S13B-4450:
Finite Frequency Measurements of Conventional and Core-diffracted P-waves (P and Pdiff)

Monday, 15 December 2014
Kasra Hosseini, Ludwig Maximilian University of Munich, Munich, Germany, Karin Sigloch, University of Oxford, Department of Earth Sciences, Oxford, United Kingdom and Simon C. Stähler, Ludwig Maximilian University of Munich, Geophysics, Munich, Germany
Abstract:
Core-diffracted waves are body waves that dive deep enough to sense the core, and by interaction with this wave guide become dispersive. They sample the core-mantle boundary and the lower third of the mantle extensively. In ray theoretical modeling, the deepest part of the ray starts to graze the core at around 97 degrees distance, but ray theory is a very poor approximation to propagation of core-diffracted waves. In reality, finite-frequency waves with their spatially extend sensitivity regions start to sense the core at significantly smaller distances already.

The actual, non-ray-like sensitivities have been difficult to model, as have been the associated synthetic seismograms. Core-diffracted waves have therefore not been used in tomography, despite abundant observations of these phases on modern broadband seismograms. Hence current global body-wave tomographies illuminate the lower third of the mantle much less well than the upper and especially the middle third.

This study aims for broadband, global waveform tomography that seamlessly incorporates core-diffracted phases alongside conventional, teleseismic waves as well as regional body-waves. Here, we investigate the properties of P-diffracted waves in terms of waveform characteristics and travel-time measurements as compared to teleseismic P-wave measured by the same methods.

Travel time anomalies, the primary data for tomography, are measured by waveform cross-correlation of data with synthetics, where the synthetics are calculated from fully numerical wave propagation in a spherically symmetric background model. These same numerical tools will be used to calculate the associated sensitivity kernels for tomography (figure, top).

Demonstrating the extent to which waveform modeling can fit real data, we assemble and discuss a global data set of 851,905 Pdiff and 2,368,452 P-wave multi-frequency cross-correlation travel times. Findings are summarized in the Pdiff travel time map (figure, bottom) in which most major structures found in previous studies are detectable, plus several new velocity anomalies.