Integration of UAV Multispectral Remote Sensing and Random Forest for Full-Growth Stage Monitoring of Wheat Dynamics

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper discusses an innovative application of UAV multispectral sensing and machine learning for monitoring the dynamics of wheat growth. Although the study is impressive, there are still several aspects to be addressed for the better scientific merits and applicability of the study.

Line-by-Line Comments:

·       Lines 7–19: The abstract outlines the paper but lacks details regarding the statistical methods employed for selecting NDRE and TVI as the best indices. Consider a brief description of the statistical approach used for these indices' identification.

·       Lines 38–45: This introduction covers issues of conventional field management but not adequately to set a basis for selection of UAV multispectral data vis-à-vis hyperspectral and satellite imaging. Consider a comparative analysis focusing on the pros and cons of UAV multispectral data vis-à-vis other remote sensing techniques.

·       Lines 56–63: The benefits of Random Forest are highlighted but lack references to prior studies validating its effectiveness in agricultural applications. Cite foundational and recent studies demonstrating the utility of Random Forest for vegetation analysis.

·       Lines 115–123: The justification for selecting the study site is absent. Explain why the chosen location is representative of broader wheat growth monitoring contexts.

·       Lines 135–150: The rationale for excluding specific spectral bands, such as blue or SWIR, is unclear. Explain why these bands were not included and present their potential effects on the research.

·       157–165 lines: Methods for radiometric calibration and atmospheric correction are not included. Describe the software and tools that can be used for reproducibility.

·       200–240 lines: The study has identified 21 vegetation indices, but multicollinearity between them is not evaluated. Analyze multicollinearity and report the results in the methods.

·       275–290 lines: Random Forest modeling description is inadequate as hyperparameter optimization details are not given. Mention the summary of tuning process—by grid search or cross-validation—and the final used hyperparameters.

·       Lines 298–305: The decision tree regression models are unable to handle the probable overfitting. Mention the discussion in terms of control over overfitting by means of pruning techniques or cross-validation.

·       Lines 350–375: The spatial analysis does not possess any statistical metrics that help measure the spatial variability Calculate and mention the spatial statistics such as Moran's I or semi-variograms.

·       Lines 410–425: Model validation focuses on MSE but omits other critical metrics like R² or RMSE. Report additional validation metrics and include confidence intervals.

·       Lines 485–500: The discussion highlights NDRE and TVI as effective indices but does not provide physiological explanations. Elaborate on why these indices outperformed others in the context of wheat physiology.

·       Lines 510–525: Scalability issues are acknowledged, but no solutions are proposed. Discuss potential adaptations for large-scale applications, such as integrating satellite data or reducing hardware costs.

·       Lines 550–565: The conclusion effectively summarizes the findings but does not propose future research directions. Highlight areas for further study, such as the inclusion of additional growth stages and advanced modeling techniques.

References: Ensure foundational studies on vegetation indices and Random Forest are cited. Moreover, add references discussing multicollinearity analysis, validation methods, and comparative studies of machine learning algorithms.

Author Response

Response to Reviewer 1 Comments

Point 1: Lines 7–19: The abstract outlines the paper but lacks details regarding the statistical methods employed for selecting NDRE and TVI as the best indices. Consider a brief description of the statistical approach used for these indices' identification.
Response 1: 
I accept this opinion.
In the revision, I addressed the reviewer’s comment by adding a description in the abstract about the statistical methods used to select NDRE and TVI as the best indices. Specifically, I included the use of Pearson's correlation and stepwise regression to identify the most effective indices. In addition, I made slight adjustments to the language to simplify and improve readability, avoiding overly technical terms while maintaining precision. These changes ensure a clear response to the reviewer’s feedback and enhance the overall clarity of the manuscript.
 
Thank you very much for your revision.

Point 2: Lines 38–45: This introduction covers issues of conventional field management but not adequately to set a basis for selection of UAV multispectral data vis-à-vis hyperspectral and satellite imaging. Consider a comparative analysis focusing on the pros and cons of UAV multispectral data vis-à-vis other remote sensing techniques.
Response 2: 
I accept this opinion.
In response to the reviewer’s comment, I have revised the introduction by incorporating a comparative analysis of UAV multispectral data with other remote sensing techniques such as hyperspectral and satellite imaging. I highlighted the advantages of UAV multispectral data, such as its high spatiotemporal resolution and flexibility, which make it particularly suitable for precise, localized monitoring in agriculture. I also discussed the limitations of hyperspectral imaging, which, despite offering detailed spectral information, is more costly and complex to process, and the drawbacks of satellite imagery, which provides larger area coverage but at a lower resolution and with longer revisit times. This comparative analysis now provides a clearer basis for the selection of UAV multispectral data in the context of agricultural monitoring, addressing the reviewer’s concerns about the justification for its use.
 
Thank you very much for your revision.

Point 3: Lines 56–63: The benefits of Random Forest are highlighted but lack references to prior studies validating its effectiveness in agricultural applications. Cite foundational and recent studies demonstrating the utility of Random Forest for vegetation analysis.
Response 3: 
I accept this opinion.
Included references to foundational and recent studies that have applied random forest models in agricultural remote sensing. The references include Breiman (2001) for the introduction of the algorithm, and more recent studies such as Gislason et al. (2006), Lu et al. (2016), and Sun et al. (2019) to support the use of random forest in vegetation analysis and agricultural applications. The revised text emphasizes the strengths of random forest and demonstrates its broad application in agriculture, providing necessary context and supporting evidence.
 
Thank you very much for your revision.

Point 4: Lines 115–123: The justification for selecting the study site is absent. Explain why the chosen location is representative of broader wheat growth monitoring contexts.
Response 4: 
I accept this opinion.
I added an explanation for why the Hebei Academy of Agriculture and Forestry Sciences Wheat Experimental Station was chosen as the study site. It emphasizes the site's representation of broader wheat cultivation contexts, such as typical environmental stresses and its significance in China's wheat production. I included a statement on how the results from this site can be generalized to similar wheat-growing regions around the world, demonstrating its broader applicability.
 
Thank you very much for your revision.

Point 5: Lines 135–150: The rationale for excluding specific spectral bands, such as blue or SWIR, is unclear. Explain why these bands were not included and present their potential effects on the research.
Response 5: 
I accept this opinion.
The revision now provides a clear rationale for excluding the blue and SWIR bands. It explains that the blue band is less sensitive for monitoring vegetation health compared to the chosen bands, and that the SWIR band was excluded because the focus was on growth rather than water stress. The text also briefly mentions the potential effects of including SWIR, such as complicating the analysis without adding significant value to the study’s objectives.
 
