Velocity Model Estimation by Full Waveform Tomography of Seismic Reflection Data

A complex subsurface can imprint seismic data with propagation effects that cannot be fully described by ray-based models. In these situations, full waveform techniques have proved to be more successful in finding accurate velocity models. Two prevalent full waveform tomographic techniques are Wave Equation Migration Velocity Analysis (WEMVA) and Full Waveform Inversion (FWI). WEMVA aims at optimizing an imaging condition in the migrated domain and is deemed to be especially suited for finding the very low wavenumber part of the velocity model. FWI on the other hand aims at matching optimally the measured seismic data and is good at adding detailed velocity information to a velocity model, but it has difficulties to cope with a poor initial model. We propose a hybrid solution, where we use the perturbation theory from WEMVA and the data matching strategy from FWI. In this way we hope to combine the virtues from both approaches into one method.