Predicting Wheat Yield by Spectral Indices and Multivariate Analysis in Direct and Conventional Sowing Systems

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript presents a technically solid and methodologically coherent study that integrates multispectral UAS data, spectral indices, and machine learning (PCA + Random Forest) to model wheat yield under direct (DS) and conventional sowing (CS) systems. The topic is highly relevant to sustainable agriculture and precision farming, and the work demonstrates a commendable balance between experimental rigor and applied significance.
Below are my detailed observations and suggestions:

1. Introduction and Background
The introduction provides a strong overview of sustainability issues and the relevance of DS and CS systems.
→ Consider adding a brief research gap paragraph explaining what previous studies (e.g., Liu et al., 2024; Zhao et al., 2023) have not addressed—particularly the integration of physiological variables and spectral indices under long-term DS.
A concise statement of novelty and contribution at the end of the Introduction would help readers immediately understand the added value of this study.

2. Materials and Methods
The methodology is well structured, covering experimental setup, spectral index computation, and model training.
→ Please specify the criteria for selecting nine spectral indices, or cite why these particular indices were used beyond availability from the sensor bands.
→ Clarify sample representativeness: why were nine sampling points per treatment chosen, and how does this ensure statistical reliability across a one-hectare plot?
→ For reproducibility, mention the Random Forest hyperparameters (number of trees, mtry selection, seed value, etc.) and provide details on how PCA components explaining 90% variance were validated.

3. Results
The results are robust and well visualized. ROC, correlation, and regression analyses are clearly presented.
→ However, some figures (e.g., Figures 6–9) could benefit from clearer labels and slightly larger font sizes for readability.
→ It would also help to summarize in one sentence the biological interpretation of why AVI and SAVI outperform other indices (linking to canopy structure or soil reflectance).
→ The table numbering should be consistently formatted (e.g., “Table 3. Values of the ROC curve…” should be referenced in the text explicitly).

4. Discussion
Excellent linkage between statistical findings and agronomic implications. The comparison with related literature (e.g., Zhao et al., 2023; Walsh et al., 2023) strengthens the discussion.
→ To enhance clarity, you could shorten some long sentences (especially those exceeding 40 words) and use topic sentences at the start of paragraphs to improve readability.
→ Consider expanding briefly on the ecological interpretation of the negative correlations observed in DS (e.g., reduced soil brightness and reflectance leading to inverse relationships).

5. Conclusions
The conclusions are accurate and supported by the data.
→ You may add a short practical implication paragraph on how these findings could inform real-world decision-making (e.g., early yield prediction, fertilizer optimization).
→ Also, briefly indicate limitations such as geographical scope (single site) or potential transferability issues to other crops or climates.

6. Language and Style
The English is overall clear and professional. A light proofreading or copy-edit (especially in Discussion) is recommended to improve conciseness and flow.
All acronyms are correctly defined, and AI-tool disclosure aligns with MDPI’s ethical policy.

7. Overall Evaluation
This is a well-structured and scientifically sound manuscript.
I recommend minor revision, primarily for language polishing, additional methodological detail (RF tuning, sampling rationale), and improved figure clarity.

Comments on the Quality of English Language

The manuscript is written in generally clear and readable English, with appropriate scientific terminology and structure. The authors have successfully conveyed complex technical content in an understandable way. However, there are occasional long and compound sentences—especially in the Introduction and Discussion—that could be simplified to improve clarity and flow.
Minor stylistic and grammatical adjustments are recommended to enhance readability, particularly in transitions between ideas and consistency in tense (past vs. present). The use of ChatGPT and Grammarly for linguistic improvement is transparently declared and aligns with MDPI’s policy.
Overall, the manuscript requires only minor English editing for conciseness and smoothness, not a full linguistic overhaul.

Author Response

Please, see attached file

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This is a well-executed study demonstrating the value of UAV-based spectral indices and machine learning for distinguishing sowing systems and predicting wheat yield. The manuscript is clear, and the results are strong. Editing based on the provided comment below and in the file would improve the work. Overall, this is a valuable contribution to research in precision agriculture.
- Please explain why SAVI and AVI perform better
- Reduce repetition in the results section
- Explain that the study only covers one growing season and location. 
- A shorter conclusion would also help with readability

Comments for author File: Comments.pdf

Author Response

Please, see attached file

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript evaluated the prediction model based on spectral indices and multivariate analysis of unmanned aerial vehicle data collection. It used multivariate analysis methods including principal component analysis (PCA) and random forest (RF) to estimate the wheat yield under two different planting patterns. The manuscript must be improved referring to the following points before publication.

1. This study mainly focused on the performance testing of the model. It lacked the data results of different time periods for the same plot, and also lacks the application model testing results outside the actual experimental sampling plots. Therefore, additional data results need to be supplemented to support the conclusion of this method.
2. For the acquisition of multi-spectral images in the field, an accurate calibration process is necessary. Please add details of the calibration process and also explain the weather conditions during the acquisition.
3. More details of the data processing procedures need to be elaborated in the methods section.
4. The terms "DS" and "CS" first appeared in the introduction, it’s recommended to use the full name, although there were subsequent abbreviations provided. And, the explanations and descriptions of the specific processes of these two planting methods should be given.
5. The key word "reduced dimensionality (PCA)" needs to be checked for any possible errors.

6. In the text, there are several instances where “R2” was not presented in the marked form.

Author Response

Please, see attached file

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have responded to the comments and made the necessary revisions. I believe that the revised manuscript has been improved and could be considered for publication in the journal.