Complex anisotropy beneath the Peruvian flat slab from frequency‐dependent, multiple‐phase shear wave splitting analysis

Abstract

[1] Flat or shallow subduction is a relatively widespread global occurrence, but the dynamics remain poorly understood. In particular, the interaction between flat slabs and the surrounding mantle flow has yet to be studied in detail. Here we present measurements of seismic anisotropy to investigate mantle flow beneath the Peruvian flat-slab segment, the largest present-day region of flat subduction. We conduct a detailed shear wave splitting analysis at a long-running seismic station (NNA) located near Lima, Peru. We present measurements of apparent splitting parameters (fast direction ϕ and delay time δt) for SKS, ScS, and local S phases from 80 events. We observe well-defined frequency dependence and backazimuthal variability, indicating the likely presence of complex anisotropy. Forward modeling the observations with two or three layers of anisotropy reveals a likely layer with a trench-normal fast direction underlying a layer with a more trench-oblique (to trench-subparallel) fast direction. In order to further constrain the anisotropic geometry, we analyzed the source-side splitting from events originating within the slab measured at distant stations. Beneath the flat-slab segment, we found trench-normal fast splitting directions in the subslab mantle, while within the dipping portion of the slab further to the east, likely trench-subparallel anisotropy within the slab itself. This subslab pattern contradicts observations from elsewhere in South America for "normal" (i.e., more steeply dipping) slab conditions. It is similar, however, to inferences from other shallowly dipping subduction zones around the world. While there is an apparent link between slab dip and the surrounding mantle flow, at least beneath Peru, the precise nature of the relationship remains to be clarified.

Response of the mantle to flat slab evolution: Insights from local S splitting beneath Peru

“…The delay times, however, are much larger in the model (~ 2.5 s) compared to our results (typically <0.5 s), which may be due in part to the differing frequency content (and therefore Fresnel zone size) between local S (~ 1.2 Hz) and teleseismic (0.05–0.5 Hz) phases used by each study. Strong frequency‐dependent delay times, with lower frequencies yielding longer delays, have already been documented beneath NNA [ Eakin and Long , ].…”

“…The observed pattern of shear wave splitting above the Peruvian flat slab is consistent with suggestions made by earlier work [ Eakin and Long , ], but the interpretation of this pattern involves constraining the precise location of anisotropy within the subduction system. There are three reasonable possibilities: the continental crust, the upper portion of the slab itself, or the mantle material in between.…”

“…There are, however, some deeper events in the catalog (Figures d–f), whose depth estimates are questionable (Figure S5), but if located correctly would have significantly longer paths through the slab, and potentially a greater contribution from slab anisotropy. While there is evidence for substantial anisotropy within the subducting Nazca slab [ Eakin and Long , ], it is not clear that a primary contribution from slab anisotropy could explain the spatial variations in splitting behavior documented in this study.…”

Constraints on Subduction Geodynamics From Seismic Anisotropy

“…Some work has been done to explain specific aspects of trench and slab morphologies; for example, it has been explored to what extent arc curvature is controlled by slab width [ Schellart et al ., ] and whether heterogeneities in the downgoing lithosphere and/or feedback between the downgoing slab and the ambient mantle play a primary role in controlling trench morphology [e.g., Morra et al ., ]. To take another example, flat‐slab subduction such as that observed today beneath Peru is often thought to be a consequence of the subduction of buoyant features on the seafloor such as ridges or seamount tracks [e.g., Cross and Pilger , ; Gutscher et al ., ], but the global correlation between such features and regions of flat or shallow subduction is relatively poor [ Skinner and Clayton , ], and other factors such as wedge viscosity [ Manea and Gurnis , ] or ambient mantle flow [ Eakin and Long , ] may play a role. A major open problem in subduction dynamics is the extent to which trench and slab morphology are affected by mantle flow above, beneath, and at the edges of subducting slabs.…”

Anisotropy in subduction zones: Insights from new source side S wave splitting measurements from India

“…Estimation of the strength of anisotropy beneath the source necessitates suppressing the effect on the receiver side, which may not be straightforward if the receiver side has complex (e.g., two layered) anisotropy [e.g., Russo , , ; Russo et al , ; Lynner and Long , , ; Eakin and Long , ]. Due to this constraint, only those stations which show simple anisotropy that is well characterized can be used [e.g., Russo , , ; Russo et al , ; Lynner and Long , , ; Eakin and Long , ]. To estimate the contribution from the source side, the shear wave splitting parameters obtained from the core refracted (like S K ( K ) S and P K S ) phases are used to correct anisotropy at the receiver side.…”