Modeling Approach

The modeling approach utilizes increasingly advanced theoretical frameworks and numerical methods in order to obtain improved models of regional seismic structure. Specifically, a large-scale regional Eurasian model will be developed from a large dataset of seismic waveforms using the path-average approximation (PAVA) and NACT (Non-linear Asymptotic Coupling Theory; Li and Romanowicz, 1995), which are well-developed normal-mode based approaches which consider 1D (PAVA) and 2D (NACT) waveform sensitivity in the vertical plane along the great-circle path between source and receiver. This model will then be refined in a smaller region using an implementation of Born single-scattering theory (Capdeville, 2005), which more accurately represents the 3D sensitivity of the seismic wavefield. Finally, we will utilize the Spectral Element Method (SEM), a numerical approach that accurately models both 3D and non-linear effects (e.g. Faccioli et al., 1996; Komatitsch and Vilotte, 1998). To conserve computational resources we will restrict the use of SEM to the upper mantle by coupling to a normal mode solution (CSEM; Capdeville et al., 2003) and applying appropriate boundary conditions.

Figure 13.47: Preliminary smooth S velocity model for the upper mantle for the Eurasia developed using PAVA and NACT approaches. Values are the percent perturbations in isotropic velocity from PREM. Model is parameterized in radial splines spaced approximately 100 km, and in level 5 spherical splines (spacing $\approx$400 km).

Figure 13.48: Comparison of performance of several mode-based approximations used in tomographic modeling. The map shows the source-receiver geometry and the velocity model, an ellipsoidal anomaly 5% slower than the background centered at 220 km depth. The top two traces are the SEM synthetics calculated from the 1D background model and the 3D model. The remaining traces show the differential waveforms obtained by subtracting the waveform produced by the 1D model. For each approximation, the differential SEM waveform is shown as a dotted line, and the waveform from the approximation is solid. Results are shown for the Path Average approximation (PAVA), Non-linear Asymptotic Coupling Theory (NACT), NACT plus a higher-order focusing approximation (NACT+F), as well as the 3D Born approximation.