The depth dependence of earthquake duration and implications for rupture mechanisms

THE duration of rupture is a fundamental characteristic of earthquakes, and is important for understanding the mechanics of faulting1,2. The complexity of the seismic source and the incoherence of the high-frequency seismic wavefield often inhibit the identification, location and timing of features in the later part of earthquake rupture. Here we sum many teleseismic records from regional seismic arrays, producing an unusually clear depiction of the earthquake source at short periods by suppressing background noise and coda generated near the receivers. The ending, as well as the beginning, of rupture is clearly identifiable for most earthquakes examined. Measurements of 130 large earthquakes show that near 100 km depth, rupture duration averages 11s when scaled to a moment of 1026 dyn cm; this decreases to 5.5 s at 650 km depth. Models of faulting suggest that duration should be inversely proportional to the shear-wave velocity and the cube root of stress drop. Thus, to explain the observed twofold decrease in duration with depth, stress drops would have to increase by a factor of four, as shear velocity increases with depth by only about 20%. However, observed stress drops show no strong trend with depth3,4, suggesting that the faulting process changes with depth.