Integrating Remote Sensing, Proximal Sensing, and Probabilistic Modeling to Support Agricultural Project Planning and Decision-Making for Waterlogged Fields
Round 1
Reviewer 1 Report
Your article was well prepared and presented. There are still some minor comments for your reference.
1. Fig.2. line 198, ‘NDIV’ should be ‘NDVI’
2. Fig.3(d) a legend was missing and so it is not clear which one is actual and which one is estimated.
3. Line 338, there is some duplicated number and English words.
4. In conclusion part, these sentences should be rewritten. They are at present descriptions and not concrete conclusions. Conclusion should be some major innovative points after the study that you found and pass on to the readers
Author Response
Response letter to the reviewers’ comments for the water manuscript
water-2223795
Dear Water Editors, Dear Reviewers,
We would like to thank you for the time taken to evaluate our manuscript (entitled “Integrating remote sensing data, proximal soil sensing, and probabilistic modeling to support agricultural project planning and decision-making for waterlogged fields” with reference number “water-2223795”). We are pleased that the editor and reviewers assessed our work positively and recognized its potential and its significance. The comments and suggestions provided by the two reviewers helped greatly to improve our manuscript and we would like to thank you all for the constructive and valuable insights. We have addressed all comments and suggestions as detailed below. In particular, the reviewers offered valuable suggestions on how to extend the discussion, improve the conclusions section, and describe the reasons for each statistical method as well as the formatting of the manuscript—which we have now addressed.
Please find below a point-by-point response to all the reviewers’ comments and how we addressed them. We have prepared the revised document and also a track-change version to facilitate the review process of the implemented changes. Reviewers’ comments are in italics, and our responses are listed always directly afterward. Suggested text that we have added in the revised manuscript is stated between “ ” and new or modified text is underlined.
Note that, the lines provided by the reviewers refer to the original version of the manuscript while the lines and references given in our responses refer to the revised manuscript.
We hope the editor and reviewers find our responses and changes to the manuscript satisfying and we once again thank you for the thoughtful input.
Yours sincerely,
The authors
Reviewer 1
Reviewer 1 comment 1: Your article was well prepared and presented. There are still some minor comments for your reference.
Our response: We thank the reviewer for the overall positive evaluation, comments, and suggestions. We have addressed those comments as detailed below.
Reviewer 1 comment 2: Fig.2. line 198, ‘NDIV’ should be ‘NDVI’
Our response: We thank the reviewer for the note. We have addressed this comment see line 347 in the revised version
Reviewer 1 comment 3: Fig.3 (d) a legend was missing and so it is not clear which one is actual and which one is estimated.
Our response: We thank the reviewer for this comment. We have revised this figure and presented the new version in the revised manuscript. Note that to avoid duplication of the legend across all panels, and to avoid redundancy in the figure legend, we kept the legend on the first panel only (see the revised figure below)
Reviewer 1 comment 4:Line 338, there is some duplicated number and English words
Our response: We thank the reviewer for this note. We have removed digit and kept only words (see revised line below). We hope this is clear to the reader.
These results suggest that an increase in one unit cropping area (1 ha)
Reviewer 1 comment 4: In conclusion part, these sentences should be rewritten. They are at present descriptions and not concrete conclusions. Conclusion should be some major innovative points after the study that you found and pass on to the readers
Our response: We thank the reviewer for this comment. Indeed we agree with the reviewer. We have rephrased the conclusions section to our best and we hope that it is satisfactory to the reviewer. As shown in the track-changed version, the following sentences have been added to the conclusion section of the manuscript. Note that, we rephrased the conclusion section without changing the initial statements that were drawn from the result.
“The results of this study suggest strong associations between vegetation indices, water indices, ECa, and susceptibility to waterlogging. This adds substantial and understudied complexity that needs to be unraveled to better understand the relationship between hydro-geochemical features and waterlogging in agroecosystems. Furthermore, the study highlights the potential data driven methods to assess and plan agricultural projects. This is essential, especially in agriculture sector where project implementation is often associated with risk and uncertainty…”
Author Response File: 
Author Response.pdf
Reviewer 2 Report
As someone who is passionate about agriculture and was raised in a farming family, I appreciate the innovative approach to a current problem, namely identifying potential sites susceptible to waterlogging on a farm scale and testing the benefits of drainage solutions.
Some details related to the appearance of the manuscript. I would recommend the authors to look carefully at the template for Microsoft Word of this journal, on this webpage https://www.mdpi.com/journal/water/instructions
I recommend that the figures (especially figure 1) should not exceed the edges of the page, and that the titles, subtitles, and equations be formatted and aligned uniformly. (I learned as an MDPI author that the equations in the MDPI templates are inserted into a table with hidden borders; this is an advice I share).
I recommend the authors to re-read the manuscript once more, with more attention to the aspects of language and phrasing (for example line 239-240 "in python. visualization..." etc.)
