Abstract

Current models predict that the seismogenic zone along subduction thrusts, where the largest earthquakes nucleate and propagate, does not extend to the forearc mantle below the crust of the upper plate. Stable sliding conditions have been shown to prevail there, particularly along several circum-Pacific margins that underwent great megathrust earthquakes (Mw > 8.5) during the twentieth century. Based on geophysical investigation, we show that the great 2004 Sumatra-Andaman earthquake (Mw = 9–9.3) contradicts these models: not only did it propagate downdip along the interface between the forearc mantle and the subducting plate, but it actually nucleated along this reportedly aseismic part of the interplate contact. Petrological models can therefore underestimate the downdip extent of rupture zones to be expected in megathrust earthquakes, and need to be revised to account for this observation, albeit unusual.

You do not currently have access to this article.