Paper submitted to IMAGE 2026, by Sophia Tanner, Hao Hu, Ying Zhang (University of Oklahoma), Guanghui Huang, Faqi Liu (TGS).
Abstract
We propose an efficient method for decomposing elastic wavefields into vectorized P-, SV-, and SH-wave modes in elastically anisotropic media while preserving amplitude and phase. Physics enhancement in wave propagation from acoustic to elastic in seismic imaging and inversion algorithms is increasingly important for providing a more complete subsurface information. However, this enhancement produces additional problems that need to be addressed for the correct application of the elastic algorithms. These are more susceptible to crosstalk artifacts from unmatched P/S modes, especially in anisotropic media where propagation and polarization directions differ. These artifacts can contaminate elastic reverse time migration and full-waveform inversion results. Existing decomposition methods commonly rely on simplifying assumptions, such as elliptical or weak anisotropy, or require substantial additional computational cost during wavefield extrapolation. Our method is derived from exact wave-mode expressions as functions of phase-propagation direction under the plane-wave solution, combined with phase-direction information, and therefore avoids restrictive anisotropic assumptions. Numerical tests using a 3D homogeneous TI model, the 2.5D SEAM Phase I model, and the 2.5D BP TI model demonstrate that the proposed method efficiently separates the total elastic wavefields into vectorized P/SV/SH wavefields with preserved amplitude and phase. The method can be incorporated into any existing anisotropic elastic modeling operators with minor additional computational cost and has strong potential to improve elastic imaging and inversion by reducing crosstalk artifacts and providing more accurate vector wavefields.
Read the full article here.