Elastic impedance variation with angle inversion for elastic parameters

Abstract

Elastic impedance (EI) and amplitude variation with offset or angle (AVO/AVA) inversion are two cardinal methods to estimate elastic parameters underground with reflection seismic data. Conventional EI inversion as a kind of pre-stack and post-stack joint inversion method has been widely applied in the industry because of its high efficiency and high stability of wavelet extraction; however, the robustness of extracting elastic parameters in conventional EI inversion is still controversial. The robustness of three-term AVO inversion has improved a lot; however, it is still challenging to extract reasonable space variant wavelets for each offset or incident angle. In this paper, a robust three-parameter estimation method, named elastic impedance variation with angle (EVA) inversion, is proposed in the Bayesian framework, which can estimate elastic parameters directly from EI. This method supposes that the parameters to be inverted are Cauchy distributed and it is implemented based on a normalized EI equation in a logarithmic domain which can reduce the nonlinearity of inversion. Application of a covariance matrix to decorrelate the parameters and constraint of well log curves introduced in an objective function enhances the robustness of EVA inversion. A model test shows that the proposed EVA inversion method enables one to estimate reasonable elastic parameters with extremely smooth initial models and moderate Gaussian noise. A real data example shows that the inverted P-wave velocity, S-wave velocity and density are identical to well log interpretation results, which shows the validity of the proposed method.