A 3D stochastic stratigraphic correlation method based on forward stratigraphic models.

Jonathan Edwards and Florent Lallier and Guillaume Caumon and Cedric Carpentier. ( 2015 )
in: 35th Gocad Meeting - 2015 RING Meeting, ASGA

Abstract

A stratigraphic model is a representation of the architecture of the stratigraphic succession of an area. It corresponds to the geometry of the stratigraphic units (3D extension, thickness variation, geometry of bounding horizons) and to their depositional relationship (the topological relationship between them: onlap, baselap, toplap, conformable, eroded). The construction of a stratigraphic model from observation data relies on two steps: first the construction of the stratigraphic correlation which is the association, in 3D, of the stratigraphic units identified along wells and the definition of their depositional relationship; second the construction of the geometry of the horizon bounding the stratigraphic units. In this article we present an algorithm that generates, in a stochastic manner, 3D stratigraphic correlation models of units identified along wells. The main improvement, as compared to previously published methods, is that the depositional relationship between stratigraphic units is now an output of the method. oreover, the presented method integrates a 3D approach while generating stratigraphic models. Wells are not correlated one to another as it was generally done in the literature but by adding a well to a set of wells for which the stratigraphic correlation has already been built. As a result, the 3D structure of stratigraphic units is taken into account while generating the stratigraphic correlation model. A prerequisite of the proposed approach is to be able to evaluate the probability of a given stratigraphic unit to be correlated to a set of stratigraphic units already correlated. We show that this probability can be translated into a multiple point statistics problem. We thus propose to infer this probability from a forward stratigraphic model generated using process based method and representing the depositional history of the studied area.

Download / Links

BibTeX Reference

@INPROCEEDINGS{EdwardsGM2015,
    author = { Edwards, Jonathan and Lallier, Florent and Caumon, Guillaume and Carpentier, Cedric },
     title = { A 3D stochastic stratigraphic correlation method based on forward stratigraphic models. },
 booktitle = { 35th Gocad Meeting - 2015 RING Meeting },
      year = { 2015 },
 publisher = { ASGA },
  abstract = { A stratigraphic model is a representation of the architecture of the stratigraphic succession of an area. It corresponds to the geometry of the stratigraphic units (3D extension, thickness variation, geometry of bounding horizons) and to their depositional relationship (the topological relationship between them: onlap, baselap, toplap, conformable, eroded). The construction of a stratigraphic model from observation data relies on two steps: first the construction of the stratigraphic correlation which is the association, in 3D, of the stratigraphic units identified along wells and the definition of their depositional relationship; second the construction of the geometry of the horizon bounding the stratigraphic units. In this article we present an algorithm that generates, in a stochastic manner, 3D stratigraphic correlation models of units identified along wells. The main improvement, as compared to previously published methods, is that the depositional relationship between stratigraphic units is now an output of the method. oreover, the presented method integrates a 3D approach while generating stratigraphic models. Wells are not correlated one to another as it was generally done in the literature but by adding a well to a set of wells for which the stratigraphic correlation has already been built. As a result, the 3D structure of stratigraphic units is taken into account while generating the stratigraphic correlation model. A prerequisite of the proposed approach is to be able to evaluate the probability of a given stratigraphic unit to be correlated to a set of stratigraphic units already correlated. We show that this probability can be translated into a multiple point statistics problem. We thus propose to infer this probability from a forward stratigraphic model generated using process based method and representing the depositional history of the studied area. }
}