Bayesian linear inversion for fluid parameters based on independent prior information

Abstract Reservoir prediction plays a vital role in exploration and development. Inversion results utilizing fluid parameters enable effective identification of fluid-bearing properties in reservoirs, with prestack seismic inversion serving as a crucial method. Deterministic prestack seismic inversion methods fail to account for the uncertainty of inversion results, leaving fluid parameters inversions susceptible to seismic data quality. Bayesian linearized inversion (BLI) can achieve high-precision fluid-parameter prediction while characterizing uncertainty through confidence intervals of the inversion results. However, conventional methods with BLI are prone to interference from the statistical correlation among the parameters to be inverted, significantly reducing the inversion accuracy of fluid parameters. To address this issue, an inversion method is proposed based on independent prior information for fluid parameters. First, Russell fluid factor equation is employed to establish the relationship between fluid factors and seismic reflection coefficients. Next, a statistical correlation decoupling method for fluid parameters to be inverted is used to calculate independent prior information. Finally, this independent prior information is integrated to update the BLI equation, achieving high-precision fluid parameters inversion. Synthetic data tests demonstrate that the proposed fluid parameters inversion method overcomes the influence of statistical correlation among fluid parameters, effectively enhancing the inversion results accuracy and stability while exhibiting certain noise resistance capability. In field data applications, the inversion results obtained using proposed method correctly characterize the fluid-bearing properties of the reservoir.