The Rupture Process of Subduction Zone Earthquakes

Eugene Humphreys, University of Oregon

Abstract

Consider the following three observation-based inferences.

1) 2-D modeling of North America plate stress finds that plate boundaries, including the Cascadia subduction zone, hold 1.5-2 X 1012 N per meter of plate boundary fault length. If this stress is distributed uniformly over 40 km of down-dip fault length, this yields ~40-55 MPa of shear stress on the megathrust. Grainsize studies of exhumed faults (including from subduction zones) yield even greater shear stress estimates (Platt & Behr, 2011).

2) The stress drop for most subduction zone earthquakes is in the range 1-10 MPa (Allmann & Shearer, 2009). Hence, subduction zone earthquakes, like other earthquakes, release only a small fraction of the shear stress present on the fault. This is consistent with the observation-based slip pulse model for earthquakes (Heaton, 1990), where slip at any point on the fault occurs for only 1-2 seconds, after which slip stops as the rupture front propagates forward.

3) If the shear stress on the fault were not low during slip, frictional heating would melt the fault zone, which would lead to neat total stress drop. Since each fault-zone melting and total stress drop are relatively uncommon, the conclusion is that shear stress during sliding is quite low (less than 5 MPa below 10 km depth). There must be a rather profound fault weakening process active during fault slip (and there are many suggestions).

We conclude that as rupture approaches a location on the megathrust, shear stress increases from an initial 40 MPa (or more) until a rupture is reached (100-300 MPa), then drops to a very low value during sliding. After 1-2 seconds, sliding stops and the shear stress on the fault recovers to near its initial value of 40 MPa (or more). The fault weakening process, being active for only 1-2 sec, is transient.

Session

Subduction zone geology