Separating non-diffuse component from ambient seismic noise cross-correlation in southern California

Xin Liu, Gregory C. Beroza, & Nori Nakata

Published August 14, 2017, SCEC Contribution #7693, 2017 SCEC Annual Meeting Poster #035

Cross-correlation of fully diffuse wavefields provides Green’s function between receivers, although the ambient noise field in the real world contains both diffuse and non-diffuse fields. The non-diffuse field potentially degrades the correlation functions. We attempt to separate the diffuse and the non-diffuse components from cross-correlations of ambient seismic noise and analyze the potential bias caused by the non-diffuse components. For the Rayleigh wave components, we assume that the cross-correlation of multiply scattered waves (diffuse component) is independent from the cross-correlation of ocean microseismic quasi-point source responses (non-diffuse component), and the cross-correlation function of ambient seismic data is the sum of both components. Thus we can separate the non-diffuse component due to physical point sources and the more diffuse component due to cross-correlation of multiply scattered noise based on their statistical independence. We also perform beamforming over different frequency bands for the cross-correlations before and after the separation, and we find that the decomposed Rayleigh wave represents more coherent features among all Rayleigh wave polarization cross-correlation components. We show that after separating the non-diffuse component, the Frequency-Time Analysis results are less ambiguous. In addition, we estimate the bias in phase velocity on the raw cross-correlation data due to the non-diffuse component.

Citation
Liu, X., Beroza, G. C., & Nakata, N. (2017, 08). Separating non-diffuse component from ambient seismic noise cross-correlation in southern California. Poster Presentation at 2017 SCEC Annual Meeting.

Related Projects & Working Groups
Seismology