Geostatistical Seismic Inversion Integrating Rock Physics Models

Abstract

We successfully applied a new methodology that inverts angle-dependent seismic reflection data directly for rock properties, including rock physics modelling within the inversion loop in a real case application. In reservoir characterization, rock physics models provide the link between geophysical signatures and petrophysical properties. Integrating the petrophysical domain within the inversion, allows the prediction of petrophysical property models (i.e. porosity, volume of shale, water saturation, and facies) to calculate the elastic response of each facies and the corresponding angle-dependent synthetic seismic data. The inversion procedure is based on a stochastic sequential simulation as the perturbation technique of the model parameter space using a cross over genetic algorithm that ensures the convergence of the methodology throughout the iterations. This methodology is divided into four main steps: stochastic sequential simulation and co-simulation of porosity, volume of shale and water saturation; classification of seismic lithofacies based on previous generated models using a Bayesian approach; calculation of angle-dependent synthetic seismic data applying a facies dependent rock physics model; comparison of the resulting seismic data against the real one and selection of the resulting seismic data to constrain the generation of new petrophysical models for the next iteration.