I have now reviewed the revised version of the manuscript by Hupkes et al. My overall feeling is that the manuscript has improved but still requires minor revision before potential publication.
The authors have addressed most of my comments and responded often convincingly to them. However, I have still some recommendations that should be taken into account before acceptance.
*When talking about fracture associations and more generally about the brittle deformation pattern in fold-and-thrust belts, I think that due credit must be given to the review paper by Tavani et al., 2015, Earth-Science reviews which is highly relevant to the topic of the manuscript, should it be about fracture occurrence and types, associations, chronology with respect to folding and even stress interpretation.
*L79 : Barbier et al. MPG, 2012 should also be considered here
*L83 : should be roughness instead of shape
*L153 Focal mechanisms of earthquakes
*L 167 remove ‘fractures’
*L188: change into principal stress axes
*The authors wrote in the rebuttal : We use orientation and discontinuity type to define discontinuity sets, and group these into associations if they are mechanically consistent (i.e. they might have formed in the same stress field). We prefer to refrain from using relative timing to define discontinuity sets, as we believe cross-cutting and abutment relations are often ambiguous, and contain little meaning if the two discontinuities formed in the same stress field that might have prevailed for several millions of years.
I disagree with this statement, since I do think that for instance multiple discontinuity sets with similar orientations may have formed at different stages of the tectonic history. This is often observed in the field on the basis of relative chronology and has been confirmed recently by U-Pb geochronology on synkinematic calcite mineralization. I strongly believe that determining the relative chronology between individual fractures (mode I fractures or faults) using cross-cutting/abutting/reactivation criteria on a statistical basis remains an extremely valuable and necessary information for tectonic reconstructions. I therefore do believe that the definition of a fracture set should also include the relative timing.
* The authors wrote in the rebuttal : It is more than likely that faults and discontinuities related to the stress perturbation around the faults influence locally the flow characteristics of the reservoir, but that is outside the scope of the current study. Interestingly enough, we do find discontinuity associations that are similar in orientation to those observed in the other outcrops
(Parmelan, Jura) network close to the seismically active Vuache fault (see stations 32, 35 and 39 on figure 10). Therefore, we define them as part of the background network, that formed prior to tilting of the strata. So even though the complex history of this fault, with several reactivation phases, the regionally consistent background network is still present within several 10s of meters away from the Vuache fault.
I may partly agree, but this must be clearly stated in the manuscript, i.e., your answer to my earlier comment must be transfered into the text. It is of utmost importance in order to unambiguously show the interested reader that this point was not neglected a priori and that you are aware of stress perturbations in the vicinity of (strike-slip) faults (rotation of principal stress axes, variation of the stress ellipsoid shape ratio) and of the development of associations of secondary fractures formed in the perturbed stress field. I therefore invite you to add a few sentences on this point and to consider citing the papers dealing with directional stress perturbations in the vicinity of the major strike-slip faults in the area investigated (Jura) by Homberg et al., JSG 1997 and Homberg et al EPSL 2004.
I would add that according to these papers, the stress may be deviated over a 4-5 km wide area around the fault itself, and not 10s of meters as stated. It is clear from the map of Fig.10 that in sites 35 and 39, the compressional stress is rotated counterclockwise, which substantiates my comments and the need to add some cautionary sentences. In fact, it is because you accept a deviation of 30° in the orientation of your discontinuity sets that you can consider them to still reflect the regional background network despite obvious stress deviations. Again, this must be clearly stated.
* The authors wrote in the rebuttal : As mentioned above, relative timing is of subordinate importance for defining discontinuity sets in our study, as we focus on associations of discontinuities and the stress field in which they are formed, rather than individual sets.
This looks like you choose to overlook the fact that similarly oriented discontinuity sets or associations of discontinuities (eg. mode I fractures) may have formed at different times as it is now demonstrated by U-Pb geochronology (eg, Zeboudj et al Geosciences, 2025), which would be documented otherwise if looking carefully and statistically at cross-cutting / abutting relationships or reactivation. May I recall that the chronology between E1 and E2 events is based on relative chronology criteria ? So even though you consider the relative timing between individual discontinuities within a given association to be of subordinate importance, you do use relative chronology between discontinuities belonging to different associations to define the sequence of development of the discontinuity associations, hence of the driving paleostress fields. So I find it weird and somewhat dangerous to depreciate the use of relative chronology in fracture studies as you seemingly do. I guess the reader would expect an even short note on that point.
