This paper illustrates a new reservoir characterization approach using seismic pattern recognition methodology based on principal component analysis, trace shape and 3D multi-attributes classification on a set of 3D seismic data volumes. We develop the reservoir interpretation workflow (<A HREF="#fig01">figure 1</A>) as follows: 1) Seismic facies maps obtained by Kohonen's Self Organizing Map (SOM) Neural Network method to seismic facies volume from a set of various seismic attributes, obtained by Hierarchical classification method; and 2) Statistical analysis of available seismic attributes by application of Principal Component Analysis (PCA) statistical method before classification. How to discriminate the amount of information that each geophysicist is dealing with, in a reservoir characterization process and what is the suitable procedure to help to discriminate between data and information? We present some results of its application on the Paleocene/Eocene reservoirs of a Campos Basin field.
seismic attributes; seismic facies; reservoir characterization
