Earthquake nucleation and rupture at a range of scales : laboratories, gold mines, and subduction zones

Abstract

We measured spectral and time-domain properties of seismic events over a size range that spans magnitudes M [approx.] -2 to 8 in order to study earthquake source processes. In addition, we conducted laboratory experiments to study interseismic behaviors that can influence earthquake nucleation and we developed a model of eathquake rupture to explain the scaling behaviors we observe. To bridge the scale gap between laboratory data and global seismic observations, we studied data from five deep gold mines in the Far West Rand region of South Africa. These mines are seismically active due to daily underground blasting and record [approx.] 1000 events per day from -2 =/< M =/< 3+ close to their sources. Frequency-magnitude relations, spatio-temporal clustering relations and observations of seismic spectra provide evidence that there are two types of events that occur in these mines, which we designate as Type A and Type B. Type-A events are fracture-dominated ruptures of previously intact rock and show an upper magnitude cutoff at M [approx.] 0.5. They are tightly clustered in space and time and occur close to active stope faces. They have scaling properties that agree with other studies of fresh-fracturing seismicity in that apparent stress decreases with magnitude and stress drop increases with magnitude.

In contrast, Type-B events are temporally and spatially distributed throughout the active mining region. They have a lower magnitude cutoff at M [approx.] 0. From frictional scaling laws and observations of source spectra, we deduce that that this lower magnitude cutoff represents the critical patch size for earthquake nucleation in this mining environment. We find that the critical patch size is on the order of 10 m with a critical slip distance on the order of 10-4 m. Type-B events have scaling properties that match extrapolations from tectonic earthquakes. For example, apparent stress and particle velocity increase with magnitude. We develop a kinematic model of increasing rupture velocity with increasing source size to account for the observed scaling of frictional shear events.