Thermal and compositional anomalies beneath the North American continent

The determination of the temperature and composition of the mantle is still a remaining challenge. This issue has important implications in understanding the stability of continental roots and evaluating their extent. Nowadays, seismic tomography models have reached a resolution allowing their interpretation in terms of thermal and compositionnal variations. Seismic velocities are primarily sensitive to temperature perturbations but chemical perturbations can not be neglected. However, velocity models can not solely constrain both unknows. And an extra set of data is needed. In this study, density anomalies are used to constrain the structure of the upper mantle beneath North America and the Caribbean region. The velocity data consist of a new regional shear wave velocity model obtained by surface wave tomography, realised by inversion of 7700 fundamental mode Rayleigh wave phase velocity measurements in the period range 40-150 s. The density data are estimated using a relative density-to-shear velocity scaling factor computed for continents by combining regionally filtered seismic and gravity data. The mineralogical variations in the mantle are expressed in terms of the global volumic fraction of iron, the parameter which has the strongest influence on density and velocity. The inferred thermal and iron content anomalies are well constrained by the data and show an age dependence down to a depth of 230+/-50 km. Below the North American craton, the mantle is colder than average and depleted in iron. Maximum values are found at 100 km with /line{δ T} = -440 K and /line{δ Fe} = -4 %, relative to the average mantle. Such a compositional depletion is also reported in other geophysical studies using heatflow measurements or xenolith observations. These chemical and thermal characteristics induce opposite buoyancy forces which could explain the longevity of cratonic lithosphere. In stable continental areas, the signal is of lower amplitudes and beneath the western Cordillera, a tectonically active region, we see no significant thermal or chemical anomaly.