Next Article in Journal
Fintech Service Quality of Saudi Banks: Digital Transformation and Awareness in Satisfaction, Re-Use Intentions, and the Sustainable Performance of Firms
Previous Article in Journal
Online, Experiential Sustainability Education Can Improve Students’ Self-Reported Environmental Attitudes, Behaviours and Wellbeing
 
 
Article
Peer-Review Record

Leveraging Remotely Sensed and Climatic Data for Improved Crop Yield Prediction in the Chi Basin, Thailand

Sustainability 2024, 16(6), 2260; https://doi.org/10.3390/su16062260
by Akkarapon Chaiyana 1, Ratchawatch Hanchoowong 2, Neti Srihanu 3, Haris Prasanchum 4, Anongrit Kangrang 1, Rattana Hormwichian 1, Siwa Kaewplang 1,*, Werapong Koedsin 5 and Alfredo Huete 6
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Sustainability 2024, 16(6), 2260; https://doi.org/10.3390/su16062260
Submission received: 1 February 2024 / Revised: 24 February 2024 / Accepted: 5 March 2024 / Published: 8 March 2024
(This article belongs to the Topic Big Data and Artificial Intelligence, 2nd Volume)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The idea and the paper are excellent, they deal with the topic that has recently been the most important for agricultural production. Global climate changes are global in nature, but there are certainly local characteristics as well, in a narrower or broader sense. Therefore, this model can serve as a basis for prediction in other latitudes, but not the results obtained in this manuscript. The biggest objection in this manuscript is the fact that the prediction refers to a period that has already passed, does it serve any purpose? The results of this paper should be improved by extending the work with the expected results in e.g. in 2030.

Author Response

All authors would like to sincerely thank the reviewer for the very positive, important, and constructive comments and suggestions. We have addressed them as follows.

  • The information of crop production in Thailand is out of date at least 2-3 years for investigation which provided by government sector. It may not on time to assess and launch policies in order to make decision or instruction. In this study, the authors proposed the new novel approach by integrating remote sensing and climatic indicators to predict crop production at the end of crop season. Therefore, this purpose of this study can serve policy makers and increase crop production in the next upcoming year.
  • To predict or project information in 2023, in this case, the information input is based one remote sensing dataset that cannot be applied for this purpose. In addition, the climate and crop area change rapidly, the information which is expected in 2023 may not be reliable. However, crop production based on land use change simulation can be done in subsequent study.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript is well-structured and informative.


Here is the detail about the questions raised.

1.    What is the main question addressed by the research?
•    The research is about the use of remote sensing, MLR models, machine learning and climatic data for prediction of crop yield prediction in the Chi Basin, Thailand.

2.    What parts do you consider original or relevant for the field? What
specific gap in the field does the paper address?
•    Prediction of factors contributing to the high yield in crops is the most relevant part of this research.

3. What does it add to the subject area compared with other published
material?
•    The study model used various variables that other studies do not consider.


4. What specific improvements should the authors consider regarding the
methodology? What further controls should be considered?
•    The methodology section is fine but the figures need improvement.

5. Please describe how the conclusions are or are not consistent with the
evidence and arguments presented. Please also indicate if all main questions
posed were addressed and by which specific experiments.
•    Conclusions are consistent with the results presented in the manuscript. Yes, all the questions posed were addressed by specific experiments.

6. Are the references appropriate?
•    Yes, the references are appropriate.

7. Please include any additional comments on the tables and figures and
quality of the data.
•    Data should represent the standard error bars in all the figures. The axis should be labelled accurately.
•    For Figure 5, the label on Y- axis is not readable. Full forms of the treatments should be provided.
•    Line 580 should be rephrased.
•    The conclusion should provide future directions of the study.

Author Response

The manuscript is well-structured and informative.

  • All authors would like to sincerely thank the reviewer for the very positive.


