The Nature of Fault Creep and Weakening in the San Andreas System Deduced from Studies of SAFOD Core (Invited)

Our understanding of the processes operative at depth in fault zones is severely hampered by the general inability to study the actively deforming fault rocks in situ. One of the main goals of the San Andreas Fault Observatory at Depth (SAFOD), a key component of Earthscope, was the recovery of core to allow petrographic, chemical, and physical examination of an active, plate-boundary fault at seismogenic depths, something that had never before been attempted. The SAFOD drill site is located 14 km northwest of Parkfield in central California, along a portion of the San Andreas Fault (SAF) that is characterized by a combination of aseismic slip and microseismicity. Innovations in the design and implementation of SAFOD (summarized in Zoback, Hickman and Ellsworth, Scientific Drilling, No. 11, March 2011) led to the identification of two actively creeping fault traces at 2.65 and 2.70 km vertical depth (~112°C). Subsequent multilateral coring operations successfully sampled the two zones of foliated gouge where creep is localized: the 2.6-m-wide central deforming zone (CDZ) and the 1.6-m-wide southwest deforming zone (SDZ). The two gouge zones are closely similar in character, consisting of porphyroclasts of serpentinite and sedimentary rock dispersed in a fine-grained, foliated matrix of Mg-rich smectitic clays (saponite × corrensite). The boundaries of the CDZ and SDZ with adjoining sedimentary rocks of the Great Valley Group are mineralogically, chemically, and texturally sharp. The Mg-rich clay minerals in the gouge zones are interpreted to be the product of fluid-assisted, shear-enhanced metasomatic reactions between the quartzofeldspathic wall rocks and serpentinite that was tectonically entrained in the SAF from a source in the Coast Range ophiolite. Laboratory friction tests indicate that gouge from the CDZ and SDZ deforms stably (i.e., creeps) at anomalously low levels of shear stress (coefficient of friction, μ ~ 0.15), which is sufficient to explain the long-term weakness of the SAF as inferred from stress and heat flow measurements. At the surface, similar rocks are found in the active trace of the SAF to distances of at least 4 km from SAFOD, and they are inferred to connect to one or both of the gouge zones at depth. The metasomatic reactions identified in the CDZ, SDZ and surface outcrops were duplicated in recent friction tests at hydrothermal conditions, in which serpentinite gouge was sheared slowly for ~15 days between quartz-bearing rocks. The serpentinite gouges in those experiments showed an immediate weakening and stabilization of shear, in contrast to the relatively strong and potentially unstable behavior of serpentinite in an ultramafic chemical system. This suggests that fault creep and weakening along the SAF may have initiated as soon as serpentinite was juxtaposed against crustal rocks. Long-term shear of the serpentinite eventually resulted in even more significant, reaction-induced weakening accompanying creep as Mg-rich clays progressively dominated the mechanical behavior of the SAF (μ ~ 0.05 for pure saponite at 100°C). The results obtained from SAFOD may also be applicable to the other creeping faults of the San Andreas system that are associated with the Coast Range ophiolite.