Thank you very much for your revision.

Point 6: 157–165 lines: Methods for radiometric calibration and atmospheric correction are not included. Describe the software and tools that can be used for reproducibility.
Response 6: 
I accept this opinion.
The revision now includes a detailed description of the radiometric calibration and atmospheric correction methods. 
 
Thank you very much for your revision.

Point 7: 200–240 lines: The study has identified 21 vegetation indices, but multicollinearity between them is not evaluated. Analyze multicollinearity and report the results in the methods.
Response 7: 
I accept this opinion.
Mentioned that the analysis was carried out using the statsmodels library in Python, which is commonly used for such statistical evaluations. I clarified that the results of the multicollinearity analysis guided the selection of indices for modeling, ensuring the robustness of the models.
 
Thank you very much for your revision.

Point 8: 275–290 lines: Random Forest modeling description is inadequate as hyperparameter optimization details are not given. Mention the summary of tuning process—by grid search or cross-validation—and the final used hyperparameters.
Response 8: 
I accept this opinion.
I added a description of the hyperparameter tuning process using grid search with cross-validation, as recommended by the reviewer. I mentioned the specific hyperparameters that were tuned (n_estimators, max_depth, and min_samples_split) and provided the final values used in the model.
 
Thank you very much for your revision.

Point 9: Lines 298–305: The decision tree regression models are unable to handle the probable overfitting. Mention the discussion in terms of control over overfitting by means of pruning techniques or cross-validation.
Response 9: 
I accept this opinion.
I added explanations about pruning and cross-validation as methods to control overfitting in decision tree models. Both techniques were emphasized to show how they contribute to model robustness and prevent overfitting.
 
Thank you very much for your revision.

Point 10: Lines 350–375: The spatial analysis does not possess any statistical metrics that help measure the spatial variability Calculate and mention the spatial statistics such as Moran's I or semi-variograms.
Response 10: 
I accept this opinion.
I added the calculation of Moran's I and semi-variograms to measure spatial variability, as requested by the reviewer. I provided a brief explanation of what Moran's I and semi-variograms are, and how they help assess spatial patterns in the data.
 
Thank you very much for your revision.

Point 11: Lines 410–425: Model validation focuses on MSE but omits other critical metrics like R² or RMSE. Report additional validation metrics and include confidence intervals.
Response 11: 
I accept this opinion.
I included R² and RMSE as additional validation metrics, as well as confidence intervals for these values. I provided a brief description of R² and RMSE, explaining their importance in model validation and how they contribute to assessing the model's predictive accuracy.
 
Thank you very much for your revision.

Point 12: Lines 485–500: The discussion highlights NDRE and TVI as effective indices but does not provide physiological explanations. Elaborate on why these indices outperformed others in the context of wheat physiology.
Response 12: 
I accept this opinion.
I elaborated on the physiological relevance of NDRE and TVI by explaining their sensitivity to chlorophyll content, photosynthesis, and canopy structure. The explanation connects how these indices reflect key physiological processes critical to wheat growth, making them particularly effective in monitoring crop health and development.
 
Thank you very much for your revision.

Point 13: Lines 510–525: Scalability issues are acknowledged, but no solutions are proposed. Discuss potential adaptations for large-scale applications, such as integrating satellite data or reducing hardware costs.
Response 13: 
I accept this opinion.
I discussed potential solutions for large-scale applications, including integrating satellite data and reducing hardware costs. I elaborated on how combining UAV and satellite data can enhance scalability and how advancements in technology could reduce costs, making the system more accessible for broader use.
 
Thank you very much for your revision.

Point 14: Lines 550–565: The conclusion effectively summarizes the findings but does not propose future research directions. Highlight areas for further study, such as the inclusion of additional growth stages and advanced modeling techniques.
Response 14: 
I accept this opinion.
I added suggestions for future research, including the inclusion of more growth stages and the use of advanced modeling techniques. I mentioned the potential for integrating additional data sources, such as weather and soil data, to improve the robustness of crop health models.
 
Thank you very much for your revision.

Point 15: References: Ensure foundational studies on vegetation indices and Random Forest are cited. Moreover, add references discussing multicollinearity analysis, validation methods, and comparative studies of machine learning algorithms.
Response 15: 
I accept this opinion.
Based on the expert's suggestions, I have added the relevant references as requested, covering foundational and recent studies on vegetation indices, Random Forest algorithm, multicollinearity analysis, model validation methods, and comparisons of machine learning algorithms. The inclusion of these references strengthens the theoretical foundation of the paper and further supports the methods and models used in this study.
Thank you very much for your revision.

Thanks again for the sincere opinions of the experts and the hard work of the editors!
Sincerely yours,

Dr. Donghui Zhang
Institute of Remote Sensing Satellite, China Academy of Space Technology, Beijing, China 100095
email: zhangdonghui@alu.cdut.edu.cn
tel: +8613381126130     Webpage: https://sciprofiles.com/profile/2204570

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

With this new version, the authors have made important changes to the original manuscript, responding clearly and accurately to the doubts raised.

A study on the use of multispectral UAV technology to monitor durum wheat growth during the main phenological phases is presented. In order to define effective and efficient management strategies, a comprehensive spatio-temporal monitoring framework was developed. The results show that the NDRE and TVI are sensitive to dynamic changes in wheat.

The manuscript is clearly and correctly organised. The integration of UAVs, multispectral indices, temporal variables and machine learning models represents a significant and innovative contribution to the field of digital precision agriculture.

The article addresses a topic of great interest in the field in question, and fits perfectly into the journal's topics.

The introduction is well laid out, personally I would expand on the discussion regarding the combination of multispectral data and machine learning systems to give readers a more detailed picture of the state of the art.

In the methodological section, important and positive changes have been made, however I feel that more detail is needed on the flight parameters of the UAV (speed, viewing angle, atmospheric conditions...). I would also integrate a more detailed explanation of the radiometric calibration criteria and atmospheric correction, to make the manuscript clearer for readers.

The results are in line with the experimental hypotheses and are presented clearly but not very concisely, I would try to shorten this paragraph.

The conclusions reflect what emerged from the results, however, there is a lack of truly practical considerations aimed at facilitating the practical implementation of the proposed methods.

The article is certainly interesting and scientifically valid, with these minor adjustments I believe it could be an important contribution to research in precision agriculture.

