UC Berkeley

Mapping Earth’s density structure is crucial for understanding how our planet, and others like it form and evolve. When accompanied by existing images of seismic wave-speed structure, and inferences about possible compositions and pressure-temperature conditions, density constraints can help to reveal some of the most fundamental properties of Earth’s interior. However, despite major advances in the resolution of seismic velocity structure, and a growing bank of geochemical and experimental reference points, the density structure of the Earth remains comparatively poorly constrained.

Estimates on Earth’s radial (depth dependent, or “1D”) structure have remained much the same as that proposed nearly 40 years ago, despite significant improvements in computational resources and a much larger catalogue of geophysical data. We map onto this radial structure lateral (or “3D”) heterogeneity, which drives the dynamics of mantle circulation. The largest such 3D structures in the Earth, the two so-called LLSVPs (Large Low Shear Velocity Provinces) situated below Africa and the Pacific Ocean, extend laterally for thousands of kilometers and extend 100s of kilometers upwards from the mantle’s base. Yet, even the most recent studies that have utilized exceptional datasets, cannot agree on whether LLSVPs are denser or less dense than their surroundings. One recent study found observed deformation due to Earth’s body tides required LLSVPs be significantly denser than their surroundings (Lau et al., 2017). Meanwhile, that same year, a study of Stoneley mode splitting functions favored they were buoyant (Koelemeijer et al., 2017). Unable to find consensus on the signs of density heterogeneity, let alone amplitude, dynamicists and experimentalists struggle to uniquely constrain the compositions and exact pressure-temperature conditions of some of the largest 3D structures in Earth’s interior.

To address both issues of 1D and 3D density structure, in this thesis, we exploit a part of the seismic response that is most sensitive to density: Earth’s free oscillations, or normal modes. Earth’s long-wavelength normal modes offer the strongest constraint on Earth’s density variations. Herein we revisit constraints on 1D density profile of the poorest constrained region – the inner core using modern catalogues of mode observations and utilizing increased computational resources. Additionally, we explore the viability of a new dataset of normal modes with unique sensitivity to the region containing the LLSVPs, i.e., the core-mantle boundary Stoneley modes. We explore: (1) the errors introduced by common approaches from their theoretical formulation; (2) the uncertainties introduced by other properties assumed when simulating their response; and (3) the expected magnitude of signal associated with proposed density heterogeneity in the lowermost mantle. Finally, we revisit constraints on LLSVP density. Using conceptual density models, we demonstrate that seemingly contrasting images of LLSVP density from previous Stoneley mode and body tide studies can be reconciled, by a unifying model of a thermal, buoyant LLSVP, underlain by a very dense, thin (~100km), compositionally distinct, base – to which Stoneley modes are largely insensitive.