Here is the detail about the questions raised.
1.    What is the main question addressed by the research?
•    The research is about the use of remote sensing, MLR models, machine learning and climatic data for prediction of crop yield prediction in the Chi Basin, Thailand.

  • The authors sincerely appreciate the reviewer’s comment questions. In response to the reviewer’s feedback, the authors have revised a sentence to make its precise. The following revised in this sentence:
    • In this study, we propose a novel approach to predict crop production at the end of the crop season by integrating remote sensing and climatic indicators. This approach addresses the issue of outdated information on crop production in Thailand, which is typically 2-3 years old and provided by the government sector. The timely availability of this information is crucial for policy makers to assess and implement policies effectively. By using remote sensing and climatic indicators, this study aims to provide up-to-date information to policy makers, helping them make timely decisions to increase crop production in the upcoming year.

 

 

 

  1. What parts do you consider original or relevant for the field?
  • Thank you for pointing this out. The authors have carefully revised and added a descriptive explanation of field observation. Corrections are highlighted in yellow color. The following added in the line 222-225:
    • Data acquisition involves field observations divided into 24 areas, covering the entire Thailand region. The method includes creating a square box for each sample, followed by rice milling to estimate rice production and convert it into units (ton/ha).
    • This information of field observation is reliable to be used. In addition, the results of prediction crop yield is highly outstanding among other research studies. It is evidence that field observation information and predicting model are consistent to each other.

What specific gap in the field does the paper address? Prediction of factors contributing to the high yield in crops is the most relevant part of this research.

  • Thank you for pointing this out.
    • As the authors mentioned that many studies applied not remote sensing but also climatic indicators. In this study, the authors are understanding the crop phenology which is influenced to crop production. We would like to apply from pixel scale to provincial scale of phenological period in order to apply remote sensing data to predict crop yield. In addition, this study filled the gap by utilizing crop drought indices to reduce the error of prediction. Therefore, this result from this study illustrated less error of prediction.

  1. What does it add to the subject area compared with other published
    material? The study model used various variables that other studies do not consider.
  • The authors sincerely appreciate the reviewer’s comment questions. In response to the reviewer’s feedback, the authors have a descriptive explanation similarly to the previous comment.
    • As the authors mentioned that many studies applied not remote sensing but also climatic indicators. In this study, the authors are understanding the crop phenology which is influenced to crop production. We would like to apply from pixel scale to provincial scale of phenological period in order to apply remote sensing data to predict crop yield. In addition, this study filled the gap by utilizing crop drought indices to reduce the error of prediction. Therefore, this result from this study illustrated less error of prediction.
  1. What specific improvements should the authors consider regarding the methodology?
  • The authors sincerely appreciate the reviewer’s comment questions. In response to the reviewer’s feedback, the authors have a descriptive explanation as followed.
    • In this study, the methodology is well defined to predict crop yield at provincial scale. This method considers the time period of crop seasonal which is the main impact of crop production to climate conditions. This method applied multiple aspects of critical models such as Linear, RF, XGBoost and SVR. Each model has different characteristic such as linear is simple to apply but it provided less relevant output. On the other hand, RF can make decision when the last layer to provided better result. XGBoost, this model is superior of RF that can learn from remaining error and make a good prediction every time till it reaches the stability. SVR can find a hyperplane in a high-dimensional space that best fits the data points while also minimizing the error. Therefore, the authors would like to make comparison which one is reliability.

What further controls should be considered?
•    The methodology section is fine but the figures need improvement.

  • The authors sincerely appreciate the reviewer’s comment. In response to the reviewer’s feedback, the authors have revised as shown in the Figure 3.

 

 

 

 

  1. Please describe how the conclusions are or are not consistent with the
    evidence and arguments presented. Please also indicate if all main questions
    posed were addressed and by which specific experiments.
    •    Conclusions are consistent with the results presented in the manuscript. Yes, all the questions posed were addressed by specific experiments.
  • All authors would like to sincerely thank the reviewer for the very positive, important, and constructive comments and suggestions.

