Dear editor, authors, 

I have reviewed the manuscript "Interpretation and Representation in Geomodels: The POKIMON Ontology for Formalizing Geomodelling Knowledge" and found it to be an interesting and needed contribution addressing the (challenging!) task of capturing the logic, interpretation and decisions behind 3D structural geological models. Hence, I suggest that it will be of interest to readers of GMD, and could be accepted for publication following some clarification and revision, as outlined below.



Major comments

1) The introduction provides a clear introduction to the challenges of reproducibility and subjectivity in geomodelling, and an overview of related prior-work developing geoscience ontologies. But, it does not really explain why ontologies are necessary in the first place ("The use of formal languages and ontologies is growing in the geosciences", line 50, is not a justification). I suggest that it would be very useful to add one or two short paragraphs outlining the common uses of geoscience ontologies and why (in broad terms) they are necessary for geoscience modelling / data management, and how they are / might be implemented in geological modelling / interpetation workflows.

2) Figures 8 to 14 are crucial results for the paper, as they illustrate the application of the POKIMON ontology to different modelling scenarios. However, I also found them to be very cryptic (at least to the uninitiated). Please explain these figures in more depth and, if possible, include geological cartoons or cross-sections highlighting the use-case and various geomodels that are being represented. Also consider colouring the edges in the graph by predicate, to help highlight different types of relation. Finally, I found the text in these figures too small to read; please make it larger (there seems to be space).



3) I found the order of subsections in Section 5 to be a little confusing; consider re-arranging to something like:

        5.1. Models

        5.2. Data, norms and constraints

        5.3. Simplification [ this is a very important topic and could be expanded on a little ]

        5.4. Algorithms and processes [ i.e. how I turn my simplified data/constraints into a model ]

        5.5. Model characteristics, content and origins [ i.e. how POKEMON describes the resulting models and their creation ]

4) It would be helpful to provide a few more "worked examples" of how POKEMON can describe specific geological logic. E.g., the simplification of a fault zone to a single fault surface, or common assumptions like dip-slip displacement, reasoning based on geological norms etc. These examples will greatly help the reader see how POKEMON can be applied, and explain how Figs. 8 - 14 were obtained and can be intepreted.

5) The "use-case" examples used in this work are all quite small-scale and geologically similar: What are the implications of this? Might larger-scale models (e.g., of the Permian Basin, or Jura mountains) have different ontological requirements? And how about very different geological settings (e.g., volcanic systems, metamorphic terranes, ...)?

Please justify why all three examples are so similar (quarry scale, lightly deformed sedimentary rocks).

Specific comments

Title: Why all-caps?

Line 11: "geoscience models" - to me this would include geodynamic simulations, weather models, etc. Is this intended? The rest of the paper seems exclusively relevant to 3D structural geological models.

Line 13: I would disagree with this statement -- most geological maps / models are accompanied by reports that explain in depth the assumptions, reasoning steps and uncertainties in a model.

Line 17: typo - "geomdelling"

Line 21: Why use "geological models" now and not "geoscience models" like in the abstract - please be consistent. I would suggest the very specific term "3D structural geological model (SGM)", to also avoid potential confusion with e.g., numerical simulations. 

Line 50: The number of ontologies is certainly growing - but are they actually being used? How / why / in what contexts?

Line 73: If we can define widely used and functional standards without a formal ontology, why do we need ontologies at all? ...I do think that ontologies are important - but this needs to be explained to the reader in this section.

Line 115: Consider adding a few paragraphs here on the simplification process - why is it necessary and how is it typically achieved. A formalised description of this simplification process seems (to me) to be one of the most valuable applications of POKEMON ("Gotta catch 'em all? Or just enough to get the job done").

Figure 1: This needs to be larger. And the text in 1D is unreadable.

Figure 2: This figure could be larger.

Line 130: It would be useful to include a short definition (table?) for each of the aspects defined here - they are not self-explanitory.

Line 231: Is "font" really that important? .... I guess you mean structure (in the information sense)?

Line 254: Conceptually / demonstratably "incorrect" models are also common? (and, can even still be useful). Please rephrase.

Section 5.1.1: Why are space and time lumped together? Isn't it common to represent only time (e.g., a chronostratigraphic chart/model) or only space (e.g., a 3D interpolation without explicit / known timing? [even if timing can often be partially inferred from the geometry ] ).

Figure 7: Consider including a somewhat more complex geological entity here - e.g., how would an axial-foliation defined by micas in Unit A and a spaced fracture cleavage in Unit B be represented? Or a weathering-related alteration that partially affects A & B in more-foliated regions near the fold hinge? Is the latter "conceptual" or "space-time"?

