Model-Based Probabilistic Inversion Using Magnetic Data: A Case Study on the Kevitsa Deposit
Round 1
Reviewer 1 Report
The manuscript “Model-Based Probabilistic Inversion Using Magnetic Data: A Case Study on the Kevitsa Deposit” deals with a possible probabilistic strategy for the integration of prior geological knowledge and geophysical observations.
The novelty of the discussed approach is not much; however, the manuscript might have a pedagogical relevance as it shows a simple case with, to some extent, interesting details.
Concerning the way the Authors’ point is presented, I have often the impression that the paper does not follow a sequential order and keeps jumping back and forth. Maybe this cannot be avoided but makes the Authors’ point hard to follow.
In addition, the terminology is sometimes obscure/imprecise (e.g. What is a “heterogeneous inversion”? What does it mean that “the chain did not converge”?).
Moreover, some minor typos are scattered around (e.g. It is not “a posterior”, but ”a posteriori”; the word "data" is plural).
The paper and the reference list are quite self-referential: what the Authors are discussing can be called petrophysical inversion (maybe, with explicit choice of the prior). For this, I modestly suggest having a look at the literature focusing on this topic and include possible relevant researches in the references. Actually, discussing and highlighting the differences with other previous works would increase the impact of the paper.
The discussion is generally quite qualitative, and conclusions are based on what sounds more like the Authors’ opinion rather than an objective fact. Thus, in a possible future version, it would be important to include more quantitative assessments.
I really hope these remarks can contribute to improving the overall quality of a paper that, in the end, I found interesting.
Kind regards.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper is written on a modern and valued topic. However, the description of the methodology needs improvement. There is a lot of abstract reasoning throughout the paper. A more detailed description of the methods used for both direct modeling and inversion is needed. Otherwise, it is not entirely clear how the obtained results can be reproduced.
Page 1 line 18. In the paper the proposed methodology is opposed to the computationally heavy grid inversion. It would be desirable to compare the computational cost in the model-based inversion and in grid inversion.
Page 1 line 27. It would be nice to confirm these statements with appropriate references.
Page 3 lines 82-86. What are the model parameters and the observed data in this paper? How are the likelihood functions constructed?
Page 3 lines 93-101. It is not entirely clear which of the listed methods was used in the paper. Maybe it is worth moving this paragraph into the introduction?
Page 3 line 103. «…typically…» It is necessary to describe exactly those methods that were used in this study.
Page 3 line 106. «…mathematical models M…» It's too abstract. What is a mathematical model in this paper?
Page 3 line 111. What are the model parameters in this paper?
Page 3 line 113. A brief description of the implicit surface representation approach should be given.
Page 3 line 116. What does rock unit mean? Especially in the context of inversion.
Page 3 lines 118-121. There is too abstract reasoning. What was in your specific task?
Page 4 lines 122-144. This is all too abstract. A more detailed description of model parameters, forward magnetic simulator, and likelihood functions is needed.
Page 4 lines 145-151. Maybe it is worth moving this paragraph into the introduction?
Page 4 line 155. How exactly was the structural geological model built?
Page 4 lines 159-161. It is worth giving a brief description of the magnetic forward simulation and the volume integrals. Did you use third-party software or your own developments?
Page 4 lines 165-166. It is worth briefly explaining what is meant by the induced magnetic field and the undisturbed field intensity.
Page 4 lines 169-179. This is too abstract. A more detailed description of the methods and algorithms used is needed.
Page 8 figure 5 (left) Why is the probability greater than 1?
Page 8 figure 5 (right) The vertical axis is unsigned.
Page 9 lines 291-295. How many cells were there? How was cell splitting done? n=?
Page 10 figure 6. It can be seen that (c), (f) and (i) are different, but is this difference so significant? Can it be quantified?
Page 11 lines 372-373. What commercial software was used?
Page 11 lines 376-386. Maybe it is worth moving this paragraph into the introduction?
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
The authors have done considerable improvement of the research methodology and added additional sections. I think that the article may be published in present form.
