Structure-oriented filtering in Crossplotting and <i>k</i>-means

Abstract

Several authors have proposed new techniques using multi-attribute analysis and machine learning. Studying the influence of different data treatments on such techniques is essential. We analyze the results by applying two clustering techniques, Crossplotting, and k-means, in filtered data. In particular, we use structure-oriented filtered seismic data before calculating seismic attributes. We use a migrated section of the Buzios field from the Brazilian pre-salt in the Santos Basin. We find that combining filtering and clustering techniques can improve salt identification.