Figure 8: Please make the text larger.

Figure 9: Please add a (or several) geological sketch(s) to this figure, to help an unfamiliar reader interpret the graph and what it is representing. Please also make the text bigger (ditto all the other figures).



Line 296: Why is there a bullet point here (outside of a list)?



Line 299: Surely geomodels can also be (demonstratably) invalid interpretations? Rephrase.



Line 349: How/why is a machine-learning action different to e.g, kriging? Both are mathematical / statistical approximations of a function?



Figure 11: Include sub-figures showing the representation process and final DCE?



Line 410: Would common modelling assumptions fit here? E.g., the assumption that a fault is dip-slip rather than oblique, or that folds are cylindrical? Please clarify how such an assumptions would be represented and potentially include an example.



Line 415: Are "data" (e.g., bedding measurements) considered to be constraints in this framework? Please clarify, and if yes then clearly state this (as bedding data and contact points are by far the most common constraints used to build 3D structural geological models).



Line 425: Please expand this section to explain how POKIMON can be used to capture the reasoning/choices behind common geological constraints or inferences - this seems critical. E.g., how would the following be represented: "Observations at location A lead me to interpret lithology X. Observations at nearby location B lead me to interpret lithology Y, which should not contact X given the current stratigraphic understanding. Hence, I have inferred that a fault exists between locations A and B."



Line 438: "Inheres"  ?



Line 445: Commonly referred to as "lumping"?



Line 530: How does scale / "order of observation" fit here? Do the conceptual and space-time entities for a Fold exist at a specific scale (regional scale fold vs parasitic fold vs micro-fold)? And the real-world entity across all scales? Or are parasitic folds separate entities in the real-world, conceptually, and in space-time (in which case, where do we stop splitting and say "this is a single fold")? How does this multi-scale nature of geological concepts/objects link to simplification / Section 5.3.2.



Line 543: Pease expand if possible (a few paragraphs would be great) - how do you envisage POKIMON being integrated into routine 3D geological modelling workflows (to capture the model-building processes)? How might the resulting information be used? How might POKIMON help us better assess uncertainty? Outlining a "vision" here will really help the reader to understand why POKIMON is useful and relevant (rather than just complex and abstract).

Please find some additional minor suggestions in the attached annotated PDF.

Once these suggestions are incorporated I would be happy to recommend this contribution for publication. POKIMON seems like a useful starting point to begin addressing a very complex problem :-)

Kind regards,

Sam Thiele

This manuscript presents the POKIMON Ontology, a new ontology based on classical and geological ontologies with the aim to describe the 3D geological modelling knowledge in all the steps of the processing.

The paper is well structured and written. Its purpose is introduced highlighting some key concepts of 3D modelling, such as modelers’ expertise, the difference between reality and models, and the possibility to have different interpretations due to the purpose of the model.

With simple case studies, the authors managed to properly describe the logical process of the POKIMON Ontology in preserving the knowledge from data sampling to the model diffusion. While the conceptual framework is clearly presented, few aspects related to the formalization choices and their practical implications would benefit from further clarification.



Here I present some topics to be explored in more detail:



1) From BFO, the authors selected the Continuant and Occurrent classes. However, it is not entirely clear how occurrents, or specifically an earthquake (as suggested by the authors at line 158), can be represented within the model. While it may be possible to constrain geological events as an assumption in model interpretation, the manuscript does not explicitly clarify this integration.

2.1) The distinction between real world and models sometimes appears ambiguous: when geologists collect datas, they are following their mental conceptual geomodel (a valid interpretation of a situation) to represent the space-time geomodel, making these two to converge on each other. It is clear which is the aim of the part, but maybe it would be helpful to better explain it.

2.2) This is also related to the “user-based action specification” (line 336), because the user is effecting the input data and the results through a conceptual (but also space-time) model, defining norms and constraints.

2.3) The introduction of “assumed origin” entities (line 403) raises some conceptual questions. As these elements are grounded in background geological knowledge to support the construction of a synthetic conceptual model, it is unclear whether they should be regarded as observed and/or interpreted origins abstracted from their specific space–time constraints. Clarifying this aspect would help to better position these entities with respect to observed data and interpretative geological reasoning within the modelling workflow.





Technical corrections:

- Line 224: DCE is referred to Description Content Entity, but at line 230 it is referred to Depiction Content Entity.

- Line 284: “[…] an Independent Continuant that is __ collection of SCEs […]”; the article is missing.

- Line 314: the word “sepcification” is spelled incorrectly.

- Line 448: “[…] and thus has insignificant impact. .” two dots are presents.