*The authors wrote in the rebuttal : We do not assume there is a regional consistent regional stress field, but the results show that in fact there is consistency with respect to orientation of σ1 that formed the discontinuity associations of the background network (see figure 10).Geochronology in the form of dating calcite veins is indeed a very interesting way to constrain timing of the formation of the background network. It will give a minimum age constraint on the forming of the discontinuities, as the timing of calcite precipitation does not necessarily coincide with the propagation of a fracture. On top of this, we don’t know how much time is involved in the creation of the background network. If we consider that they are formed under sub-critical conditions, it might have taken 10s Ma.Therefore, it is expected that different ages will come out for the discontinuity association.
I cannot grasp the logic behind the last sentence regarding the present study. Calcite mineralization is undoubtedly syn-kinematic when associated from slickenfibers from striated mesoscale faults or when calcite has grown as fibers within opened veins. In case of blocky calcite textures, I agree that calcite mineralization may be delayed with respect to opening at the scale of an individual vein, but one can safely consider that, for a vein set, calcite precipitation is coeval with vein opening over the time span corresponding to the fracturing event forming the vein set, and is thus syn-kinematic in a broad sense. This means that in the time interval required to form a discontinuity association you are interested in, calcite mineralizations can be considered as syn-kinematic for veins and faults (see Zeboudj et al Geosciences 2025) within the uncertainty of U-Pb calcite geochronology, so the range of consistent ages obtained for such given fracture association may give a hint to the duration of the related fracturing event/driving stress field. U-Pb calcite geochronology has also revealed that individual fractures as well as fracture associations may be wrongly interpreted to be coeval, hence formed in the same stress field, if the interpretation is done on the sole basis of their geometry and orientation, so going back and forth between absolute dating and 'classical' field-based approach (a fracture set defined on the basis of similar orientation, deformation mode AND relative chronology) is to date the more efficient way to meaningful tectonic reconstructions. I acknowledge absolute dating of fractures is beyond the scope of the study, but, again, a cautionary note would be very useful.
* The authors wrote in the rebuttal : But in the current study, we don’t have any absolute time constraints on the formation of DAs. That is why we focus on relative timing with respect to tilting/bending of the strata.
Yes (see above), but in fact you analyzed fractures with an Andersonian state of stress in mind, which is already an hypothesis in folded and faulted domains (again, see Tavani et al., 2025). As I wrote previously, the classical fold test must be preferred.
* The authors wrote in the rebuttal : We added the references for stylolite as paleostress marker in the Introduction. Although it is a highly interesting topic, using stylolites to reconstruct magnitudes of paleostress is outside of the scope of the current manuscript, and therefore references are not included. However, for tectonic stylolites, this may be challenging, as the assumption that σ2 and σ3 are of same magnitude is no longer valid, as for bed-parallel stylolites.
Yes, I agree, but in the case of tectonic stylolites you should be able to define whether the stylolites truly developed under a reverse or a strike-slip stress field, and therefore if they are mechanically consistent (or not) with the discontinuity association of interest to your tectonic reconstructions. Note that the references I provided pertain to tectonic stylolites only. I acknowledge this point goes beyond the scope of your manuscript.
* The authors wrote in the rebuttal : In the Results section, when defining the regional events, we appreciate the fact that bed- perpendicular stylolites are indicative of a strike-slip or reverse regime.
Just to make things clear, you appreciate tectonic stylolites may be indicative of strike-slip or reverse regime because they are seemingly associated in the field with strike-slip or reverse faults, respectively. But this is some kind of circular reasoning. What I mean is that the real stress regime associated with stylolite development could be documented independently from the fracture associations by the approach mentioned above, the only to date to the best of my knowledge, providing a possible test on the validity of relating stylolites to the same association than conjugate faults and veins. However, I acknowledge that this point requires also the estimate of the depth of formation of the tectonic stylolite, and that the approach would need much more analyzing effort than collecting fracture data in poorly deformed sedimentary rocks.
* The authors wrote in the rebuttal : We did not focus on the fold and/or fault that might be present in the well, because that is not the scope of the current manuscript. We focus on characterizing the background network based on outcrop observations. It is by no means the aim to present a ‘complete’ interpretation of the BHI that explains all features observed in the well.
Frankly speaking, you cannot simply brush this comment aside by claiming the occurrence of the fold is not the scope of the manuscript. Since you choose to show two wells GeO-01 and GeO-02, you cannot overlook simply the fold occurrence in GeO-01.
