Automatic facies mapping on DOMs and images : comparisons of different approaches

Sophie Viseur. ( 2017 )
in: 2017 Ring Meeting, pages 150342, ASGA

Abstract

Numerical outcrops often serve as support for interpreting facies. The facies may be expressed by a change of the outcrop geometry, its colour or other physical properties such as the LIDAR intensity. The differences are highlighted by a sudden change of values or patterns, or both. Several approaches have been proposed to extract strata limits or fracture traces and planes. However, few approaches were devoted to partition outcrops into facies maps. In this paper, statistical approaches are proposed to extract in an automatic or a semi-automatic way the facies encoding from the numerical outcrop data. The proposed approach relies on comput- ing statistical differences between DOM properties (geometry, colours, etc.). They may be applied onto different kinds of outcrop data: pointsets or triangulated surfaces textured with photos. Using different case studies, the results of these approaches are shown and discussed in terms of performance, accuracy and use. ∗Speaker

Download / Links

BibTeX Reference

@INPROCEEDINGS{Viseur2017a,
    author = { Viseur, Sophie },
     title = { Automatic facies mapping on DOMs and images : comparisons of different approaches },
 booktitle = { 2017 Ring Meeting },
      year = { 2017 },
     pages = { 150342 },
 publisher = { ASGA },
  abstract = { Numerical outcrops often serve as support for interpreting facies. The facies may be expressed by a change of the outcrop geometry, its colour or other physical properties such as the LIDAR intensity. The differences are highlighted by a sudden change of values or patterns, or both. Several approaches have been proposed to extract strata limits or fracture traces and planes. However, few approaches were devoted to partition outcrops into facies maps. In this paper, statistical approaches are proposed to extract in an automatic or a semi-automatic way the facies encoding from the numerical outcrop data. The proposed approach relies on comput- ing statistical differences between DOM properties (geometry, colours, etc.). They may be applied onto different kinds of outcrop data: pointsets or triangulated surfaces textured with photos. Using different case studies, the results of these approaches are shown and discussed in terms of performance, accuracy and use. ∗Speaker }
}