Reinterpreting Stress Orientations Near the San Andreas Fault

The strength of the San Andreas Fault (SAF) and the orientation of stress in its vicinity are controversial. Two end-member models have been proposed: the strong fault model (SAF strength equivalent to laboratory samples) and the relatively weak fault model (SAF an order of magnitude weaker than the surrounding crust). These two models predict maximum compressive stress axes at low angle ( ∼30° ) or at high angle ( ∼80° ) to the fault strike, respectively. Several recent studies have attempted to test these models by inverting the focal mechanisms of small earthquakes for stress orientation near the SAF, but are inconsistent as to which model is supported. Particularly at odds are two studies in southern California (Hardebeck and Hauksson, 1999, 2000; and Townend and Zoback, 2001) that use identical focal mechanism data but reach opposite conclusions. We investigate whether the disagreement between studies comes from discrepancies in the observed stress orientations, which would indicate flawed methodology, or from differences in interpretation. Townend and Zoback proposed that the stress orientations reported in their study are different from those of Hardebeck and Hauksson due to the use of different schemes for spatially binning the seismicity for inversion. We test this idea by comparing the results of the two studies over the entire region, and find that the stress orientations are actually very similar, usually to within the uncertainty of the inversion results. The largest differences occur in regions with few earthquakes, as the two techniques mainly differ in how stress orientations are assigned to the areas between earthquake clusters. The inconsistency between the studies therefore lies in the interpretation. The stress orientations reported by both studies are often at intermediate angles ( ∼40-60° ) to the SAF, not consistent with either the high angle or the low angle model, which understandably has confused the interpretation of these results. We perform two additional stress inversions on a high-quality focal mechanism data set for southern California, obtained using a new focal mechanism technique (Hardebeck and Shearer, 2002.) In one inversion, we stack stress orientation vs. distance profiles across all segments of the SAF, to cancel out signals not related to the SAF. In the other, we bin the seismicity using cluster analysis. In both experiments, we again obtain intermediate angles near the SAF. Similar intermediate angles have also been observed in central and northern California (Provost and Houston, 2001, 2003). Neither the strong fault nor the relatively weak fault model satisfactorily describes the SAF. Alternative models are needed which can explain the intermediate stress angles along much of the fault. One such model is a slightly weak SAF, with a coefficient of friction approximately half that of strong faults, or containing pore fluids at elevated but sub-lithostatic pressure. Alternatively, observed stress rotations caused by earthquakes suggest low deviatoric stress magnitude at depth, and variable stress orientation through the seismic cycle due to tectonic loading and seismic release. In this low-stress model, all active faults must be weak, and the observed stress orientations along the SAF reflect its loading state and earthquake history, not its strength.