The authors wrote in the rebuttal : in this study we focus on the background network and we prefer to turn the observation around: the same percentage of features in both wells can be related to the predicted background network. This underlines the regional character of the background network, or in other words, that this portion of the fractures may be extrapolated to the entire reservoir. As mentioned above, it was outside of the scope of the current study to investigate how the features that are not related to the background network compare between the two wells.
I may understand your point, but first in L 394 you wrote there was no fold ! Second, since you also wrote that a fold create some localized fracturing, the reader expects that crossing a fold will increase the number of fractures encountered in the well. How do you explain that the same percentage of features in both wells are related to the predicted background network, regardless of the folded or flat-lying character of strata ?
*The authors wrote in the rebuttal : In this study, we document DAs in the field. It turns out there is a regional consistency in the orientation of DAs, regardless of the distance to regional faults such as the Vuache. We acknowledge that local stress perturbation in flat-lying strata may occur, but with the methodology presented in the current MS, they would not be documented, as they don’t reveal a regional pattern.
I disagree. Local stress perturbations are likely documented but they are hidden behind the 30° deviation you accept for considering the discontinuity association to belong to the background pattern. Clearly, the stress is rotated counterclockwise on the map of figure 10 (sites 35 and 39). See also my earlier comment above. Again, a cautionary note is needed.
I hope that these new comments will help the authors clarify further their interpretations, introduce some cautionary notes about their choices and possible pitfalls of their approach and provide a more comprehensive description of the hypotheses behind their approach without omitting some specific points which would easily open room for criticism.
I am looking forward to seeing the revised manuscript published soon. |
General comments
Optimizing the use of outcrops and image log data to characterize subsurface fractures is a topic of widespread scientific and practical interest. This paper provides a useful example of fracture description that is relevant to geothermal applications. The approach is to obtain kinematically meaningful fracture and stylolite relations in various outcrops to identify regionally persistent patterns of fractures. Then the patterns are used to guide image log interpretation in two wells. Using fracture relationships for kinematic interpretation of fracture and stylolite patterns has long precedent (see the review by Hancock, 1985) but the specific approach used here of high grading a suite of key features from several outcrops regionally and the application to comparing outcrop fractures to the subsurface and as an aid to interpreting image logs for geothermal studies is sufficiently innovative and topical to be of interest. The work is within the scope of this journal. The illustrations are good, and the text is mostly well prepared. At 28 pages the MS is succinct.
The MS as it currently stands needs to be improved with moderate revision if the work is to have impact. The Introduction does not make the aims, assumptions, and claims sufficiently clear. Some of the material here can be reorganized. And the claims need to be made explicit. The Methods mix in arguments about interpretation that belong in the Discussion but leave out specifics of what methods were used and contains vague statements about the attributes of the outcrops. The use of well data is not evident from the Introduction or Methods even though two of the MS conclusions focus on claims about the subsurface. A short new but information-rich Methods section is needed. Stylolites are prominent in the MS title and the methods use fracture and stylolite relations, but because of the ‘discontinuity’ terminology the role of stylolites is not apparent until well into the MS. Although the approach of selecting key fracture relations seems to have worked, I don’t think the MS does a sufficient job of discussing the caveats—why did the method work here? What could go wrong if the approach is attempted elsewhere? And the MS needs to do a better job comparing or at least contextualizing the results with other recent attempts to use outcrops to guide fracture assessment for geothermal applications. And some parts of the latter part of the Discussion could be condensed as currently written they do not seem to be well linked to the results.
I appreciate the attempt in the MS title to grab the readers’ attention. But the implicit claim of the title: ‘A Fracture Never Comes Alone: Associations of Fractures and Stylolites…’ is not sustainable. In the literature there are many fractures described that are not associated with stylolites. I don’t think that the 1988 fracture review by Pollard and Aydin even mentions stylolites. Plenty of fractures ‘come alone’. In many cases where stylolites are present, they are bed-parallel structures that are not necessarily associated with fracture formation and that do not have usable kinematic significance for fracture analysis. The association of fractures are stylolites is not unique to this example, but the pattern is also not universal. The kind of stylolite appealed to here (MS fig. 1), effectively a kind of widely spaced disjunctive cleavage, is a feature of certain carbonate rocks (and rarely, sandstones) and settings, but the MS (and MS title) never makes this much narrower scope clear. The title of the MS ought to be revised to reflect this. For stylolites, in addition to the Hancock review, the authors should consider the classic paper on the topic by Marshak and Engelder (Marshak, S., & Engelder, T. 1985. Development of cleavage in limestones of a fold-thrust belt in eastern New York. Journal of Structural Geology, 7(3-4), 345-359.) Posing the issue as one of kinematic analysis rather than in terms of stresses (or paleostresses) may have some advantages (e.g., Friedman, 1964; Groshong, 1988; Marrett and Peacock, 1999). Marrett, R., & Peacock, D. C. (1999). Strain and stress. Journal of Structural Geology, 21(8-9), 1057-1063.
