Very shallow source of the October 2010 Mentawai tsunami earthquake from tsunami field data and high-rate GPS

"Tsunami earthquakes," which produce very large tsunamis compared to those expected from their magnitude, have long puzzled geoscientists, in part because only a handful have occurred within the time of modern instrumentation. The Mw 7.8 Mentawai earthquake of 25 October 2010, which occurred seaward of the southern Mentawai islands of Sumatra, was such an event. This earthquake triggered a very large tsunami, causing substantial damage and 509 casualties. Detailed field surveys we conducted immediately after the earthquake reveal maximum runup in excess of 16 m. The Sumatra GPS Array (SuGAr) recorded beautiful 1-sec data for this event at sites on the nearby islands, making this the first tsunami earthquake to be recorded by a dense, high-rate, and proximal GPS network, and giving us a unique opportunity to study these rare events from a new perspective. We estimate a maximum horizontal coseismic GPS displacement of 22 cm, at a site ~50 km from the epicenter. Vertical displacements show subsidence of the islands, but are on the order of only a few cm. Comparison of coseismic offsets from 1-sec and 24-hr GPS solutions indicates that rapid afterslip following the earthquake amounts to ~30% of the displacement estimated by the 24-hr solutions. The coseismic displacements are smaller than expected, and an unconstrained inversion of the GPS displacements indicates maximum fault slip of ~90 cm. Slip of this magnitude will produce maximum seafloor uplift of <15 cm, which is clearly not enough to produce tsunami runup of 16 m. However, investigation of the model resolution from GPS indicates that we are limited in our ability to resolve slip very close to the trench. We therefore deduce that to obtain the adequate level of slip and seafloor uplift to trigger the tsunami, the rupture must have occurred outside the resolution of the GPS network, i.e., at very shallow depths close to the trench. We therefore place prior slip constraints on the GPS inversion, based on preferred values from tsunami modeling of the field data. In the constrained inversion, the small GPS displacements force any slip close to the islands back down to much lower values than the a priori estimates, leaving only a very narrow and shallow strip of high slip close to the trench. In this presentation we will show several possible models that include slip on either the megathrust itself or a shallow splay fault, with maximum slip of ~7 m and ~4 m, respectively. This very shallow slip raises questions about the seismic hazard potential of a region of the fault that is often considered to be aseismic. Particularly, these results suggest that when model resolution is not adequate for making determinations of the updip limit of the seismogenic zone of subduction faults, it may be best to assume that it extends all the way to the trench.