The scaling of pulse duration to seismic moment is estimated for earthquakes along an interplate thrust zone, from digital waveforms recorded by short-period and broad-band instruments of the East Aleutian (Shumagin) Seismic Network. We measure pulse duration using an empirical Green's function technique based on damped time-domain deconvolution. From several thousand events, 22 earthquakes with magnitudes 3.0-7.0 and depths 23-56 km are found to give reliable estimates of pulse duration. Durations are also determined directly from one-parameter nonlinear inversions, for a variety of simple functional forms of source time functions. Symmetric source pulses (boxcar or triangle shapes) fit waveforms better than an asymmetric model [t exp(-2t/D)] for most (62 per cent) of the waveform pairs, while the asymmetric model fits best for only 8 per cent of the data. Pulse duration increases with the size of events, from 0.1 to 10s over the seismic moment (M(0)) range of 10(14) to 3 x 10(19) Nm. When normalized by the cube root of seismic moment, pulse durations show similar to 8 x variation; comparable static stress drop estimates range from 0.2 to 135 MPa. Contrary to predictions of some laboratory and theoretical studies, earthquakes at the deepest part of the thrust zone do not show significantly higher stress drops than do shallower events. Rupture properties, however, show a strong dependence on earthquake size. The three largest events (M(0) > 5 x 10(18) Nm) have the three longest normalized durations, on average 3.8 times longer than those for smaller events. The durations require smaller events to have 10-100 x larger static stress drops, or similar to 4 x faster rupture velocities, or some combination of the two. Possibly, the largest events rupture both strong and weak patches while smaller events just rupture strong patches on the fault surface. The characteristic dimension that separates large from small events, 3-15 km, is comparable to characteristic wavelengths of Pacific basin bathymetry and may reflect the influence of the subducted sea-floor upon fault-zone heterogeneity.
Rx977Times Cited:13Cited References Count:54