The MS’s generalization of fracture and stylolite assemblages from figure 1 (from Hancock 1985) also needs to be presented with more nuance. Although all the structures shown in the figure can form together, the literature suggests that in many cases only one or some of the structures are present (see regional studies by Engelder; and the recent geothermal-related outcrop study by Elliott et al.). And even if all the elements are present, they may not be contemporaneous or even related at all. Such an assemblage relationship needs to be demonstrated, not assumed. In other words, opening-mode fractures do not necessarily bisect arrays of small faults or en echelon fracture arrays (or transect stylolites). I think the authors appreciate this, but the point isn’t clear from the text.
Specific comments
The Introduction could also do a better job of leading up to the claims. If I have the claims right, these steps should be:
With each bullet point equal to a paragraph with background citations.
The focus on needing to find evidence for ‘geomechanical drivers’ seems like it could be hard to support and is probably not needed. Showing why fractures formed is notoriously challenging and the evidence for doing so is probably lacking in this instance. Where is the precise fracture timing and depth information or the rock mechanical properties history? For the approach used in this MS to work, that kind of geomechanical argument is probably not needed. The authors use orientation and relative timing information from geometric relations to classify faults, opening-mode fractures, and stylolites into kinematically compatible groupings and show that over a wide area these groupings have consistent patterns. The regionally consistent patterns can then be used to improve interpretation of fractures visible on image logs in some wells.
The writing could use work to make the text clearer. I’ve marked some of these concerns below keyed to lines in the text. The use of terms, particularly in the Introduction, is confusing and the text there and elsewhere ought to be rationalized for clarity. I don’t think there is a meaningful difference between ‘fractures’ and ‘fracture sets’ and ‘discontinuities’ and ‘discontinuity sets’ except that the latter two are less widely used. And, on line 86, the text finally mentions that stylolites are also being measured. The paper title uses ‘fracture’ and nothing would be lost by sticking with this and related terms throughout, or ‘fractures and stylolites’.
Comments keyed to lines in the text
1-8 (Abstract) The text here is all correct but it seems out of place in an Abstract, which I think ought to get to the findings more directly. This text seems more suitable to the Introduction or the start of the Discussion. Consider condensing or moving this text.
The Abstract could start with the text in line 7 (with edits): “We present a method that uses associations of fractures and stylolites, which we call discontinuity sets, to link outcrop and subsurface structures. Discontinuity sets are associations of kinematically compatible structures—faults, opening-mode fractures, and stylolites—that can form broadly contemporaneously. Relative timing can be obtained from crossing and abutting relations. Although such associations are commonly described in outcrop fracture studies they are rarely used to link outcrop observations to structures in geothermal targets or to help guide classification of sparse structural observations made using image logs. We use the orientations and type of discontinuity associations as indicators to map out principal paleostress trajectories of regional discontinuity-forming events that created a background discontinuity network…’ (See the comments on the Introduction structure above).
8 ‘robust’? It seems like a plausible link, but what do you mean by robust? Maybe you could claim robustness if you had independent timing information. Seems overstated.
21-22 This line struck me as sounding a bit circular: ‘[fractures] control…in fractured reservoirs…’ It’s also potentially confusing and convoluted since most of the ‘discontinuities’ are fractures but this isn’t stated. Can this line be revised to be more straightforward? Also, it might be worthwhile to mention that not all geothermal reservoirs are fractured.
22 Berre et al. and Medici et al. both talk about ‘fracture networks’. So why call these ‘discontinuity networks’? These are also review papers about modeling, where the features conducting fluid are assumed to be open fractures. That seems to differ from what you are dealing with: partly or fully sealed fractures (veins) and stylolites.
