yes, this has been corrected.
Comment 12: Study area: why that region and period were selected for study? Specify crop cultivar, size of the area, type of soil. Make clear the importance of the crop for the selected region.
Response: As recommended, the Study Area section have been detailed (lines 128–132) to include detailed justification for the location and period, specific cultivars used (CIM-Kosi, CIM-Unnati, CIM-Kranti), plot dimensions, soil type (loamy sand with pH), and the regional economic importance of Mentha arvensis.
Comment 13: Figure 1: use (a), (b), (c) to facilitate understanding to the readers. The same for other Figures.
Response: Have added panel labels (a), (b), (c) to Figure 1 and ensured consistent labeling for all multi-panel figures as suggested.
Comment 14: line 127: why those settings were chosen?
Response: This setting was chosen by taking reference from existing study. We have added the reference in line 157.
Comment 15: line 141: planting or plantation?
Response: It should be ‘planting’ and have corrected this in line 167.
Comment 16: line 156: Table 3 instead of table.
Response: Corrected to ‘Table 3’ in line 187.
Comment 17: line 160: specify the version of Arc GIS.
Response: The version is ArcGIS Pro version 3.1.7 and has been inorporated in line 191.
Comment 18: Ground truth: are 10 plants representative?
Response: The sampling is representative, as random sampling approach has been used in this study to select ten plants per plot, ensuring spatial and physiological variability. For each selected plant, the third fully expanded leaf from the apical meristem was sampled to maintain consistency. The average values obtained for 10 plants per plot were used for machine learning model calibration and validation to estimate NFI.
Comment 19: Model development: specify the parameters used for modeling. Add information about the programs/softwares and packages used to implement it. Explain how split train and test, its number of points, etc.
Response: The modeling parameters, hyper-parameter tuning in each model explanation section have been detailed now as suggested i.e. for SVR in line 229-232, RF in line 241-245, GBR in line 253-358. We have also mentioned about the software (Python 3.10, scikit-learn), packages used, the 80:20 train-test split, and the number of data points in lines 204–210.
Comment 20: Figure 2: make sense NFI > 1?
Response: NFI values do exceed the value of 1 because it the ratio of chlorophyll content to flavonol content in the leaf. High chlorophyll values and low flavanol values in plant leaf may result in value higher than 1.
Comment 21: Feature selection: make clear the features selected for each model. Why spectral bands weren't considered individually?
Response: We have revised it accordingly (in lines 262–265 and 272-274). We have used vegetation indices in this study because of the hypothesis that vegetation indices mathematically combine spectral bands through ratios, differences, or normalization techniques, effectively minimizing confounding effects caused by soil background, variable illumination, and canopy structural variation, thereby providing more robust proxies for physiological and biochemical traits relevant to nitrogen estimation.
Comment 22: line 265: what procedures did you adopt to avoid overfitting?
Response: To avoid overfitting model hyper-parameters were optimized using Grid Search with 5-fold cross-validation which is explained in detail from line number 207 to 214.
Comment 23: Table 4: explain better what is "Combined".
Response: We have replaced the word combine with “whole growing period” so that it can be self-explanatory.
Comment 24: Figures 4: include 1:1 regression line.
Response: We have revised the figure accordingly to include a 1:1 reference line for visual comparison.
Comment 25: Figure 5: Apply same improvements (1:1 line, same x-y scale, use (a), (b), (c)).
Response: Figure 5 has been updated accordingly with consistent scaling, a 1:1 line, and panel labels.
Comment 26: line 295.
Response: We have revised figure accordingly.
Comment 27: line 295: Figure 5.
Response: Corrected in line 341. also, we have corrected the numbering of the figure previously it was written wrong as figure 5.
Comment 28: line 349-363: I am not sure about the stated, mainly because you had just one area and season for development and application of the model.
Response: We thank the reviewer for this valuable observation. We acknowledge the limitation that the model was developed and tested using data from a single experimental site and cropping season, which may affect its generalizability under varying environmental and management conditions. In response, we have revised from 416-421 to explicitly state this limitation and to clarify that the present study should be regarded as a proof-of-concept rather than a fully generalized solution. Additionally, we have highlighted the need for future research involving multi-location and multi-season trials to validate and refine the proposed approach.
Comment: 29 Conclusion: focus to answer the objective of the paper.
Response: We have revised the Conclusion (lines 431–442) to clearly summarize how the study meets the stated objective and what its key contributions are for nitrogen monitoring in Mentha arvensis.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsDear authors,
the paper is well-structured and methodologically sound, but it requires improvements in language clarity, formatting and interpretation depth.
In the introduction consider explicitly stating the research gap in NFI prediction for aromatic crops and why Mentha arvensis is a valuable case study.
In the materials, include geographic coordinates in decimal format and a map legend in Figure 1 and explain why ExG threshold was set to 0.1.. Was it empirically determined some how?
In vegetation indices and modeling, improve figure captions (especially Figures 3–5) for clarity and standalone readability.
In discussion avoid general statements like "This aligns with previous studies..." without proper context.
In conclusions, consider explicitly stating what stakeholders (e.g., farmers, agronomists) can do with this information.
Last, but not least, I strongly recommend language editing to enhance flow and academic tone.
Best,
Author Response
Comment 1: In the introduction consider explicitly stating the research gap in NFI prediction for aromatic crops and why Mentha arvensis is a valuable case study.
Response: we have revised it and clearly explicit the research gap and importance of NFI prediction in Mentha arvensis from line 103-115.
Comment 2: In the materials, include geographic coordinates in decimal format and a map legend in Figure 1 and explain why ExG threshold was set to 0. 1.. Was it empirically determined somehow?
Response: We have updated the figure according to the comment and Yes ExG was determined empirically we have mentioned this in line 175.
Comments 3: In vegetation indices and modeling, improve figure captions (especially Figures 3–5) for clarity and standalone readability
Response: we have revised the caption of the figures accordingly.
Comment 4: In discussion avoid general statements like "This aligns with previous studies..." without proper context.
Response: we have revised the discussion to replace generic statements with explicit references and context. Line 357-361 and 396-399
Comment 5: In conclusions, consider explicitly stating what stakeholders (e.g., farmers, agronomists can do with this information.
Response: We have mentioned it accordingly inline 440-442
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsI think the citation of authors isn't according to the instructions from periodic.
Author Response
Comment 1: I think the citation of authors isn't according to the instructions from periodic.
Response: Thank you for pointing the citation issue, we have revised the citation using the software “Endnote”, wherein the format mentioned (ACS style) has been applied on the manuscript.
No other changes have been done in the revised manuscript
Author Response File: Author Response.pdf