I would appreciate if the authors would elaborate more on the technologies used. In section 1.2, the authors explain quite well why they use Bayesian statistics models in the experiment, based on other studies. On the other hand, about the clustering part, the use of the classic k-means algorithm is not mentioned as much (about this, I recommend the authors to decide what to call it in the text: K-means or Kmeans...)
I think it would be helpful if the authors provided some justification for using k-means rather than, say, other alternatives like perhaps Mean-Shift Clustering, DBSCAN, K-Medians, etc. In addition, I think the readers of the journal could benefit from authors’ expanding upon their discussion of how they used Elbow technique, which calculates the sum of squares error or the cluster inertia to find out the goodness of the split, or the Silhouette analysis, for calculating the Silhouette Coefficient. Other posibility to determine the value of k (the optimal number of clusters) could have been the Gaussian Mixture model, or perhaps another interesting method to be used could have been the Davies-Bouldin Index, etc.
I suggest that there should be a stronger emphasis in the manuscript on what the authors contribute new in addition to the most recent studies and publications in the subject by other researchers (perhaps even in a new subsection called Discussions).
It would be also interesting to read in this manuscript the authors' plans for continuing the research, perhaps by enhancing what they have accomplished thus far or introducing something entirely new (a little bit more than a single sentence at the end of the Conclusions section).
Author Response
Reviewer 2
Reviewer 2 comment 1: As someone who is passionate about agriculture and was raised in a farming family, I appreciate the innovative approach to a current problem, namely identifying potential sites susceptible to waterlogging on a farm scale and testing the benefits of drainage solutions.
Our response: We thank the reviewer for the overall positive evaluation of the manuscript.
Reviewer 2 comment 2:Some details related to the appearance of the manuscript. I would recommend the authors to look carefully at the template for Microsoft Word of this journal, on this webpage https://www.mdpi.com/journal/water/instructions
Our response: We thank the reviewer for this note and recommendation. We have checked the template once again and adapted the formatting of the manuscript accordingly. See the revised version of the manuscript.
Reviewer 2 comment 3: I recommend that the figures (especially figure 1) should not exceed the edges of the page, and that the titles, subtitles, and equations be formatted and aligned uniformly. (I learned as an MDPI author that the equations in the MDPI templates are inserted into a table with hidden borders; this is an advice I share).
Our response: We thank the reviewer for the comments and suggestions. We agree with the reviewer that the early version of the manuscript had some formatting issues. We have addressed those issues in the revised manuscript. See for example figures, equations, and text throughout the manuscript.
Reviewer 2 comment 3: I recommend the authors to re-read the manuscript once more, with more attention to the aspects of language and phrasing (for example line 239-240 "in python. visualization..." etc.).
Our response: We thank the reviewer for the note. We have cross-checked the manuscript and addressed related issues as suggested by the reviewer.
Reviewer 2 comment 4: I would appreciate if the authors would elaborate more on the technologies used. In section 1.2, the authors explain quite well why they use Bayesian statistics models in the experiment, based on other studies. On the other hand, about the clustering part, the use of the classic k-means algorithm is not mentioned as much (about this, I recommend the authors to decide what to call it in the text: K-means or Kmeans...). I think it would be helpful if the authors provided some justification for using k-means rather than, say, other alternatives like perhaps Mean-Shift Clustering, DBSCAN, K-Medians, etc.
Our response: We thank the reviewer for these comments and suggestions. We agree with the reviewer, and these are indeed critical questions. In the revised version of the manuscript, we have extended the Statistical analysis section and added a paragraph describing the K-means algorithm used in this study (please see the paragraph below, in track changes and revised manuscript). We have harmonized the terminology and kept “K-means” instead of “Kmeans” throughout the manuscript. As the aim of this article was to test simple but straightforward methods that have been shown to work in other research fields, we were unable to compare multiple statistical learning techniques as suggested by the reviewer. Instead, we elaborated on the choice of each method applied in this paper and how it is suitable for our case study. We also acknowledge that the comparison of multiple evaluation methods for unsupervised models is indeed an interesting and large topic that deserves its section in a manuscript. However, this was beyond the scope of this manuscript. Furthermore, we have extended the analysis section and added our reasoning about the choice of the method with a focus on the K-means algorithm (see the paragraph below and lines 221-238 in the revised version of the manuscript).
“To identify potential sites for waterlogging in the investigated field, we used an unsupervised learning technique “the K-means clustering algorithm”. K-means clustering is the partitioning algorithm that assigns each data point in the dataset to only one of the adjacent clusters using a measure of distance or similarity. K-means clustering has been identified as a simple, intuitive, and elegant approach for partitioning a data set into K-distinct, non-overlapping clusters. To perform K-means clustering, the user first specifies the desired number of clusters K, then the K-means algorithm will assign each observation to exactly one of the K clusters (James et al., 2013). To simplify the learning and computation time, in this study, we set the desired number of clusters to 2, because in our case study scenario, we expected each part of the field to fall into one of the following categories: Potential waterlogged sites on well-drained sites. There are many unsupervised learning algorithms in the recent literature. However, We chose the K-means clustering for the following reasons: First, we chose the K-means clustering because we were interested in minimizing the within-cluster variance. Second, before conducting the analysis, our environmental covariates (i.e. predictors) were normalized making the K-means algorithm suitable for this type of data. Finally, as we had predefined the number of clusters suitable for our problem, here the K-means clustering algorithm was also an ideal choice.”
