Upper mantle structure of the South American continent and neighboring oceans from surface wave tomography

We present a new three-dimensional SV-wave velocity model for the upper mantle beneath South America and the surrounding oceans, built from the waveform inversion of 5850 Rayleigh wave seismograms. The dense path coverage and the use of higher modes to supplement the fundamental mode of surface waves allow us to constrain seismic heterogeneities with horizontal wavelengths of a few hundred kilometres in the uppermost 400 km of the mantle. The large scale features of our tomographic model confirm previous results from global and regional tomographic studies (e.g. the depth extent of the high velocity cratonic roots down to about 200–250 km). Several new features are highlighted in our model. Down to 100 km depth, the high velocity lid beneath the Amazonian craton is separated in two parts associated with the Guyana and Guapore shields, suggesting that the rifting episode responsible for the formation of the Amazon basin has involved a significant part of the lithosphere. Along the Andean subduction belt, the structure of the high velocity anomaly associated with the sudbduction of the Nazca plate beneath the South American plate reflects the along-strike variation in dip of the subducting plate. Slow velocities are observed down to about 100 km and 150 km at the intersection of the Carnegie and Chile ridges with the continent and are likely to represent the thermal anomalies associated with the subducted ridges. These lowered velocities might correspond to zones of weakness in the subducted plate and may have led to the formation of “slab windows” developed through unzipping of the subducted ridges; these windows might accommodate a transfer of asthenospheric mantle from the Pacific to the Atlantic ocean. From 150 to 250 km depth, the subducting Nazca plate is associated with high seismic velocities between 5°S and 37°S. We find high seismic velocities beneath the Paraná basin down to about 200 km depth, underlain by a low velocity anomaly in the depth range 200–400 km located beneath the Ponta Grossa arc at the southern tip of the basin. This high velocity anomaly is located southward of a narrow S-wave low velocity structure observed between 200 and 500–600 km depth in body wave studies, but irresolvable with our long period datasets. Both anomalies point to a model in which several, possibly diachronous, plumes have risen to the surface to generate the Paraná large igneous province (LIP)