Mapping the lowermost mantle using core-reflected shear waves (original) (raw)

1994, Journal of Geophysical Research

A map of laterally varying D" velocities is obtained for the region from 50øS to 50øN in latitude and 70øE to 190øE in longitude. Velocities are found using an analysis of the differential travel time residuals from 481 ScS-S and 266 sScS-sS phase pairs. The long-period data are taken from the Global Digital Seismograph Network digital waveform catalog for the time period of January 1980 to March 1987. Each differential travel time is found by a cross correlation of the S phase ground displacement, corrected to simulate differential attenuation, with all following phases. Travel times are corrected for ellipticity and mantle heterogeneity outside of their D" paths, and the remaining residuals are interpreted as the result of D" heterogeneity. Ray-tracing tests are made to check the validity of converting travel time residuals into velocity path anomalies.The resulting map reveals significant long-wavelength D" structure including a 3% low-velocity region beneath northeastern Indonesia, surrounded by three identified high-velocity zones beneath northwestern Pacifica (+4%), Southeast Asia (+3%), and Australia (+3-5%). This structure is of continent/ocean spatial scales and is most likely created by dynamic processes dominant in the lower mantle. The low-velocity region may have both chemical and thermal origins and is very possibly the site of an incipient lower mantle plume where mature D" rock which has been heated by the core has become gravitationally unstable and begun to rise. A chemical component possibly exists as a chemical boundary layer is dragged laterally toward the plume site, much the way continents are dragged toward subduction zones. The high-velocity zones possibly result from the downward convection of cold lower mantle plumes, which pond at the core-mantle boundary. These seismic anomalies may also contain a chemical signature from faster iron-poor materials brought down through the lower mantle or the additional presence of SiO2 stishovite, perhaps in its higher-pressure polymorph. 13,667 13,668 WYSESSION ET AL.: MAPPING THE LOWERMOST MANTLE certain the viscosity, temperature, and conductivity of the lower mantle, we cannot yet apply them to the real mantle. Wysession et al. [ 1992, 1993] identified an unusual feature at the base of the mantle beneath Indonesia. Profiles of corediffracted P and S waves revealed a large D" low-velocity region (-3%) beneath eastern Indonesia, which was adjacent to a region of higher than average CMB velocities to the west under Southeast Asia. The geometrical limitations of the diffracted data did not allow for a more detailed image of the features. This lowvelocity feature also appeared in the ScS study of Woodward and Masters [ 1991 b] and the tomographic model that incorporated it, SH8/WM13 [Woodward et al., 1993], and it was our aim to obtain a better resolution of it using a data set independent from the diffracted waves. Just as regional SS-S studies such as that by Kuo et al. [1987] served as models for the global SS-S study of Woodward and Masters [1991a], so has the global ScS-S study of Woodward and Masters [ 199 lb] served as a model for our study, which is similar in concept. The differences in our work involve the inclusion of sScS-sS residuals, the method of determining the travel time residuals, and the visualization and interpretation of the data. Another impetus for our work was the ScS-S CMB study of Lavely et al. [ 1986], which used analog Word-Wide Standard Seismograph Network data to look at CMB variations beneath the North Atlantic and Indian Oceans. The use of core-reflected waves has long been established as a powerful means of examining the base of the mantle [BaumgardtThe CMB features that we find have lateral dimensions of -1000-2500 km or more in diameter, and the data are very consistent within each feature. This long-wavelength coherence of the data was somewhat surprising since the Fresnel zone resolution width of the ScS waves is near 400 km for 10-s periods (570 km for 20-s periods), permitting resolution of narrower features. Recent studies using short-period P waves have also seen D" features with widths of the order of a few hundred kilometers [I,•dale and Benz, 1993; Kriiger et al., 1993], and core-phase scattering studies have found CMB inhomogeneities with wavelengths as small as 10 km [Bataille et al., 1990]. We do not yet know what NSF-EAR-05368 and the W. M. Keck Foundation. The authors would like to thank Bob Woodward for help in using the whole mantle model SH8/WM13. The waveform cross-correlation program was adapted from that of An-Ning Zhu and Doug Wiens. Helpful discussions were held with Craig R. Bina and John Vidale. The manuscript was greatly improved with the help of comments by Thorne Lay, George Sutton, and an anonymous reviewer. Acknowledgment is also given in memoriam for helpful dialogues with Durk Doornbos.