Distinct lateral slab geometry and structures in the Costa Rica subduction zone revealed by teleseismic receiver functions

Slab geometry and structures are critical to understanding subduction processes, regional tectonics, and arc volcanism. Located at the convergent plate boundary between the Cocos, Nazca, and Caribbean plates (locally the Panama microplate), the Costa Rica subduction zone is featured by the aseismic subduction of the Cocos Ridge initiated at ~2-3 Ma, and the spatial coincidence with the arc volcanoes that ceased activities since ~5-8 Ma and the uplift of the Talamanca Mountain since ~3 Ma. These phenomena were interpreted by the flat subduction of the Cocos Ridge that has a thick ocean crust. However, this interpretation has been challenged recently by geophysical imaging, which suggests alternative models involving the steep Cocos slab, the doubly convergent Caribbean plate, and the stagnant Nazca slab fragment, leaving the dominant factor driving the regional tectonics enigmatic.

Here, we propose a new teleseismic receiver function (RF) method, Dip Direction Searching Plus (DDS+), to detect weak RF signals associated with dipping interfaces. DDS+ estimates dip directions by fitting the back-azimuthal variations in both radial and transverse RFs. Applying DDS+ to teleseismic data recorded by 17 broadband seismic stations across Costa Rica, we identify positive RF phases with clear back azimuthal variations, indicating dipping interfaces with dip directions of ~N8˚-57˚E (±12˚ on average) beneath 11 stations. The dip direction estimates are consistent to the Cocos Slab2 model. The estimated depths of these interfaces (~13-113 km; ±2.8 km on average) align with the Cocos Slab2 model and the intra-slab seismicity, suggesting the phase are probably Ps conversions from the Moho of the Cocos plate. While the Cocos Moho extends to the depth of ~110 km beneath the northern Talamanca Mountain, it is absent at stations to the south where the slab is expected to subduct beyond 50 km depths. Additionally, we observe a mysterious positive RF phase indicating an interface at ~40-60 km depths in the mantle. This phase was interpreted as either the subducting Caribbean plate Moho (southwestern dipping) or the stagnant Nazca plate Moho (flat) beneath the Talamanca Mountain. Our result reveals no prominent dipping features for this phase, therefore favoring the stagnant Nazca plate Moho interpretation.

Different from previous studies debating continuously flat or steep Cocos subduction, our analysis indicates a steeply dipping Cocos slab to the north and a flat (or truncated) geometry to the south. Therefore, the flat Cocos subduction model cannot explain the volcanic cessation and Talamanca uplift across the entire region. Instead, we propose that the stagnant Nazca slab fragment plays a key role in barricading mantle magma upwelling and thus ceases the arc volcanism. Our study provides new insights into the slab geometry and structures within the Costa Rica subduction zone and the dominant factor shaping the orogenesis and volcanism.