Direct elastic FWI updating of rock physics properties

Qi Hu, Scott Keating, Kristopher A. Innanen

Quantitative estimation of rock physics properties is of significant interest in reservoir characterization, and most current workflows with this subject are based on amplitude variation with offset (AVO) analysis. With the expectation of more accurate results, we propose to directly estimate rock physics properties using elastic full-waveform inversion (FWI). We implement this by incorporating the rock physics model, which builds a link between elastic and rock physics properties, into the FWI workflow to reformulate the model parameterization using rock physics parameter classes. We consider three rock physics models: the Han empirical model, the Voigt-Reuss-Hill (VRH) boundary model, and the Kuster and Toksz (KT) inclusion model. Each is used to formulate a model parameterization of porosity, clay content, and water saturation (P-C-S). We employ the truncatedNewton optimization method to update the model by iteratively minimizing the differences between synthetic and observed data. With numerical examples, our method shows considerable promise for recovering rock physics properties. It also possesses advantages over a sequential approach, which first invert for elastic attributes, then recover rock physics properties from them. We note that the Han model gives the most accurate results, whereas the KT model generally recovers the models with the largest errors. These large errors likely originate from the higher degree of nonlinearity of the KT model. The analytical radiation patterns of the P-C-S parameterization illustrate that the perturbation of water saturation has a minor effect on seismic data, this explains why water saturation is more challenging to recover compared with the porosity and clay content in our examples.