Seismic and multi-parameter 1D reference models of the upper mantle

<p>1D reference Earth models are widely used by the geoscience community and include global, regional and tectonic-type reference models. Seismic 1D profiles are used routinely as reference in imaging studies. Multi-parameter models can also include density, composition, attenuation, lithospheric thickness and other parameters, of interest in a broad range of studies. The recent growth in the number of seismic stations worldwide has yielded a dramatic increase in the global sampling of the Earth with seismic data and presents an opportunity for an improvement in the global and tectonic-type reference models. Concurrent developments in computational petrology have provided methods to constrain self-consistent multi-parameter Earth models with seismic and other data. Here, we use a large global dataset of Love and Rayleigh fundamental mode, phase-velocity measurements, performed with multimode waveform inversion using all available broadband data since the 1990s, and compute phase-velocity maps at densely spaced periods in a broad, 17-310 s period range. We then invert the phase velocity curves averaged globally and across 8 tectonic environments (4 continental: Archean cratons, stable platforms, recently active continents, and active rift zones; and 4 oceanic: old, intermediate and young oceans, and backarc regions) for 1D reference models of the upper mantle. For each tectonic type, a multi-parameter 1D model is computed in a petrological inversion, where the lithospheric thickness and temperature at the bottom of the lithosphere and in the underlying mantle are the inversion parameters, and steady-state conductive lithospheric geotherms are assumed. Lithospheric and asthenospheric compositions are taken from geochemical databases, and seismic velocities, densities and Q are computed from composition, temperature and pressure using computational petrology and thermodynamic databases. The models quantify the age dependence of the lithospheric thickness and temperature in continents and oceans. Radial anisotropy is also determined and shows notable variations with depth and with tectonic environments. For most tectonic types, the smooth, accurate observed phase velocity curves can be fit by the 1D models with a misfit under 0.1-0.2% of the phase velocity value. Additionally, we compute models with minimal complexity of seismic velocity structure, also fitting the data but without a sub-lithospheric low-velocity zone as in the thermal multi-parameter models. These purely seismic models, similar in appearance to ak135, do not correspond to realistic geotherms but provide useful reference for seismic imaging studies in different environments.</p>