Author Response

Response to Reviewer 2 Comments

Point 1: The introduction is well laid out, personally I would expand on the discussion regarding the combination of multispectral data and machine learning systems to give readers a more detailed picture of the state of the art.
Response 1: 
I accept this opinion.
Based on the expert's suggestion, I have expanded the discussion in the introduction regarding the combination of multispectral data and machine learning systems. This enhancement provides readers with a more detailed and comprehensive understanding of the current state of the art in this field. I believe this additional information enriches the background and context of the study, helping to emphasize the relevance and potential of integrating these technologies for crop monitoring and precision agriculture.
Thank you very much for your revision.

Point 2: In the methodological section, important and positive changes have been made, however I feel that more detail is needed on the flight parameters of the UAV (speed, viewing angle, atmospheric conditions...). I would also integrate a more detailed explanation of the radiometric calibration criteria and atmospheric correction, to make the manuscript clearer for readers.
Response 2: 
I accept this opinion.
Based on the expert's suggestion, I have added more details to the methodological section, particularly regarding the flight parameters of the UAV, including speed, viewing angle, and atmospheric conditions during data collection. Additionally, I have provided a more detailed explanation of the radiometric calibration criteria and atmospheric correction process, clarifying these steps to ensure the manuscript is more transparent and accessible for readers. These additions help to enhance the comprehensibility and reproducibility of the methodology, providing a clearer understanding of the data acquisition and processing techniques used in the study.
Thank you very much for your revision.

Point 3: The results are in line with the experimental hypotheses and are presented clearly but not very concisely, I would try to shorten this paragraph.
Response 3: 
I accept this opinion.
Based on the expert's suggestion, I have revised the results section to make it more concise while retaining the essential findings. I focused on streamlining the explanation without losing important details, ensuring that the key results are communicated clearly and efficiently. This revision should improve the readability of the manuscript and make the presentation of the results more succinct.
Thank you very much for your revision.

Point 4: The conclusions reflect what emerged from the results, however, there is a lack of truly practical considerations aimed at facilitating the practical implementation of the proposed methods.
Response 4: 
I accept this opinion.
Based on the expert's suggestion, I have revised the conclusion to include more practical considerations regarding the implementation of the proposed methods. I have emphasized how the techniques can be effectively applied in real-world agricultural settings, including the potential for integrating UAV-based monitoring with existing farming practices and technologies. Additionally, I discussed how the methods can be scaled for larger areas and the necessary steps to ensure cost-effectiveness, making the approach more accessible to farmers and practitioners. These additions aim to provide a clearer pathway for the practical adoption of the proposed methods in precision agriculture.
Thank you very much for your revision.

Point 5: The article is certainly interesting and scientifically valid, with these minor adjustments I believe it could be an important contribution to research in precision agriculture.
Response 5: 
I accept this opinion.
Thank you very much for your positive feedback. I appreciate your recognition of the scientific validity and relevance of the article. I have made the suggested adjustments and believe that these improvements enhance the clarity and practical applicability of the study. I hope the revisions will contribute to advancing research in precision agriculture and provide valuable insights for both the scientific community and practitioners in the field.
Thank you very much for your revision.

Thanks again for the sincere opinions of the experts and the hard work of the editors!
Sincerely yours,

Dr. Donghui Zhang
Institute of Remote Sensing Satellite, China Academy of Space Technology, Beijing, China 100095
email: zhangdonghui@alu.cdut.edu.cn
tel: +8613381126130     Webpage: https://sciprofiles.com/profile/2204570

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Agriculture
Integration of UAV Multispectral Remote Sensing and Random
Forest for Full-Growth Stage Monitoring of Wheat Dynamics
Manuscript ID: agriculture-3460378
 
For Author  
Based on a review of the provided manuscript titled " Integration of UAV Multispectral Remote Sensing and Random Forest for Full-Growth Stage Monitoring of Wheat Dynamics" here are my comments and recommendations for each section:
Comments:
Abstract
-    The study's primary goals, procedures, and conclusions are succinctly outlined in the abstract. It may be enhanced, though, by: (i) Making it clear in the first phrase how important the results are for agricultural methods. (ii) To make a bigger effect, include detailed quantitative findings about the predicted performance and modeling accuracy. (iii) Offering suggestions for real-world applications or ramifications of precision agricultural research.
Introduction
-    A thorough history of the significance of wheat and the difficulties in growing it is given in the introduction. However, it may be improved by: (i) Adding more recent research on UAV uses in agriculture to the literature review, emphasizing any gaps that your work fills. (ii) Outlining the research questions or hypotheses in detail at the conclusion of the introduction to help the reader comprehend the goals of the study.
Materials and Methods
-    Though generally well-structured, this section might need the following improvements: (i) More information about the UAV data capture procedure, including flying characteristics and circumstances during data collection. (ii) Outlining the justification for the use of particular vegetative indicators and their connection to the traits of wheat growth. (iii) Providing a more thorough description of the statistical and data processing techniques applied, especially with regard to the random forest model and its validation.
Results
-    The findings are clearly shown in the results section, however take into account the following: (i) Using visual aids like charts or graphs to better depict important linkages and patterns. This would improve comprehension and reader engagement. (ii) Making certain that every figure and table is cited within the text and has a brief explanation of its significance. (iii) Offering a more thorough evaluation of the model's performance that incorporates metrics other than MSE, including R-squared values or cross-validation findings.
Discussion
-    Although the commentary does a good job of interpreting the data, it may be strengthened by: (i) Including comparisons with previous research to put the findings in a more contextualized setting. Talk about the ramifications of these comparisons and whether the results are consistent with or different from those of earlier research. (ii) More clearly addressing the study's shortcomings, such as possible sources of bias or inconsistent findings. (iii) Making recommendations for future research avenues that might expand upon current work, especially with regard to the use of other modeling methodologies or the integration of additional data sources.
Conclusions
-    The conclusion does a good job of summarizing the main conclusions, but it might be improved by: (i) Emphasizing the research's wider implications for resource management and precision agriculture. (ii) Outlining particular suggestions for agricultural professionals in light of the study's findings. (iii) Outlining specific research topics that might build upon the results that have been given.
References
-    Review the reference list for completeness and accuracy. Ensure that all cited studies are relevant and recent.
-    Authors should be added few recent references in introduction, data, methods and discussions.
-    Change the format of references according to the requirement of the journal.
-    References cited in the text must appear in the list of references.
-    You will find some new related references, which should be added to the literature review.
-    Digital Object Identifier (DOI) for the references should be added.
-    Some References are cited in the body but their bibliographic information is missing. Kindly provide its bibliographic information in the list.

