Pre-stack Three-Parameter Seismic Inversion Method Based on Bayesian Theory

Abstract

The traditional pre-stack seismic inversion method can only obtain two parameters of the longitudinal and transverse wave velocities. The traditional inversion method is stable, but has a low accuracy. Additionally, it is difficult to obtain the density information for the traditional method to predict the saturation of the reservoir fluid. In contrast, the pre-stack three-parameter inversion method can reveal the physical properties and hydrocarbon characteristics of underground reservoirs reliably. In order to obtain the longitudinal wave velocity, the shear wave velocity and the density simultaneously, and to improve the inversion stability and resolution, the AVO pre-stack three-parameter inversion is carried out in the Bayesian framework by combining the different scales information, such as well logging, seismic and geologic data. The likelihood function and the prior distribution are used to form the objective function. The prior probability distribution of the inversion parameters is obtained by the well data, and the likelihood function is obtained by the seismic data. The inversion objective function under the Bayesian framework is equivalent to introducing the regularization item into the traditional inversion problem, which can make the inversion of the seismic wave velocities and density of more stable. The inversion test based on the model and the actual data prove that the inversion method described in this paper can improve the multi-solution problem of inversion by using the information of different frequency bands. At the same time, it has a high stability and a certain practical value.