Identification of Hydrothermal Alteration Minerals for Exploring Gold Deposits Based on SVM and PCA Using ASTER Data: A Case Study of Gulong

Round 1

Reviewer 1 Report

thank you for significant revisions to original manuscript. Applications are now clearly described, and validated by field work.

Reviewer 2 Report

The modifications you made are significant and the readability of the paper is very good now, all the best.

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

Congratulatios for your work!

In general, the work is quite good, I miss the treatment of textural information. You have performed a good analysis of the intrapixel information but you have not completely ignored the interpixel information.
Although the results of the classification are good, using textural information (geostatistical or statistical methods) you would have improved the accuracy of the delimited areas.
It may include textural information that is outside the scope of your work but surely it would have improved your results.

Acoording you paper, I agree the mixed pixel decomposition method with hyperplane optimized remove the influence of vegetation interference information from the interior of the pixel. And on the other hand, the classical method PCA and SVM have provide a good extraction model of altered mineral.

Both methodologies are synergistic and provide good results but  the lack of textural information treatment don't give us the if the results can be improved. Anyway, good job!

Reviewer 2 Report

Figure 1: Add latitude/longitude. Add scale. Add index map, fix the legend. Q-D means nothing. p-S means nothing. “gamma” means nothing. What is in the red box? The figure quality is a fuzzy JPEG. It must be improved. Is the ASTER image for the same area as the map?? Your study area box on map is a tiny area. So the study area on the ASTER image, I imagine, is a tiny box?? Where on the map is the river on the ASTER image? The map contributes little to my understanding of the geology. I need a map of the study area! Not all of China

68-76. The explanation of the geology is inadequate. What does this poor description have to do with the small study area shown on the map? The geology description must be complete and simplified, without geologist slang.

Line 78: why do you need 6 ASTER images? Why did you do “mosaic and subset”? Much work is needed here to explain.

95: ALL pixels are mixed, not just those that have some vegetation.

3.2: PCA does NOT highlight or separate alteration minerals if they are at very low occurrences in the scene. PCA explains 95% of the variance. Looked-for mineral signatures are left in the “noise”

131-134: How did you use ratios? How did you select ratios? Which ratios did you use? Were these the inputs to PCA?

194-196: How did you separate pyrite, sericite and chlorite from PCAs?

211-239. This entire discussion is FAR too technical. You must provide simpler explanations for the methods you use, and send all of the technical details to an appendix. Start with the earlier discussion of hyperplanes.

Figure 5. What does this show? This looks like figure 1, with slightly different colors. Explain in detail

249: how did you pick these bands and ratios?

253: PC4 reflects alteration iron mineralization”. This statement is entirely unsupported.Same for next paragraph on sericite and chlorite

Figure 9: where is this??????????Is there any ovelap with the ASTER image?? Where do all the lines come from? What are the symbols?

295: ACA: why were some classified areas labeled as false? This analysis is just jargon and provides no explanation whatsoever.

Field samples: Where are the laboratory analyses to justify the statements of minerology?

Reviewer 3 Report

Thanks for this interesting paper, the methodology explained in this paper is very interesting and useful, although the results requires some additional clarifications. It is also important to add some recommendation at the end, to show the implications and to discuss the application of this new methodology. I strongly recommended to do English proofreading to correct some minor spelling and grammar mistakes. e.g., in the abstract ".....interference bas been removed" spell check. The figure captions needs some additional explanation, otherwise they are not showing much, add a scale to the figures. -Figure 14-, I suggest adding another image, (a) original image without any reprocessing, and (b) the current reprocessed, to show the difference and to explain the impact of this study outcomes.

 

The conclusion:

……, “and also provides the possibility for extracting other weak information in this paper” this sentence does not make any sense as it is currently written, you need to rephrase it. The second point of the conclusion dose nor really reads well, and it needs rephrasing to clarify the meaning. Point three, also consider rewriting it, to make your finding clearly visible.

All the best

Reviewer 4 Report

This manuscript propose a methodology for gold ore deposits identification. Authors developed an intensive computational effort in order to suppress the effect of the dense vegetation of the study area and to map spatial distribution of indicative minerals. In this sense, I consider that the manuscript is interesting and the mathematical explanation of the methodology is comprehensive.

Unfortunately, an intensive review of the manuscript is needed. Three main negative issues should be addressed:

No discussion of the results has been included. Your results and methodology was not compared with any previous research. Without a proper discussion, the scientific significance of the manuscript is irrelevant. Accuracy assessment procedure is not properly explained (a specific section should be included in the material and methods section). Some datasets were not properly employed for independent training and validation. The methodology employed to develop some training/validation dataset is unknown. The spatial distribution of the final field survey is unknown. I am concerned because the computational effort of this manuscript is outstanding but after all these efforts, the final results were not able to provide a quantitative assessment of the relative abundance of the three indicative minerals. However, the results will be highly valuable after a better explanation of the all the accuracy assessment process. The presentation of the cartographic information is poor. Basic cartographic elements were omitted. The spatial extension of the second geological map is inconsistent respect to the other maps.

 

I include some additional comments and suggestions.

Line 19. An errata: has instead of bas.

Line 67 figure caption. Reginal or regional? Include RGB band combination of the ASTER image. Include the meaning of the geological formations (I.e., Q-D, p-S, gamma,...).

Figure 1 et seq. Basic cartographic elements such as scale bar or north arrow should be included in each map/satellite image. What is the source of the geological cartography?

Lines 77 to 82. Additional information of the satellite images and the preprocessing methods is needed. Include the number of images and acquisition dates. Did you employ all the spectral bands? Did you apply a topographic correction? What water index was employed?

Lines 202 to 205. A comprehensive accuracy assessment section should be included in the material and methods section. How were obtained the training samples (i.e., field survey, aerial photography,...)? What is their spatial distribution? In order to avoid spurious results, training and validation samples shouldn’t be the same.

Lines 236 to 242. What about the accuracy with an independent validation dataset. Additional details of the employment of the GA results for vegetation suppression are needed. What do you mean with “the final assignment“? Additional evidences of effectiveness of your vegetation suppression methodology are needed. You may include a set of spectra of the ASTER images at random points before and after the removal of the vegetation effect. A discussion of your results (comparing your approach respect to previous research) is mandatory.

Line 245. Review the numeration of the section.

Line 246 to 248. Justify your statement with proper references.

Line 253. You should justify why PC4 is  related with pyrite mineralisation. The same for sericite and chlorite.

Line 262. Relate your results with the geology of the study area and discuss them. The same for sericite and chlorite.

Figure 9. Include basic cartographic elements and the source of the map. The shape of the map is not the same of the satellite images. Are you presenting a subset of the study area? The computational work of the manuscript is very interesting but the treatment of the satellite images and other cartographic elements is weak.

Lines 286 to 292. Comprehensive information of the accuracy assessment should be included in an specific section. This kind of information is scarce and disperse throughout the manuscript.

Lines 286 to 292. In found 14 gold deposits in the map of figure 9. How do you obtained 1800 training/validation samples? Further details are needed.

Lines 314 to 345. All previous computational efforts are conditioned by this field survey. This is the critical point of the manuscript. You should include a map of the sampling locations in order to get knowledge of their spatial distribution and geological characteristics. Are you able to provide information of the relative abundance of the three indicative minerals (i.e., pyrite, sericite and chlorite) at each sampling location? Your current field validation is semi-quantitative. Discuss your results.