Iterative multiparameter elastic waveform inversion using prestack Kirchhoff approximation

Hassan Khaniani, John C. Bancroft, Eric von Lunen

Elastic Full Waveform Inversion (FWI) is an iterative method that simultaneously uses traveltime and amplitude of the seismic data to recover subsurface elastic properties. To date, despite the advancements in mathematical aspects of FWI, this method has not found much application in commercial seismic data processing, mainly because of its computation cost. Conventional elastic FWI methods require a depth imaging algorithm for forward modeling (e.g., Finite Difference Time Domain (FDTD)) and a depth migration for the inversion (e.g., Reverse Time Migration (RTM)).

Our aim is to propose a revised "standard strategy" for the inversion of elastic properties from the linearized reflected elastic waves. We use the direct relationship between the scattering potential of the Born approximation with the reflectivity function of the asymptotic Kirchhoff approximation. To estimate the gradient function of algorithm, we implemented the direct inversion strategy of Beylkin and Burridge (1990) to obtain an iterative well-known standard AVO inversion. Both of the forward and inverse operators use prestack time imaging methods that map the migrated P-to-S traveltime to P-to-P traveltime. We obtain two registered volumes in a pseudo-depth eliminating the need of ray tracing for registration issues. For complex structures, one can add ray tracing to the algorithm.