The Use of C-Band and X-Band SAR with Machine Learning for Detecting Small-Scale Mining

Round 1

Reviewer 1 Report

[1] The work utilises remote sensing data such as Sentinel-1 and Kompsat in small scale mining in Ghana. While this could be a useful work, I find that motivation is highly lacking.  For instance, you can not say that ‘SAR has not yet been considered …. of SSM in Ghana (line 57)’ to claim the novelty as well as motivation of this work. In addition, a great deal of works utilising remote sensing and GIS in mining sector of Ghana are conducted by Dr Abdul-Wadood. He also examined environmental impacts of mining in the same work. These works are albeit missing without any reason. Use of his works could be valuable though for your work. Furthermore, SAR has a few advantages of which all weather imaging capacity is the first and foremost. However, utilizing different bands could be a big issue which is evidently enhance uncertainty. I therefore, like to ask for outlining clear motivation and enhancing international significance of this work. As such you can dedicate a paragraph or a few lines, supported by existing works, in the intro section to demonstrate how SAR is being used in various locations, except Ghana, for mapping floods, wetlands, coastal shoreline etc. Below examples could be of great value to go ahead. Why and how C-band and X-band with varying polarisation can be useful in SSM used in the work is definitely missing. This requires serious attention. Machine learning is seen as a panacea to all problems, I however differ with this view. Thus, you have to have a revamping of the intro section to motivate readers with international significance exemplified with the following works. https://doi.org/10.5897/IJPS.9000357; https://www.jstage.jst.go.jp/article/jjshwr/19/1/19_1_44/_pdf/-char/ja;  https://doi.org/10.3390/rs70607615; https://doi.org/10.1109/JSTARS.2014.2347171

[2] Line 55: instead of solution, use ‘panacea’

[3] In this study you used VV, HV etc polarisation. I believe use of varying polarisation could introduce uncertainty. How did you mitigate this issue? I like to see

[4] Line 150 missing source reference. Line 165: you did use a software to carry out this work, hence I don’t see a real contribution in terms of methods

[5] I would refine fig 3 to show a real and a useful solution with a flowchart used

[6] Your method does not specify whether a filter was used to remove speckle of SAR data? This must be clearly noted

[7] What was the logic of using multi-class versus binary water. I am sceptical regarding the method used hence reported results of this study

[8] These days OA, PA, UA are in fact dead accuracy metrices, you need to work with quantity and allocation disagreements (Pontius and Millones, 2011) to make your work more robust

[9] A good discussion section but missing real contribution relative to the literature (above works could be of help though). Also, conclusion does not reflect problems at hand to solve.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

General description:

Small-scale mining activities in tropical regions as hazardous both environmentally and socially are in the focus of the present work. To detect map small-scale mining a water body’s classification of SAR imagery using five machine-learning classifiers is performed by analyzing multi-temporal filtration of three SAR datasets.

The experimental results prove the water bodies in the SAR images as a proxy to illegal mining activities.

Remarks:

Page 4, row 150, It is written Error! Reference source not found. It means that something is missed.

Conclusion:

Large number of experimental results obtained by different satellite systems are provided and analyzed in the present study. All stages of the classification process using different classification approaches are thoroughly described and graphically illustrated.

Results of the measurements and classifications sound convincing and can be applied to detect and prevent illegal small-scale mining activities.

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 2 Comments

Point 1: Page 4, row 150, It is written Error! Reference source not found. It means that something is missed.

Response 1: Thank you for pointing out the error, it is now fixed.

Reviewer 3 Report

The authors proposed a method for Small-scale mining detection using SAR images. This paper is overall interesting. However, the contribution and innovation should be more clearly clarified. I have the following comments:

 

1. Motion error[1,2] is a key factor that may degrade the quality of SAR images. Is it possible for the proposed method to detect small scale mining with high accuracy if the SAR images are unfocused with motion errors? Including experimental results may not be feasible, but it should be clearer to have a discussion on this point.
2. Some other deep learning papers in SAR application should be introduced as [3].
3. The authors should emphasize the unique problem in small scale mining detection. What is the difference compared with other object detection task?
4. A language revision is necessary to fit some minor grammatical errors, e.g. the misuse of prepositions affects the readability of article.
5. The authors should pay attention to the format of references.

[1] Y. Ren, S. Tang, P. Guo, L. Zhang and H. C. So, "2-D Spatially Variant Motion Error Compensation for High-Resolution Airborne SAR Based on Range-Doppler Expansion Approach," in IEEE Transactions on Geoscience and Remote Sensing, doi: 10.1109/TGRS.2020.3048115.

[2] J. Chen, B. Liang, J. Zhang, D. -G. Yang, Y. Deng and M. Xing, "Efficiency and Robustness Improvement of Airborne SAR Motion Compensation With High Resolution and Wide Swath," in IEEE Geoscience and Remote Sensing Letters, doi: 10.1109/LGRS.2020.3031304.

[3] W. Pu, "Shuffle GAN With Autoencoder: A Deep Learning Approach to Separate Moving and Stationary Targets in SAR Imagery," in IEEE Transactions on Neural Networks and Learning Systems, doi: 10.1109/TNNLS.2021.3060747.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Thanks 

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.