Using multicomponent ambient seismic noise cross-correlations to identify higher mode Rayleigh waves and improve dispersion measurements

Ambient seismic noise cross-correlation with three-component sensors yields a nine-component empirical Green's tensor, in which four components of the radial-vertical plane contain Rayleigh waves. We exploit the retrograde elliptical nature of particle motion of the fundamental mode Rayleigh wave to correct the phase of the four radial-vertical components and stack them to obtain an average fundamental mode Rayleigh-wave time-series. This technique can suppress incoherent noise and wave packets that do not follow the targeted elliptical particle motion. The same technique can be used to isolate the first higher mode Rayleigh wave that follows prograde elliptical particle motion. We first demonstrate the effectiveness of the method on synthetic waveforms and then apply it on noise cross-correlations computed in Central California. Using this method, we isolate 1st higher mode Rayleigh waves on noise cross-correlations in the Great Valley, California, which provides new phase velocity constraints for estimating velocity structure in the sedimentary basin. We also obtain improved estimates of fundamental mode Rayleigh-wave dispersion for surface-wave tomography. The waveforms stacked assuming retrograde particle motion return at least similar to 20 per cent more group velocity dispersion measurements satisfying a minimum signal-to-noise ratio (SNR) criterion than the individual components for periods similar to 4-18 s. For equivalent group velocity measurements, SNR for the stacked estimate of the fundamental mode Rayleigh wave is on average 40 per cent greater than that measured on the individual components at periods less than 10 s. The technique also provides an easy way to detect large errors in sensor orientation.