Teleseismic Waves


Exploration arrays such as the Long Beach network normally do not record teleseismic waves very well because of the mismatch between the high-frequency sensors and the relatively low-frequency signal. However, for events that are large enough (i.e. the Tohoku, Japan event of 2011) or ones that have short-duration high-frequency source time functions they in fact can be detected. Here we show some observations and preliminary analysis of the Tohoku earthquake and a 6.3 deep event in the Tonga Trench recorded on the Long Beach Array. These events show that we can detect local structure using teleseismic earthquakes.
Teleseismic waves originate from earthquake that are farther than 30 degrees from the network. If they are closer than 90-degree distance then their raypath is entirely in the mantle and crust of the earth. As shown in the cartoon, sources that are 80 degrees away have rays that are within 10-degrees of vertical when they arrive at Long Beach.
P-waves from the 2011 Mw=9.1 Tohoku event were recorded by the Long Beach array. In the figure above the data are time aligned by the initial arrival. The red line on the map plot shows the orientation of the line of receivers displayed in the lower part of the panel. The seismograms show considerable variations in the amplitude in both the initial waves and those following. It is unlikely these are due to variations near the source because the small size of the Long Beach array projects to a tiny area in the source region. It is more likely that these are caused by local structure near the LB array. The plot in the upper right shows the rms energy over a 60-sec time window, and the two zones marked as A and B show enhanced amplitudes in both the raw traces in the rms amplitude measurements. The figure below shows a map view of the rms energy.
A map view of the rms amplitudes of the P-wave and its 60-sec coda from the Tohoku earthquake are shown. The letters A and B mark the same features as in the previous figure. 'B' can be identified with the main production area of the Long Beach Oilfield and it the also the location of the topographic high of Signal Hill. 'A' does not associate with any known feature, but curiously the seismic reflection data is very poor in this area. Note the feature sub-parallel with the Newport-Inglewood Fault Zone, which is a multi-strand fault system.
The plots shows the map view variations in the rms amplitude and time of P-waves from a deep (550 km) Mw=6.3 earthquake in the Tonga region. The amplutude shows similar variations as with the Tohoko event, although the Signal Hill anomaly is not quite as prominant, but the variations parallel to the NIFZ are stronger. Note that the array data does not extend as far as the coast for this event. The times are variations relative to the initial P-wave, and in this case the Signal Hill anomaly is very prominent even after a generous correction for topography.
The above section is a time-distance plot of the P-wave from the Tonga earthquake. The horizontal dimension of the plot is approximately 10 km and the section is about 82 degrees from the event. The section is time aligned on the intial P-wave, and the average trace (stacked trace shown below the section) is deconvolved. The clear feature in the code is a pair of horizons with a different moveout than the initial P-wave. That moveout is 6.1 km/sec which indicates that the event is due to scattering the crust somewhere in the vicinity of the LB array. Note that it is not a free surface multiple because it has the same polarity as the P-wave. It strength is about 20% of the P-wave (which is strong for a scattered wave).

A Model For the Multipathing The apparent moveout on the scattered phase is 6.2 km/sec which means the scattering must occur in the crust on the receiver-side of the raypath. The time separation of ~3 secs., which indicates that the scatterer is not too far from the Long Beach array. To the left is an ad hoc model that can explain these two observations. The sharp change in the Moho is expected because of the ocean-continent transition and has been hinted at in other studies (ten Brink et al, 2000; Nazereth and Clayton, 2003). In the lower left is a finite-difference simulation. The black box represents the zone under the LB array. The data in this box are deconvolved with the average trace in the box and the result is shown to the right. A comparison of this simulation and the observation is shown in the right panel. The match is approximately correct but the model needs to be tuned to get the amplitudes to match.

References
  • Nazereth, J. and R. Clayton, (2003), Crustal structure of the Borderland-Continental Transition Zone of southern California adjacent Los Angeles, J. Geophys. Res., 108 (B8), doi:10.1029/2001JB000223.
  • ten Brink, U., S. Zhang, T. Brocher, D. Okaya, K. Klitgord, and G. Fuis, (2000), Geophysical evidence of the evolution of the California Inner Continental Borderland as a metamorphic core complex, J. Geophys. Res., 105, 5825-5857.