A Pre-stack Three-term AVO Inversion Method Based on Integrated Norm Regularization

Abstract

Comparing with conventional post-stack methods, pre-stack inversion is better for subsurface reservoir or fluid prediction. However, the actual pre-stack inversion problems for seismic exploration are ill-posed and are difficult to be steadily and rationally solved. Although pieces of prior information for AVO parameters have been employed to regularize the ill-posed inversion process via many methods, such as Cauchy distribution under Bayesian framework, they are often not integral to solve mathematical matrixes under geology constraining. In this paper, based on the basic theories of AVO inversion and Bayesian, an integral norm regularization including matching degree to the input seismic data, spatial consistency and the statistical relationships of three interest parameters is developed to formulate AVO inversion. In particular, the conventional pre-stack inversion method without regularization, pre-stack inversion based on Cauchy distribution prior constraint and the proposed integrated norm regularization are separately applied in both the numerical and field data. It is found that the proposed three-term AVO inversion method shows strong anti-noise ability and robustness. And it is more beneficial for describing real geology and achieving better density results. This research will mainly find its efficiency in AVO inversion enhancement for reservoir/fluid prediction.