Global and tectonic-type physical reference models of the upper mantle
- 1Bullard Laboratories, Department of Earth Sciences, University of Cambridge, Cambridge, United Kingdom (yx362@cam.ac.uk, sl2072@cam.ac.uk)
- 2Institut de Ciències del Mar, ICM-CSIC, Barcelona, Spain (cciviero@icm.csic.es)
- 3Department of Physics of the Earth and Astrophysics, Universidad Complutense de Madrid, Madrid, Spain (jfullea@ucm.es)
Physical 1D-average reference models of the Earth offer valuable summaries of the radial variations in rock properties and a reference for geophysical studies. PREM, in particular, has been used widely for >40 years and comprises Vp, Vs, density, radial anisotropy and attenuation profiles, while also fitting the Earth’s mass and moment of inertia. Many of PREM’s features have proven remarkably accurate, despite the limited amount of data used to construct it, but some features are inconsistent with now available data. Also, the upper mantle structure differs so much between Earth’s different tectonic environments that a global average is not quite representative of any of them. The recent growth in seismic station coverage yields very dense data sampling, globally and over different tectonic environments. Here, we use a large global dataset to construct ten 1D, multi-parameter, reference models of the upper mantle, for the globe and for 9 basic tectonic types: cratons; stable platforms; Phanerozoic continents with normal (<46.5 km) and thick (>46.5 km) crust; rifts and continental hotspots; old oceans; intermediate oceans; young oceans; backarcs.
The dataset comprises Love and Rayleigh-wave phase velocities, measured using waveform inversion and all available data since 1990s; surface heat flow measurements; topography/bathymetry. With tomography-based tectonic regionalization, we identify areas within each tectonic environment and compute average dispersion curves in the 20-30 to 310 s period range, which constrain shear velocity and anisotropy in the entire upper mantle.
We then use computational-petrology-based inversion to calculate 1D physical models for the globe and the 9 basic tectonic types. Our non-linear gradient search converges to true best-fitting models. The main unknowns in the inversion are the depth of the lithosphere-asthenosphere boundary (LAB); the geotherm from the LAB down to 400 km depth; radial anisotropy (0-800 km). The steady-state geotherm in the lithosphere is computed from the LAB depth and the radiogenic heat production and thermal conductivity profiles by solving the conductive heat transfer equation. Rock composition and the geotherm determine the density, seismic velocities and attenuation down to 400 km. Seismic velocities in the crust, transition zone (410-660 km) and shallow lower mantle can vary to fit the data. Density below 410 km and all parameters in the core and most of the lower mantle are from PREM. Like PREM, our reference models honour the Earth's mass and moment of inertia.
Small phase-velocity errors and relative data-synthetic misfits (<~0.1%) are necessary to resolve radial trade-offs in the upper-mantle structure. We achieved this by obtaining very accurate dispersion curves and by meticulously tuning the inversion, its parameterisation and regularisation.
The best-fitting models have slightly depleted lithospheric mantle and fertile asthenosphere for most tectonic types. In Archean and Proterozoic continents, the mantle lithosphere is more depleted. No other compositional heterogeneities are required to fit the data. Isotropic-average seismic velocities decrease monotonically from the Moho to the LAB. The geotherms follow the mantle adiabatic temperature gradient in the asthenosphere. Our results provide useful, accurate new reference models for global and regional seismic imaging and other geophysical studies.
How to cite: Xu, Y., Lebedev, S., Civiero, C., and Fullea, J.: Global and tectonic-type physical reference models of the upper mantle, EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-5459, https://doi.org/10.5194/egusphere-egu23-5459, 2023.