2:15 PM - 2:30 PM
[SCG42-03] Reference models of Earth’s bulk structure and heterogeneity from full-spectrum tomography
★Invited Papers
Keywords:Mantle, Surface Waves, Body Waves, Normal Modes, Geodesy, Core
Radial reference Earth models describe average (1-D) physical properties indicative of the bulk composition and dominant processes within the Earth's interior. Tomographic (3-D) models represent contiguity, strength and extent of heterogeneity in intrinsic properties like temperature, composition, grain size, or crystal structure. In contrast to classical radial reference models, we account for the intertwined theoretical and observational effects of lateral heterogeneity on the inferences of bulk structure. Reference datasets and modeling concepts introduced here provide an accurate way to calibrate, validate and infer the average properties of a heterogeneous Earth.
Progress in modeling the Earth’s interior is driven by diverse seismic data, ranging from full seismic waveforms and derivative measurements of body waves (~ 1 – 20 s), surface waves (~ 20 – 300 s), and normal modes (~ 250 – 3300 s). Full-spectrum tomography employs these observations to constrain physical properties – seismic velocity, anisotropy, density, attenuation and the topography of discontinuities. We construct a reference bulk Earth dataset with normal-mode eigenfrequencies and quality factors, surface-wave dispersion curves, body-wave impedance constraints and travel-time curves, and astronomic-geodetic observations. This dataset represents better geographic coverage, wider variety of techniques, reduced uncertainties, and broader frequency range than those used in earlier radial models. New observations indicate reduced seismic velocities in both top and bottom boundary layers of the mantle and strong gradients in the outermost outer core.
We describe three concepts that account for the influence of lateral heterogeneity in the modeling of average elastic, anelastic and density variations from reference datasets. First, horizontal wavelength of the heterogeneity is widely assumed to be much greater than that of the corresponding normal mode in both ray-theoretical and finite-frequency modeling of traveling waves. Discrepancies in local eigenfrequencies due to the unaccounted for volumetric sensitivity with this approximation exceed the data uncertainty for both long-period fundamental spheroidal (Rayleigh waves, T ≧ 200 s) and toroidal modes (Love waves, T ≧ 100 s). Second, non-linear effects from the strongly heterogeneous crustal structure are substantial even for waves that are within the validity limits of the local-eigenfrequency approximation. Non-linear crustal effects on the propagation phase velocity at short periods (T ≦ 100 s) can be substantial compared to the average dispersion observed globally (≧ 40%); contributions from lateral heterogeneity cannot therefore be averaged out linearly in derivations of radial models. Third, current ray coverage of traveling waves results in a substantial bias towards the continental structure in the Northern Hemisphere from available seismic networks.
A new radial reference model (NREM1D) is constructed to account for nonlinear effects due to strong crustal variations and geographic bias in sampling heterogeneities. All physical parameters in NREM1D vary smoothly between the Moho and 410-km discontinuity, thereby excluding the 220-km discontinuity imposed a priori in some earlier models. Radial anisotropy in the upper mantle, with revised estimates (1--2% vSH>vSV) in the mantle lithosphere (24.4 – 80 km) and peak anisotropy (~3.5% vSH>vSV) corresponding to a deforming asthenosphere (~ 150 km), is necessary to fit recent average dispersion curves between 25--250 s. The upper mantle is more attenuating than deeper regions with the lowest Qµ ~ 80 between 150--180 km depth. NREM1D is the first anisotropic model that predicts arrival times of major mantle and core phases in agreement (±0.5 s) with widely used isotropic velocity models that were optimized for earthquake location. NREM1D is readily extendable due to its modular construction and represents accurately the average properties of Earth's bulk constituents.
Progress in modeling the Earth’s interior is driven by diverse seismic data, ranging from full seismic waveforms and derivative measurements of body waves (~ 1 – 20 s), surface waves (~ 20 – 300 s), and normal modes (~ 250 – 3300 s). Full-spectrum tomography employs these observations to constrain physical properties – seismic velocity, anisotropy, density, attenuation and the topography of discontinuities. We construct a reference bulk Earth dataset with normal-mode eigenfrequencies and quality factors, surface-wave dispersion curves, body-wave impedance constraints and travel-time curves, and astronomic-geodetic observations. This dataset represents better geographic coverage, wider variety of techniques, reduced uncertainties, and broader frequency range than those used in earlier radial models. New observations indicate reduced seismic velocities in both top and bottom boundary layers of the mantle and strong gradients in the outermost outer core.
We describe three concepts that account for the influence of lateral heterogeneity in the modeling of average elastic, anelastic and density variations from reference datasets. First, horizontal wavelength of the heterogeneity is widely assumed to be much greater than that of the corresponding normal mode in both ray-theoretical and finite-frequency modeling of traveling waves. Discrepancies in local eigenfrequencies due to the unaccounted for volumetric sensitivity with this approximation exceed the data uncertainty for both long-period fundamental spheroidal (Rayleigh waves, T ≧ 200 s) and toroidal modes (Love waves, T ≧ 100 s). Second, non-linear effects from the strongly heterogeneous crustal structure are substantial even for waves that are within the validity limits of the local-eigenfrequency approximation. Non-linear crustal effects on the propagation phase velocity at short periods (T ≦ 100 s) can be substantial compared to the average dispersion observed globally (≧ 40%); contributions from lateral heterogeneity cannot therefore be averaged out linearly in derivations of radial models. Third, current ray coverage of traveling waves results in a substantial bias towards the continental structure in the Northern Hemisphere from available seismic networks.
A new radial reference model (NREM1D) is constructed to account for nonlinear effects due to strong crustal variations and geographic bias in sampling heterogeneities. All physical parameters in NREM1D vary smoothly between the Moho and 410-km discontinuity, thereby excluding the 220-km discontinuity imposed a priori in some earlier models. Radial anisotropy in the upper mantle, with revised estimates (1--2% vSH>vSV) in the mantle lithosphere (24.4 – 80 km) and peak anisotropy (~3.5% vSH>vSV) corresponding to a deforming asthenosphere (~ 150 km), is necessary to fit recent average dispersion curves between 25--250 s. The upper mantle is more attenuating than deeper regions with the lowest Qµ ~ 80 between 150--180 km depth. NREM1D is the first anisotropic model that predicts arrival times of major mantle and core phases in agreement (±0.5 s) with widely used isotropic velocity models that were optimized for earthquake location. NREM1D is readily extendable due to its modular construction and represents accurately the average properties of Earth's bulk constituents.