A74. Bird, P. [2008] Fault slip rates in the western U.S. from a joint fit to geologic offsets, GPS velocities, and stress directions (abstract), Eos Trans. AGU, 89(52), Fall Meeting Supplement, Abstract T23E-08.

I merge the SCEC, WGCEP, PBO, & WSM community datasets in deformation models for the western US.  In California I use: (1) fault traces, dips, and slip senses from WGCEP Fault Models 2.1 or 2.2; (2) fault offset rates and uncertainties obtained by Bird [2007, Geosphere, 3(6)] from offsets in the USGS Paleosites Database; (3) a 2006 California joint GPS solution for interseismic benchmark velocities by Shen, King, Wang, & Agnew; and (4) stress-direction indicators from World Stress Map.  In other western United States I use: (5) my collection of fault traces and offset rates as documented in Bird [2007]; and (6) selected GPS velocities from PBO.  All are fit by weighted least-squares in kinematic F-E program NeoKinema.  As described previously, this program (a) interpolates stress directions to determine their uncertainties, (b) attempts to minimize off-fault strain-rates and align them with stress, and (c) iteratively corrects geodetic velocities from short-term to long-term using local dislocation-in-halfspace corrections.

All datasets can be fit at a common level of 1.8 standard errors (RMS or N2 norm).  If “acceptable” fit is defined as N2 < 2 for all datasets, there is a range of acceptable models, defining a range of long-term fault slip rates and (anelastic) continuum strain-rates.

In preferred model GCN2008060, the mean long-term slip rates for trains of the San Andreas fault are (SE to NW): Coachella 15 mm/a, San Gorgonio Pass-Garnet Hill 6, San Bernardino South 12, San Bernardino North 19, Mojave South 16, Mojave North 17, Big Bend 15, Carrizo 25, Cholame 26, Parkfield 31, Creeping 29, Santa Cruz Mt. 23, Peninsula 18, North Coast 16, & Offshore 9 mm/a.  Up to Cajon Pass, these all agree with 2007 WGCEP [2008], but my Mojave N & S and Big Bend rates are much slower, my Carrizo & Cholame rates are marginally slower, and my North Coast & Offshore rates are much slower.

These differences are due to greater amounts of permanent (anelastic) straining off the mapped fault traces in NeoKinema, relative to the elastic-microplate models of 2007 WGCEP [2008].  I have not been able to lower the RMS continuum strain rate in these models below 5´10^-16 /s (=1.6%/Ma).  Such distributed straining results from gaps and geometric incompatibilities in the fault network and from geologic/geodetic discrepancies.  This straining probably also occurs on faults (which are not part of WGCEP Fault Models), and it probably also produces earthquakes.