Scattering Intensity Analysis and Classification of Two Types of Rice Based on Multi-Temporal and Multi-Mode Simulated Compact Polarimetric SAR Data
Round 1
Reviewer 1 Report
I appreciate authors for incorporating and modifying all my concerns in this revised version. I don't have any further comments.
Author Response
Dear reviewer
Thank you again for your kind consideration and valuable suggestions.
Reviewer 2 Report
Thanks to the author to provide the new version and cover letter. They answer my concern one by one and modified the manuscript. I have no more questions just may need to check the English error and the quality of images 10 and 13 is not good.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 3 Report
The authors have addressed most of my questions, except for the following two.
- “Rice mapping using SAR data is not a new topic and has been done by numerous studies, including those based on CP SAR. Although the authors introduced the deficiencies of previous study, I did not see a considerable improvement made by this study because of the lack of comparison between different approaches. For example, if you believe CP parameters of multiple transmitting modes perform better than CP parameters of one transmitting mode (lines 90-91), a comparison is necessary to demonstrate such improvement”.
As a response to this comment, the authors made a comparison between their study with previous studies. The comparison shows that the approach proposed by the authors produced the highest accuracy. Nevertheless, the comparison is unconvincing because these studies were carried out in different areas and the validation samples were totally different. If you want to demonstrate that CP SAR performs better than FP SAR for rice classification, you can make a comparison between CP SAR and FP SAR using your own data instead of comparing your results with those done several years before.
- “The authors done a lot of work to find optimal parameters and construct a decision tree model for rice mapping. However, all this work seems to be based on the visual interpretation of statistical data. In fact, those work could be easily done using machine learning techniques, such as decision tree, random forests, based on training data. Compared with visual interpretation, the results obtained by those methods should be more accurate, objective, and convincing.”
Regarding this question, the authors insisted that the manually created tree produced a good classification. I do not have a problem with this statement. What I mean is that does your tree can produce better results than a tree created by a decision-tree algorithm based on training data?
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 4 Report
The authors have improved the manuscript significantly and have answered all the concerns of reviewer. I support its publication.
Author Response
Dear reviewer
Thank you again for your kind consideration and valuable suggestions.
Reviewer 5 Report
I think author(s) could not understand review comment-02. Compact polarimetric modes namely π/4 (Souyris et al. 2005) and CTLR (Raney 2007) are well described in the existing literature. Souyris et al. 2005 has defined the π/4 CP mode with transmission at 450degree (π/4) and reception in linear horizontal (H) and vertical (V) polarization and associated physical interpretation of the targets while using this mode. However, in the current study, authors have referred CP modes such as “Left circular transmission and π/4 reception”, “Right circular transmission and π/4 reception” etc.. (Line 152-155 and Figure-03). My question is: How do authors arrive at these CP modes? How one can interpret these polarization interactions with the targets? What kind of symmetry assumptions followed in these modes. Please explain in detail.
Figure-01: Keeping field photo of T-H and D-J rice in figure-01 do not add any value until they point out any specific locations. Similarly in figure-03.
What is Shoal feature class, please describe it’s physical characteristics in detail. Why is it mixing with the rice classes in initial decision tree nodes?
Figure-02 does not show any peculiar differentiation between the two rice types. Explain.
The quality of Figure-03 is poor. Increase the font size of the text.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 3 Report
The authors have addressed all my concerns.
Reviewer 5 Report
Thanks for the clarifications on transmitting and receiving polarization formulation. I am fine with the satisfactory responses to other comments as well.
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
Please find the attachment.
Comments for author File: 
Comments.pdf
Reviewer 2 Report
The topic of the manuscript is important and scientific soundness remote sensing application in agriculture and interesting. After reading the comments and questions are as follow:
- Introduction not clearly mentioned about research goal and what is the main input of this research?
- Fig. 1 quality is not good such a blurry.
- Why do you use the 2015 field sample and writing after about 7 years?
- Does two types of rice in the same phenological and plantation calendar?
- How did you decide on 6 CP SAR?
- Why did you choose the decision tree not other than a random forest to may get a better classification?
- How to decide about the threshold of backscattering in divided classes in DT? seems only two types of rice are very similar in phenology.
- Did you take any samples in irrigation time? which field is covered by water? If so it becomes mixed by a permanent water body how did you separate them?
- How many samples were used for classification and accuracy assessment? did you use field sample data from each field date for a kappa coefficient?
- Your accuracy assessment seems a little doubtable for the upper 90%.
- Why TH get higher accuracy than DJ?
- Was the test site includes only 4 LC classes? not road, vegetation, irrigation canal etc?
- Please present a bit more about field data, phenology and LC types.
Reviewer 3 Report
Lines 16-18, “it is necessary to explore the feature recognition and classification ability of CP parameters under different transmitting and receiving modes to different ground objects”. I do not think this research gap can be addressed by conducting a study on rice classification. In other words, the research gap and this study poorly matched. This mismatch makes the significance and contribution of this study questioned.
Rice mapping using SAR data is not a new topic and has been done by numerous studies, including those based on CP SAR. Although the authors introduced the deficiencies of previous study, I did not see a considerable improvement made by this study because of the lack of comparison between different approaches. For example, if you believe CP parameters of multiple transmitting modes perform better than CP parameters of one transmitting mode (lines 90-91), a comparison is necessary to demonstrate such improvement.
The authors done a lot of work to find optimal parameters and construct a decision tree model for rice mapping. However, all this work seems to be based on the visual interpretation of statistical data. In fact, those work could be easily done using machine learning techniques, such as decision tree, random forests, based on training data. Compared with visual interpretation, the results obtained by those methods should be more accurate, objective, and convincing.
I suggest the authors to provide detailed expatiations of the physical meaning or scattering mechanisms denoted by the CP parameters. It would be helpful for readers to understand why a specific parameter contributes to rice classification.
Lines 441, “The classification results can provide more accurate information for rice growth monitoring and yield estimation in the study area”. However, I did not see any work on yield estimation.
Please give a full name of T-H and D-J when they appear for the first time in the paper, even in the abstract.
Reviewer 4 Report
This article provides a multi-temporal and multi-mode simulated compact polarimetric SAR-based methodology for the scattering intensity analysis and classification of two types of rice.
Considerations
Authors must follow the specifications of the Microsoft Word template or LaTeX template to prepare their manuscript. Abstract: A single paragraph of about 200 words maximum.
In the introduction, highlights are missing to attracting the attention of the reader for novelty and relevance of the proposed contribution.
Conclusions could explore experimental results in a more consistent approach and give an assertive response to the proposed objectives.
Reviewer 5 Report
Please find the attached file.
Comments for author File: Comments.pdf
