Recordings of the 2004 Parkfield earthquake on the General Earthquake Observation System array: Implications for earthquake precursors, fault rupture, and coseismic strain changes

The 2004 Parkfield earthquake generated a unique set of near-field, high-resolution colocated measurements of acceleration, volumetric strain, and velocity at 11 stations in the General Earthquake Observation System (geos) array. The recordings indicate no precursory strain or displacement was discernable at sensitivities of 10⁻¹¹ strain and 5 × 10⁻⁸ m 25 sec prior to the earthquake at distances of 0.5 to 12 km of fault rupture. Coherent fault-parallel and fault-normal displacement pulses, observed along the fault north of the epicenter, are consistent with model predictions for “fling,” directivity, and displacement for right-lateral, strike-slip fault rupture. The fault-parallel and fault-normal pulses imply apparent rupture velocities of 2.86 ± 0.15 and 3.03 ± 0.24 km/sec, respectively. Unprecedented high-resolution volumetric-strain recordings on opposite sides of the fault show that dynamic strains radiated from ruptured segments of the fault are more than an order of magnitude larger than final coseismic strain offsets associated with fault slip, suggesting that dynamic radiated strain may have contributed to the triggering of failure on unruptured segments. High-resolution recordings show that coseismic strain offsets occur abruptly over time intervals of less than 10 sec near the time of arrival of the dominant radiated fault-parallel and fault-normal displacements. Subsequent measurements show that the strain offsets continue to increase by as much as 69% in 5 min and 300% in 24 hr over that measured during initial fault slip at depth. Estimates of local material parameters from simultaneous measurements of volumetric strain and acceleration confirm seismic calibration factors previously measurable in situ only at tidal periods.