Stochastic association of fault evidences using graph theory and geological rules Proposed theoretical framework Structural interpretation input

in: 2017 Ring Meeting, pages 1--20, ASGA

Abstract

We propose a graph-based framework to describe the problem of fault network interpretation from spatial fault evidences. This framework allows us to discuss the very large uncertainties that interpreters generally face. It also leads us to propose a method to sample structural uncertain- ties by producing several fault network interpretations that honor observations and validity rules. Our methodology relies on graph-based algorithms to generate clusters of fault evidences which may belong to the same fault. A clique partitioning approach is used for automatically inferring associations based on user-defined geometric and kinematic geological rules. This strategy is hierar- chical as it starts by detecting the major structures, as commonly done in structural interpretation. The obtained associations can then be used to manually or automatically construct 3D structural models. We validate our methodology by testing fault associations generated from synthetic sparse data extracted from a reference model. We also apply this methodology to sample uncertainties on the association of fault sticks picked on a set of 2D cross-sections imaging a faulted basement offshore Morocco.

Download / Links

BibTeX Reference

@INPROCEEDINGS{Godefroy2017,
    author = { Godefroy, Gabriel and Bonneau, Francois and Caumon, Guillaume and Laurent, Gautier },
     title = { Stochastic association of fault evidences using graph theory and geological rules Proposed theoretical framework Structural interpretation input },
 booktitle = { 2017 Ring Meeting },
      year = { 2017 },
     pages = { 1--20 },
 publisher = { ASGA },
  abstract = { We propose a graph-based framework to describe the problem of fault network interpretation from spatial fault evidences. This framework allows us to discuss the very large uncertainties that interpreters generally face. It also leads us to propose a method to sample structural uncertain- ties by producing several fault network interpretations that honor observations and validity rules. Our methodology relies on graph-based algorithms to generate clusters of fault evidences which may belong to the same fault. A clique partitioning approach is used for automatically inferring associations based on user-defined geometric and kinematic geological rules. This strategy is hierar- chical as it starts by detecting the major structures, as commonly done in structural interpretation. The obtained associations can then be used to manually or automatically construct 3D structural models. We validate our methodology by testing fault associations generated from synthetic sparse data extracted from a reference model. We also apply this methodology to sample uncertainties on the association of fault sticks picked on a set of 2D cross-sections imaging a faulted basement offshore Morocco. }
}