| Two types of seismic parameters are predicted using new algorithms. The first are anisotropic parameters that are solved for by full-wavefield inversion of seismic data, and the second are the image positions of free surface multiples in the depth migrated domain.;For the first time, full-wavefield inversion has been applied to anisotropic parameter estimation. By supplementing the traveltimes with amplitude information, the complete set of anisotropic parameters for scalar and elastic media can be uniquely determined, using only surface seismic data; this is not possible for the pre-existing methods, which utilize only traveltime information.;For scalar media, 3D near-offset P-wave surface data are sufficient to recover all the elliptical anisotropic parameters. Inversion for elastic anisotropic parameters requires 3D, wide-azimuth, three-component data containing both P and P-S reflections. The present implementations of both the scalar and the elastic inversions are only applicable to flat-layer models, but can solve for multiple layers with various anisotropic symmetries, in layer stripping mode. Conventional isotropic normal moveout (NMO) analysis of P wave alone can provide an adequate starting model for scalar, and with some assumption, for elastic inversions; this is another distinct advantage compared to the travel-time based methods, which require NMO analysis of both P and P-S converted waves.;Full-waveform inversion can handle both primary and multiple reflections at the same time, but is not often used because of its high cost. Cost-efficient ray-based multiple predictions provide an alternative solution to the multiple problem. The image positions of the multiples are predicted using a two-step procedure of modeling and migration. By superimposing the results onto the original migrated volume, multiple events can be identified. The proposed method uses ray-shooting to reduce the computational cost, and its unique mid-point approximation delivers better results than the pre-existing 1D and 3D methods, especially where dips are large. The new method can be used to facilitate the identification of the multiple events, which otherwise might be mistaken as geological interfaces. This development is shown to be practical, accurate and efficient for application to real data. |
