Synthetic modelling of core-mantle boundary topography

Abstract

In this study, the resolvability of core-mantle boundary topography was studied using synthetically generated seismic body waves. The full-waveform synthetic seismograms calculated at 2130 stations due to four earthquakes on the equator in the PREM model were compared to the seismograms for an Earth consisting of PREM and the core-mantle boundary (CMB) topography model of Li et al. (1991a) scaled to a peak-to-peak amplitude of 8km. The ray theoretical predictions for travel time delays due to CMB topography were compared to delays measured from cross-correlation of full waveform synthetic seismograms generated using SPECFEM3D. The assumption that delays caused by mantle heterogeneities and delays caused by CMB topography are linearly additive was investigated by comparing the separate travel time differences due to mantle heterogeneities and CMB topography to the time difference due to their combined effects. Single and double P and S-wave reflections were studied, as well as Pdiff and Sdiff. We found that ray theory correctly predicts that uplifted topography results in early arrivals of PcP, ScP, ScS, ScSScS, ScSScP, ScPPcP and PcPPcP, provided that the arrivals do not interfere with other phases, that reflection coefficients are higher than 0.1 and that angles of incidence are sufficiently sharp. For all core reflections, delays predicted by ray theory are equal or larger in magnitude than the delays observed in synthetic seismograms, by factors 1.0 to 2.5. For the diffracted phases Pdiff and Sdiff, we found a relationship between positive CMB topography and late arrivals. This delay as a response to uplifted topography can be explained by longer arclengths at larger core radii combined with slow propagation velocities sampled at the underside of the CMB. Linear decomposition is valid for all phases mentioned above.