A workflow using self-organized mapping to predict rock properties from seismic reflection data

摘要

We present a workflow that integrates self-organized mapping classification results with rock physics based on well log data. The facies are extracted from reflection seismic data and rock physical models are obtained from well logs to calculate density, and P- and S-wave velocities in terms of the well log impedance. We apply these models and the SOM classes of the different facies to convert 2D images of impedance to 2D images of Vp, Vs, density, and Poisson ratio. We verify the workflow by creating 2D synthetic seismograms. Modeling software is applied to calculate (at great speed) complete seismograms that capture lateral variability of the reservoir model. The results show good agreement between observed and 2D synthetic seismograms. In addition, the results demonstrate that the SOM classification method enables to extract facies from seismic data, and allow us to integrate the lithology at the borehole scale with the 2D seismic based-impedance. This analysis produces accurate reservoir properties for the three geological units in the interwell region.