Bi-objective optimization for the inversion of seismic reflection data: Combined FWI and MVA

Abstract

The inversion of seismic reflection data is challenging due to the oscillatory nature of the seismic data, the nonlinear relation between the data and the Earth's model parameters, and the relative sensitivity of the different subsets of the data to the model parameters. Full-waveform inversion solves this inverse problem in the data space but is known to have shortcomings, especially because of the multi-modality and the numerous locally optimal solutions that its objective function allows. Migration velocity analysis indirectly relates to the inversion of seismic reflection data. Used in velocity model building, the method yields a macro model of the Earth's subsurface but constrains this model only to be kinematically correct. Migration velocity analysis has also its share of well-known pitfalls. Interestingly, the two methods have complementary characteristics which potentially mitigate the drawbacks of each technique. To take advantage of the capability of full-waveform inversion and migration velocity analysis, a bi-objective optimization strategy combines the two methods to solve the inversion of seismic reflection data. A numerical example illustrates the benefits of this new strategy.