Author Response

Response to Reviewer 3 Comments

Point 1: Abstract
-    The study's primary goals, procedures, and conclusions are succinctly outlined in the abstract. It may be enhanced, though, by: (i) Making it clear in the first phrase how important the results are for agricultural methods. (ii) To make a bigger effect, include detailed quantitative findings about the predicted performance and modeling accuracy. (iii) Offering suggestions for real-world applications or ramifications of precision agricultural research.
Response 1: 
I accept this opinion.
Based on the opinions of experts, I have made revisions and improvements to the abstract. Firstly, I emphasized the importance of research results for agricultural methods at the beginning, highlighting how multi spectral monitoring based on drones can enhance the application value of precision agriculture. Secondly, I included specific quantitative results, including the predictive performance and accuracy of the model, to provide a more convincing presentation of the results. Finally, I put forward practical application suggestions for the research, explaining how these methods can achieve precise intervention in crop management and be combined with existing agricultural practices to improve resource utilization efficiency. These modifications make the abstract more influential and practical, helping readers better understand the practical significance and application potential of the research.
 
Thank you very much for your revision.

Point 2: Introduction
-    A thorough history of the significance of wheat and the difficulties in growing it is given in the introduction. However, it may be improved by: (i) Adding more recent research on UAV uses in agriculture to the literature review, emphasizing any gaps that your work fills. (ii) Outlining the research questions or hypotheses in detail at the conclusion of the introduction to help the reader comprehend the goals of the study.
Response 2: 
I accept this opinion.
(i) I have included more recent studies on the use of UAVs in agriculture, highlighting advancements in the technology and identifying existing gaps that this study aims to address. This provides a more comprehensive view of the current state of the field and underscores the contribution of this research.
 
(ii) I have clearly outlined the research questions and hypotheses at the end of the introduction, giving the reader a more precise understanding of the study’s objectives and how it aims to contribute to the field of precision agriculture.
 
 
Thank you very much for your revision.

Point 3: Materials and Methods
-    Though generally well-structured, this section might need the following improvements: (i) More information about the UAV data capture procedure, including flying characteristics and circumstances during data collection. (ii) Outlining the justification for the use of particular vegetative indicators and their connection to the traits of wheat growth. (iii) Providing a more thorough description of the statistical and data processing techniques applied, especially with regard to the random forest model and its validation.
Response 3: 
I accept this opinion.
(i) UAV Data Capture Procedure: I have included additional details about the UAV data collection process, such as the flight characteristics (speed, altitude, viewing angle) and the environmental conditions (weather, atmospheric factors) during data acquisition. This provides a clearer understanding of how the data was captured and its relevance to the study.
 

 

 
(ii) Justification for Vegetative Indicators: I have expanded on the choice of vegetation indices, explaining their relevance to wheat growth traits such as chlorophyll content, biomass, and photosynthetic activity. I also outlined how these indices were selected based on their ability to capture key aspects of wheat physiology.
 
(iii) Statistical and Data Processing Techniques: I have provided a more detailed description of the statistical and data processing methods, especially regarding the random forest model. This includes an explanation of how the model was trained, the hyperparameters tuned, and the validation process used to assess its performance, such as cross-validation and performance metrics.
Thank you very much for your revision.
 
Point 4: Results
-    The findings are clearly shown in the results section, however take into account the following: (i) Using visual aids like charts or graphs to better depict important linkages and patterns. This would improve comprehension and reader engagement. (ii) Making certain that every figure and table is cited within the text and has a brief explanation of its significance. (iii) Offering a more thorough evaluation of the model's performance that incorporates metrics other than MSE, including R-squared values or cross-validation findings.
Response 4: 
I accept this opinion.
(i) Visual Aids: I have added charts and graphs to better illustrate the key relationships and patterns observed in the data. These visual aids should help improve reader engagement and comprehension by providing a clearer depiction of the findings.
 
 
(ii) Citing Figures and Tables: I have ensured that all figures and tables are properly cited within the text. Each figure and table now includes a brief explanation of its significance, helping readers understand how the data supports the findings.
 
 
 
(iii) Model Performance Evaluation: I have provided a more thorough evaluation of the model's performance, incorporating additional metrics such as R-squared values and cross-validation results. This offers a more comprehensive view of the model’s accuracy and robustness, beyond just the MSE.
 
Thank you very much for your revision.

Point 5: Discussion
-    Although the commentary does a good job of interpreting the data, it may be strengthened by: (i) Including comparisons with previous research to put the findings in a more contextualized setting. Talk about the ramifications of these comparisons and whether the results are consistent with or different from those of earlier research. (ii) More clearly addressing the study's shortcomings, such as possible sources of bias or inconsistent findings. (iii) Making recommendations for future research avenues that might expand upon current work, especially with regard to the use of other modeling methodologies or the integration of additional data sources.
Response 5: 
I accept this opinion.
(i) Comparison with Previous Research: I have added a discussion comparing the findings of this study with those of previous research. This provides a more contextualized understanding of the results, highlighting whether they align with or differ from earlier studies and discussing the implications of these comparisons.
 
(ii) Study Limitations: I have explicitly addressed the potential limitations of the study, including possible sources of bias, environmental variability, and any inconsistencies in the findings. This helps to present a balanced interpretation of the results.
 
(iii) Future Research Directions: I have included recommendations for future research, particularly in terms of exploring alternative modeling techniques and integrating additional data sources, such as weather, soil, and hyperspectral imaging data. These suggestions aim to build upon the current work and enhance the applicability of UAV-based crop monitoring.
 
Thank you very much for your revision.

Point 6: Conclusions
-    The conclusion does a good job of summarizing the main conclusions, but it might be improved by: (i) Emphasizing the research's wider implications for resource management and precision agriculture. (ii) Outlining particular suggestions for agricultural professionals in light of the study's findings. (iii) Outlining specific research topics that might build upon the results that have been given.
Response 6: 
I accept this opinion.
(i) Wider Implications: I have emphasized the broader implications of the study's findings for resource management and precision agriculture. Specifically, I highlighted how UAV-based multispectral data can be leveraged to optimize resource use, improve crop health monitoring, and enhance decision-making in precision agriculture.
 
(ii) Suggestions for Agricultural Professionals: I have included specific recommendations for agricultural professionals, focusing on how the findings can be practically applied to optimize crop management, such as using the identified vegetation indices for early-stage crop monitoring and targeted interventions.
 
(iii) Future Research Directions: I have outlined several specific research topics that could build upon the results of this study, including exploring the use of alternative modeling techniques, integrating additional data sources like soil and weather information, and expanding the use of UAV-based monitoring to other crops and agricultural systems.
 
Thank you very much for your revision.

