Full Zoeppritz Equation-based Elastic Parameters Bayesian Generalized Linear Inversion

Abstract

In the view of elastic parameters estimation, the linear approximations of Zoeppritz equation are no longer suitable for the data with large incident angles. Therefore full Zoeppritz-equation-based generalized linear inversion is proposed in this paper. To constrain the whole inversion process, Bayesian theory is also introduced not only to add priori information but also evaluate the uncertainty of the inversion results. During the inversion process, the exact gradient is applied instead of one-order difference, which improves the reliability of the inversion results. Tests on synthetic data show that the elastic parameters are almost perfectly inversed. Comparing with the inversion results by Aki’s linear approximation, our algorithm can generate more stable and accuracy estimations. We apply the method to process the real data from a carbonate reservoir in southern China, the elastic parameters are basically recovered and match the logging data well.