Formulating full waveform inversion as a least squares migration-type problem
Scott Keating, Kristopher A. Innanen
The imaging of seismic reflectors can be iteratively improved by using a least squares migration (LSM) approach. Full waveform inversion (FWI) is usually posed as an inverse problem for the long to intermediate scale features of a seismic model, with little contribution from reflectors. Here, we show that full waveform inversion can be formulated to perform a similar role to LSM by choosing an appropriate choice of model parameterization. This approach specifically recovers the small wavelength features of the subsurface and does not require a complicated objective function or data filtering. Numerical tests show that this approach can be more effective than conventional FWI in recovering seismic reflector information.