On Regional Differences in Dispersion of Mantle Rayleigh Waves

Abstract

Summary Rayleigh waves generated by the Peru-Bolivian border earthquake of 1963 August 15 have been analysed at 35 WWSSN stations in the period range between 150 and 300 seconds. Thirty-one of these stations were grouped into six narrow azimuth windows as a result of an analysis of observed phase velocities. In each of these windows the phase velocity dispersion shows similar characteristics, and the composition along the paths (in terms of the shield, oceanic and tectonic fractions of a path) was approximately the same. An error analysis indicates that differences in phase velocities obtained for individual windows are statistically meaningful, thus proving that the Rayleigh wave dispersion measured on world-circling paths shows regional variations. These six areal windows cover a third of the Earth's surface. Group velocities were measured for the six azimuthal windows by summing the auto-correlograms of individual seismograms for stations within each azimuth range. A statistical test has shown that the measurements of phase and group velocities are mutually consistent proving for the first time that the lateral inhomogeneities in crust and mantle can also be detected by measurements of group velocities for world-circling paths. The simultaneous least-squares condition for both phase and group velocity observations for each window was used to derive a polynomial approximation for the frequency of free oscillations as a function of order number measured as the excess over a reference model. ‘Pure path‘ phase and group velocities derived according to the approach of Toksoz and Anderson indicate substantial differences in dispersion for each basic type of mantle. Pure path phase velocities derived in this report differ considerably from the corresponding results of an analysis by Kanamori, whereas world-wide averages are very similar for both reports. Group velocities can be separated much more effectively than phase velocities, especially at the longest periods. This effect may be explained by world-wide variations of depths to the so-called 400 km and 650 km discontinuities. These variations appear to be unrelated to the regionalization adapted in this paper.