Predicting porosity, water saturation, and shale volume with high-resolution seismic inversion using Hopfield network; Upper Assam Basin, India

Quantitative seismic interpretation involves prediction of reservoir properties from seismic data. In this study, reservoir parameters such as porosity, water saturation, and shale volume are estimated for evaluation of hydrocarbon potential within some reservoir intervals in the Upper Assam Basin, India. This is accomplished by inverting for acoustic impedance from seismic amplitude followed by conversion to desired petrophysical properties using mathematical relationships established during well-log analysis. Unlike conventional seismic inversion approach, the unsupervised Hopfield neural network is integrated to the optimization process to provide reliably higher-resolution estimates. The predicted properties were validated with recorded well data showing excellent matches. The well-log analyses revealed that at target intervals, the low impedance values represent clean and porous sandstones. Based on this observation, deeper targets are inferred to be prospective with noticeably low impedance zones of clean and porous gas saturated sandstones. The shallow intervals, however, are inferred to be of stiffer materials showing higher impedance values representing shaly and less porous sandstones.