Abstract
We use seismic tomography to investigate the state of the supraslab mantle beneath northern Chile, a part of the Nazca-South America Plate boundary known for frequent megathrust earthquakes and active volcanism. We performed a joint inversion of arrival times from earthquake generated body waves and phase delay times from ambient noise generated surface waves recorded by a combined 360 seismic stations deployed in northern Chile at various times over several decades. Our preferred model shows an increase in Vp/Vs by as much as 3 per cent from the subducting slab into the supraslab mantle throughout northern Chile. Combined with low values of both Vp and Vs at depths between 40 and 80 km, we attribute this increase in Vp/Vs to the serpentinization of the supraslab mantle in this depth range. The region of high Vp/Vs extends to 80–120 km depth within the supraslab mantle, but Vp and Vs both increase to normal to high values. This combination, along with the greater abundance of ambient seismicity and higher temperatures at these depths, suggest that conversion from basalt to eclogite in the slab accelerates and that the fluids expelled into the supraslab mantle contribute to partial melt. The corresponding maximum melt fraction is estimated to be about 1 per cent. Both the volume of the region affected by hydration and size of the wave speed contrasts are significantly larger north of ∼21°S. This latitude also delimits large coastal scarps and the eruption of ignimbrites in the north. Ambient seismicity is more abundant north of 21°S, and the seismic zone south of this latitude is offset to the east. The high Vp/Vs region in the north may extend along the slab interface to depths as shallow as 20 km, where it corresponds to a region of reduced seismic coupling and overlaps the rupture zone of the recent 2014 M8.2 Pisagua earthquake. A potential cause of these contrasts is enhanced hydration of the subducting oceanic lithosphere related to a string of seamounts located on the Iquique Ridge of the Nazca Plate.