Investigating Source Mechanisms of Deep-Focus Earthquakes at the Peru-Brazil Border with Regional and Teleseismic Data

The occurrence of deep-focus earthquakes (h > 300 km) is restricted to a handful of regions worldwide, generally associated with subduction zones. In particular, the South American subduction zone hosts two narrow belts of deep-focus seismicity with depths greater than ~500 km along the Peru-Brazil border and Bolivia/northern Argentina. This subduction zone has a thermal parameter of Φ < 2500 km and is regarded as a warm end-member. Only in 2015, the USGS catalog listed up to 25 deep-focus events in the Peru-Brazil belt, with magnitudes and depths ranging from 4.0 to 7.6 Mw and 515 to 655 km, respectively. Notably, this sequence included a well-investigated doublet of two 7.6 Mw events occurring 5 min apart trailed by a number of aftershocks of magnitude 4.0 Mw or larger. Published focal mechanisms for the main doublet display predominantly double-couple components that closely agree with the GCMT solution (E1: 350°, 39°, -80° and E2: 350°, 30°, -81°), suggesting shear failure at those depths. Mechanisms capable of shear instability at those large depths traditionally include dehydration embrittlement, transformational faulting, thermal runaway or a combination of those. Aiming at investigating the physical mechanism responsible for these deep-focus events, we are using a combination of regional and teleseismic recordings from the Brazilian Seismographic Network (RSBR) and other regional and national networks in the continent to determine focal mechanisms for deep-focus earthquakes (M > 4) that occurred between 2014 and 2022. The mechanisms are being determined through a Cut and Paste approach, which compensates for inaccuracies in the velocity model through independent relative time shifts between observations and predictions for P, SV and SH wave trains sampling both the upper and lower hemispheres of the focal sphere. The results on the 2015 doublet, using the full dataset (regional and teleseismic stations), indicated two very similar normal faults fully consistent with the GCMT solutions, at the preferred depths of 616 (E1) and 621 (E2) km. Preliminary inversions using only regional networks (RSBR) for 15 smaller earthquakes (4.3 < M < 7.1) also yield normal mechanisms with T axes oriented roughly E-W. This apparent uniformity of the focal mechanisms for the South-American deep-focus earthquakes, with near-vertical P axes and near-horizontal (east-west-oriented) T axes, strongly suggests vertical compression along the subducting plate is the main source of stress driving deep-focus seismicity. Down-dip compression is expected from either buoyancy forces, equilibrium phase transformations or a metastable olivine wedge (MOW); however, how earthquakes are nucleated at those depths is harder to explain. Transformational faulting within the MOW has been the preferred mechanism in cold slabs, while in warm slabs its presence has been more debated due to wedge size being expected to decrease with temperature. Transformational faulting in other metastable minerals such as enstatite is our preferred alternative, as dehydration embrittlement and thermal runaway seem to lack the capacity of triggering earthquakes at those large depths.