Point 7: References
-    Review the reference list for completeness and accuracy. Ensure that all cited studies are relevant and recent.
-    Authors should be added few recent references in introduction, data, methods and discussions.
-    Change the format of references according to the requirement of the journal.
-    References cited in the text must appear in the list of references.
-    You will find some new related references, which should be added to the literature review.
-    Digital Object Identifier (DOI) for the references should be added.
-    Some References are cited in the body but their bibliographic information is missing. Kindly provide its bibliographic information in the list.
Response 7: 
I accept this opinion.
Based on the expert's suggestions, I have revised the reference list to ensure its completeness and accuracy, reformatting it according to the journal's requirements, and adding recent relevant references along with DOI information.
Thank you very much for your revision.

 

Thanks again for the sincere opinions of the experts and the hard work of the editors!
Sincerely yours,

Dr. Donghui Zhang
Institute of Remote Sensing Satellite, China Academy of Space Technology, Beijing, China 100095
email: zhangdonghui@alu.cdut.edu.cn
tel: +8613381126130     Webpage: https://sciprofiles.com/profile/2204570

 

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The introduction effectively establishes the importance of wheat as a staple crop and the challenges posed by global climate change. It highlights the potential of UAV-based remote sensing in advancing precision agriculture, offering a strong foundation for the study's objectives. However, the rationale behind focusing on multispectral data over other advanced sensing technologies, such as hyperspectral imaging, is not fully justified. Furthermore, the transition from broader challenges in agriculture to the specific research aim is somewhat abrupt, leaving the context underexplored. What do you think?

Why was multispectral imaging prioritized over potentially more precise methods like hyperspectral imaging, especially given the study's focus on full-cycle wheat growth monitoring?

The methods section provides a detailed account of UAV sensor specifications, data acquisition procedures, and ground-truth measurements. The use of vegetation indices and random forest modeling is well-explained, but the justification for selecting specific indices and their relevance to wheat physiology could be more explicit. The exclusion of certain growth stages, such as germination and grain filling, raises concerns about the comprehensiveness of the monitoring approach. How do the authors ensure that excluding early and late growth stages does not compromise the model’s applicability for full-cycle monitoring?

The results section presents a wealth of data on wheat height and chlorophyll content, along with their correlations with vegetation indices. The figures and spatial distribution maps are informative but lack a clear explanation of how these insights translate into actionable agricultural practices. Additionally, the variability in results across different growth stages is noted but not adequately discussed in terms of its implications for model reliability. To what extent do environmental factors, such as light variations (cloudy etc) or microclimatic differences, affect the accuracy and consistency of the vegetation indices used?

The discussion interprets the findings in the context of precision agriculture but falls short in critically evaluating the limitations of the methodology. For instance, while the superiority of the random forest model is emphasized, there is minimal exploration of its potential biases or limitations in generalizability. The discussion of abnormal field areas and their implications for agricultural interventions is compelling but could benefit from more concrete examples or case studies. How do the observed discrepancies between UAV and ground-truth measurements influence the practical utility of the proposed framework in real-world agricultural settings?

Remote sensing vegetation indices have been widely used in various fields, from assessing vegetation health and environmental changes to managing crop production under several conditions such as temperature extremes, water availability, and variations in light intensity or quality (e.g., https://doi.org/10.3390/ijpb15030058). These indices are invaluable tools for detecting plant responses, which often manifest through changes in pigment composition and photosynthetic efficiency. Highlighting this broader utility would strengthen the rationale for optimizing vegetation indices tailored for wheat and other crops.

The conclusions highlight the study’s contributions to precision agriculture and its potential applications. However, the lack of specific recommendations for integrating UAV technology into existing agricultural systems diminishes the practical relevance of the findings. While future directions are mentioned, they remain general and do not address key aspects such as cost-effectiveness or farmer adoption.

Overall, the manuscript's narrative is dense, with technical details sometimes overshadowing broader implications. A more balanced approach to discussing technical and practical aspects would enhance accessibility for a wider audience.

The discussion on vegetation indices and their utility is robust, but the additional emphasis on how they compare with alternative metrics would strengthen the study's validity.

Some references appear outdated, particularly in the rapidly evolving field of UAV remote sensing. Incorporating more recent studies would enhance the manuscript's relevance.

Thank you.

Author Response

Response to Reviewer 4 Comments

Point 1: The introduction effectively establishes the importance of wheat as a staple crop and the challenges posed by global climate change. It highlights the potential of UAV-based remote sensing in advancing precision agriculture, offering a strong foundation for the study's objectives. However, the rationale behind focusing on multispectral data over other advanced sensing technologies, such as hyperspectral imaging, is not fully justified. Furthermore, the transition from broader challenges in agriculture to the specific research aim is somewhat abrupt, leaving the context underexplored. What do you think?
Response 1: 
I accept this opinion.
Based on the expert's suggestions, I have made the following revisions to the Introduction:
(i) Justification for Focusing on Multispectral Data: I have included a more detailed explanation of why multispectral data was chosen over other advanced sensing technologies like hyperspectral imaging. I highlighted the advantages of multispectral data, such as its cost-effectiveness, ease of use, and sufficient sensitivity for monitoring key vegetation traits in wheat, making it a practical choice for precision agriculture.
 
(ii) Smoother Transition to Research Aim: I have refined the transition between discussing the broader challenges in agriculture and the specific research aims. I included additional context to smoothly guide the reader from the global agricultural issues to the focused objectives of the study, ensuring a more coherent flow.
 
 
Thank you very much for your revision.

Point 2: Why was multispectral imaging prioritized over potentially more precise methods like hyperspectral imaging, especially given the study's focus on full-cycle wheat growth monitoring?
Response 2: 
I accept this opinion.
Based on the expert's suggestion, I have expanded on the rationale for prioritizing multispectral imaging over hyperspectral imaging.
 
Thank you very much for your revision.

Point 3: The methods section provides a detailed account of UAV sensor specifications, data acquisition procedures, and ground-truth measurements. The use of vegetation indices and random forest modeling is well-explained, but the justification for selecting specific indices and their relevance to wheat physiology could be more explicit. The exclusion of certain growth stages, such as germination and grain filling, raises concerns about the comprehensiveness of the monitoring approach. How do the authors ensure that excluding early and late growth stages does not compromise the model’s applicability for full-cycle monitoring?
Response 3: 
I accept this opinion.
(i) Justification for Selecting Vegetation Indices: I have clarified the rationale for selecting the specific vegetation indices used in this study. I explained their direct relevance to wheat physiology, particularly how indices like NDRE and TVI are sensitive to key physiological traits such as chlorophyll content and canopy structure, which are critical for monitoring wheat growth at various stages.
 
