Acoustic impedance inversion using anisotropic total variation with structure-guidance and smoothly clipped absolute deviation penalty

Abstract

The post-stack seismic acoustic impedance (AI) inversion is an important method for estimating subsurface lithological parameters. Post-stack seismic inversion based on total variation (TV) regularization is widely used to recover AI from noisy seismic data. However, TV has inherent biases and does not incorporate geological structure information, leading to lower inversion accuracy. To address this issue, we propose a new seismic AI inversion method. It is based on anisotropic total variation with structure-guidance and smoothly clipped absolute deviation penalty (SGSATV). First, smoothly clipped absolute deviation (SCAD) penalty is introduced into the anisotropic total variation (ATV) framework. Then, dip information from post-stack data is extracted using structural tensors and then incorporated into the inversion framework. Finally, an objective function with seismic matching term, SGSATV constraint, and initial model constraint is constructed. For the optimization problem, the objective functional is solved within the split Bregman framework. Numerical examples and field data demonstrate that the proposed method preserves geological edges and achieves higher inversion accuracy in geologically complex areas.