24 These La Bruna et al. papers are great (only one of them is in the reference list). But they are about outcrop fractures or fracture attributes that might influence flow rather than, as the sentence implies, being about studies that demonstrate with evidence such as production data that fractures actually influence flow. Many papers describe features that might influence flow, so it’s not clear why these two papers would be singled out (an e.g., at least is needed). But what you want to support the statement is one of the papers that uses well data to make this point. The number of papers that demonstrate that some aspect of subsurface fractures influence flow is pretty small, owing to the problems of characterizing subsurface fracture arrays. A paper that uses production data evidence wrt fractures is Solano et al. 2011 SPE Res Eval Eng. I suggest that the line and referencing be modified to reflect this. One example in the literature that links a specific fracture attribute to flow response is open versus sealed fractures (e.g. Weisenberger et al. 2019 Petroleum Geoscience). So if your fractures are fully or partly sealed this ought to be addressed in the Discussion.
28 For definitions of fracture sets see the review by Hancock 1985, J. Struct. Geol. Another aspect of sets is ‘relative timing’. Why omit it here?
29 I suggest that you tone down the geomechanical aspect here as unneeded. All you need to know, or assume, is that the structures are kinematically compatible and broadly contemporaneous. That’s what figure 1 shows. You use the relative timing between structures, from crossing and abutting relations, and their orientation with respect to tilted beds to group structures. The claim here is that fractures ought to be separated by ‘geomechanical driver’ and although this approach has precedent going back at least to Nelson’s 1985 book using a mechanism or ‘driver’ is a problematic way to classify fractures since the cause of fractures is notoriously hard to specify. Fold- and fault-related fractures and regional fractures have been recognized in the literature since at least the 1950s (as call outs to the literature ought to reflect) but unless you already know what the distribution and timing of fractures is how does an appeal to a ‘geomechanical driver’ help? If you are looking at a fracture in core (or a trace on an image log) you probably will not be able to accurately classify the fracture as ‘fold related’ or ‘regional’. See the discussion of equifinality in Revs. Geophys. 2019. Maybe the driver material belongs in the Discussion.
The entire paragraph from 27 to 34 seems out of place.
Background or regional fractures are not necessarily more evenly or uniformly distributed than other types of fractures. The literature has excellent examples of clustered fractures within regional sets. See the 2018 J. Struct. Geol. theme issue on spatial arrangement for examples.
30 ‘regional’ fractures have been recognized in the literature at least as far back as Balk, 1936. Balk, R. (1936). Structure elements of domes. AAPG Bulletin, 20(1), 51-67. And there are studies that identify regional fracture patterns in outcrop and compare them to sparse core observations.
31 ‘the’ background set. This seems to imply that there might just be one regional set. But regional studies (like papers be Engelder from outcrops in NY) document multiple regional sets.
40 The fracture sampling issue needs to be mentioned. Part of this concerns gaps in fracture observations that are inevitable when using wellbores to sample dispersed features like fractures and another is the problem of putting the sparse fracture samples into broader context: in other words, how easy is it, for example, to specify that a trace on an image log corresponds to certain features seen in outcrop? Part of this latter issue is how similar fractures look that formed by different processes (a situation called ‘equifinality’ where these issues are extensively discussed in a recent review: Laubach et al., 2019, Reviews of Geophysics). Since this MS proposes a solution to this issue by isolating specific kinds of kinematically meaningful relationships from the outcrop and using those geometric and relative timing inferences to guide image log interpretation, it would strengthen the argument to describe this sampling issue explicitly.
46-53 This paragraph struck me as vague and having a mixed message. The previous paragraph established that wellbore data has limitations. If you have fracture/stylolite relations in core or visible on image logs, that tells you something about the structures in the subsurface that would not obviously be improved by seeing that relationship in a distant outcrop. A useful thing about outcrops is being able to see features that can never be directly observed in the subsurface, like length or connectivity, which by their nature cannot be captured by wellbore probes.
And the referencing could be more extensive. There have been several studies that specifically address the issue of how to compare outcrop fractures to the subsurface, including specifically for geothermal applications. Note them. Or cover the topic, with references, in the Discussion.
What do you mean by ‘analogy’ and there is more involved in a useful comparison that just similar rock types, age, and structural setting (including diagenesis/rock property history).
For one thing, outcrops by definition have different loading histories than rocks that are still in the subsurface. It’s well established that uplift and unloading commonly do produce fractures (e.g., Engelder, 1985; English, 2012) as do a wide range of near subsurface and geomorphic processes (e.g., Eppes et al., 2024, Earth Surface Dynamics 12, 35-66. https://doi.org/10.5194/esurf-12-35-2024) so these differences may not be trivial. The first step in outcrop fracture studies aimed at guidance for the subsurface is usually trying to identify these.
