Description: We present the first detailed 2D seismic tomographic image of the trench-outer rise, fore- and back-arc of the Tonga subduction zone. The study area is located approximately 100 km north of the collision between the Louisville hot spot track and the overriding Indo-Australian plate where ~80 Ma old oceanic Pacific plate subducts at the Tonga Trench. In the outer rise region, the upper oceanic plate is pervasively fractured and most likely hydrated as demonstrated by extensional bending-related faults, anomalously large horst and graben structures, and a reduction of both crustal and mantle velocities. The 2D velocity model presented shows uppermost mantle velocities of ~7.3 km/s, ~10% lower than typical for mantle peridotite (~30% mantle serpentinization). In the model, Tonga arc crust ranges between 7 and 20 km in thickness, and velocities are typical of arc-type igneous basement with uppermost and lowermost crustal velocities of ~3.5 and ~7.1 km/s, respectively. Beneath the inner trench slope, however, the presence of a low velocity zone (4.0–5.5 km/s) suggests that the outer fore-arc is probably fluid-saturated, metamorphosed and disaggregated by fracturing as a consequence of frontal and basal erosion. Tectonic erosion has, most likely, been accelerated by the subduction of the Louisville Ridge, causing crustal thinning and subsidence of the outer fore-arc. Extension in the outer fore-arc is evidenced by (1) trenchward-dipping normal faults and (2) the presence of a giant scarp (~2 km offset and several hundred kilometers long) indicating gravitational collapse of the outermost fore-arc block. In addition, the contact between the subducting slab and the overriding arc crust is only 20 km wide, and the mantle wedge is characterized by low velocities of ~7.5 km/s, suggesting upper mantle serpentinization or the presence of melts frozen in the mantle.

Description: The deep structure of the south-central Costa Rican subduction zone has not been studied in great detail so far because large parts of the area are virtually inaccessible. We present a receiver function study along a transect of broadband seismometers through the northern flank of the Cordillera de Talamanca (south Costa Rica). Below Moho depths, the receiver functions image a dipping positive conversion signal. This is interpreted as the subducting Cocos Plate slab, compatible with the conversions in the individual receiver functions. In finite difference modeling, a dipping signal such as the one imaged can only be reproduced by a steeply (80°) dipping structure present at least until a depth of about 70–100 km; below this depth, the length of the slab cannot be determined because of possible scattering effects. The proposed position of the slab agrees with previous results from local seismicity, local earthquake tomography, and active seismic studies, while extending the slab location to greater depths and steeper dip angle. Along the trench, no marked change is observed in the receiver functions, suggesting that the steeply dipping slab continues until the northern flank of the Cordillera de Talamanca, in the transition region between the incoming seamount segment and Cocos Ridge. Considering the time predicted for the establishment of shallow angle underthrusting after the onset of ridge collision, the southern Costa Rican subduction zone may at present be undergoing a reconfiguration of subduction style, where the transition to shallow underthrusting may be underway but still incomplete.