Obtaining robust estimates of the Vs/Vp ratio in the Earth’s lowermost mantle

Seismic tomography provides valuable insights into the structure, composition and evolution of the mantle. However, the origin of structures like the Large-Low-Velocity-Provinces (LLVPs) in the lowermost mantle remains debated. Their velocity anomalies have been interpreted to be due to purely thermal or also compositional variations, with implications for mantle circulation, the evolution of the core and the Earth’s heat budget.

To uniquely interpret seismic structures such as the LLVPs, it is crucial to constrain the relationships between different seismic observables, e.g. the ratio between shear-wave velocity (Vs) and compressional-wave velocity (Vp) variations. Joint inversions of seismic velocities have been performed, but their velocity amplitudes may be biased, uncertainties are typically not provided, and the resolution of Vs and Vp structures generally differs in existing models.

To overcome these issues, we make use of the recently developed SOLA method (Zaroli, 2016), which is based on a Backus-Gilbert philosophy. Instead of finding a model with a particular data fit, we aim to construct model averages of the true Earth with uncertainties, whilst having a control on the model resolution. This direct control on resolution enables us to build Vs and Vp models that sample the same parts of the mantle, and therefore to robustly constrain the Vs/Vp ratio.

Here, we test this philosophy by applying the SOLA method to normal modes. These free oscillations of the Earth are particularly useful to study the relationships between seismic velocities as they are directly sensitive to multiple physical parameters, including Vs, Vp as well as density. We illustrate our approach and discuss the trade-off between uncertainties and resolution using synthetic tests for both Vs and Vp, before showing real data inversions. Finally, we discuss the implications of our results for the Vs/Vp ratio in terms of mantle temperature and composition.