Earthquake Stress Drop, Source Spectra from Laboratory Experiments, and the Apparent Unsteadiness of Frictional Slip

During frictional failure experiments, high sample rate slip, stress, and slip velocity measurements were used to determine seismic source properties. The experiments were conducted over a range of coseismic total energy release. Near‐field measurements were used to estimate stress drops using spectral approaches that are analogous to the standard seismological moment corner determined $Δσ$ and acceleration amplitude $Δσarms$ stress drops. Rapid frictional sliding produces slip velocity amplitude spectra that vary with $frequency−1$ so long as the energy dissipated by friction approaches the total energy release. Under these conditions, $Δσ$ and $Δσarms$ are similar in size. Because these are nearly direct measurements of fault slip velocity, and they produce flat acceleration spectra above the corner frequency, friction in these experiments seems to produce dynamic acceleration and deceleration at all resolved frequencies. The physical origin of this unsteady motion is not known exactly but likely reflects rapid dynamic variations in fault strength or applied stress. On the other hand, when the total energy stored before failure is artificially made somewhat larger relative to that which can be dissipated by friction, the velocity spectra vary with $frequency−2$ immediately below the corner. At these conditions, $Δσ$ and $Δσarms$ diverge. The excess stored energy is preferentially partitioned into radiated energy upon failure. Collectively the experiments imply that to produce unsteady, white noise accelerations that are observed for natural earthquakes, requires that at least 95% of the energy released does not escape the source region to be radiated to the far field.