3D restoration: Achievements and perspectives

Restoration techniques consist in suppressing post-sedimentary tectonic deformations by sequential flattening of topmost stratigraphic horizons and propagation of retro-deformation to underlying rock units. They aim at reducing uncertainties through structural model consistency checks.

Originally applied on cross-sections, restoration has been extended to surfaces and volumes. These approaches compute deformation either by preserving thickness, area and volume, or by minimizing volume variation or deformation energy. Whether based on geometric optimization or on the finite-element (FE) method, restoration offers new possibilities in quantitative structural modeling, but also raises practical problems.

One such new possibility is the use of restoration to determine unknown structural parameters using inverse theory. For instance, transverse fault displacement, which is very difficult to determine from subsurface data may be evaluated by the following procedure: the value of this throw component is iteratively perturbed about some reference configuration, then surface restoration and retro-deformation analysis are used to assess the likelihood of each possible configuration.

On the practical problem side, the rheological assumption (elastic behavior) on the materials in finite-element restoration may lead to wrong interpretations. Particularly, salt is known to be a visco-plastic material, and thus, is not implied in reversible processes. Consequently, it cannot be restored using classical techniques. We explore the possibility of a mixed approach for the restoration of salt diapir. This approach combines multi-surface restoration preserving layer thickness, and volume conservation of salt stocks.

Another problematic point in FE-based volumetric restoration lies in the conforming mesh generation. Indeed, although tetrahedral meshes can theoretically fit the complexity of all structural models, it is extremely difficult in practice to find a good compromise between the number of tetrahedra and the mesh quality, especially when dealing with unconformities and onlap geometries. Therefore, instead of using conforming tetrahedral meshes, we propose to embed triangulated horizons in a tetrahedral volume model, and to transfer boundary conditions from the horizons onto the nodes of the tetrahedral volume. This new procedure also allows for using an arbitrary surface as a target for the volume restoration. If a paleotopography is known for the model, it may then be used as the target surface.