A New Method for Crop Type Mapping at the Regional Scale Using Multi-Source and Multi-Temporal Sentinel Imagery
Round 1
Reviewer 1 Report (Previous Reviewer 2)
The introduction and the method sections are improved, but the result and especially in the discussion section needs to rewrite. I am not convinced why their methodology has a high RMSE compared to the statistical reports (which I felt is the flaw of this paper). I have provided my edits in the attached pdf.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report (New Reviewer)
The study “A new method for crop type mapping at the regional scale using multi-source and multi-temporal Sentinel imagery” proposed a crop classification method using two different-in-nature satellite data sets, i.e., Sentinel-1 (SAR) and Sentinel-2 (optical). The study used a machine learning technique Random Forest (R.F.), for the decision system. The model’s output was validated against ground observations for various crop types over Henan, China. The proposed method shows an overall accuracy of 84.15% which is acceptable. The authors discussed all the data sets and methods in detail and effectively. The results are well presented and explained. I recommend the acceptance of the manuscript after the following minor corrections.
1. Writing of the introduction needs extensive improvement. In many places, sentences could be better structured and easier to understand.
2. First paragraph of the introduction has no rational connection between sentences. Please rewrite.
3. Please check reference 1. World Health Organization is not a person to abbreviate ash Organization W.H.
4. Figure 8 is illegible; please redraw it with clarity.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report (New Reviewer)
The manuscript titled “A new method for crop type mapping at the regional scale using multi-source and multi-temporal Sentinel imagery” proposed a crop type mapping method based on the combination of multiple single-temporal feature images and the time-series feature image derived from Sentinel-1 (SAR) and Sentinel-2 (optical) satellite imagery on the Google Earth Engine (GEE) platform. The study is significant to agriculture and food security. However, there are some issues to be discussed :1) about “the time-series feature image “, “…. Then, the 6 median feature images that cover the entire study area were aggregated to obtain the time-series feature image.” in Section “2.3.2 of this paper. Classification Based on Time-Series Feature Image”, however, the time-series feature images were aggregated by monthly median from ground reference data ,…, and Google Earth based on Figure 4. Flowchart. The two definitions of "time series feature image" are different, please explain. 2) About “Scheme 1 ,2 and 3”, how do schemes 1 and 2, respectively, correspond to optical and radar data in” Table 5. Accuracy …..”? Are schemes 1 and 2 here defined the same as described in section 2.3.4 ? 3) Some expressions need to be modified and perfected to avoid ambiguity.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report (Previous Reviewer 2)
I still felt that the authors could work on the conclusion section to improve it. Could you please work on some of the edits in the attached pdf file?
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 3 Report (New Reviewer)
The revised version of the manuscript “A new method for crop type mapping at the regional scale using multi-source and multi-temporal Sentinel imagery” has a great improvement in the content expression. However, there are some issues should be discussed: 1) whether the amount of available observations of Sentinel-1and Sentinel-2 after de-cloud processing affects the accuracy of classification ? 2) How applicable or extendable is the proposal method in this paper？It is recommended to discuss in the discussion section.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
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
I have enjoyed reading the manuscript entitled “A new method for crop type mapping at the regional scale using multi-source and multi-temporal Sentinel imagery” that was submitted to Remote Sensing. The authors propose three schemes Scheme 1：Classification based on time-series feature image. Scheme 2：Classification based on multiple single-temporal feature images. Scheme 3：Classification by voting the probability images of scheme 1 and scheme 2 336 pixel by pixel based on the CA-score. Finally, the authors found that Scheme 3 in better mapping crop in Henan Province. Since the final results are not shared, I cannot evaluate the accuracy of the classification results. From my understanding of the crop planting situation in Henan Province, the classification results are not as accurate as the author said.
Before publication, this manuscript could be further improved by taking the following proposals into consideration.
1. Please share the final classification results so that the effectiveness of the method can be evaluated according to the geographic latitude and longitude coordinates
2. The manuscript is devoted to exploring the potential of S1 and S2 fusion for crop classification. However, I think the data fusion and the RF classification methods used in the manuscript are very common, so innovations in the manuscript should be re-highlighted.
3. L61-63 Author's lack of investigation into the large number of Sentinel 1/2 based research efforts for crop mapping. In the study area of this paper and the Huaihe River Basin, a lot of research work has been carried out based on the mapping of Sentinel 1/2 and Landsat time series images
4. Line 130, As we all know, the planting system in most areas of Henan Province is double-season planting, so why did you choose to focus on the autumn harvest crops?
5. L143-146 In the Henan flatlands, due to the household contract responsibility system, fragmented planting is very serious
6. The quality of the figures in the manuscript needs to be improved.
7. Line 203, is the ground reference data sample points or a sample area? From Figure 1, the reference data seems to be sample points, but the sample area can better train and verify the classification results.
8. Figure 4 should have been presented in a more concise manner.
9. Line 237, I don't think 50 trees seem to be enough to train a classifier, can you provide how the classification results vary with the number of trees?
10. I can't see the advantage of the results from the third classification method from the comparison in Figure 6. It seems more appropriate to choose a smaller or other area.
11. Line 421, which one is Scheme 4?
12. Line 427, the discussion should be a deeper description of the research results, and it seems more appropriate to put the introduction of the classification method in the first section.
13. In this study, the authors can first classify the time-series feature images, and then select pixels with higher CA-score from the classification results (the pixels are considered to be correctly classified) as additional training samples for the classification based on multiple single-temporal SAR and optical images. Than compare with this study classified scheme
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
The introduction was easy to follow until the last paragraph of the Introduction part. There was not much discussion of why the authors chose the approach. The Results and the discussion are poorly written. I would like the authors to review the paper with a significant revision.
- The main question that the author addressed, "Through comparative analysis, the proposed classification method had excellent performance and can achieve accurate mapping of multiple crop types at a 10 m resolution at large spatial scales," was not explained that well in the gap statement.
- Based on my experience, the authors put a lot of effort into Introduction but only a little into the result and discussions section. Was this approach novel or in addition to some approach?
- Some of the graphs needed to be appropriately formatted for scientific publications.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The article proves to be relevant in exploring the combination of multiple images of unique temporal features and the image of time series features derived from Sentinel images in improving the classification of crop types, as well as the accuracy evaluation metric. The article is of interest to readers of this journal.
I found the article very interesting in which the authors discuss a new method of mapping crop types from Sentinel images, to be used in model 1: Classification based on time-series feature image, model 2: Classification based on multiple single-temporal feature images and in model 3: of the combination of the two previous ones from the classification by voting on the probability images. The article is of interest to the readers of this journal, but revisions must be made. As a suggestion the following should be considered for the improvement of this article:
a) Despite the theme of using Sentinel and RF images in the study area having been widely realized, I judged the combination of models used in the research to be of medium originality.
b) In the methodology, it is necessary to define more clearly the criteria for the selection of sentinel images used, as well as the steps performed in the work. In figure 4 the flowchart is confused, it must be redone.
c) In conclusion it will be necessary to define more clearly the implications of your discoveries for science.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
There have been some changes from the first review. However, some edits I provided earlier have not been answered yet. I felt that the authors contradict their statement in one or two instances. Still, more work on the literature review is needed to improve this manuscript. I would urge the authors to go through the edits and make changes.
