HR: 14:20h
AN: S43D-03 [Abstracts]
TI: Array Processing of Teleseismic Body Wave Phases Recorded by the Transportable Array
AU: * Pavlis, G L
EM: pavlis@indiana.edu
AF: Department of Geological Sciences, Indiana University, 1001 East 10th Street,
Bloomington, IN 47405,
AU: Vernon, F L
EM: flvernon@ucsd.edu
AF: Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics,
Scripps Institution of Oceanography
University of California, San Diego, 9500 Gilman Drive, LaJolla, CA 93092,
AB:
We have begun application of a new array processing procedure to teleseismic body waves recorded by the
USArray. Unlike standard slant stack processing used in high-frequency arrays this procedure begins by
assuming the source is already known. This allows an abstraction of the data as an arrival time gather in which
time zero for each seismogram is the predicted arrival time of the phase of interest. We use a novel, robust
stacking algorithm to compute a beam and align the data. The analyst is required to pick an initial trace to use as
a starting estimate of the array beam. This trace is used to provide initial alignment of data by cross-correlation
with this initial beam estimate. A median stack is then used to provide a robust starting estimate of the array
beam. The final beam is computed as a weighted stack with a penalty function derived from coherence of each
trace with the array beam used to improve the robustness of the algorithm. An iterative loop continues to refine the
array beam estimate until lags computed by cross-correlation with the array beam do not change. This procedure
is found to normally converge in 2 to 5 iterations and is remarkably effective in automatically discarding problem
data. The primary outputs of the procedure are an array beam, arrival time residuals, and relative amplitudes at
each station relative to the array beam. We have successfully applied this procedure to P, S, pP, PP, and Pdiff
body wave phases, but other phases could be handled with the same program. We find remarkable coherence of
the initial P and S phase in the 30 to 90 degree range routinely used for body wave tomography. Large events
commonly show coherence of 0.8 or higher over the entire array. Smaller events often require subarray
processing that requires a least squares procedure to produce a consistent set of residuals for the entire array.
This different behaviour for small events is largely due to a well-known frequency dependence in spatial
coherence. Small events are often only well recorded at short periods which are correlated over shorter distance
scales than longer periods that dominate larger events.
DE: 7203 Body waves
DE: 7208 Mantle (1212, 1213, 8124)
DE: 7290 Computational seismology
SC: Seismology [S]
MN: 2007 Fall Meeting