Loci and maximum size of thrust earthquakes and the mechanics of the shallow region of subduction zones

Earthquakes do not extend updip along the plate interface at most subduction zones all of the way to the plate boundary at the trench axis or deformation front. Rather, the shallowest part of that interface moves mainly through stable, aseismic slip. That part of the plate boundary, referred to here as the aseismic zone, occurs along the base of an accreted wedge of young sediments. The probable primary cause for the existence of this aseismic zone is the stable slip properties of the unconsolidated and semiconsolidated sediments in that zone. Subducted sediment is progressively dewatered and underplated to the base of the overriding plate. Through this process, more consolidated rocks eventually come into contact at depth across both sides of the plate boundary. From the point where sufficiently hard rock is found across the plate interface, that interface will change its slip behavior to unstable stick-slip sliding, which is characteristic of consolidated material under most conditions. This type of motion is accommodated seismically as episodic slip in large earthquakes. The location of this transition to seismic behavior, referred to as the seismic front, marks the deep end of the aseismic zone and the top of the seismogenic zone, i.e., that part of the plate interface that moves primarily in thrust earthquakes. Several convergent plate margins are discussed to illustrate the seismic front and the aseismic zone. We find that the seismic front defined by smaller earthquakes that occur during the interval between large events is nearly the same as that for large events as inferred from the locations of their aftershocks. The location of the seismic front is important for making estimates of the maximum possible size of thrust earthquakes along the plate boundary because it delimits the trenchward limit of the potential rupture area of these events. The size of large thrust earthquakes is proportional to rupture area and to at least the cube of the downdip width, W, of that area. Thus by defining the location of the seismic front for convergent margins it is possible to make better estimates of W and hence to deduce the maximum size of future interplate earthquakes. The average repeat time of such events is also related to W and so better estimates of average repeat time are possible from improved knowledge of W. The hypothesis that the location of the seismic front is related to the maximum depth of subduction of unconsolidated sediment has implications for forearc mechanics. We use a mechanical analysis, laboratory modeling, and multichannel seismic information to develop a simple model explaining the growth of forearcs. The outer-arc high or trench-slope break that occurs arcward of most accretionary wedges represents an abrupt change in the critical taper of the accretionary wedge. We argue that that change is caused by a large arcward increase in the strength of the material within the overriding plate. That stronger material is called the backstop. Our laboratory modeling experiments indicate that a backstop with a trenchward-dipping upper surface results in the development of an overlying structure that includes all of the primary morphologic features observed in modern forearcs. In our modeling an accretionary wedge develops trenchward of the backstop, an outer-arc high develops above the trenchward toe of the backstop, and farther arcward a passive forearc basin forms above the stronger material of the backstop. This model is consistent with the observation that earthquakes do not extend updip along the plate interface all the way to the trench axis. We hypothesize that plate motion is accommodated seismically along the base of the consolidated backstop and mostly or entirely aseismically trenchward of the toe of the backstop along the base of the accretionary wedge. At several margins the seismic front is approximately coincident with the outer-arc high, supporting this interpretation.