HR: 0830h
AN: U51B-0005    [PDF]
TI: Lower Mantle Anisotropy Beneath the Cocos Plate
AU: * Rokosky, J M
EM: jrokosky@es.ucsc.edu
AF: University of California, Santa Cruz, 1156 High Street, Santa Cruz, CA 95064
AU: Lay, T
EM: tlay@es.ucsc.edu
AF: University of California, Santa Cruz, 1156 High Street, Santa Cruz, CA 95064
AU: Garnero, E J
EM: garnero@asu.edu
AF: Arizona State University, Dept. of Geological Sciences Box 871410, Tempe, AZ 85287
AB: Shear wave anisotropy exists across a broad region in the lowermost mantle beneath Central America and the Caribbean, causing splitting of ScS and diffracted S phases. We use shear wave signals from 16 intermediate and deep South American earthquakes recorded by the dense network of broadband seismographic stations in California to perform a high resolution analysis of the anisotropic structure in a localized region above the core-mantle boundary beneath the Cocos Plate. Our data include 162 ScS splitting measurements and 230 ScS-S differential travel times. Broadband ground displacement recordings are corrected for near-receiver lithospheric anisotropy and for a small phase shift in the ScSV reflection coefficient. ScS splitting measurements are made between the peak amplitudes on the transverse and longitudinal components; the SH and SV signals appear to be decoupled, with SH typically corresponding to the fast polarization arrival. Deconvolution of a source wavelet obtained by stacking of the SH signals for a given event is used to equalize the signals between events and to improve the temporal resolution of splitting. The deconvolved spike-trains generally support complete separation of the ScSH and ScSV wavefields, compatible with vertical transverse isotropy (VTI) at near-grazing incidence. Splitting measurements are also made between ScS peaks in the deconvolved traces. The two sets of splitting measurements are assigned quality ratings based on consistency between raw and deconvolved measurements and our overall confidence in the ScS signals. Only data of high quality are retained. In general, the process of deconvolution improves the precision of splitting measurements in cases where the receiver coda is strong or the source wavelet is complex. The average ScSV-ScSH differential arrival time is 1.2 s, with little distance or spatial dependence. Transverse component ScS-S differential travel times were calculated relative to PREM and corrected for a mantle shear wave tomography model (Grand, 2003) to a depth 250 km above the core-mantle boundary. The residuals indicate a strong lateral gradient in positive shear velocity anomalies, increasing toward the north beneath the Cocos Plate. The anisotropy measurements have a weak negative correlation with velocity anomalies overall, but show little evidence for a spatial gradient. The magnitude of splitting does not appear dependent on path length within the D" region, suggesting a relatively thin region of anisotropy. S wave triplication arrivals from a shear velocity discontinuity 250 km above the core-mantle boundary show minor splitting in deconvolved traces for a few high quality signals. This appears independent of ScS splitting in the same traces, suggesting that the anisotropic layer in which the ScS splitting originates lies below the D" discontinuity.
DE: 7207 Core and mantle
SC: U
MN: 2003 Fall Meeting