This case study shows the benefit of using multiple seismic trace attributes and the pattern recognition capabilities of neural networks to predict reservoir architecture and porosity distribution in the Pegasus Field, West Texas, and net pay and reservoir property distribution in the Zafiro Field, offshore Equatorial Guinea. The study used the power of neural networks to integrate geologic, borehole, and seismic data. Illustrated are the improvements between the new neural network approach and the more traditional methods of estimating rock properties from seismic data, such as seismic trace inversion, amplitude mapping, and AVO studies. Our procedure is straight forward but does require careful quality control to ensure reliable predictions from the seismic data. Network training, test, and validation data sets provide calibration of seismic attributes with well log data, optimize the network parameters, and estimate the performance of the system to predict hidden representative data. Comprehensive statistical methods and interpretational/subjective measures ensure that only attributes providing true relationships and a physical basis are used in the prediction of rock properties from seismic attributes. The result is a 3-D volume of seismically derived rock properties for the reservoir interval of interest. In effect, we are transforming the seismic trace attributes into seismic-scale petrophysical logs. The advantage of this transformation is the additional interwell information this method provides. The additional reservoir detail allows for optimum placement of horizontal wells and improved field development.

End_of_Record - Last_Page 27---------------