Adaptively parametrized surface wave tomography: methodology and a new model of the European upper mantle

摘要

In this study, we aim to close the gap between regional and global traveltime tomography in the context of surface wave tomography of the upper mantle implementing the principle of adaptive parametrization. Observations of seismic surface waves are a very powerful tool to constrain the 3-D structure of the Earth's upper mantle, including its anisotropy, because they sample this volume efficiently due to their sensitivity over a wide depth range along the ray path. On a global scale, surface wave tomography models are often parametrized uniformly, without accounting for inhomogeneities in data coverage and, as a result, in resolution, that are caused by effective under- or overparametrization in many areas. If the local resolving power of seismic data is not taken into account when parametrizing the model, features will be smeared and distorted in tomographic maps, with subsequent misinterpretation. Parametrization density has to change locally, for models to be robustly constrained without losing any accurate information available in the best sampled regions. We have implemented a new algorithm for upper mantle surface wave tomography, based on adaptive-voxel parametrization, with voxel size defined by both the ‘hit count' (number of observations sampling the voxel) and ‘azimuthal coverage' (how well different azimuths with respect to the voxel are covered by the source-station distribution). High image resolution is achieved in regions with dense data coverage, while lower image resolution is kept in regions where data coverage is poorer. This way, parametrization is everywhere tuned to optimal resolution, minimizing both the computational costs, and the non-uniqueness of the solution. The spacing of our global grid is locally as small as ∼50 km. We apply our method to identify a new global model of vertically and horizontally polarized shear velocity, with resolution particularly enhanced in the European lithosphere and upper mantle. We find our new model to resolve lithospheric thickness and radial anisotropy better than earlier results based on the same data. Robust features of our model include, for example, the Trans-European Suture Zone, the Panonnian Basin, thinned lithosphere in the Aegean and Western Mediterranean, possible small-scale mantle upwellings under Iberia and Massif Central, subduction under the Aegean arc and a very deep cratonic root underneath southern Finland