Array Measurements of P  and PcP  Slowness Residuals with
Implications for Lateral Heterogeneity in the
Lower Mantle beneath the Caribbean Sea
and Other Geophysical Studies

 Advisor: Professor Eugene T. Herrin
Doctor in Philosophy conferred May 15, 1999
Dissertation completed April 28 , 1999

 Recent technology advances made possible the deployment of small seismic arrays (aperture less than 25 km) throughout North America.
TXAR ( Lajitas, Texas) and YKA   ( Yellowknife, Canada) are examples of such arrays.  Before using the arrays for location and magnitude calculation for regional and teleseismic events, they must be calibrated and the effects of subsurface structure removed from the data.
 Calibration studies at TXAR used a modified version of the correlation method in order to estimate azimuth and horizontal phase velocity.
The Moho discontinuity beneath TXAR was determined to the first order to strike along an azimuth of 109( ( NW-SE) and dip 11( to the northeast.
 Seismic reflections from the Earth's core recorded at the TXAR and YKA arrays in North America from events in the Caribbean Islands,
Venezuela and the Mid-Atlantic Ridge have observed slowness more than 70% greater than predicted by the standard Earth model, IASPEI91.
P  waves turning in the lower-most mantle in the same region also have anomalous slowness.  The slowness anomalies are not accompanied by significant travel time residuals and appear to be caused by lateral inhomogeneities in the velocity structure of the lower mantle.
 Ray tracing through elliptical velocity anomalies located in a region of the lower mantle beneath the Caribbean Sea found to be anomalous in previous tomographic studies suggested that integrated velocity anomalies in the range of 25% to 30% relative to IASPEI91 are required in order to bend the core reflected rays enough to match the observed slowness and travel time.
 This study also presents an application of the wavelet transforms as an alternative technique for location of local and near-regional events using a single array and the travel times of P  and Lg phases.  The complexity of Lg waveforms makes it difficult to consistently determine a unique Lg  arrival time.  An automatic method for timing Lg arrivals once the phase was roughly associated by analyst or using automatic detection algorithms was tested.  Wavelet transforms were used to decompose the Lg  signal into its components localized both in time and scale.  A threshold detector was then applied to the resulting time series to determine the Lg arrival time.  The Lg  arrival time was automatically picked with a standard deviation of less than 1.5 seconds ( less than 10 km location error) for well known locations.