Portal de Periódicos da CAPES

Array analysis of seismic scattering

1990; Seismological Society of America; Volume: 80; Issue: 6B Linguagem: Inglês

10.1785/bssa08006b2242

1943-3573

Anton M. Dainty, M. Nafi Toksöz,

Seismic Imaging and Inversion Techniques

Abstract The use of high frequency arrays permits investigation of the spatial as well as the temporal character of the seismic wavefield in the frequency band 1-10 Hz. The authors have carried out investigations into seismic scattering at NORESS, FINESA, and ARCESS. Two principal methods have been used, frequency-wavenumber analysis of coda and coherency analysis of seismic phases. For regional seismograms, frequency-wavenumber analysis of Lg coda shows that Lg-to-Lg scattering dominates and that scattering is initially concentrated in the forward direction, changing to isotropic backscatter later in the coda. The P coda is more complex, although all cases studied show that energy is concentrated in the on-azimuth direction. However, different phase velocities are found for the three different cases studied, indicating that P coda is made up of several different contributions whose balance varies in different situations. In previous work, we have found that the decay of coherency with spatial separation scales as the wavelength for the Lg phase of regional events in the frequency band 1-10 Hz, declining to about 0.5 within one wavelength. Analysis of synthetic seismograms for realistic models of the area suggests that this loss of coherency is not due to dispersion, reaffirming a scattering interpretation. However, the decline in coherency is isotropic, seemingly contradicting the finding of forward scattering found in the frequency-wavenumber analysis. This may reflect a sensitivity of the coherency analysis to scattering within or very close to the array, while the frequency-wavenumber analysis favors scatterers at greater distance.