(ii) Exclusion of Early and Late Growth Stages: I have addressed the exclusion of early (germination) and late (grain filling) growth stages by discussing how the selected growth stages in the study were representative of the key developmental phases of wheat. I also explained that while these stages are important, the focus on critical mid-growth stages such as tillering, jointing, and flowering provides sufficient insight into the overall growth trends of wheat. To ensure full-cycle applicability, I have emphasized that future studies could incorporate these stages to further validate the model’s robustness and applicability across all stages.
 
 
Thank you very much for your revision.

Point 4: The results section presents a wealth of data on wheat height and chlorophyll content, along with their correlations with vegetation indices. The figures and spatial distribution maps are informative but lack a clear explanation of how these insights translate into actionable agricultural practices. Additionally, the variability in results across different growth stages is noted but not adequately discussed in terms of its implications for model reliability. To what extent do environmental factors, such as light variations (cloudy etc) or microclimatic differences, affect the accuracy and consistency of the vegetation indices used?
Response 4: 
I accept this opinion.
(i) Actionable Agricultural Practices: I have provided a clearer explanation of how the insights from wheat height and chlorophyll content, along with their correlations with vegetation indices, can be translated into actionable agricultural practices. Specifically, I discussed how these indices can be used to guide interventions such as targeted fertilization and irrigation management, which can improve crop health and optimize resource use.
 
 
 
(ii) Variability Across Growth Stages: I have expanded on the variability observed across different growth stages, addressing how it impacts the reliability of the model. I explained how changes in environmental conditions during each growth stage can affect the vegetation indices and model predictions, and how understanding these variations is crucial for ensuring model robustness.
 
(iii) Environmental Factors and Model Accuracy: I have discussed how environmental factors, such as light variations (e.g., cloudy days) and microclimatic differences, can influence the accuracy and consistency of vegetation indices. These factors can cause fluctuations in reflectance values, which may affect the reliability of the indices. I emphasized the importance of accounting for such variability in future studies and the need for robust calibration and atmospheric correction techniques to mitigate these effects.
 
 
Thank you very much for your revision.

Point 5: The discussion interprets the findings in the context of precision agriculture but falls short in critically evaluating the limitations of the methodology. For instance, while the superiority of the random forest model is emphasized, there is minimal exploration of its potential biases or limitations in generalizability. The discussion of abnormal field areas and their implications for agricultural interventions is compelling but could benefit from more concrete examples or case studies. How do the observed discrepancies between UAV and ground-truth measurements influence the practical utility of the proposed framework in real-world agricultural settings?
Response 5: 
I accept this opinion.
(i) Critical Evaluation of Methodology: I have included a more detailed discussion of the limitations of the random forest model, particularly in terms of its potential biases and limitations in generalizability. I addressed the fact that while the model performs well under the conditions of this study, its effectiveness may vary with different environmental conditions, crop types, or geographic regions. I also mentioned the need for future research to assess its robustness across a broader range of agricultural settings.
 
(ii) Concrete Examples for Abnormal Field Areas: I expanded on the implications of detecting abnormal field areas, by providing more concrete examples and case studies. I explained how these insights can be applied in real-world agricultural interventions, such as precision fertilization and irrigation scheduling, to improve crop management.
 
(iii) Discrepancies Between UAV and Ground-Truth Measurements: I discussed the discrepancies between UAV-based measurements and ground-truth data, particularly focusing on how these differences may affect the practical utility of the proposed framework in real-world agricultural settings. I explained how variations in environmental conditions, sensor calibration, and data processing could influence the accuracy of UAV-based measurements and emphasized the need for robust validation procedures to ensure the framework’s practical applicability.
 
Thank you very much for your revision.

Point 6: Remote sensing vegetation indices have been widely used in various fields, from assessing vegetation health and environmental changes to managing crop production under several conditions such as temperature extremes, water availability, and variations in light intensity or quality (e.g., https://doi.org/10.3390/ijpb15030058). These indices are invaluable tools for detecting plant responses, which often manifest through changes in pigment composition and photosynthetic efficiency. Highlighting this broader utility would strengthen the rationale for optimizing vegetation indices tailored for wheat and other crops.
Response 6: 
I accept this opinion.
Under the guidance of this paper, the manuscript content has been revised (https://doi.org/10.3390/ijpb15030058). Based on the expert's suggestion, I have expanded the discussion to highlight the broader utility of remote sensing vegetation indices. I included a reference to their wide application in various fields, such as assessing vegetation health, monitoring environmental changes, and managing crop production under conditions like temperature extremes, water availability, and variations in light intensity. By emphasizing the versatility of vegetation indices in detecting plant responses through changes in pigment composition and photosynthetic efficiency, I have reinforced the rationale for optimizing indices specifically tailored for wheat and other crops. This broader context strengthens the justification for the use of vegetation indices in precision agriculture. Let me know if you need any further adjustments!
 
 
Thank you very much for your revision.

Point 7: The conclusions highlight the study’s contributions to precision agriculture and its potential applications. However, the lack of specific recommendations for integrating UAV technology into existing agricultural systems diminishes the practical relevance of the findings. While future directions are mentioned, they remain general and do not address key aspects such as cost-effectiveness or farmer adoption.
Response 7: 
I accept this opinion.
(i) Recommendations for Integrating UAV Technology: I have added specific recommendations for integrating UAV technology into existing agricultural systems. This includes practical steps for farmers to adopt UAV-based monitoring, such as training programs, partnerships with UAV service providers, and examples of how UAV data can complement traditional farming practices like soil testing and crop scouting.
 
(ii) Addressing Cost-Effectiveness and Farmer Adoption: I have discussed the cost-effectiveness of UAV technology, emphasizing how advances in UAV hardware and software can make this technology more affordable for farmers. I also outlined the potential barriers to adoption, such as the need for technical knowledge and initial investment, and suggested strategies for overcoming these challenges, such as subsidies or cooperative farming models.
 
 
(iii) More Specific Future Directions: I have refined the future research directions to address key aspects like the integration of UAV data with other precision agriculture tools, the scalability of UAV applications for large farms, and methods to improve the ease of use and accessibility of UAV systems for farmers.
 
Thank you very much for your revision.

Point 8: Overall, the manuscript's narrative is dense, with technical details sometimes overshadowing broader implications. A more balanced approach to discussing technical and practical aspects would enhance accessibility for a wider audience.
The discussion on vegetation indices and their utility is robust, but the additional emphasis on how they compare with alternative metrics would strengthen the study's validity.
Some references appear outdated, particularly in the rapidly evolving field of UAV remote sensing. Incorporating more recent studies would enhance the manuscript's relevance.
Thank you.
Response 8: 
I accept this opinion.
Based on the expert's suggestions, I have made the following revisions to improve the manuscript:
(i) Balanced Approach: I have worked to provide a more balanced approach by discussing both technical details and their broader implications in parallel. This allows the manuscript to remain accessible to a wider audience, ensuring that the practical applications of the study are equally emphasized as the technical findings.
 