I suggest you provide a broader assessment of how exposed rocks are judged to be appropriate analogs for the subsurface target (see papers by Agosta et al., 2010; Sanderson, 2016; Ukar et al., 2019). Possibly in the Discussion. A range of factors go into selecting a good analog for a subsurface geothermal target, including matching rock types and—broadly—structural history (Bauer et al., 2017; Busch et al., 2022, Peacock et al., 2022; Elliott et al., 2024). Some studies have questioned the viability of using outcrops for making specific predictions about key subsurface parameters l (Peacock et al., 2022) whereas others claim that such assessments are possible in some instances (Elliott et al., 2024). Since what you are doing is a contribution to solving this problem, the Discussion is a good place to contextualize your work. Many of the other approaches such as using chemical aspects of the fracture system (e.g. Elliott et al. 2025) seem like they would be a good compliment to your approach.
55 Genetic relations between fractures and stylolites have long been appreciated. See references in Groshong (1975). And that multiple fracture orientations can form in a single deformation goes back at least to Stearns. See also: Olson, J. E., 2007, Fracture aperture, length and pattern geometry development under biaxial loading: a numerical study with applications to natural, cross-jointed systems. In Couples, G & Lewis, H., eds., Fracture-Like Damage and Localization, Geological Society of London, Special Publication. 289, 123-142.
Groshong Jr, R. H. (1975). Strain, fractures, and pressure solution in natural single-layer folds. Geological Society of America Bulletin, 86(10), 1363-1376.
68 ‘carbonate rocks’; just saying carbonates sounds slangy.
80 (Methods section) This section is confusing. The second part of it (2.2) seems like it belongs in the Discussion. In 2.2 you are making the case that your outcrop data can be linked to the subsurface. This is an interpretation, not a method. The point is best addressed in the Discussion.
The first part of the Methods (2.1) also needs to be clarified. Mixed in here are incomplete descriptions of the outcrop sizes and what can be measured in them, data collection methods like circular scanlines that may have only been collected at one outcrop (line 100), a distribution of 10×10 m outcrop stations that is supposed to be “…as evenly as possible over the studied area”, and a method for selectively extracting kinematically significant fracture/stylolite relations. These elements need to be separated out and described clearly and quantitatively. Some aspects, like outcrop sizes, maybe ought to be in the Geological Setting. A useful approach would be to build a table and use that as a guide to revising the section. The Methods section also should mention that you had access to and described two wells (line 220).
My suggestions above for the Introduction are based in part on the impression I had from this section that the outcrops you had to work with are small, and not amenable to the type of analysis of large clean outcrops as for example in Elliott et al. 2025.
81 How do you use this approach if you don’t have independent measures of the (paleo)stress directions? This is a problem with emphasizing the stress or paleostress aspects. You just needed to observe kinematically significant structural relations, like orientation and crosscutting or abutting relations and map the patterns regionally. In the 1990s a similar approach was used to map coal fracture patterns regionally in outcrop and guide interpretation of core, image log, and production data from coalbed methane wells (see this review paper: 1998, Characteristics and origins of coal cleat: a review: International Journal of Coal Geology, 35, 175-207, their figures 1, 15, 16, and 19).
81-90 Much of this material seems like it belongs in the Discussion.
81 ‘…mode I and mode II fractures, vein arrays…’ I suggest that you rethink your terminology here. The mode terminology seems to add unnecessary jargon. A simpler descriptive terminology appropriate to natural examples is to just call these features ‘opening-mode fractures and faults’. Besides, the mode I, II etc terminology refers to where you know the crack mouth opening displacement, which is why these terms are typically found in experimental or theoretical treatments where this aspect can be observed or specified. According to Pollard and Aydin: “Broadly speaking, joints are associated with the opening whereas faults are associated with the shearing modes. Because the mode may vary along the fracture front and may involve mixtures of modes I, II, and III, however, one should not be too categorical about these associations.” The terms joint and vein have connotations about mineral deposits that are unhelpful (see Rev. of Geophys. 2019). Veins can form in several ways (they can be filled opening mode fractures or dilatant parts of faults, in addition to some being replacement deposits), but mixing this term that relates to mineral deposits with the mode terminology is confusing. Why not just say ‘mineral deposits in the fractures’ if that is what you mean? Both opening-mode fractures and faults commonly contain mineral deposits.
83 “Discontinuity sets are defined on the basis of both orientation and discontinuity type.” I believe that you said this already (line 28). In any case, relative timing is also typically a component of defining sets.
86 Here is the first indication that your ‘discontinuities’ include stylolites. That you are using stylolites ought to be mentioned earlier. And why not just say throughout ‘fractures and stylolites’ instead of the awkward ‘discontinuities’?
