Automated Multi-Scale and Multivariate Geological Logging from Drill-Core Hyperspectral Data

Round 1

Reviewer 1 Report

This paper presents an open source toolbox based on the CWT method that can complement and supplement the manual drill-core logging process. The advantage of the method is that it offers a choice of scales and it is not limited to a specific form of data and a specific region. The work in is sounding and interesting. Some revision are recommended as follows:
1. It is recommended to clarify the synthetic data generation method in order to make the synthetic data more convincing. 
2. The "Figure1" in Figure 1 is redundant. 
3. Figure 1 is not mentioned in the text. 
4. The font size in the figures should be uniform. For example, Figure 3 has a variety of different font sizes, and Figure 8 has a different font size than the other figure.

Author Response

Thank you so much to the reviewer for his/her time and suggestions. 

1. It is recommended to clarify the synthetic data generation method in order to make the synthetic data more convincing. 

We thank the reviewer for the comment and for highlighting this point. We have extended the explanation of the generation of the synthetic dataset in the Synthetic dataset section. In addition, references to Figure 1 were included to illustrate the generation of the dataset. Furthermore, a note was added specifying that the code for the generation of synthetic dataset has been included and can be found under the data availability statement.

2. The "Figure1" in Figure 1 is redundant. 

Thank you very much for noticing this. It has been corrected.

3. Figure 1 is not mentioned in the text. 

The mention to Figure 1 have been added in the synthetic dataset subsection.

4. The font size in the figures should be uniform. For example, Figure 3 has a variety of different font sizes, and Figure 8 has a different font size than the other figure.

We thank the reviewer for raising this point. It has been corrected. All text in the images has been standardized and the font sizes have been adjusted.

Reviewer 2 Report

The authors propose an interesting method to transform hyperspectral image-derived mineral maps into vertically coherent geological units.
The proposed method extracts the local first principal component near each observation and then segments the resulting series of scores using a continuous wavelet transform. The correlation pattern between the variables is thus segmented. In addition to identifying geological boundaries, the scalogram also identifies mineralogically distinct domains.
This approach performs well on both synthetic and real multivariate datasets. In short, this method automatically converts hyperspectral mineral maps into vertically coherent geological units that can be used in 3D geological modelling workflows. The data's multivariate nature is exploited to identify lithological changes that would otherwise be missed.
The proposed methodology assists and complements the manual drill-core logging process.

The authors should recheck and improve the description of the method, aiming to eliminate some shortcomings and inaccuracies. 
The topic is interesting and the article overall seems well written and worthy of publication. 
However, I suggest a re-reading to eliminate the very few remaining imperfections, as:
- Figure 1 should be referred into the text before being placed.
- Line 344: instead of (Fig. 6-9) should be (Figs. 6-9).
It would also be useful to include a map to show Elvira's position in the context of the Iberian Peninsula.
Finally, it would be advisable to furter differentiate the content of this manuscript from the recent article by the same authors at doi:
http://dx.doi.org/10.1016/j.oregeorev.2021.104514

Author Response

We are grateful to the reviewer for his/her time and suggestions.

1. The authors should recheck and improve the description of the method, aiming to eliminate some shortcomings and inaccuracies. 

We appreciate the comment, the wording has been adapted to describe the method more clearly. In addition, references to the general workflow (Figure 3) and the illustrated workflow (Figure 4) have been included.

2. I suggest a re-reading to eliminate the very few remaining imperfections, as: Figure 1 should be referred into the text before being placed.

Thank you very much for noticing this. It has been corrected.

3. Line 344: instead of (Fig. 6-9) should be (Figs. 6-9).

We thank the reviewer for raising this point. the correction has been made.

4. It would also be useful to include a map to show Elvira's position in the context of the Iberian Peninsula.

We thank the reviewer for this recommendation. We agree on this point, and this will surely improve the manuscript. A map with the location of the Elvira deposit within the general geological scheme of the Iberian Pyrite Belt has been added as the Figure 2.

