Seismic interferometry versus spatial auto-correlation method on the regional coda of the NPE

Discussion and Conclusions

Consistent with the observation of de Ridder (2008), we have observed that the cross-spectra of the NPE coda contain coherent energy that corresponds to the surface-wave. This observation shows that for the coda of the NPE, the SI approach is analogous to the SPAC approach. Both methods provide different information; in addition to an estimate of group velocity for the surface wave from SI, we have also estimated phase velocity from the cross-spectra. Only the energy between 1 and 4 Hz can be fitted unambiguously with $J_0$ functions, because higher frequencies contain too much spurious energy that is not equipartitioned, and lower frequencies do not contain sufficient energy. We also show that by analysing common-midpoint gathers, we can begin to see spatial variation of phase velocities along the array. This suggest the possibility to study 3D structure using cross-correlation techniques, as has also been proposed by Harmon et al. (2008) and Prieto et al. (2008a).

The relatively accurate fit of the damped Bessel function for Figure 8 suggests that most of the energy is part of the fundamental mode of the surface wave. Small discrepancies can be explained by many factors; the energy might not be completely equipartitioned, there could be some energy in higher modes, and very likely there are heterogeneities on a scale smaller than the smoothing length that invalidate our assumptions of a horizontally layered earth. Another effect of scattering is attenuation (Hong et al., 2005). The effects of scattering attenuation and intrinsic attenuation are not easily distinguishable at this point and further research is needed.

The source position of the NPE, transversely oriented to the direction of the array, seems very unfavorable for SI or SPAC. But small-scale heterogeneities in the earth rendered the coda wavefield equipartitioned to first order. Since the recording is sign-bit, this property is fully exploited, because amplitude changes with azimuth are suppressed. Moreover, the energy is relatively well distributed in both directions along the array, since the array is nearly transverse with respect to NTS.

Although the wavefield does not contain sufficient energy for imaging, subsurface information can still be extracted by inverting the dispersion curves for a subsurface velocity profile. The recording of the NPE and geometry of the array in Railroad Valley provided the rare opportunity to study the properties of a small earthquake-like event in the transverse direction, with high spatial resolution.

Seismic interferometry versus spatial auto-correlation method on the regional coda of the NPE