Tsunamigenic potential due to frontal rupturing in the Sumatra locked zone

- Seismic reflection profile of the last gap along Sumatra subduction zone.
- Mechanical modeling used to estimate seismic and tsunamigenic risks.
- A low effective friction is required to reproduce the observed deformation.
- Low effective friction could be due to dynamic weakening during frontal propagation.
- High risk of large tsunami in Mentawai area.

The Sumatra subduction zone is one of the most seismically active subduction zones. Although there have been three Mw≥8.4 earthquakes in the region, including the disastrous 2004 Mw=9.2 Sumatra-Andaman earthquake, a 500 km long patch around Mentawai Islands is still locked and could produce a large megathrust earthquake. If the rupture propagates to the subduction front, as it most likely occurred during the 2004 earthquake, it may lead to a devastating tsunami. Here, we present high-resolution reflection seismic data from the Sumatra locked zone that shows the subduction interface down to 20 km depth. The seismic data also show that the wedge is composed of two layers: a shallow layer formed by mixed to landward vergent thrusts, termed as pop-ups, and a deeper layer showing sub-horizontal reflectors. The lower layer is most probably formed by duplexes, whose roof serves as a pseudo-décollement for the mixed to landward thrust systems. Based on the seismic results, we perform mechanical modeling in order to understand the formation of these structures and to retrieve the associated frictional properties. We first show that the activation of the pseudo-décollement requires (1) either a sudden increase of effective friction along the plate interface or an irregular geometry of the plate interface, (2) a lower effective friction along the pseudo-décollement than along the plate interface. We then show that low effective frictional values (≤0.1) are required to reproduce the observed frontal structures. The low effective friction along the pseudo-décollement could either be due to the presence of a long-term high pore pressure layer or to dynamic weakening associated with earthquakes. Since similar structures are present in the 2010 tsunami earthquake area, we favor the dynamic weakening hypothesis. According to the mechanical modeling, if the next rupture propagates up to the toe rupturing the three most frontal pop-up structures, we could expect at least 5.5 to 9.2 m of frontal horizontal displacement and a frontal uplift of 2 to 6.6 m along the frontal thrusts. This would amplify the uplift of the water column and, as a consequence, generate a large tsunami.