5. Finally, it would be advisable to further differentiate the content of this manuscript from the recent article by the same authors at doi: http://dx.doi.org/10.1016/j.oregeorev.2021.104514

We appreciate the comment, we agree on this point. That is why we have added in figure 3 (general workflow) the scope and topics covered by the previous and current work, clearly differentiating between them. The main difference being that the previous work was based on mineral quantifications prediction from drill-core hyperspectral data. On the other hand, in this paper we present an open-source tool and a methodology to perform borehole segmentation from multivariate input data, producing geologically meaningful domains at different scales.

Reviewer 3 Report

The manuscript nicely written and constructed.  I have some suggestions for more improving the manuscript as follows:

1. Novelty and significant of the study should be more highlighted in the introduction.
2. Specific objectives need to be added to the last part of the introduction.
3.  Geology section need to be added to the manuscript, which explains geological features, lithologies and mineralization in the study area. Please add a geology map and show geographical location of the study area.
4. Please provide flowchart of the methodology applied in your research, which showing start and end-product. 
5. Some  more recent publications (2021-2022) need be added to reference list and context
6. The results, discussion and conclusions have be written nicely.

Author Response

We are grateful to the reviewer for his/her time and suggestions.

1. Novelty and significant of the study should be more highlighted in the introduction.

We thank the reviewer for the comment, this suggestion will surely increase the quality of the manuscript. We have now highlighted some of the novelties of the study in the introduction.

2. Specific objectives need to be added to the last part of the introduction.

We appreciate the comment. However, specific objectives have been listed in the middle of the introduction. such as: “To properly obtain geologically meaningful domains that can be integrated into 3D geological modelling, the multivariate dataset of downhole mineral abundance has to be segmented into lithological domains. This requires taking into account (i) the multi-scale nature of the geological data and (ii) the fact that some lithological changes are only apparent in the correlation of the variable.”

3. Geology section need to be added to the manuscript, which explains geological features, lithologies and mineralization in the study area. Please add a geology map and show geographical location of the study area.

We thank the reviewer for raising this point. A map with the location of the Elvira deposit within the general geological scheme of the Iberian Pyrite Belt has been added as the Figure 2. However, the idea of this paper is to present a methodology and workflow applied through an open-source tool and both the synthetic and real data sets are designed to illustrate the method and not to present a case study.

Nevertheless, a reference is always made to: http://dx.doi.org/10.1016/j.oregeorev.2021.104514  which is one of our recently published works including  an exhaustive description of the regional geological settings of the IPB  and also the geological section, lithologies and mineralization of the Elvira deposit is thoroughly explained.

4. Please provide flowchart of the methodology applied in your research, which showing start and end-product. 

We thank the reviewer for this recommendation, the general flowchart of the method has been added in figure 3, where panel a describes the input data and the method can be observed step by step up to panel m which describes the end-product results. In addition, figure 4 illustrates the same workflow, step by step, but incorporating the results of the synthetic dataset. Also, the wording in the description of the methodology has been improved.

5. Some more recent publications (2021-2022) need be added to reference list and context

We thank the reviewer for this comment and for highlighting this point. We have now provided several updated references such as:

Zaitouny, A.; Ramanaidou, E.; Hill, J.; Walker, D.M.; Small, M. Objective Domain Boundaries Detection in New Caledonian Nickel531Laterite from Spectra Using Quadrant Scan.Minerals2022,12, 1–18.  https://doi.org/https://doi.org/10.3390/min12010049.53218.

Booysen, R.; Lorenz, S.; Thiele, S.T.; Fuchsloch, W.C.; Marais, T.; Nex, P.A.; Gloaguen, R. Accurate hyperspectral imaging of 511 mineralised outcrops: An example from lithium-bearing pegmatites at Uis, Namibia. Remote Sensing of Environment 2022, 269. 512 https://doi.org/10.1016/j.rse.2021.112790.

Hill, E.J.; Fabris, A.; Uvarova, Y.; Tiddy, C. Improving geological logging of drill holes using geochemical data and data 531 analytics for mineral exploration in the Gawler Ranges, South Australia. Australian Journal of Earth Sciences 2021, 0, 1–27. 532 https://doi.org/10.1080/08120099.2021.1971763.