Bayesian lithology analysis with seismic-driven likelihood to anisotropic medium: a case study from Northwest Australia

Abstract An isotropic medium is generally assumed for reservoir characterization although it has been reported that velocity anisotropy often has a non-negligible impact. One of the difficulties in incorporating anisotropy is that the exact anisotropy parameters are generally unknown. To overcome the problem, a seismic-driven approach to Bayesian probability analysis is proposed, which defines the likelihood from seismic inversion data. Comparing this approach for lithology prediction to one based on well data, it is found that an isotropic well-data approach does not work well but is improved by taking anisotropy into account. A larger improvement is achieved by the seismic-driven approach which provides a robust result which is insensitive to the anisotropy assumption.