Teleseismic Traveltime Tomography of Northern Sumatra

Abstract

We imaged the mantle structure beneath northern Sumatra by inverting high‐quality seismic arrival time data and using a newly developed eikonal equation‐based teleseismic tomography method. Traveltime differences between neighboring stations were reliably extracted by cross‐correlating teleseismic waveforms, which were recorded by 26 stations from January 2009 to January 2018. Both P and S wave tomographic results show the oblique subduction of the Indo‐Australian oceanic lithosphere beneath the Sunda plate. The maximum penetration depth of the subducted slab into the mantle varies roughly from 400 km in the north to 800 km in the south. The plunging fold of the subducted slab, which mimics the shape of the trench and the volcanic arc, has less curvature than that reported in the literature; additionally, our results suggest that the gentle slab fold can be traced to a depth of about 500 km. Furthermore, a slab tear may exist beneath northern Sumatra.

Slab Morphology Beneath Northern Sumatra Revealed by Regional and Teleseismic Traveltime Tomography

“…After filtering the waveform data within the frequency range from 0.02 to 0.2 Hz, the cross‐correlation traveltime differences between all possible station pairs were measured. In teleseismic traveltime tomography, it is always assumed that the heterogeneity outside the study domain is negligible (Liu et al, ; Rawlinson et al, ). To satisfy this assumption, every station pair must be sufficiently close.…”

“…All the tomographic results discussed in the subsequent sections consist of relative velocity perturbation, that is, the difference between M b and M 0 divided by M 0 . The least square problem for both regional and teleseismic traveltime tomography is expressed as follows (Liu et al, ; Puspito et al, ): where A is the sensitivity matrix, X is the discrete model perturbation vector, C d is the a priori data covariance matrix, C m is the a priori model covariance matrix, and ε is the damping factor for velocity inversion. In regional traveltime tomography, b represents the traveltime difference vector; similarly, b denotes the vector containing relative traveltime differences in teleseismic traveltime tomography.…”

“…The fundamental principle of this strategy is to control the maximum model perturbation at each iteration and steadily update the model using a relatively small updating step. More details of this approach can be found in Liu et al (). Notably, the inversion strategy of setting an upper bound is not sensitive to the number of inversion steps, as the number of inversion steps changes according to the upper bound.…”

“…A similar slab model with folding was proposed by Hall and Spakman (), but the curvature of the folded slab is relatively small when compared with the model of Pesicek et al (); Pesicek, Thurber, Widiyantoro, et al (). Recently, Liu et al () developed a slab model with even gentler folding by using teleseimic traveltime tomography.…”

“…It is obvious that the debate on the morphology of the slab beneath northern Sumatra is still ongoing. To resolve this dispute, we apply an eikonal equation‐based traveltime tomography method (Liu et al, ; Tong et al, ) to image the crust and mantle structure beneath northern Sumatra by inverting high‐quality regional and teleseismic traveltime data from the Agency for Meteorology, Climatology and Geophysics of Indonesia (BMKG). After relocating the regional earthquakes in an initial 3‐D model, the regional data are utilized to image the structures of the crust and uppermost mantle.…”