Both laboratory and in situ studies of crustal rocks far from faults have typically shown them to have high coefficients of friction (0.65-0.85, dimensionless). Zoback et al. [1993] find such behavior to 6 km depth in the KTB borehole in Germany. However, there is something different about major faults.
In the last Report, Hickman [1991] summarized the evidence for a weak San Andreas fault: the lack of a heat-flow anomaly, and of proven hydrothermal circulation to remove the heat by advection. The lack of any heat-flow anomaly in the new Cajon Pass scientific borehole next to the San Andreas fault limits the vertical integral of shear traction to 1011 N/m [ Lachenbruch and Sass, 1992].
In many places, a nearly
perpendicular/parallel
relationship between principal
stress axes and the fault plane
limits the possible shear
tractions to low values. In-situ
stress measurements in the Cajon
Pass hole show that crustal
blocks have high friction
internally, but that the most-compressive horizontal principal
stress direction (
)
is perpendicular to the San
Andreas, precluding any dextral
shear on that fault today [
Zoback and Healy, 1992]. Next
to the southern San Andreas
fault, Pliocene-Quaternary folds
are forming with axes parallel to
the fault, and extension along
these axes [ Burgmann,
1991]. On the part of the San
Andreas fault that slipped in the
Loma Prieta earthquake, stress
directions from aftershocks imply
no measurable shear traction
remaining after slip on the main
fault [ Zoback and Beroza,
1993]. On the San Francisco
peninsula, for 35 km north of the
Loma Prieta rupture, fault plane
solutions also show
almost
perpendicular to the San Andreas
fault [ Olson and Zoback,
1992]. Mount and Suppe
[1992] collected borehole-elongation data and showed that
in both California and Sumatra,
the most compressive
(
) direction is 70-90
from major strike-slip
faults, requiring them to be very
low-friction surfaces. In the
Sumatran arc the minimum angle
between slip vectors and the
trench is 65-75
, implying
that the dextral fault along the
arc is no stronger than the
water-lubricated subduction shear
zone [ McCaffrey, 1992].
In one of the most exciting
developments of the last four
years, Hauksson [1994]
showed that the regional stress
field in the area of the Landers
earthquake permanently rotated 7-20
in that event, with the
local rotation approximately
proportional to the local fault
slip (and stress drop). This
proves that the seismic stress
drop was a large fraction of the
initial shear stress, and
therefore that the fault was weak
even at the start of slip. This
method of quantitative shear
stress estimation is unique in
the large volume of crust that it
samples; unfortunately, it can
only be applied where a large
earthquake falls within a well-established seismic network.
A nonlinear thin-plate finite-element model of California and its faults [ Bird and Kong, 1994] was optimized with respect to geologic, geodetic, and stress data, with the result that friction on major faults is only 0.12-0.17. Bird [1992b] found typical fault friction in Alaska to be at least this low, with the same method.
A hint that fault weakness may persist for very long times is given by the reactivation of Cretaceous thrust faults in Wyoming-Utah as Quaternary normal faults [ West, 1993].