--> Pattern Recognition Methodology and Application in E

Datapages, Inc.Print this page

Pattern Recognition Methodology and Application in E&P Interpretation Flow


Antonio Corrao1, L. Ferri2, M. Fervari2, P. Rocchini2

(1) ENI SpA, AGIP E&P Division, Milan, Italy (2) ENI SpA-Agip Division, S. Donato Milanese (Milan), Italy

 A pattern recognition methodology has been developed to optimise the interpretation of complex multi-reflection seismic objects in a 3D domain.

This paper describes the methodology and the applications of implemented semi-automatic techniques that can support the following interpretation activities in the E&P flow:

1) seismic volume pre-screening

the computation of a set of pattern and statistical analysis attributes provide new non conventional parameters that highlight the different seismic texture in the seismic data

2) preliminary pattern description

the description of seismic pattern is performed by the interpretation of a set of cross-plots of couples of attributes; the methodology allows the generation of seismic facies volumes through the definition of “polygons” on those cross-plots which are more suitable for identifying interesting 3D seismic pattern.

3) target oriented pattern classification

more detailed interpretation work is required for characterising the pattern associated with a specific seismic object. This refining work provides a classification of seismic pattern in order to identify in 3D domain the target seismic object.

4) geobody detection

the connectivity algorithm applied to the result of target oriented pattern classification allows the identification of one or more geobodies characterised by “multi-reflection” seismic facies.

5) detailed characterisation of seismic feature in 2D domain

the methodology supports also a data flow focused on 2D domain for a detailed description of the internal pattern of geobodies.

The application of the presented semi-automatic methodology can improve the result of 3D seismic interpretation, offering an approach oriented to the detection of relevant stratigraphic features.