90-100 The size and degree of exposure of the outcrops is hard to parse from this description. You mention stations that are 10x10 m but in line 101 you seem to use circular scanlines with radius 1 m and say that only one outcrop had ‘quality pavements’ to allow circular scanlines. Describe what the outcrops are like, probably in the last part of the Geological Setting.
99 ‘seven’ (small number convention). Check the MS throughout.
106 (section) The argument that the data you collected can be used to link the outcrop and the subsurface belongs in the Discussion (and may occur in the claims at the end of the Introduction).
There are other studies in the literature that have the goal of identifying outcrop analog fractures that can be used as guides to geothermal reservoir extrapolation. Some of these provide different perspectives on the issue and ought to be mentioned in the Discussion to give balance to your conclusions: Elliott, S.J., Forstner, S.R., Wang, Q., Corrêa, R., Shakiba, M., Fulcher, S.A., Hebel, N.J., Lee, B.T., Tirmizi, S.T., Hooker, J.N., Fall, A., Olson, J.E., Laubach, S.E., 2025. Diagenesis is key to unlocking outcrop fracture data suitable for quantitative extrapolation to geothermal targets. Frontiers in Earth Science 13, 1545052.
131 ‘excellent exposures’ is vague. How big, how complete is the exposure? Are fractures that formed in the subsurface readily separated from surface-related fractures here? How?
140-146 (In the Geological Setting) It would be useful to mention, even if qualitatively, how the structural and burial history or outcrops and rocks in the subsurface differ. Also mention the current state of stress/ stress regime (could cite world stress map papers). In some areas surface fractures relate to current stresses (see the pop ups described by Engelder in the 1980s; references in Elliott et al. 2025).
149 (In the Results) It might be helpful to start by describing the structural elements that are present in the entire area.
156 Note and consider the strong condemnation of the term ‘shear fracture’ in the Pollard and Aydin 1988 GSA Bulletin review. Maybe ‘small displacement faults’?
168 Consider adding a star or other mark to the stratigraphic column to show which unit is being analyzed.
170 Dissolution along the fractures. Does this play a part in the interpretation? This may be of interest to readers concerned with some of the deep carbonate fractured reservoirs in China, where this kind of dissolution is a key element. Is there any evidence of this process in outcrop? This seems like it could be part of your Discussion.
198 ‘is composed of’ but ‘comprises’. You use this weird English convention correctly in line 190.
224 (figure 6) Nice way to do the scales on these images.
225 These wells need to be anticipated in the Introduction and Methods.
232 Help the reader understand the Doesberg 2023 reference (an unpublished MS thesis). Did you do image log interpretation or just use some kind of compilation from this reference? Line 236 makes it seem like you interpreted the images. You might be interested in how Wang et al. 2023 handled references to reinterpreted archival image log data: Wang, Q., Narr, W., Laubach, S.E., 2023. Quantitative characterization of fracture spatial arrangement and intensity in a reservoir anticline using horizontal wellbore image logs and an outcrop analog. Marine & Petroleum Geology 152, 106238. https://doi.org/10.1016/j.marpetgeo.2023.106238
239 How can you know that veins are ‘invisible’ if you don’t have core? Filled fractures do commonly show up on image logs.
239 ‘feat’ > feature?
250-252 Hmm. What if these picks are wrong? Is this discussed further?
281-285 This is confusing.
285 ‘On the contrary’ > ‘in contrast’
303 I assume that by ‘the only way’ you mean given the type of data that has been collected to date? Maybe instead ‘a practical, widely used, and relatively inexpensive way’? But one with several important drawbacks.
304 Maybe start the line with ‘In the subsurface of the Geneva basin…’ to make it clear that this is a location specific issue.
307-321 In 307 you say that image log bias has rarely been investigated, but this isn’t really the case, although I guess it depends on what you mean by ‘bias’. There have been many studies of the capabilities and limitations of image logs. Bias is a systematic distortion of a result due to some factor. Unless you mean the bias of a specific analyst, the problem is one of inherent ambiguity rather than bias. The kind of reproducible rules, such as in the Andrews et al. reference, are good. But excellent discrimination rules were worked out in the 1990s based on wells with both image logs and core; these are the basis for commercial log picks. There have been many core-to-log comparisons published since 1988 and they mostly come to the same sad conclusion that there is a lot of inherent ambiguity in this aspect of image log interpretation. The reason for this is that many features on image logs look alike. Drilling induced fractures may not have the characteristic shapes and distributions that would allow rules to reliably differentiate them, mineral deposits in natural fractures can be microns thin and undetectable on image logs, and in some case in core inspection. Or fill in sealed fractures can be eroded out. Open natural fractures are not necessarily aligned with current day SHmax. The problem of correctly differentiating drilling and natural fractures or open and sealed fractures has been the focus of several studies since the late 1980s. This section of text can probably be reduced to a short paragraph.
