Rupture propagating at the forbidden speed: is fault zone structure playing a role?

Several high-resolution studies indicate the presence of slow seismic velocities along major fault segments (e.g., San Andreas, San Jacinto, Landers, Hector Mine, Nojima and North Anatolian faults). We find through dynamic rupture simulations that earthquake ruptures can be profoundly affected by waves trapped in fault zones. Key rupture properties such as rise time (the slip duration at a given point), slip rate and rupture speed are highly dependent on the velocity contrasts and thickness of such structures. Rise time is controlled by the travel time of reflected waves. Slip rate is influenced by the slow seismic velocities of fault zones. Supershear ruptures are induced by head waves that propagate along the interface between fault zones and host rock. Such supershear transitions in fault zones are possible even when stresses on the fault are too low to allow supershear ruptures in a homogeneous medium. A prominent feature of ruptures in fault zones is that they can stay at a speed close to the S wave speed of the host rock. However, ruptures in a homogeneous medium can not propagate steadily at a speed between the Rayleigh and S wave speeds. Waveform analyses of the 2009 Mw 4.6 Inglewood earthquake and of an aftershock in the 2003 Big Bear sequence both reveal a rupture speed close to S wave. Here we test the hypothesis that these events occur inside fault zones by modeling the high-frequency waves of ML 2-ML 3.5 clusters inside a region smaller than 1km in Big Bear. Preliminary analyses of waveforms at each station show a complex behavior of S waves despite more coherent P waves. This is consistent with our synthetic tests which show that S waves are much more affected by the structure for events inside fault zones.