Department of Geology and Geophysics
University of Connecticut
Storrs, CT 06269-2045
Email:
cormier@geol.uconn.edu
poster/oral: oral
| Modeling of the elastic wavefield in three-dimensional structures having an arbitrary distribution of scale lengths continues to be challenge in seismology. A range of well-developed approximate analytic and numeric techniques can always be suitably chosen to solve any problem provided one understands the limitations of each technique. Asymptotic ray theory and its related extensions for diffraction can accurately model the wavefield in regions having high velocity and density gradients provided that wavelengths are much shorter than any scale length of the medium (either v/grad(v) or radius of curvature of a discontinuity). For modeling in the vicinity of the gradients of subducting slabs or fault zones, ray theory and its extensions, however, will often only be accurate at frequencies 20 Hz and above, limiting its practical use to travel time reconnaissance. When the ratio of wavelength to scale length becomes small enough to approximate a heterogeneity as a point discontinuity, ray theory, together with the Rayleigh-Born approximation, can applied to predict scattered wavefields. Its accuracy and validity, however, are difficult to assess without comparison to known analytic and numeric solutions, illustrated by predictions for the coda shape of scattered precursors to PKIKP by small-scale heterogeneities in the lowermost mantle. With the steady progress of Moore's Law and increased community familiarization with parallel computing, the future of three-dimensional modeling belongs to numeric techniques. Two-dimensional modeling of the teleseismic wavefield is now feasible up to 2 Hz to antipodal distances in 1 to 2 days of computation. Three-dimensional modeling is similarly feasible in the same frequency band at either shorter range for applications to strong-ground motion or at teleseismic range for frequencies on the order of 0.1 Hz. An unanticipated benefit of these new abilities is that our description of earth models can now include realistic small-scale fabric. Pseudospectral synthetics for the interaction of the wavefield with possible fabrics for the lowermost mantle and inner core show that the effects of small-scale fabric can be mistaken for the effects of intrinsic attenuation or the effects of larger scale structure. |