Extraction of Cotton Information with Optimized Phenology-Based Features from Sentinel-2 Images

Round 1

Reviewer 1 Report

The proposed topic is interesting in terms of the importance of cotton cultivation in China's economic structure. The consideration of monitoring the growth processes of certain crops, especially those that are key to current food security, are strategic on a global scale, but also, and perhaps more importantly, on a local scale, as cited by the authors, referring to the consequences of land use following the work of Jonathan A. Foley et al.

Certainly, the growth and maturation process of any crop takes place in different phases, and remote sensing from satellite imagery is more accurate at regional rather than local scales, as the authors point out. From this point of view, they seek and use methodologies already applied for these processes, but in an orderly and well-structured method.

The use of different indexes in each phenological stage indicates a search for the best possible data for each of them, therefore, the combination of indexes and their justification is an important contribution to the work.

It is a rigorous work, very well explained, that can be applied as a model for other works, due to the clear sequencing of the processes. It includes multiple indexes adapted for this crop, valuing and explaining the reason for this decision.

In the graphic and cartographic section, the maps, diagrams and diagrams are timely and help to understand the spatialization of the problem.

Author Response

Thank you for your positive comments, which helped us to rationalize the article more clearly. In this round of revisions, we have carefully revised the text and reorganized the layout of figures in this revised manuscript to make our method more understandable and our results more convincing. Thank you again for your affirmation!

The main revised and added parts are highlighted in red palatino linotype(10) in the manuscript.

Reviewer 2 Report

Comments:

This manuscript is interesting with a time-series multiple phenological feature optimization method (Tmpf-OM) proposed to improve the extraction of cotton fields. There are some problems mainly on the organization and writing. A minor revision is needed before acceptance for publication.

L21-L22: change “were served” to “served”

L25: what did the words “strictly selected” mean?

L34: change “better” to “greater”

L38: “RMSE no more than 14.62 Kha” does not tell readers anything. Please use relative RMSE.

L42: change “agriculture” to “crops”

L56: What does “personal errors” mean?

L63 and elsewhere: please check the guidelines on how to cite an article with three and more authors.

L65: extract cotton area?

L71: change “same” to “similar”

L76, L175, L178, and elsewhere: define all abbreviations you used first time in the manuscript.

L104: “some accuracy” is not clear.

L110: “fewer and more” is not clear.

L124 – L125: change “. (2)” to “; (2)”; “. (3)” to “; and (3)”

L132: change “2. Research data” to “2. Study area and data”

L174: This section title is not correct.

L175: change “CLCD” to “China land cover dataset (CLCD)”

L180: This section title is not correct.

L182: “Accuracy for Landsat 8 when judged against errors of commission were 99.7%.” is not clear.

L208: It should be “2.4.2”

L209: Cheng “are” to “was”

L216: “Each crops”?

Figure 3: Some of the abbreviations used in this figure were not defined.

L229: should be “mosaiced”.

L240: ” Rare land classes in this paper are water bodies, impervious surfaces, and forests.” May be changed to “There are small portions of this area being water bodies, impervious surfaces, and forests.”

L256 – l268: You should define all abbreviations when they were first time used.

L269: please clearly tell readers how did you get 2307 pure cotton pixels. Did you select them from 6777 cotton samples? How?

L330: SATELLITE INDEX?

Table 1: You should show and define the indices earlier and biophysically tell readers general reasons for selecting them.  Then, in this section you could specify the reasons of selecting them. Your organization and writing in the current way make readers hard to read and understand your article.

 L345 – L349: please correct all the subscripts.

L370: You defined RF at several places.

In section 4.1: I could not find where you defined the variables such as LSWI_3, LSWI_4, EVI_sin3, NDVI_sin3, etc.

L428: “Figure 5 displays all features (except for 428 terrain features) distribution of cotton and non-cotton…” may be changed to “Figure 5 displays the feature separability of cotton and non-cotton (except for terrain features).

L455: see above.

In section 4.2: A test of statistically significant differences of the accuracies from the scenarios may be needed to convince readers.

L540: Using the words such as “good”, “better” and “best” is not good idea.

L597: MDI?

L639: Sentinel-1?

Author Response

Please check the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

This article ‘Extraction of cotton information with optimized phenology- 2 based features from Sentinel-2 images’ developed a time-series multiple phenological feature optimization method by integrating the VIs during five typical phenological stages, time-series dynamic change features and terrain features. However, there are still many problems that are not clearly described. I would recommend to reconsider it for publication after another round of review. 

Major Comments:

1.      The subject of this article is not very clear. If the subject of the article is “developed a time-series multiple phenological feature optimization method (Tmpf-OM) to improve the identification accuracy of cotton fields”, then the comparison of the method with the current traditional method should be added; If the main purpose of the article is to construct more reasonable characteristics, then it is necessary to explain which features perform better in the end, and explore the reasons behind their excellent performance.

2.      L291: In Section "3.1.1. Five representative phenological stages features", there is a problem with the narrative. It should be detailed and focused. It is suggested that the writing of chapters can be rearranged according to the structure of the total score, and after a general introduction to the five typical phenological stages, the selection of indices can be introduced one by one. For the amount of images selected for the final time series analysis, it is recommended to reduce the description of this section considering that it is not the focus of the article.

3.      L414-417: “It should be noted that NDVI and EVI can only detect three phenological stages of cotton with poor stability…”. This sentence is more like an explanation of the poor effectiveness of using only FC1 feature classification, and it may be more appropriate to move that description into the analysis.

4.      The analysis of the entire results section was weak and did not highlight the focus of the article. It is suggested to first add a discussion of the differentiation of cotton from other feature types by different characteristics; Secondly, for the analysis of the classification characteristics of the north and south regions in Fig5-8, it is recommended to consider reorganizing the chart combination to achieve a comparative analysis of the classification characteristics of the north and south regions.

5.      L513: In this accuracy evaluation section, it will be better if considering adding accuracy comparisons with other traditional methods.

6.      The discussion part id partially redundant and the focus is not clear. It will be better if add the analysis of the reason why the north and south regions get different optimal features. 

7.       L582-592: According to the current article chapter setting, the chapter title is the advantages of the method, and it is more reasonable to move this section to section 4.2.    

-----

1. L145-148: The elements appearing in Fig 1 should be clearly marked with their meanings, and at the same time, considering that the text mentions that the entire study is divided into north and south parts, a clear display or textual description of the results of the definition of the north and south regions should be given in the figure.
2.  L333: “FLS&BDS” recommended to modify to ” FLS, BDS” ;

3.      L345-349: “ai,bi1…”  Pay attention to the format;

 

4.      L438: The meaning of ordinates in Fig 5 should be indicated, same problem in Fig 7.  

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

all the problem  I concerned is revised correctly.