Toward a three-dimensional tomographic model of the Pacific upper mantle with full resolution and uncertainties

Tomographic models suffer from unevenly distributed noisy data and therefore have complicated resolution and uncertainties that can hinder their interpretation. Using linear Backus & Gilbert inversion, it is possible to obtain tomographic models with resolution and uncertainties in a single step. Using such a method, we aim to produce a three-dimensional tomographic model of the Pacific upper mantle from surface-wave data. To linearise the forward problem, we use finite-frequency theory to describe the sensitivity of surface-wave phase-delays to the three-dimensional shear-wave velocity. We build a data-base of phase-delay measurements for surface-waves that cross the Pacific Ocean. We estimate the data uncertainties caused by measurement errors using a multitaper technique and those caused by poor knowledge of the seismic source and crust by a Monte-Carlo method. Using the Backus & Gilbert approach, the phase-delay dataset, and the data uncertainty estimates, we obtain a model of the shear-wave velocity of the Pacific upper mantle together with its three-dimensional resolution and uncertainties. These allow us to discuss, using robust statistical arguments, the existence and the three-dimensional organisation of structures we expect to see in the Pacific upper mantle, such as plume-like upwellings or small-scale sub-lithospheric convections.