Attempts to invert FEAVO-affected data for a velocity model

The main papers discussing aspects of solving the FEAVO problem are (), (), (), (), and (). All are iterative inversions that try to find the velocity model that will result in the given raw data. With the exception of (), this operator is some form of plane-wave decomposition using a straight-ray assumption; none uses an actual differential (``wave-equation'') operator. () applies corrections in order to account for differences between infinite-frequency rays and the ``fat rays'' associated with the physics of wave propagation. All of them, with the exception of (), invert either for the traveltime effects associated with AVO, or for the amplitudes, but not for both simultaneously.

None of the previous attempts is completely successful in producing a velocity model that satisfies the initial goals of the problem. Using only a two-dimensional midpoint-offset map of FEAVO effects, instead of recognizing that they correlate across depth (as equation and Figure show), they introduced too many degrees of freedom in the inversions. Considering only straight rays was incorrect even in the case of the universally encountered v(z) variation. All methods require some form of picking, which results in endless headaches. None makes successful use of the entire quantity of information by simultaneously considering both the traveltimes and the amplitudes for all the reflectors. Only one () incorporates the information given by standard velocity analysis (which is not significantly affected by FEAVO effects).