Slab geometry and a diffuse plate boundary beneath Sumatra: constrained using a new receiver function analysis method

Abstract

Geometry and structure of the subducting plate boundary are key to understanding geodynamic processes of subduction and related geological phenomena. Located between the obliquely converging Indo-Australian and Sunda plates, the Sumatran subduction zone is featured by a strongly deformed slab coupling with the overlying plate, and complicated slab-mantle interactions, leading to frequent occurrence of great megathrust earthquakes (e.g., 2004 Mw9.2 and 2005 Mw8.7 events) and extremely intensive magmatism (e.g., Toba supervolcano). Previous seismic studies reveal a rugged slab surface with seamounts, and slab folding and tearing beneath Sumatra, both of which govern the features of earthquake rupture and magma generation associated with fluid release and mantle wedge hydration. However, the details of the slab geometry (e.g., along-strike variation of dip direction and dip angle) and the “slab dehydration-mantle hydration” process across the subducting plate boundary remain poorly known, due to limited data coverage and resolution of these studies.To better reveal the geometry of the slab and the feature of “slab dehydration-mantle hydration” during the oblique subduction, in this study, we develop a Dip Direction Searching (DDS) method to constrain the dipping structure of slab and the nature of the slab upper boundary. In this method, we estimate dip directions of velocity discontinuities by grid search based on the back azimuthal variation of radial receiver functions (RFs). DDS is a single-station-based method thus applicable in the areas with sparse seismic instruments. Synthetic tests demonstrate that the DDS method has higher resolution (with uncertainty of several degrees) in dip direction estimation than traditional RF analysis approaches and is applicable to the cases with strong white noise contamination, incomplete/uneven back azimuthal coverage, <5%-10% crustal and mantle anisotropy, and their compound effects. The method also provides constraints on the thickness and depths of dipping layers.Applying the DDS method, we find a dipping Low Velocity Layer (LVL) commonly beneath the forearc areas and constrain its depths, thickness, and dip directions. The depth and dip direction estimates are highly consistent with the Slab2 model, indicating that the LVL is at the subducting plate boundary. We interpret the lower boundary of the LVL as the subducting oceanic Moho, which is less deformed so its dip direction can represent the dip direction of the whole slab. The slab dip direction gradually increases from 47±5.3˚ in southern Sumatra to 70±10.7˚ in northern Sumatra, indicating an along-strike bending of slab, which is possibly related to the oblique subduction. We find that the dip directions at the upper and lower boundaries of the LVL differ up to 23˚ beneath central Sumatra, indicating the two boundaries are locally unparallel. The thickness of the LVL is estimated to be 10-14 km, larger than those of regular oceanic crusts (~7 km). These observations imply that the LVL is composed by not only the oceanic crust but also a low-velocity serpentinized mantle layer at the top. Therefore, the upper boundary of the LVL represents the serpentinization front, indicating a diffuse plate boundary.