On the Noise Level of the Ambient Noise Cross-Correlation Function and Its Application

Thursday, 18 December 2014
Ying-Nien Chen1, Yuancheng Gung2 and Ling-Yun Chiao1, (1)Institute of Oceanography, National Taiwan University, Taipei, Taiwan, (2)Department of Geosciences, National Taiwan University, Taipei, Taiwan
Retrieving the Empirical Green’s function (EGF) between two receivers by cross-correlating continuous records is now a well-recognized technique and the derived EGFs have been applied to various fields in seismology. However, little attention has been given to a more quantitative description on the noise behavior of the noise-derived cross-correlation functions (CCF), for its signal-to-noise ratio (SNR) can be improved easily by increasing the total correlation time. In our early work, using data recorded from Taiwan and Korea, we have designed a procedure to quantify the noises within the noise-derived CCFs and their spatial/directional properties. Because the effective amplitudes and the noise levels of CCFs are closely related to the distribution of the ambient sources and the population of the uncorrelated noises, respectively, here we move forward to further investigate the source mechanisms of the ambient noises by using their distinctive sensitivities to the surrounding sources.

It is known that the short period (3~5 seconds) secondary microseisms (SPSM) are much more active in Taiwan Straight and such asymmetric energy excitations are well demonstrated in the amplitude asymmetry of CCFs. Interestingly, we find that, there is nearly no directional dependence of the noise levels of CCFs, namely, the effects of uncorrelated sources in the CCFs are essentially isotropic.

With the same data set, we also investigate the excitation mechanisms of (1) the intermediate period (7-9 seconds) microseisms, which are one of the spectral peaks in the ambient noises, and (2) the coda train in the noise-derived CCFs. Based upon our analysis, we argue that (1) the 7~9 period ambient noises are contributed mainly by the near-coast primary microseisms rather than the far-field long period secondary microseisms, and (2) instead of the distant ocean waves, local scatter effects are the major sources of the CCF coda train.