Indonesia Conference Directory

Hydrocarbon exploration requires comprehensive understanding of geological and geophysical properties of subsurface reservoir. Therefore, subsurface rock properties determination, which include total and effective porosity, clay content, and water saturation, is a very important, particularly for reservoir target. In term of spatial coverage, subsurface rock properties is then distributed by using seismic multiattribute, which is performed in two approaches – linear multivariate regression and probabilistic neural network (PNN). Linear multivariate regression seismic multiattribute assumes that the relation of seismic attributes and reservoir property is linear, while probabilistic neural network, seismic multiattribute assumes non-linear relationship. This research employed linear multivariate regression using internal attribute of seismic data to predict several rock properties. Transformation of seismic internal attribute to rock properties was attained by a series of weights derived by least-squares minimization. To estimate the reliability of the derived multiattribute transform, cross validation is used where each well is systematically removed from the training set, and the transform is rederived from the remaining wells. The prediction error for the hidden well is then calculated. The validation error, which is the average error for all hidden wells, is used as a measure of the likely prediction error when the transform is applied to the seismic volume. This method was applied to Globigerina Limestone reservoir in Madura Strait that resulted in good prediction of reservoir properties. Result of this research is used for quantification of remaining lead and prospect in the field area. Furthermore, the predicted subsurface rock properties is used as input to optimalize in developing well in the proven fields.