 
(ii) Comparison with Alternative Metrics: I have included additional discussion on how the selected vegetation indices compare with alternative metrics, such as hyperspectral indices and satellite-based data. This comparison strengthens the study's validity by placing the findings in the context of existing research and illustrating the advantages and limitations of different measurement methods.
 
(iii) Incorporating Recent Studies: I have updated the literature review to include more recent studies, particularly in the rapidly evolving field of UAV remote sensing. This inclusion ensures that the manuscript remains relevant and reflects the latest advancements in technology and methodology.
Thank you very much for your revision.

 

Thanks again for the sincere opinions of the experts and the hard work of the editors!
Sincerely yours,

Dr. Donghui Zhang
Institute of Remote Sensing Satellite, China Academy of Space Technology, Beijing, China 100095
email: zhangdonghui@alu.cdut.edu.cn
tel: +8613381126130     Webpage: https://sciprofiles.com/profile/2204570

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript contains a very interesting study on the integration of UAV multispectral remote sensing and Random Forest modeling for wheat growth monitoring. The revisions have considerably improved the clarity and structure of the manuscript; previous concerns have been addressed, with some more detailed suggestions below to further refine the study.

1.      Introduction (Lines 38–45): The added comparison of UAV multispectral data with other remote sensing techniques is valuable. However, consider strengthening the justification by briefly discussing how cost-effectiveness and operational feasibility make UAV preferable for small-scale precision agriculture.

2.      Study Area (Lines 115–123): The site justification is much clearer now. Further elaboration on its representativeness with some quantitative measure, for instance, percentage of wheat-growing regions under similar climatic conditions, would strengthen it further.

3.      Methods (Lines 157–165): Description of radiometric calibration and atmospheric correction has improved. To help reproducibility, stating the software version used for atmospheric correction would be nice.

4.      Multicollinearity Analysis(Lines 200–240): It would be helpful to include the VIF results. Briefly explain why your chosen threshold for multicollinearity is, in fact, a valid threshold to determine statistical independence among the retained indices.

12. Random Forest Model(Lines 275–290): The hyperparameter tuning section is now better explained. A little sentence adding why grid search was chosen over other tuning approaches (e.g., Bayesian optimization) would make it clear.
13. Overfitting Control (Lines 298–305): The discussion of pruning and cross-validation is useful. To strengthen robustness, it would be nice to clarify whether k-fold cross-validation was performed and the number of folds.
14. Spatial Analysis (Lines 350–375): It is very nice that Moran's I and semi-variograms have been added. It would be nice to also provide a short interpretation of the Moran's I value with regard to what a moderate positive spatial autocorrelation might mean in practical agricultural contexts.
15. Validation Metrics (Lines 410–425): R ² and RMSE strengthen the model evaluation quite a lot. Including confidence intervals of those values in the results directly would be better statistically speaking.
16. Discussion (Lines 485–500): The new physiological explanations for NDRE and TVI are very valuable, but elaborate further on whether this effectiveness is cultivar-specific (variations amongst different wheat) or constant for all wheat.
17. Scalability Discussion (Lines 510–525): Nice to see the inclusion of satellite data. It would be nice to have a brief discussion of possible trade-offs between resolution and real-time monitoring capacity when transitioning from UAVs to satellites.
18. Conclusion (Lines 550–565): Nice to see the addition of future research directions. Consider specifying whether future research will look into machine learning alternatives to Random Forest, such as Deep Learning approaches.

Author Response

Response to Reviewer 1 Comments

Point 1: The manuscript contains a very interesting study on the integration of UAV multispectral remote sensing and Random Forest modeling for wheat growth monitoring. The revisions have considerably improved the clarity and structure of the manuscript; previous concerns have been addressed, with some more detailed suggestions below to further refine the study.
Response 1: 
I accept this opinion.
I appreciate the positive feedback regarding the study and the improvements made to the manuscript. I am grateful for your recognition of the enhanced clarity and structure. We have carefully considered your detailed suggestions and have incorporated them into the revised version to further refine and strengthen the study. Thank you very much for your time and effort in reviewing the manuscript.
Thank you very much for your revision.

Point 2: Introduction (Lines 38–45): The added comparison of UAV multispectral data with other remote sensing techniques is valuable. However, consider strengthening the justification by briefly discussing how cost-effectiveness and operational feasibility make UAV preferable for small-scale precision agriculture.
Response 2: 
I accept this opinion.
Thank you for the insightful suggestion. I agree that strengthening the justification for UAVs in small-scale precision agriculture will provide additional clarity. In response, we have revised the manuscript to include a brief discussion on the cost-effectiveness and operational feasibility of UAVs, emphasizing their advantages for small-scale operations due to lower costs, ease of deployment, and high flexibility compared to satellite or hyperspectral technologies. We believe this addition better highlights UAVs' potential for precision agriculture in smaller farming contexts.
 
Thank you very much for your revision.

Point 3: Study Area (Lines 115–123): The site justification is much clearer now. Further elaboration on its representativeness with some quantitative measure, for instance, percentage of wheat-growing regions under similar climatic conditions, would strengthen it further.
Response 3: 
I accept this opinion.
Thank you for your constructive suggestion. We agree that providing a quantitative measure of the site's representativeness will further strengthen the justification. In response, we have added information regarding the percentage of wheat-growing regions under similar climatic conditions, which demonstrates the relevance of the study area to other wheat-producing regions. This addition enhances the generalizability of the findings to broader agricultural contexts.
 
Thank you very much for your revision.

Point 4:  Methods (Lines 157–165): Description of radiometric calibration and atmospheric correction has improved. To help reproducibility, stating the software version used for atmospheric correction would be nice.
Response 4: 
I accept this opinion.
Thank you for the suggestion. I agree that providing the software version used for atmospheric correction would improve the reproducibility of the methods. In response, we have included the version of the software used for atmospheric correction in the revised manuscript. We believe this addition enhances the transparency and reproducibility of the study.
 
Thank you very much for your revision.

