The paper is technically sound and generally well-written. It proposes and showcases tools for the assessment of geological models in the minerals industry, from the green-field scale to the mine scale, therefore it is of interest to the readership of Geoscientific Model Development.

I recommend publication subject to the following revisions:



Main comments:

1) Lines 30-31: when the authors refer to measurement errors, what about geological mapping or logging errors (due to the geologist's criterion, for example) that may not cancel out after replication of the measurement? Some comments on this issue would be welcome.

2) Line 197: isn't it too strong to assume "isotropy"? I believe that anisotropic variations are common in geological modelling

3) Lines 207-208: the experimental variogram is not always well-behaved at short distances, to the weighting may render the indicator in Eq. (3) highly sensitive to the short-distance behavior and nugget effect.

4) In addition to the presented indicators, would contact relationships between lithocodes (measured through transition probabilities, transiograms or cross-to-direct indicator variogram ratios) be worthy of interest? Again, some comments would be welcome.



Minor comments

1) There is a mix of US (e.g.: "minimize", "summarize") and UK ("summarises", "anonimised", "modelling") English

2) I am not familiar with the word "voxet" (seemingly, a set of voxels): this could be defined to avoid confusion

3) Line 41: the date of the reference is 1978, not 1976

4) Line 45: there is a question mark before Sambridge

5) Lines 72-72: what/where are the Centre for Exploration Targeting, Loop researcher and related networks?

6) Figure 6: does the fourth row represent the "average normalized range and standard deviation", or the "average squared normalized range and variance" (the caption of the subfigures is not consistent with the figure caption)

7) Line 167: can be computed

8) Lines 196-199: "s" is a vector, but is "h" a distance or a vector? Is Z a "random variable" or a "random field"? Notation should be revised for consistency

9) Line 203: inside the norm, it should be s_j - s_k, rather than Z(s_j)- Z(s_k)

10) Line 236: considered classes

11) Figure 10: the caption in the top right subfigure should be \Gamma(p)^c

12) Line 324: overlap

13) Lines 526 and 533: who are "et al."?

Dear Editor 

Determining the uncertainty in geological models is a crucial steps as it directly affects the next steps of geoscience projects.

In this Manuscript the authors have proposed some indicators for determining the uncertainty in the geological models. This is interesting and well written manuscript. I found this manuscript to be a very interesting read, specially from practical point of view, and one that is certainly worthy of publication

I have just some small  comments that I would like to be addressed by the authors:

Beside the indicators that you have already mentioned in manuscript, What other parameters one can used as an indicators (local or global)  for determining  uncertainty in geological models?

You have proposed different uncertainty indicators.  Which of proposed indicators would be preferred to use firstly?  Which of them are more reliable and adequate for determining uncertainty?

Line 294: I don't consider Wikipedia as a scientific reference. Please eliminate it from you manuscript.

Overall, I thought this was a very well written manuscript. Any tool(s) that encourage EDA, involve exploring uncertainty, and make this accessible are needed in our industry. My main concern is the difference between uncertainty quantification and visualization tools. The authors should make it clear that the focus here is EDA and visualization, while the uncertainty quantification (local or global) is not really the focus. Input data/parameter uncertainty (p19) is a good thing to focus on and is different to local/global uncertainty, which is not the focus here; for example, this usually requires some form of model, authors mention a ‘Gaussian world’.

Major Changes:

I personally have not completed research based on surveys, but there are standard methods for presenting and developing these surveys. There are many details of surveys that are important for understanding/interpreting the results, some of these include: Which individuals (not names, but their background/industries/etc) were solicited to complete the survey? How many responded vs. how many were asked? Is 35 enough? What industries responded or were asked? Does this represent a reasonably diverse cross section through the industry or were only Australian professional in mining surveyed? Were these geomodelers, managers, junior/senior geologists/engineers? How was bias minimized? How were the questions decided upon? How were the multiple-choice answers selected to make sure the project goals were achieved? The majority of the justification in the work is based on conclusions drawn from these survey results, please expand on the details.

I am not sure how Section 3 follows from the survey Section 2. There are many existing software/tools/scripts/etc available to quantify uncertainty and then visualize uncertainty (which I would consider different goals). Based on Q10, it seems like the biggest issue is underestimation of global uncertainty, but this is not addressed. Poor transition between sections 2 and 3.

P8 Why are these 3 scenarios considered? Is this problem specific? Is this a good analysis for all datasets?

I am concerned readers would confuse the models generated in Figure 5 with more typical continuous or categorical geomodels (i.e.. a Gaussian world). While there is certainly value to the proposed models, these would have limitations compared to industry best practice, please make this clear.

Not quite sure what is meant by ‘underlying scalar field derived from implicit modeling’. Implicit modeling is a class of techniques, seems like something specific is used in Figure 5. Explain what is meant by this.

P12 Why are you discussing connectivity? Is this for a mining application (the survey was mining focused) or petroleum or hydrology? Would help to know what groups was solicited for survey results and the target audience for your tools.

Some minor comments:

Figure 2: I think the authors mean realization (not ‘real’)

Figure 2: replace (or add) the type of data for ‘data 1’ and ‘data 2’ (I’m guessing drillhole/well samples and remote/production data).

Figure 8: is this the variogram for the ‘model’? Elaborate on the appropriateness of this. Clearly this is not the variogram for the underlying sample data.