References
James, G.; Witten, D.; Hastie, T.; Tibshirani, R. Unsupervised Learning. In An Introduction to Statistical Learning: with Applications in R; James, G., Witten, D., Hastie, T., Tibshirani, R., Eds.; Springer Texts in Statistics; Springer: New York, NY, 2013; pp. 373–418 ISBN 978-1-4614-7138-7.
Reviewer 2 comment 5: In addition, I think the readers of the journal could benefit from authors’ expanding upon their discussion of how they used Elbow technique, which calculates the sum of squares error or the cluster inertia to find out the goodness of the split, or the Silhouette analysis, for calculating the Silhouette Coefficient. Other posibility to determine the value of k (the optimal number of clusters) could have been the Gaussian Mixture model, or perhaps another interesting method to be used could have been the Davies-Bouldin Index, etc.
Our response: We thank the reviewer for this comment and suggestions. We agree with the reviewer that this deserves a further explanation in the analysis section of the manuscript. This comment and the associated answer are related to Reviewer 2 comment 4. Part of the answer is also applicable to this comment. Furthermore, we have extended the analysis section and added further details on the methods used to evaluate the model in our manuscript (see the paragraph below and lines 239-251 in the revised version).
“To evaluate the model and select the number of optimal clusters that fit our data points, we used the silhouette and Elbow methods (Thorndike, 1953; Rousseeuw, 1987; Goutte et al., 1999). For each technique, we run 11 iterations with each iteration representing the number of clusters. For the first evaluation method, an average silhouette coefficient based on 11 iterations was calculated to identify the optimal number of clusters representing our data points. The entire clustering was then displayed by combining the silhouettes into a single plot, allowing an assessment of the relative quality of the clusters and an overview of the data configuration. There are many evaluation techniques for validating clustering algorithms. Since both K-means and silhouette analyses are based on Euclidian distance metric, we decided to use the silhouette coefficient for model evaluation, to simplify and harmonize the entire analysis. In the second phase, the elbow method was used. The method consists of plotting the explained variation as a function of the number of clusters and picking the elbow of the curve as the number of clusters to use. We chose the elbow method because it is widely used in other unsupervised analyses such as principal component Analysis. The results of the two techniques (Fig. A2) provide a robust approach for evaluating the output of our cluster analysis”
Reference
Thorndike, R.L. Who Belongs in the Family? Psychometrika 1953, 18, 267–276, doi:10.1007/BF02289263.
Rousseeuw, P.J. Silhouettes: A Graphical Aid to the Interpretation and Validation of Cluster Analysis. Journal of Computational and Applied Mathematics 1987, 20, 53–65, doi:10.1016/0377-0427(87)90125-7.
Goutte, C.; Hansen, L.K.; Liptrot, M.G.; Rostrup, E. Feature-Space Clustering for FMRI Meta-Analysis. Hum Brain Mapp 2001, 13, 165–183, doi:10.1002/hbm.1031.
Reviewer 2 comment 6: I suggest that there should be a stronger emphasis in the manuscript on what the authors contribute new in addition to the most recent studies and publications in the subject by other researchers (perhaps even in a new subsection called Discussions).It would be also interesting to read in this manuscript the authors' plans for continuing the research, perhaps by enhancing what they have accomplished thus far or introducing something entirely new (a little bit more than a single sentence at the end of the Conclusions section).
Our response: We thank the reviewer for this comment. We have extended our discussion and conclusions sections (see the revised version in lines 433-452).
“In summary, agriculture scientists and resource managers are experiencing advances in agricultural technologies often related to precision farming. However, each individual tool is used in an isolated manner without a clear connection. For example, in agriculture, especially water management, research often focuses on monitoring with limited research on management and decision-making. This study attempted to connect different approaches to support decision-making in the agriculture sector. From monitoring to planning to decision-making. This study revealed that the combination of remotely sensed data, Bayesian modeling, and A/B test present promising area for future research in agriculture resource management. Combining these tools can provide a deep understanding of the impact of waterlogging on farm production and productivity as well as on the cost-benefit ratio. As freely available data, data collection, and analytical tools emerge, there is great potential for these methods to significantly contribute to sustainable water and agriculture management. Furthermore, this study opens up questions that are worthy of further investigation. (1) it is not clear how different factors interact to influence decision-making specifically for water management, but also in agriculture resource management. Here a combination of Bayesian network and causal inference can provide further insights. (2) There is a need to integrate multiple data sources such as social, environmental, and economic aspects of farmers, long-term datasets, and quantifying uncertainties associated with each factor, which can improve the accuracy and reliability of predictions related to waterlogging management and other agricultural challenges.”
Author Response File: Author Response.pdf
