the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
3D geological modelling of igneous intrusions in LoopStructural v1.5.10
Fernanda Alvarado-Neves
Laurent Ailleres
Lachlan Grose
Alexander R. Cruden
Robin Armit
Abstract. Over the last two decades, there have been significant advances to improve the 3D modelling of geological structures by incorporating geological knowledge into the model algorithms. These methods take advantage of different structural data types and do not require manual processing, making them robust and objective. Igneous intrusions have received little attention in 3D modelling workflows, and there is no current method that ensures the reproduction of intrusion shapes comparable to those mapped in the field or in geophysical imagery. Intrusions are usually partly or totally covered, making the generation of realistic 3D models challenging without the modeller's intervention. In this contribution, we present a method to 3D model igneous intrusions considering geometric constraints consistent with emplacement mechanisms. Contact data and inflation and propagation direction are used to constrain the geometry of the intrusion. Conceptual models of the intrusion contact are fitted to the data, providing a characterisation of the intrusion thickness and width. The method is tested in synthetic and real-world case studies and the results indicate that the method can reproduce expected geometries without manual processing and with restricted datasets. A comparison with Radial Basis Function (RBF) interpolation shows that our method can better reproduce complex geometries such as saucer-shaped sill complexes.
- Preprint
(2528 KB) - Metadata XML
- BibTeX
- EndNote
Fernanda Alvarado-Neves et al.
Status: open (until 03 Nov 2023)
-
RC1: 'Comment on gmd-2023-85', Anonymous Referee #1, 28 Aug 2023
reply
This manuscript presents an approach of modeling intrusions more realistically and based on geological conceptual understanding. It is compared with an industry standard, namely radial basis function interpolation. I very much agree that the RBF standard is lacking and creates unrealistic shapes. In reality therefore, people often add fake control points and use this as a hack to create realistic shapes. Hence, there is definitely room for doing something better.
The major problem I have with the manuscript is the conditioning to borehole data. The method proposed, partly based other papers, is not a valid approach. Since the definition of the model is an implicit one, and since the model is now a parametric model, a Bayesian approach is needed, with a prior and a likelihood function/distribution. All other methods will create conditioning artifacts. This is true for all object-based methods (see REF A,B), and is a disadvantage to all object (and implicit) methods. The proposed conditioning using sequential Gaussian simulation of a signed distance function is not a valid approach because the variogram of the signed distance function is not defined, nor inferred anywhere in the manuscript. If the signed distance constraints are too large, then the Gaussian model perturbation implied will make the model shape deviate from the intended shape, hence defeating the purpose. Also the signed distance function is likely not a Gaussian function, unless, maybe, the signed distance is very small, meaning that the model already constrains well to the data (the easy problem). This is not a realistic assumption. Additionally, the actual true signed distance is not observed in the borehole data, only the pseudo-signed distance, the latter is not a valid random function (RF), because it depends on the borehole location (see Ref C). That means one cannot use SGS (a RF approach) to do conditioning.
The examples are also not convincing because they appear all to cover the geological shape very well. More rigorous and more challenging test need to be run such as for example, much less data (e.g. 5 boreholes). In exploration/resource definition, this is likely the case, namely, only very few boreholes exist and much more uncertainty around the location of the intrusion is expected. In that case, uncertainty quantification is needed, for example to assess planning of new boreholes. Because of the invalid conditioning, the method will therefore also not properly create uncertainty models.
My recommendation therefore is between major revision or return the manuscript for a longer effort and more in depth effort on conditioning.
Ref A Holden L, Hauge R, Skare Ø, Skorstad A (1998) Modeling of fluvial reservoirs with object models.
Math Geol 30:473–496.Ref B Skorstad A, Hauge R, Holden L (1999) Well conditioning in a fluvial reservoir model. Math Geol
31:857–872.Ref C Fouedjio F, Scheidt C, Yang L, Wang Y, Caers J. Conditional simulation of categorical spatial variables using Gibbs sampling of a truncated multivariate normal distribution subject to linear inequality constraints. Stochastic Environmental Research and Risk Assessment. 2021 Feb;35:457-80.
Citation: https://doi.org/10.5194/gmd-2023-85-RC1
Fernanda Alvarado-Neves et al.
Data sets
Case studies 1,2, 3 and 4 datasets and Jupyter notebooks Fernanda Alvarado-Neves, Laurent Ailleres, Lachlan Grose, Sandy Cruden, Robin Armit https://doi.org/10.5281/zenodo.8189191
Model code and software
LoopStructural Lachlan Grose, Laurent Ailleres, Gautier Laurent, Roy Thomson, Yohan de Rose https://zenodo.org/record/7734926
Fernanda Alvarado-Neves et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 212 | 47 | 7 | 266 | 4 | 3 |
- HTML: 212
- PDF: 47
- XML: 7
- Total: 266
- BibTeX: 4
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1