Full waveform inversion (FWI) is a data matching procedure which attempts to use the whole wavefield to obtain quantitative information about the subsurface.
If the exact physical properties of the subsurface are known, it should be possible to generate synthetic data which exactly match the measurements made in a seismic survey, provided that the physics are well understood. In FWI, this is used as the basis for an optimization problem: the discrepancy between measured data and synthetic data generated from a subsurface estimate are minimized. The subsurface estimate which minimizes this discrepancy should have physical properties very similar to those of the subsurface.
FWI has been demonstrated to be an effective tool for high resolution velocity model building, but continues to face serious challenges in the form of computational expense and obstacles in the numerical optimization. Additionally, the significant potential for FWI to recover a more complete picture of the subsurface, extending beyond velocity to other physical parameters, has been largely untapped.
CREWES is currently working on several projects to address the challenges and extend the scope of FWI. These include work on multiparameter FWI, cycle skip mitigation, nonlinear FWI, strategies for coping with unknown physics in FWI and reflection waveform inversion.
FIG 1. Velocity structure (left) and example synthetic shot gather (right) for model estimate before inversion.
FIG 2. True velocity structure (left) and example measured shot gather (right).
FIG 3. Velocity structure (left) and example synthetic shot gather (right) for model estimate after inversion.