The effect of variations in arrival angle on surface wave phase velocity models: observations and improvements

We use a two-station method based on ray theory to make phase-anomaly measurements for Love and Rayleigh waves at 25-100 seconds period across the extent of the USArray, and invert these measurements for phase velocity models at discrete periods. To investigate some of the limitations imposed by ray theory, such as the assumption of propagation as a single plane wave, as well as those imposed by the assumption of great-circle path propagation, we make estimates of the arrival angle for seismic waves recorded across the Transportable Array. We make single-station phase measurements using the method of Ekström, Tromp, and Larson (1997), and use equal-phase contours of these phase measurements to visualize the wavefront for a given event. We find varying amounts of deviation from the plane-wave assumptions inherent in traditional two-station methods. Using both the USArray data and synthetic seismograms calculated from a 3-D Earth model, we estimate the arrival angle at each station using the geometry of these wavefronts and calculate a corrected inter-station phase value to test the effects of the deviations on two-station measurements. In most cases, corrections are small, in agreement with previous studies, and the synthetic arrival angle estimates match well with those made from real data. To test the effects of far-field anomalies on propagation, we estimate the average arrival angle for the entire regional array for a single event by fitting an apparent source location for the observed phase measurements. This is done for events around the globe, and yields a best-fit arrival angle that varies with frequency and shows systematic variations with event back azimuth. A comparison of our wavefront estimates of arrival angle with measurements of polarization based on 3-component seismogram particle motion will further address the deviations from the great-circle path and characterize a wave propagation path, allowing us to improve our phase-velocity models for the western US.