Retracted: Self-consistent models of Earth’s mantle and core from long-period seismic and tidal constraints

Abstract

Summary In this study we inverted a large set of normal-mode centre frequencies and quality (attenuation) factors, including geodetic data (mass, moment of inertia and tidal response), using self-consistently-built models of the radial elastic and anelastic seismic structure of the Earth. The mantle models are constructed using petrologic phase equilibria in combination with a laboratory-based visco-elastic model that connects dissipation from seismic to tidal periods, whereas seismic properties for a well-mixed and homogeneous core are computed using equations-of-state. Relative to the preliminary seismic reference model (PREM), we find that for the models to fit the observations, mantle P- and S-wave velocities have to be slightly faster and slower, respectively, while outer-core P-wave velocity is slower on account of a different velocity gradient, whereas inner-core velocity structure is similar, within the uncertainties of the inferred model parameters. In terms of density, we find that the mantle is less dense and the outer core more dense than PREM, while the inner core is similar to PREM. These changes are driven in part by the astronomic-geodetic data. The laboratory-based visco-elastic model considered here resolves the anelastic response of Earth’s mantle from long-period seismic (∼100 s) to tidal (18.6 yrs) periods, accounting for both normal-mode and tidal dissipation measurements. To study the impact of the inferred mantle seismic velocity structure, we computed P- and S-wave travel times and compared these to the observations of globally-averaged P- and S-wave travel times from the reprocessed ISC catalog that resulted in an excellent match. In an attempt to further refine the seismic P-wave velocity structure of the outer core, we also considered multiple core-mantle-boundary underside-reflected body wave travel time data. While the match to the underside reflections clearly improves as a result of a steeper velocity gradient in the outer core relative to the normal-modes- and astronomic-geodetic-data-only case, subtle differences nevertheless persist that appear to support a change in velocity gradient in the outermost core, evocative of a stably stratified layer. Finally, as a potential means of refining core composition, we considered the density contrast across the inner-core boundary (ICB) based on our inverted models. The most probable ICB density difference found here is 0.3–0.45 g/cm3, which is in the lower range of earlier body-wave- and normal-mode-based predictions. This suggests that the compositional heterogeneity associated with light-element partitioning, which is considered the principal driving mechanism for the compositional convection that powers the geodynamo, may be less effective than previously thought, calling for exsolution of solids from the liquid outer core as a possible additional source of energy. This would also help address the problem of a young inner core.