Abstract
The frictional resistance along fault zones during slip is an important parameter in characterizing the mechanics of earthquakes and faulting. However, on the basis of both geomechanical and thermal observations, many plate boundary faults, including the San Andreas Fault (SAF) in California, have been interpreted to slip at shear stresses considerably less than predicted by laboratory-derived friction laws and for hydrostatic pore pressure. An understanding of whether large fault zones truly are "weak" and the potential causes for such weakness are thus key unknowns in the physics of faulting. Here I discuss how integrating hydrogeologic modeling with a wide range of observations and interdisciplinary knowledge provides important insight into these questions.

First, using numerical models of coupled fluid flow and heat transport, and by comparing model results with observational constraints, I show that redistribution of heat by groundwater flow is an unlikely explanation for the lack of a near fault increase in heat flow associated with frictional heating on SAF that supports large shear stresses during slip. I also discuss how drilling into faults quickly after a large earthquake has the opportunity to provide the most clear in situ measure of the average frictional resistance during earthquake slip and how interpretations of low friction from existing data other large fault zones are likely robust.

I then evaluate longstanding hypotheses that invoke regional sources of fluid resulting from metamorphic dehydration reactions within the crust or upper mantle as mechanisms for generating large fluid overpressures within the fault in order to explain its weakness. I calculate reasonable fluid source terms for both crustal and mantle dehydration following the creation of the SAF and show that crustal dehydration sources are too small and short-lived to generate large sustained overpressures, but that there likely exists a much larger and long-lived source of fluids from mantle devolatization. Incorporating these sources of fluid in models with realistic permeability anisotropy, controlled by critically-stressed faults and fractures within the country rock, shows that large localized fluid pressures can be focused within a SAF acting as a hydrologic barrier. The results illustrate that realistic fluid sources and simple permeability architectures have the potential to generate large, localized, and sustained pore pressures that could account for both the absolute and relative weakness of the SAF and possibly other major plate boundary fault zones, and in a manner consistent with a broad range of geological, geochemical, geomechanical and thermal observations including those from the San Andreas Fault Observatory at Depth.