Abstract: Characterization of Lithological Log Responses in Turbidite Series Using Neural Networks
HASSIBI, MAHNAZ and IRAJ ERSHAGHI
In Turbidite sequences, patterns observed on lithological logs can show considerable differences across the reservoir structure. Recognition of markers and sub-markers, with consideration of continuity of layers is a major challenge. Neural networks offer powerful capabilities for pattern recognition and pattern grouping. In this study, application of a number of techniques in neural network including Back-Propagation, Self Organize and Adaptive Resonance Theory type of networks are examined for lithological signals observed for Turbidite series. New approaches in the application of transforms are also presented to examine the similarity of responses. Finally the application of the proposed method to the well log responses from the Tar zone in the LB unit is presented. With the proposed approach, geologists are provided with a powerful tool to decipher well log responses for characterization of Turbidite series.
Search and Discovery Article #90945©1997 AAPG Pacific Section Meeting, Bakersfield, California