Bayesian imaging of the Hawaiian ULVZ from Sdiff postcursors

Abstract:

The Hawaiian ultra low velocity zone (ULVZ) on the core-mantle boundary (CMB) has been the subject of significant debate over recent years. Attempts to uniquely constrain the shear velocity reduction, as well as ULVZ morphology, are met with difficulty owing to significant computation associated with full waveform modelling and lack of good source-receiver coverage, leading to studies assuming a simplified cylindrical anomaly.

Postcursors of S core-diffracted waves (Sdiff) are sensitive to strong anomalies in the S-wave velocity structure near the CMB, with resulting delayed arrival times of greater than 30 seconds observed for data sampling beneath Hawaii.

Using 2D multi-arrival wavefront tracking simulations (Hauser et al, 2008), we can model the full multi-pathing behaviour of Sdiff postcursors, caused by ULVZs, and compute arrival times for a given input velocity structure. These are then compared to arrival time data, picked from full waveform synthetics or event data by deconvolving Instaseis synthetics (van Driel et al, 2015) of Sdiff arrivals with the waveforms. Capitalising on short computation times on the order of seconds as opposed to hours required for full waveform calculations, we set up a Bayesian inversion algorithm to generate a probabilistic tomographic model of the ULVZ below Hawaii using an efficient level-set parameterisation.

We test our inversion set up by calculating reference arrival times using full waveform synthetics for the true event and station distribution possible for Hawaiian imaging for a number of known ULVZ velocity input models. This allows us to characterise the trade-offs between between size and shear velocity reduction, and the resolvability of the ULVZ morphology. In particular, we observe a strong linear relationship between radius of the ULVZ and the shear velocity reduction, causing a ring of strong standard deviations around the anomaly.

We then perform the Bayesian inversion on a real data set of Sdiff measurements sampling CMB structure beneath Hawaii. Using this novel approach, we seek fundamental new constraints on the physical parameters, and their associated errors, of the ULVZ below Hawaii.