Weakness of the San Andreas Fault revealed by samples from the active fault zone

Understanding the strength and slip behaviour of tectonic faults is a central problem in earthquake physics and seismichazard assessment1–8. Many major faults, including the San Andreas Fault, are weak compared with the surrounding rock, but the cause of this weakness is debated1. Previous measurements of the frictional strength of San Andreas Fault rocks are too high to explain the observed weakness1,9–13. However, these measurements relied on samples taken at a distance from the active fault or from weathered surface samples. Recent drilling into the San Andreas Fault9 has provided material from the actively slipping fault at seismogenic depths. Here we present systematic measurements of the frictional properties and composition of the San Andreas Fault at 2.7 km depth, including the wall rock and active fault. We find that the fault is weak relative to the surrounding rock and that the fault rock exhibits stable sliding friction behaviour. The fault zone contains the weak mineral smectite and exhibits no frictional healing—bonds in the material do not heal after rupture. Taken together, the low inherent strength and lack of healing of the fault-zone material could explain why the San Andreas Fault slips by aseismic creep and small earthquakes in central California, rather than by large, destructive earthquakes. The apparent weakness of major tectonic faults, as suggested by heat-flow measurements (see for example refs 10,11) and stress orientations (see for example ref. 12), has beenwidely debated1. One of the fundamental unknowns is the absolute magnitude of shear stress acting on faults. Earthquake stress drops are generally between 1 and 10 MPa; however, the shear stress that is relieved during an earthquake is thought to be a small percentage of the absolute fault strength. If the major faults that bound Earth’s tectonic plates are strong, shear stresses at seismogenic depths are likely to be 100MPa or higher, whereas this number could be a factor of 10 ormore lower if these faults are frictionally weak. A number of studies have concluded that the San Andreas Fault (SAF) is mechanically weak10–12. Several mechanisms have been proposed, including the presence of weak minerals such as talc2 or various clays3, the formation of fabric within the fault zone4,5 or pore pressure far in excess of hydrostatic6–8. Each of these mechanisms is supported by different aspects of a currently incomplete understanding of fault properties at seismogenic depths. One central problem in resolving the debate about the strength of the SAF is that there are no measurements of frictional strength for carefully located samples from the actively slipping fault at seismogenic depths. Existing data are limited to shear zones in the country rock away from the active fault and/or to samples from weathered surface outcrop exposures (Fig. 1a), and indicate friction coefficients of ∼0.30–0.55 (refs 15–17), which are considerably higher than those required to satisfy weak-fault models10–12.