Multi-Resolution Collaborative Fusion of SAR, Multispectral and Hyperspectral Images for Coastal Wetlands Mapping

Round 1

Reviewer 1 Report

Authors propose a multi-resolution collaborative fusion framework on three different remote sensing images (SAR, Multispectral and Hyperspectral Images) for coastal wetlands mapping. My comments are as follows:   

1.                   The technical contribution seems limited. Authors need to elaborately justify their technical novelty/contribution.  

2.                   Apart from Ziyuan (ZY)-1 02D hyperspectral, Sentinel-2 multispectral, and Sentinel-1 SAR images, more datasets should be tested. 

3.                   More ablation assessments are required as the proposed multi-resolution collaborative fusion framework involves several existing methodological components.      

4.                   I recommend comparing the proposed scheme with the recent progressive works.    

5.                   Why RF and SVM are used for the classification purpose? Why not other classifiers e.g., KNN, Decision Tree etc.?

6.                   Which kernel function is used with SVM? Why? Authors should discuss the tunning of the associated hyperparameters in the kernel function.

7.                   Authors should discuss the tunning of the associated hyperparameters in the RF classifier.

8.                   The evaluation is only empirical. In order to assess the validity and the utility, it would be desired to state formally its guarantees and to substantiate it by an evidence stronger than just experimental evaluation. 

 

Author Response

  We are particularly grateful to you for your approval and constructive comments to improve our paper. According to the comments, we have tried our best to revise the manuscript, and an item-by-item response follows. The modified parts have been highlighted in red color in the revised manuscript. Thanks for your time.

  Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

I assess the prepared manuscript very positively. The authors were asked to professionally prepare the results and discuss research methods. There is no reference to other works in the Discussion.

Comments:

Study area: Missing citation for description. Give quotes of some works that characterize this area.

Results: Figure 6. for figure "b" CNMF - content does not match the legend. I am aware of the difficulties, but if you can do something about it, it will be perfect

Discussion: The authors probably forgot to discuss their results against the background of other already published research results. Let the authors try to prove that the interesting results obtained are reproducible with others.

Author Response

  We are particularly grateful to you for your approval and constructive comments to improve our paper. According to the comments, we have tried our best to revise the manuscript, and an item-by-item response follows. The modified parts have been highlighted in red color in the revised manuscript. Thanks for your time.

  Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The topic of the manuscript deals with the fusion of hyperspectral, multispectral, and SAR imagery for coastal wetland areas. That topic fits well within the journal's scope and has various interesting aspects to cover. While some have been covered, several others have been somewhat short or omitted altogether.

As the authors correctly point out the co-registration of all the image sources is essential to assure high quality of the results. This part really lacks detail. It is not clear what the source and resolution of the DEM used for the pre-processing were. How many ground control points have been finally used? How many control points to verify the results? What was the co-registration accuracy? Sub-pixel accuracy is a little vague. Did you consider using other ground control? Some other measurements that would establish the coastline location? Of particular interest is the coregistration of the hyperspectral imagery, since that tends to be the most difficult one.

The authors entirely ignore tidal effects. How big are the tidal changes in the area of interest?

The authors compare their proposed method with a large number of other methods. The idea is a good one. However, It is not clear to the reader which of these reference methods is THE standard method for coastal wetland mapping. Another comparison I would be looking for is ground truth for coastal profiles. Alternatively, you could use some high-resolution airborne data to establish a more absolute reference.

The figures could use some work. The captions are not always self-explanatory. Sometimes, units are missing.

 

Author Response

  We are particularly grateful to you for your approval and constructive comments to improve our paper. According to the comments, we have tried our best to revise the manuscript, and an item-by-item response follows. The modified parts have been highlighted in red color in the revised manuscript. Thanks for your time. 

  Please see the attachment.

Author Response File: Author Response.pdf