Stratigraphy-guided Ensemble Smoother Multi-Data Assimilation for poststack seismic inversion

Abstract

Abstract Poststack seismic acoustic impedance inversion is of vital significance in reservoir prediction. Nonetheless, conventional poststack seismic inversion approaches linearize the relationship between seismic response and acoustic impedance, which restricts their adaptation to weak reflection interfaces only. The ensemble smoother with multiple data assimilation (ESMDA) has emerged as a powerful tool for generating a credible set of prior ensemble members by matching simulated seismic responses to available observations. Yet, in standard implementations of ESMDA, single-trace inversion methods often encounter spatial discontinuities and instability. This is mainly because of the independence among prior ensemble members at each trace. Generally, ESMDA consists of two parts: generation of prior ensemble members and update of ensemble members. To enhance the spatial continuity of the ESMDA inversion results, stratigraphy information is incorporated into the prior ensemble members through stratigraphy-guided sequence Gaussian simulation. In this way, spatial continuity is introduced into the prior ensemble members, and inversion is performed to update these models. Tests on both synthetic and field data demonstrate the effectiveness of the proposed ESMDA inversion scheme. It can provide high-resolution inversion results with robust spatial continuity.