Effective wavefield extrapolation in anisotropic media: Accounting for resolvable anisotropy
AuthorsAlkhalifah, Tariq Ali
KAUST DepartmentEarth Science and Engineering Program
Environmental Science and Engineering Program
Physical Science and Engineering (PSE) Division
Seismic Wave Analysis Group
Online Publication Date2014-04-30
Print Publication Date2014-09
Permanent link to this recordhttp://hdl.handle.net/10754/563513
MetadataShow full item record
AbstractSpectral methods provide artefact-free and generally dispersion-free wavefield extrapolation in anisotropic media. Their apparent weakness is in accessing the medium-inhomogeneity information in an efficient manner. This is usually handled through a velocity-weighted summation (interpolation) of representative constant-velocity extrapolated wavefields, with the number of these extrapolations controlled by the effective rank of the original mixed-domain operator or, more specifically, by the complexity of the velocity model. Conversely, with pseudo-spectral methods, because only the space derivatives are handled in the wavenumber domain, we obtain relatively efficient access to the inhomogeneity in isotropic media, but we often resort to weak approximations to handle the anisotropy efficiently. Utilizing perturbation theory, I isolate the contribution of anisotropy to the wavefield extrapolation process. This allows us to factorize as much of the inhomogeneity in the anisotropic parameters as possible out of the spectral implementation, yielding effectively a pseudo-spectral formulation. This is particularly true if the inhomogeneity of the dimensionless anisotropic parameters are mild compared with the velocity (i.e., factorized anisotropic media). I improve on the accuracy by using the Shanks transformation to incorporate a denominator in the expansion that predicts the higher-order omitted terms; thus, we deal with fewer terms for a high level of accuracy. In fact, when we use this new separation-based implementation, the anisotropy correction to the extrapolation can be applied separately as a residual operation, which provides a tool for anisotropic parameter sensitivity analysis. The accuracy of the approximation is high, as demonstrated in a complex tilted transversely isotropic model. © 2014 European Association of Geoscientists & Engineers.
SponsorsI thank KAUST for its support. I also thank the SWAG group within KAUST for their support. I thank the associate editor, Sergey Fomel, and an anonymous reviewer for their comments and suggestions that helped to improve this paper.