Point 5:  Multicollinearity Analysis(Lines 200–240): It would be helpful to include the VIF results. Briefly explain why your chosen threshold for multicollinearity is, in fact, a valid threshold to determine statistical independence among the retained indices.
Response 5: 
I accept this opinion.
Thank you for the suggestion. I agree that including the Variance Inflation Factor (VIF) results will improve the clarity of the analysis. In response, we have added the VIF results to the revised manuscript. Additionally, we have provided a brief explanation of why the chosen threshold for multicollinearity (VIF > 10) is a valid criterion for determining statistical independence among the retained indices. This threshold is widely accepted in statistical analysis as an indicator of high multicollinearity, which can lead to unreliable regression coefficients. By maintaining VIF values below this threshold, we ensure the statistical independence of the retained indices.
 
Thank you very much for your revision.

Point 6:  Random Forest Model(Lines 275–290): The hyperparameter tuning section is now better explained. A little sentence adding why grid search was chosen over other tuning approaches (e.g., Bayesian optimization) would make it clear.
Response 6: 
I accept this opinion.
Thank you for your suggestion. I agree that briefly explaining why grid search was chosen over other tuning approaches will add clarity. In response, we have included a sentence explaining that grid search was selected because of its simplicity and robustness in finding the optimal hyperparameters through exhaustive search over a predefined parameter grid. While other methods like Bayesian optimization can be more efficient, grid search was preferred in this case due to its straightforward application and reliable results, especially when computational resources and time allowed for the exhaustive search.
 
Thank you very much for your revision.

Point 7:  Overfitting Control (Lines 298–305): The discussion of pruning and cross-validation is useful. To strengthen robustness, it would be nice to clarify whether k-fold cross-validation was performed and the number of folds.
Response 7: 
I accept this opinion.
Thank you for your suggestion. I agree that clarifying the number of folds used in the k-fold cross-validation will further strengthen the robustness of the methodology. In response, we have specified that 10-fold cross-validation was performed to evaluate the model’s performance on different subsets of the data. This process ensures that the model generalizes well and reduces the risk of overfitting.
 
Thank you very much for your revision.

Point 8:  Spatial Analysis (Lines 350–375): It is very nice that Moran's I and semi-variograms have been added. It would be nice to also provide a short interpretation of the Moran's I value with regard to what a moderate positive spatial autocorrelation might mean in practical agricultural contexts.
Response 8: 
I accept this opinion.
Thank you for your suggestion. I agree that providing a brief interpretation of the Moran's I value will enhance the understanding of the results. In response, we have added an explanation regarding the Moran's I value. A moderate positive spatial autocorrelation, as indicated by the Moran's I value of 0.65, suggests that areas with higher wheat growth tend to cluster together, while regions with lower growth also form clusters. In practical agricultural contexts, this implies that field management interventions such as irrigation, fertilization, or pest control may need to be applied differently across the field, targeting areas of similar growth characteristics to optimize resource allocation and improve crop yield.
 
Thank you very much for your revision.

Point 9:  Validation Metrics (Lines 410–425): R ² and RMSE strengthen the model evaluation quite a lot. Including confidence intervals of those values in the results directly would be better statistically speaking.
Response 9: 
I accept this opinion.
Thank you for your valuable suggestion. I agree that including the confidence intervals for R² and RMSE would enhance the statistical robustness of the model evaluation. In response, we have added the confidence intervals for both R² and RMSE in the revised manuscript. This addition provides a clearer understanding of the uncertainty in the model's predictions and strengthens the overall assessment of the model’s accuracy and reliability.
 
Thank you very much for your revision.

Point 10:  Discussion (Lines 485–500): The new physiological explanations for NDRE and TVI are very valuable, but elaborate further on whether this effectiveness is cultivar-specific (variations amongst different wheat) or constant for all wheat.
Response 10: 
I accept this opinion.
Thank you for your valuable suggestion. I agree that elaborating on whether the effectiveness of NDRE and TVI is cultivar-specific will enhance the discussion. In response, we have added a section exploring the potential variations in the effectiveness of these indices across different wheat cultivars. While NDRE and TVI have shown strong correlations with wheat growth parameters in this study, their performance may vary depending on the cultivar. Physiological differences such as leaf morphology, growth patterns, and nutrient uptake between wheat cultivars could influence how well these indices reflect growth characteristics. Therefore, further research investigating these indices across multiple cultivars is needed to determine whether their effectiveness is consistent or if it varies with different wheat varieties.
 
Thank you very much for your revision.

Point 11:  Scalability Discussion (Lines 510–525): Nice to see the inclusion of satellite data. It would be nice to have a brief discussion of possible trade-offs between resolution and real-time monitoring capacity when transitioning from UAVs to satellites.
Response 11: 
I accept this opinion.
Thank you for your insightful suggestion. I agree that discussing the trade-offs between resolution and real-time monitoring capacity when transitioning from UAVs to satellites will add depth to the scalability discussion. In response, we have included a brief discussion on this topic. While UAVs offer high spatial resolution and real-time monitoring capabilities, satellite-based remote sensing typically provides broader coverage but at lower spatial resolution. This trade-off is important when considering the scalability of the monitoring system; for large-scale agricultural applications, satellites can offer cost-effective, widespread coverage, but the lower resolution may limit the ability to detect fine-scale variations in crop conditions. On the other hand, UAVs provide high-resolution data but may be less feasible for continuous or large-area monitoring due to logistical constraints. Thus, selecting between UAVs and satellites depends on the specific goals of the study, the spatial resolution needed, and the required real-time data frequency.
 
Thank you very much for your revision.

Point 12:  Conclusion (Lines 550–565): Nice to see the addition of future research directions. Consider specifying whether future research will look into machine learning alternatives to Random Forest, such as Deep Learning approaches.
Response 12: 
I accept this opinion.
Thank you for your valuable suggestion. I agree that specifying whether future research will explore machine learning alternatives to Random Forest, such as Deep Learning approaches, would provide more clarity and direction for the study. In response, we have added a brief mention of this in the conclusion. Future research will indeed consider exploring other machine learning methods, including Deep Learning techniques, to further improve the predictive capabilities and performance of wheat growth monitoring systems. These approaches may offer enhanced model accuracy, especially in handling large and complex datasets, and could provide new insights into crop dynamics that traditional methods might miss.
 
Thank you very much for your revision.

 

Thanks again for the sincere opinions of the experts and the hard work of the editors!
Sincerely yours,

Dr. Donghui Zhang
Institute of Remote Sensing Satellite, China Academy of Space Technology, Beijing, China 100095
email: zhangdonghui@alu.cdut.edu.cn
tel: +8613381126130     Webpage: https://sciprofiles.com/profile/2204570

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I wish success to the authors in their study.
With Best Regards

Comments for author File: Comments.pdf

Author Response

Thank you for your hard work. Sincere wishes