Project Abstract
|
We test the amplitude of station-to-station ambient-field Green’s functions in southern California ob-tained from three different approaches: correlation, deconvolution, and correlation of coda of correlation (C3). These approaches produce nearly identical Green’s functions if the amplitude is discarded. In regions where the 1D velocity structure is a good approximation the correlation and C3 approaches yield the approximate geometric spreading effect of surface wave due to a point source at the surface, while deconvolution does not. In the Los Angeles basin, however, all three approaches fail to predict the basin amplification although the Green’s functions have a remarkable waveform similarity with the ones calculated by a finite element method in the SCEC CVM 4.0 showing clear basin amplification. The virtual coda of correlations, consisting of multiply scattered waves between stations, produce Green’s functions of greatest symmetry between causal and anticausal time lags, rendering the C3 approach the best possible candidate to extract robust amplitude of the Green’s functions. Our tests show that the amplitude from the C3 approach is insensitive to the length of virtual coda used and temporal/frequency normalizations. New signal processing techniques are needed if robust amplitude of the Green’s functions can be obtained from the ambient seismic field. |