  1. Are the references appropriate?
    •    Yes, the references are appropriate.
  • All authors would like to sincerely thank the reviewer for the very positive, important, and constructive comments and suggestions.

  1. Please include any additional comments on the tables and figures and
    quality of the data.
    •    Data should represent the standard error bars in all the figures. The axis should be labelled accurately.
  • Thank you for pointing this out. The authors disagree to put all standard error bars in all figures. In this study, the authors provided the assessment metric such as RMSE.
  • RMSE refers to crop prediction error. Therefore, the authors would not like to add standard error bars in this case.
  • For Figure 5, the label on Y- axis is not readable. Full forms of the treatments should be provided.
  • The authors sincerely appreciate the reviewer’s comment. In response to the reviewer’s feedback, the authors have revised as shown in the Figure 5.

 

 

  • Line 580 should be rephrased.
  • Thank you for pointing this out. The authors have carefully reviewed your comment.
    • In this line is automatically generated by publisher side. The authors cannot revise this comment.
    • The conclusion should provide future directions of the study.
  • Thank you for pointing this out. The authors have carefully revised and added a descriptive explanation of field observation. Corrections are highlighted in yellow color. The following added in the line 573-574:
  • This method also provided timely data that can be used for decision making during the crop growth season. The discoveries of the proposed study may also be exploited to plot crop yields and its gaps at the provincial level in Thailand and neighboring countries, helping farmers and policymakers make informed decisions. However, land use change is a major concern for crop production prediction, and it will be included in subsequent studies to improve predictive assessment.

Reviewer 3 Report

Comments and Suggestions for Authors

Predictions of crop production in the Chi basin are of major importance for decision support tools in countries such as Thailand, which aim to increase domestic income and global food security by implementing the appropriate policies. The research aims to establish a predictive model for predicting crop production for an internal crop growth season prior of harvest at the province scale for fourteen provinces in Thailand's Chi basin between 2011 and 2019. Authors provide approaches for reducing redundant variables and multicollinearity in remotely sensed and meteorological data to avoid overfitting models using correlation analysis and variance inflation factor. Temperature condition index (TCI), normalized difference vegetation index (NDVI), land surface temperature (LSTnight), and mean temperature (Tmean) were the resulting variables in the prediction model with a p-value < 0.05 and a VIF < 5. Baseline data (2011–2017: June to November) were used to train four regression models: eXtreme Gradient Boosting (XGBoost), random forest (RF), and XGBoost achieved R2 values of 0.95, 0.94, and 0.93.  All 3 significantly outperformedmultiple linear regression.

The testing dataset (2018–2019) displayed a minimum root mean square error (RMSE) of 0.18 ton/ha applying the XGBoost model. Accordingly, it is estimated that between 2020 and 2022, the total crop production in the Chi basin region would be 7.88, 7.64, and 7.72 million tons, respectively. The proposed model is proficient at improving crop yield prediction accuracy when compared to a conventional regression method and that it may be deployed in different regions to develop more informed decisions about agricultural practices and resource allocation.

The work is clear and well-founded, and the results support the conclusions. I just suggest a few corrections.

2.4. Features Selection: Correlation Analysis (CA) and Variance Inflation Factor

Why were these particular methods chosen? What others exist for this type of data?

 

What difference is there between them?

 

215 2.5. Regression Model 

A brief description of models would be valuable.

 

Comments on the Quality of English Language

80 Recently, a data-driven remote sensing approach has become efficient to measure in measuring crop conditions and predicting crop yield production  from a distance without being physically present in the study area.

 

210 0.05 and a VIF < 5 [45] (Figure 3Error! Reference source not found.). The selected variables in this step will...