The point I guess is that the image logs are widely used but have mostly intractable limitations, so the kind of outcrop inferences and guidance for log interpretation you provide can be helpful in trying to get reliable data from the logs. Your discussion ought to talk about how general your guidance might be or is it specific to this unit or rock type in this basin.
316-320 This section of text describes an important contribution of this MS. But the message seems buried. A clearer description is needed.
327 ‘barren’ and ‘mode I’ are not equivalent things. And image logs cannot tell if a fracture is barren or not. The mineral deposit veneers on some natural fractures are microns thin and require an SEM to detect, so they (and even thicker deposits) are invisible to current image log technology.
A point that I don’t see considered is that the outcrop images you show seem to be mostly sealed fractures. Are these fractures filled with calcite deposits (the Results ought to describe this). If the fractures (or at least some of them) in outcrop are calcite filled, that at least is some evidence they are not near surface features but are representative of subsurface deformation. Do you mention this? And if they are sealed, how do they contribute to fluid flow? Or show up as open on image logs? If sealed, is their main role as weaknesses for reactivation during stimulation (Cao et al. point to this as a major uncertainty)? Earlier in the text you mention dissolution along fractures in this basin. Is this an issue worth discussing?
323-341 I agree with the points here, but this section of text could use some work for clarity.
340 (section of Discussion). I think this section ought to be condensed such that it focuses of issues you cover in your Results.
360 If you include the effects of fracture abundance in your Results or geological background you should describe what porosity and permeability the host rock has. If host-rock permeability is appreciable then closely spaced fractures (if open) could affect overall permeability due to flow through the host rock between fractures (Philip et al. 2005, SPE Res. Eval. Eng.) If the host rock is impermeable, but the open fractures are not interconnected then the closeness of the fractures to each other should matter. There is a large literature on connectivity and flow (e.g. Long and Witherspoon 1985). Connectivity is not necessarily a function of fracture abundance. But you don’t describe connectivity in your outcrop description. Maybe the best move is to make this entire section much shorter and just say that once you have established that the outcrops are representative of the subsurface with your outcrop to image log comparison, you could go back to the outcrops to get this other information that would be useful for modeling.
363 You mention ‘saturation’ without putting this concept into context. Maybe best to just leave it out. Where in your Results is there evidence one way or the other to argue for some degree of saturation?
396 The conclusion “Outcrop study is a time and cost-efficient method to obtain a first-order evaluation of the contribution of the background network in the subsurface”. I’m sure that this is a true statement. But you have not done a time or cost analysis or a value of information assessment, so I question whether this is a valid conclusion. Maybe the remark belongs at the end of the Discussion along with some ballpark estimates of costs and time of field data acquisition and the potential value of improved image log interpretation. For an example of this and a spreadsheet that can be used to make your calculation, see: Almansour et al. 2020. Value of Information analysis of a fracture prediction method. SPE Reservoir Evaluation & Engineering, 23 (3), 811-823. doi: 10.2118/198906-PA.
Check the figure captions for the word ‘legenda’; should be ‘legend’.
The titles in the reference list are formatted inconsistently.
Some reference mentioned in the review
Elliott, S.J., Forstner, S.R., Wang, Q., Corrêa, R., Shakiba, M., Fulcher, S.A., Hebel, N.J., Lee, B.T., Tirmizi, S.T., Hooker, J.N., Fall, A., Olson, J.E., Laubach, S.E., 2025. Diagenesis is key to unlocking outcrop fracture data suitable for quantitative extrapolation to geothermal targets. Frontiers in Earth Science 13, 1545052.
Rysak, B.R., Gale, J.F.W., Laubach, S.E., Ferrill, D.A., Olson, J.E., 2022. Mechanisms for the generation of complex fracture networks: observations from slant core, analog models, and outcrop. Frontiers in Earth Science, v. 10, Section Geohazards and Georisks. In Li, Y, Rutter, E.H., Shang, J. and Ji, Y., Eds., Special Issue, Recent Advances in Mechanics and Physics of Rock Fractures across Scales. doi.org/10.3389/feart.2022.848012