Facies and fracture detection using Multiple-Point statistics.

@inproceedings{ViseurGM2013,
 abstract = { Numerical outcrop data, such as areal photographies, spectral images or Lidar point clouds, are increasingly used in geosciences as a support for geological feature mappings and this, at different scales of investigation. The interpretations of such observed structures may be tedious and time consuming. It is then interesting to develop methods for interpreting the geological structures in a semi-automated way from different and several kinds of data sources. The developed methods must be thus general enough to be applicable on several and different data types.
In this paper, it is proposed to use the Multiple-Point Statistics (MPS) technology as a semi-automated interpretor. First, a reference geological interpretation is defined on a small area of the data to interpret. This way, both the data values (image colours, Lidar intensity, etc.) and the interpreted structures (facies, fractures, etc.) are determined at each data point. Second, this reference model serve as training image with a set of secondary data (the image colours, lidar intensity) for a MPS application. Therefore, at each data point, the MPS method simulates the most probable facies or category (e.g. fracture versus matrix) in function of the geological patterns described in the training image and the secondary data (image colours, lidar intensity) available at each data point. This strategy has been applied on outcrop images, using different components of color space. The different results are compared each other as well as with available manual interpretations. },
 author = { Viseur, Sophie },
 booktitle = { Proc. 33rd Gocad Meeting, Nancy },
 title = { Facies and fracture detection using Multiple-Point statistics. },
 year = { 2013 }
}