Enhanced Detection of Artisanal Small-Scale Mining with Spectral and Textural Segmentation of Landsat Time Series

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

A very thorough treatment of your data and methods. My only concern is why yo chose to use data from 2019 or earlier. There are at least 3 more years of data available that you should have used.  And why did you not use Sentinel-2 data? The bands are similar to Landsat, the spatial resolution is higher!, and the US Geological Survey/NASA have freely available harmonized Landsat and Sentinel data sets, where the radiometry is matched between the two data sets.

Author Response

We would like to thank the reviewer for their comments. We have incorporated most of them, which we highlight in our response (attached), and we believe this has made the manuscript much stronger. In few cases where we disagreed with the reviewer, we provide justification for retaining the original content.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This study effectively addresses the issue of extracting small-area features by integrating indices, texture, and temporal information. The content of the article is relatively comprehensive. However, in terms of innovation, this article appears to be somewhat limited for an extraction-oriented paper. The use of fusion of indices, texture, and time for land cover classification (Yan S., 2019) and detection has been extensively researched in recent years. Furthermore, some studies have also calculated numerous indices and texture information, yet they have not adequately discussed the computational workload and redundancy in the data classification process.(Liu D., 2021).Therefore, I hope the authors can think more deeply about where this paper demonstrates innovation in areas other than the research subject (ASMs).

1. The introduction section adequately addresses the applied technological issues. However, there are notable shortcomings in terms of clearly defining the scientific problems being addressed. Additionally, the literature review is not sufficiently clear, and the language used lacks fluidity and coherence. for examples: Lines 34–35: This description is very subjective and not comprehensive enough; Lines 62-63: Please pay attention to the strictness and accuracy of expression.

2. Figure 1, particularly Figure A and Figure C, exhibit significant confusion and lack of clarity. The annotations are disorganized, and in some cases, the figure boundaries are lost, blending with the background. 

3. 3.2 Spectral transformations and 3.3 Textural transformations. Calculating various similarity indices, as well as the use of texture information in later data classification, can easily lead to data redundancy and exponentially increased computational workload during data processing. Can this study identify the optimal indices and texture information by setting up control groups? 

4. In Figure 3, does the curve correspond to the NDVI curve of the actual area? If so, please provide the corresponding RGB false-color optical remote sensing image of that area.

Shi, Y. (2019). Urban land use and land cover classification using multisource remote sensing images and social media data. Remote Sensing, 11(22), 2719.

Liu, D.,  (2021). Investigation of the capability of multitemporal RADARSAT-2 fully polarimetric SAR images for land cover classification: A case of Panyu, Guangdong province. European Journal of Remote Sensing, 54(1), 338-350.

 

 

 

Comments on the Quality of English Language

Extensive editing of English language required

Author Response

We would like to thank the reviewer for their comments. We have incorporated most of them, which we highlight in our response (attached), and we believe this has made the manuscript much stronger. In few cases where we disagreed with the reviewer, we provide justification for retaining the original content.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors The paper investigated a method for detecting artisanal small-scale mines in the Amazon Rainforest using Landsat. Overall, it's an interesting work, which may be published under a couple of highlights – corrections:

Abstract

·   Lines 6-8:  Rewrite the long sentence.
·   Line 8: Add ‘To address this issue, we….’.
·   Line 11: Specify the number of textural and spectral features.
·   Define tacronym when first used

Introduction
·   Consider rewriting longer sentences into shorter ones.
·   Lines 62-64:  Based on what do you argue this statement?
· Overall, the introduction requires improvement as it does not clearly demonstrate the research's significance. Moreover, all paragraphs lack coherence.
·   Additionally, the literature review should include more important studies on monitoring and detecting artisanal and small mining using remote sensing and machine/deep learning. Highlight how your approach is different from existing methods.

Study area

·  Provide more details about the elevation ranges, climate, land use and land cover types, pollution issues, and the ecological significance of the area beyond its role in ASMs activities.
·   Figure 1 requires improvement.

Materials and methods

·  For Figure 2, consider adding colour to differentiate each step.
·   Line 96: Specify which sources?
·   Line 107: ‘ low-water period’ might not be clear to readers, explain more.
· What are the sources of uncertainties associated with Landsat imagery interpretation?
·Lines 130-145: ensure that acronyms are defined when first used and include references for each statement.
· What are the advantages and limitations of using LandTrendr for detecting ASM?
· Line 172: ‘eight control parameters are required’ provide more details.
·  Line 177: Explain the reasoning for the selection.
· Line 197: Which tool/software did you use to perform variable selection with Random Forests?
· Provide a table indicating the RF tuning parameters used in this study
·  Line 2016-2019: What was the sampling design?
· I suggest providing some aerial views of ASMs in the study area.

Results

· Figure 4. To be redone. I suggest enhancing the figure's resolution, which appears to be slightly blurry.
· Figure 5. The graph is illegible.
· Overall, the approach looks like describes the results obtained. Thus, critical analysis is required.

Discussion

· The discussion needs to be more compelling. The authors are advised to rewrite it.
· In terms of practical applications, what are the implications of the findings?

Author Response

We would like to thank the reviewer for their comments. We have incorporated most of them, which we highlight in our response (attached), and we believe this has made the manuscript much stronger. In few cases where we disagreed with the reviewer, we provide justification for retaining the original content.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

thank you for your edits and response

Author Response

No changes requested by this reviewer.

Reviewer 2 Report

Comments and Suggestions for Authors

1. The format needs to be checked and improved more thoroughly. The keywords of this article seems to be missing, and the reference format is not the rise template format; Line 61, 267: there are two points at the end of a sentence; Line 16: ensemble of ensemble or ensemble of ensembles? 

2. I've  wondered why only chose 30m resolution Landsat in this study, as the article mentions 30m resolution tends to monitor a mix of pixels from small-scale mines rather than the higher resolution, publicly available Sentinel-2 data after 2015, Is it because of the need to monitor changes in long time series?  Do we need to integrate sentinel-2 and Landsat series after 2015? 

3. Figure 1 I don’t kown this lower right scale corresponds to that part of the three figures

4. I hope you will double check the figure 3. and give the detailed geographic coordinates of the point, not  Landsat scene path 228, row 64.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Comments on the Quality of English Language

The format needs to be checked and improved more thoroughly. The keywords of this article seems to be missing, and the reference format is not the rise template format; Line 61, 267: there are two points at the end of a sentence; Line 16: ensemble of ensemble ?

Author Response

Our point-by-point response is attached. Thank you again for your feedback.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Thanks for your revisions, I have no additional comments.

Author Response

The reviewer made no additional comments.