A standard technique for locating events with T phases is to pick the peak energy of T phases as the arrival time, then proceed as if it was an unscattered phase originating at the epicenter. The peak energy arrival time, however, can shift to different parts of the wave train due to incoherent scattering. We show that a 50% reduction in variance relative to picks of peak arrival times can be achieved by fitting an assumed functional shape to the log of the entire envelope of the T phase.
We test the stability of this approach by comparing relative event locations based on this method with those determined by cross-correlating waveforms of Rayleigh and Love surface waves recorded teleseismically using a swarm of earthquakes at the northern end of the Easter microplate as an example. Relative event locations show that there is no systematic bias in T-phase locations. The T-phase location and detection can be extended to much smaller events than are detectable by surface waves.
We estimate T-phase maximum amplitudes of about 50 events within the swarm from the amplitude of the fitted functions rather than directly from the seismograms. Twenty-four of these events are large enough to estimate their magnitudes by surface-wave analysis. The empirical analysis shows the log of T-phase maximum amplitude and surface-wave magnitude (MS) exhibit a relatively uniform linear relationship with much less scatter than in previously published studies of T-phase amplitude versus magnitude. The reduction in scatter is due to both the stability of the functional estimate of amplitude and the use of a swarm of earthquakes with similar depth, mechanism, and source-receiver geometry.
The observed T-phase coda for shallow earthquakes can be synthesized using a simple model of multiple-reverberation seafloor scattering. The results show that scattering of energy at a rough seafloor from multiple reverberations in the water column is significant in T-phase generation even where the water depths are relatively uniform.