[S-SS07] Seismic wave propagation: Theory and Application

[SSS07-11] Vector wavelet propagation through random elastic media: Hybrid Monte Carlo simulation using the spectrum division (Theory)

*Haruo Sato¹, Kentaro Emoto² (1.Tohoku Unvieristy, 2.Kyushu University)

Keywords:seismic waves, body waves, random media, scattering, velocity structure, Monte Carlo simulation

Envelope broadening and coda wave excitation of a small earthquake are used to measure the random heterogeneous structure of the solid Earth. Radiative transfer theory (RTT) is useful for this purpose, in which the scattering coefficient by the Born approximation is a key component. In this talk, we consider von Karman-type random media and the case where the center wavenumber of a propagating wavelet is higher than the corner wavenumber. As the center wavenumber increases, the phase shift across the correlation length of the random fluctuation increases. Then, the Born approximation causes extremely large forward scattering beyond the scope of the perturbation method. In such a case, the Eikonal approximation can be applied and lead to envelope broadening, but it cannot explain the excitation of coda waves. We propose a hybrid Monte Carlo simulation using spectral division to overcome this difficulty. We divide the power spectral density function of random elastic media into high- and low-wavenumber spectral components with respect to the center wavenumber of the wavelet. Applying the Born and Eikonal approximations to each of these components, we statistically evaluate wide-angle scattering and narrow-angle ray-bending. We suppose a linear correlation among fractional fluctuations of P- and S-wave velocities and mass density, which maintains the linear polarization of a vector wave through the scattering process. Monte Carlo simulation incorporating the two types of scattering processes is used to synthesize three-component RMS velocity amplitude time traces for the radiation from a moment tensor source. In parallel, the propagation of a wavelet in random elastic media is computed by the finite difference simulation, which will be used as a reference to validate the proposed Monte Carlo simulation. In this talk, the theoretical construction of the method will be introduced, and the evaluation of the proposed Monte Carlo simulation will be presented in the subsequent talk.