Synthetic inversions for density in the Earth’s interior from seismic waveform and geodetic data

Abstract:

Density plays an important role in the Earth’s interior. On one hand, it is the driving force behind convection; on the other, in imaging efforts, it serves to discriminate between thermal and compositional heterogeneities. Despite its importance, the Earth's 3D density structure remains difficult to infer from surface observations. Gravity measurements, for instance, are sensitive to density but suffer from inherent and strong non-uniqueness. The travel times of seismic body and surface waves are only weakly sensitive to density, and trade-offs with velocity structure are often large.

New opportunities for the recovery of density in the Earth may arise from recently developed full-waveform inversion methods that go beyond traditional traveltime tomography.

In this study, we assess to which extent, and in which way, full-waveform inversion can be combined with gravity and other geodetic measurements to better constrain global-scale density structure. To this end, we perform synthetic tests in 2D in which we systematically assess the effect on density recovery of different (joint) inversion schemes, types of data used and constraints applied.

Our synthetic inversions are based on numerical wave propagation and adjoint techniques to compute seismic Fréchet sensitivity kernels, which are combined with synthetic gravity data to drive gradient-based optimisation schemes. Additional constraints such as a predefined velocity structure, or Earth’s mass and inertia tensor, may be imposed using a projected descent scheme.