Do not consider the "Reconsider after major revisions" qualification as a negative assessment. It simply reflects that I do not find an option for what I think you should do.
Basically, you have been lucky with three positive and positively critical reviewers. I feel your response is late (you should respond immediately after receiving a review so as to favor discussion) and somewhat cursory. Specifically, I find that:
1) the criticism by ref 1 about the simplistic nature of the metric deserves a more appropriate discussion in the paper (why, which other alternatives did you consider, why did you choose ...)
2) I share the concern of referee 2 about the paper structure. Consider his suggestions to make it easier to read.
I will await the revised version of the manuscript to decide on whether to send it again to referees or accept it as is.

Dear authors,
The paper is good. You were lucky in having three thorough positive reviewers and you have responded in kind, leading to an improved paper. Moreover, the topic you address is important (I have often wondered about many of the questions you respond). I am left with many questions, but this may reflect the complexity of your goal. Therefore, I conclude that it has a great potential (although you never know whether a paper is going to have a large impact, and while uncertainty is recognized as important by all, few actually care much; I’d recommend you to present your results in congresses). I share below with you some of the possible improvements that come to my mind after reading the final version (the list is long, but most comments are very easy to handle).
Do not take these comments as compulsory in any way, but rather as an invitation to have a final thorough reading while feeling free to write your thoughts out and hopefully transform a good paper into an excellent one (but, please, if you change things, let me know what you have changed to facilitate my life!).
Excellent job!
Jesus
Comments (mostly editorial)
Line 56 “only 410 boreholes” sounds weird for most of us (I have never had 410 borehole descriptions!)… especially considering that in the abstract you do not mention the spatial extent of the study area. When I realize that the surface area is 45 km2, or some 10 “good” boreholes per km2, I conclude that the density is exceptionally high. While it is clear that the geological complexity is such that a deterministic description is not possible, I still believe that “only 410 boreholes” is inappropriate.
The fact that “The borehole lithology logs infer local changes in the immediate vicinity of the boreholes” depends on the geological setting, which you have not yet described (in the end, I believe the statement is true in your case).
Line 59: “Finally, the importance of the TI was studied. An example was presented where an alternative geological model from a neighboring area was used to simulate hydrostratigraphic models” does not read well (it is not clear to me what the “finally” refers to). Perhaps better “The importance of the TI was studied. An example was presented …” or “The importance of the TI was studied through an example where an alternative geological model from a neighboring area was used to simulate hydrostratigraphic models”
Line 77-78. The statement “E.g. indicating different levels of uncertainty in different subparts of the model domain” should perhaps be in parentheses (e.g., indicating different levels of uncertainty in different subparts of the model domain).
Line 83. My standard references for MPG (I prefer MPG to MPS, which suggest discrete multivariate distributions) are those of Strebelle (2002) and Mariethoz and Renard (2010). You cite both in the text later, but I find surprising your choice here. The first time I read your paper I had to go to the references to ascertain that you were indeed talking about what I call MPG.
Lines 88-100: The structure of a paper is always arguable, but I believe that this paragraph belongs better in the “methods” section than in the introduction. Usually, the introduction is used for revising the state of the art, so as to motivate the objectives of the paper. In your case, the motivation and objectives are clear and well established, but this paragraph looks awkward in the introduction as you discuss methods and the site, both of which are described later. I realize that this type of change is tedious and may introduce other “losses” of logic, but I still invite you to consider revising the structure of the introduction and methods. The same can be said about line 111, or lines 121-122. In all these cases, you identify sources of uncertainty and say how you are going to address it. The end result is that, after reading the introduction, the reader is left with a mixture of loss of confidence in geological modeling and a somewhat vague feeling that you are going to somehow assess them. I invite you to consider revising all sources of uncertainty to conclude that the objective of the paper is to assess their impact on the uncertainty of the geological model and, then, describe in a compact manner how you are going to do it (perhaps in an itemized way).
Line 104: “provide” instead of “provides”
Line 105: “Resolution decreases with depth, and diminishes at a specific depth, which is dependent on the geophysical method” reads weird. Perhaps better “Resolution decreases with depth, specially beyond a specific depth, which is dependent on the geophysical method”
Line 153: I do not know what you mean by “tunnel valleys”
Figure 1: Readers that do not know SKYTEM may be confused about Figure 1C. I assume that the blue lines are the trajectories of SkyTEM flights, you may want to say so (instead of soundings) in the caption and identify them with a blue line (instead of a dot) in the legend.
Line 176: I would not say “The TI thus provides a conceptual geological understanding”. If by conceptual understanding, one means a qualitative description of the site and its origins, the TI provides neither. To me, the “conceptual geological understanding” is what you provide in lines 147-150. Although the TI can be based on geological understanding (and, as such, it is indeed a conceptual statement), the TI provides simply a quantitative description of the expected (often only small scale) spatial variability patterns of the random function of interest (in your case textural description).
Section 3: I realize that you have made an effort at improving the structure of Section 3, and I think that further improvement is possible (structure is important for Cartesian people like me and, I assume, most of the readers understanding section 3). Specifically, snesim and DS are two methods for doing MPS, while the tau approach is a method to combine conditional probabilities. Therefore, including the tau-model as a subsection of snesim is misleading. You may choose: 3.1 MPS; 3.1.1 snesim; 3.1.2 DS; 3.2 tau model. You may also choose: 3.1 MPS: snesim; 3.2 MPS: DS; 3.3 tau model. The latter may be better for your goal because you appear to have chosen (for reasons that are not clear to me to use DS and snesim for different goals).
The tau model: the description of the tau model can also be improved. The one you make is somewhat abstract (Journel style), which may be appropriate for a statistical journal but should be made clearer for a hydrological journal: Possible improvements include:
- The tau value can be used to qualify the degree of redundancy among data sets (see www.ccgalberta.com/ccgresources/.../2003-130-naivebayes.pdf)
- Use a bold face for Di data events (as used in eq. 1). Also, explain what data events are. At first, I thought it referred to actual data sets (e.g., core logs), but later it appears to refer to probability grids (a concept you may define as well)
- In line 210, you should define x_i as distances from a given realization of A to its true value.
- Line 212: the statement “where the tau values are assigned as follows: [τ1,τ2]” does not really describe how tau values are assigned.

Line 259: K was already defined in line 248 (you do not need to define it again).
Line 276: You do not need the comma after (2003).

Consider addressing comment 2 of reviewer 1 (actually your response) in the manuscript.

Excellent work!
Just a few minor typos.