The Generation Of A Rock And Fluid Properties Volume Via The Integration Of Multiple Seismic Attributes And Log Data

We present a systematic seismic reservoir characterization<br>workflow that integrates log and seismic data using an artificial neural network.<br>Seismic attributes are examined both qualitatively and<br>quantitatively to determine the best discriminators of rock<br>and fluid properties. These attributes are systematically<br>classified using an artificial neural network, the Kohonen<br>self-organizing map (K-SOM) algorithm. Ultimately the<br>classified litho-facies volume is calibrated to available well<br>control by applying the K-SOM technology to well-derived data.<br>The product is a seismic-scale rock and fluid properties reservoir model that is consistent with borehole and surface seismic data.<br>The workflow is applied to the characterization of a Vicksburg-age reservoir in South Texas.