Author Response

Comments and Suggestions for Authors

Predictions of crop production in the Chi basin are of major importance for decision support tools in countries such as Thailand, which aim to increase domestic income and global food security by implementing the appropriate policies. The research aims to establish a predictive model for predicting crop production for an internal crop growth season prior of harvest at the province scale for fourteen provinces in Thailand's Chi basin between 2011 and 2019. Authors provide approaches for reducing redundant variables and multicollinearity in remotely sensed and meteorological data to avoid overfitting models using correlation analysis and variance inflation factor. Temperature condition index (TCI), normalized difference vegetation index (NDVI), land surface temperature (LSTnight), and mean temperature (Tmean) were the resulting variables in the prediction model with a p-value < 0.05 and a VIF < 5. Baseline data (2011–2017: June to November) were used to train four regression models: eXtreme Gradient Boosting (XGBoost), random forest (RF), and XGBoost achieved R2 values of 0.95, 0.94, and 0.93.  All 3 significantly outperformed multiple linear regression.

The testing dataset (2018–2019) displayed a minimum root mean square error (RMSE) of 0.18 ton/ha applying the XGBoost model. Accordingly, it is estimated that between 2020 and 2022, the total crop production in the Chi basin region would be 7.88, 7.64, and 7.72 million tons, respectively. The proposed model is proficient at improving crop yield prediction accuracy when compared to a conventional regression method and that it may be deployed in different regions to develop more informed decisions about agricultural practices and resource allocation.

The work is clear and well-founded, and the results support the conclusions. I just suggest a few corrections.

  • All authors would like to sincerely thank the reviewer for the very positive, important, and constructive comments and suggestions.

 

 

 

 

 

2.4. Features Selection: Correlation Analysis (CA) and Variance Inflation Factor

Why were these particular methods chosen? What others exist for this type of data?

 What difference is there between them?

  • Thank you for pointing this out. The authors have carefully revised and added a descriptive explanation of field observation. Corrections are highlighted in yellow color. The following added in the line 277-283:
    • They are various factors that can apply to reduce the overfitting results such as removing one of the correlated variables, combing correlated variables, principal component analysis (PCA). In this case, the variable indicators to predict crop yield are limited indicators. However, removing or combining correlated variables are not consistent with this procedure. Likewise, PCA is another approach that can transform multiple correlated variables into one variable to be used as predictor in the model [41]. However, the limitation of this study is a number of variables. Thus, the appropriate approach is necessary to apply.

 

2.5. Regression Model 

A brief description of models would be valuable.

  • Thank you for pointing this out. The authors have carefully revised and added a descriptive explanation of field observation. Corrections are highlighted in yellow color. The following added in the line 326-331:
    • Regression models are statistical methods used to investigate the relationship between one or more independent variables and a dependent variable. Regression models aim to estimate the effect of the independent variables on the dependent variable. These models assume a functional form for this relationship, such as linear or nonlinear. The model parameters are estimated using statistical techniques, and the model's goodness of fit is assessed using various metrics, such as R2 and RMSE.

 

Comments on the Quality of English Language

80 Recently, a data-driven remote sensing approach has become efficient to measure in measuring crop conditions and predicting crop yield production from a distance without being physically present in the study area.

  • Thank you for pointing this out. The authors have carefully revised this comment. The following added in the line 92:

210 0.05 and a VIF < 5 [45] (Figure 3Error! Reference source not found.). The selected variables in this step will...

  • Thank you for pointing this out. The authors have carefully revised this comment.

Reviewer 4 Report

Comments and Suggestions for Authors

Dear Authors, I really appreciate your research and it is very interesting and valuable. In my opinion, your work is written very carefully and you presented the results and conclusions in a comprehensive manner. I have no objections to the manuscript as written and believe it can be published in its current form

Author Response

  • All authors would like to sincerely thank the reviewer for the very positive.

Once again, the authors appreciate the reviewer for their insightful comments to improve the paper quality. We hope our responses are written well enough to address the reviewer’s comments and suggestions.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

In view of the author's reply and the much improved text of the manuscript, I suggest publishing the paper in your journal.

Back to TopTop