Science Reports
Scientific Drilling Into the San Andreas Fault Zone
—
An Overview of SAFOD’s First Five Years
by Mark Zoback, Stephen Hickman, William Ellsworth,
and the SAFOD Science Team
doi:10.2204/iodp.sd.11.02.2011
14
Scientific Drilling, No. 11, March 2011
Science Reports
Abstract
The San Andreas Fault Observatory at Depth (SAFOD)
was drilled to study the physical and chemical processes
controlling faulting and earthquake generation along an
active, plate-bounding fault at depth. SAFOD is located near
Parkfield, California and penetrates a section of the fault that
is moving due to a combination of repeating microearth-
quakes and fault creep. Geophysical logs define the San
Andreas Fault Zone to be relatively broad (~200 m), contain-
ing several discrete zones only 2–3 m wide that exhibit very
low P- and S-wave velocities and low resistivity. Two of these
zones have progressively deformed the cemented casing at
measured depths of 3192 m and 3302 m. Cores from both
deforming zones contain a pervasively sheared, cohesion-
less, foliated fault gouge that coincides with casing deforma-
tion and explains the observed extremely low seismic veloci-
ties and resistivity. These cores are being now extensively
tested in laboratories around the world, and their composi-
tion, deformation mechanisms, physical properties, and
rheological behavior are studied. Downhole measurements
show that within 200 m (maximum) of the active fault trace,
the direction of maximum horizontal stress remains at a
high angle to the San Andreas Fault, consistent with other
measurements. The results from the SAFOD Main Hole,
together with the stress state determined in the Pilot Hole,
are consistent with a strong crust/weak fault model of the
San Andreas. Seismic instrumentation has been deployed
to study physics of faulting—earthquake nucleation, propa-
gation, and arrest—in order to test how laboratory-derived
concepts scale up to earthquakes occurring in nature.
Introduction and Goals
SAFOD (the San Andreas Fault Observatory at Depth) is
a scientific drilling project intended to directly study the
physical and chemical processes occurring within the San
Andreas Fault Zone at seismogenic depth. The principal
goals of SAFOD are as follows: (i) study the structure and
composition of the San Andreas Fault at depth, (ii) deter-
mine its deformation mechanisms and constitutive proper-
ties, (iii) measure directly the state of stress and pore pres-
sure in and near the fault zone, (iv) determine the origin of
fault-zone pore fluids, and (v) examine the nature and signif-
icance of time-dependent chemical and physical fault zone
processes (Zoback et al., 2007).
Detailed planning of a research experiment focused on
drilling, sampling, and downhole measurements directly
within the San Andreas Fault Zone began with an interna-
tional workshop held in Asilomar, California in December
1992. This workshop highlighted the importance of
deploying a permanent geophysical observatory within the
fault zone at seismogenic depth for near-field monitoring of
earthquake nucleation. Hence, from the outset, the SAFOD
project has been designed to achieve two parallel suites of
objectives. The first is to carry out a series of experiments in
and near the San Andreas Fault that address long-standing
questions about the physical and chemical processes that
control deformation and earthquake generation within active
fault zones. The second is to make near-field observations of
earthquake nucleation, propagation, and arrest to test how
laboratory-derived concepts about the physics of faulting
Figure 1
. Map of the Parkfield segment of the San Andreas Fault
showing the epicenters of the 1966 and 2004 Parkfield earthquakes
and the SAFOD drillsite (Rymer et al., 2006). The air photo shows
the terrain in the area of the SAFOD drill site and the epicenter of
the 1966 Parkfield earthquake.
M
I D
D L
E
M
O
U N
TA
I N
C
H
O
LA
M
E
H
IL
LS
C
H
O
LA
M
E
V
A
LL
E
Y
41
46
46
FRESNO CO
Cholame
Parkfield
Gold
Hill
2004
Epicenter
1966
35°52'30"
35°45'
36°00'
500
500
750
750
1000
1000
SAFOD
120°30'
12
0°2
2'3
0"
12
0°1
5'
SAN ANDREAS FAULT
SOUTHWEST FRAKTURE ZONE
10 Km
0
AREA
OF MAP
CA
LIF
OR
NIA
1966
SAFOD
Scientific Drilling, No. 11, March 2011
15
Science Reports
Science Reports
scale up to earthquakes occurring in nature. In the years
following the Asilomar workshop, dozens of planning
meetings were held to synthesize the research questions of
highest scientific priority that were deemed to be operation-
ally achievable. Numerous other meetings were also held
related to site selection and to detailed operational plans for
drilling, sampling, downhole measurements, and long-term
monitoring.
When planning of the EarthScope initiative got underway
at the National Science Foundation (NSF) in the late 1990s,
the project was named SAFOD and became one of the three
components of EarthScope along with the Plate Boundary
Observatory (PBO) and USArray. In 2002, a 2.2-km-deep
Pilot Hole was funded by the International Continental
Scientific Drilling Program (ICDP) and was drilled at the
SAFOD site. The main SAFOD project started when NSF
funded the EarthScope proposal in 2003, with substantial
cost sharing and operational support for SAFOD provided by
the U.S. Geological Survey (USGS), ICDP, and other agen-
cies.
The SAFOD operational plan was designed to address a
number of first-order scientific questions related to fault
mechanics in a hostile environment where the mechanically
and chemically altered rocks in the fault zone are subject to
high mean stress, potentially high pore pressure, and ele-
vated temperature. Some of these questions are listed below.
• What are the mineralogy, deformation mechanisms, and
constitutive properties of fault gouge? Why do some
faults creep? What are the strength and frictional
properties of recovered fault rocks at in situ conditions
of stress, fluid pressure, temperature, strain rate, and
pore fluid chemistry? What determines the depth of
the shallow seismic-to-aseismic transition? What do
mineralogical, geochemical, and microstructural
analyses reveal about the nature and extent of water-
rock interaction?
• What is the fluid pressure and permeability within and
adjacent to fault zones? Are there super-hydrostatic
fluid pressures within some fault zones, and through
what mechanisms are these pressures generated
and/or maintained? How does fluid pressure vary
during deformation and episodic fault slip (creep and
earthquakes)? Do fluid pressure seals exist within or
adjacent to fault zones, and at what scales?
• What are the composition and origin of fault-zone fluids
and gases? Are these fluids of meteoric, metamorphic,
or mantle origin (or combinations of the three)? Is
fluid chemistry relatively homogeneous, indicating
pervasive fluid flow and mixing, or heterogeneous,
indicating channelized flow and/or fluid compart-
mentalization?
• How do stress orientations and magnitudes vary across
fault zones? Are principal stress directions and magni-
tudes different within the deforming core of weak fault
zones compared to the adjacent (stronger) country
rock, as predicted by some theoretical models? How
does fault strength measured in the near field com-
pare with depth-averaged strengths inferred from
heat flow and regional stress directions? What is the
nature and origin of stress heterogeneity near active
faults?
• How do earthquakes nucleate? Does seismic slip begin
suddenly, or do earthquakes begin slowly with accel-
erating fault slip? Do the size and duration of this pre-
cursory slip episode, if it occurs, scale with the magni-
tude of the eventual earthquake? Are there other
precursors to an impending earthquake, such as
changes in pore pressure, fluid flow, crustal strain, or
electromagnetic field?
• How do earthquake ruptures propagate? Do they propa-
gate as a uniformly expanding crack, as a slip pulse, or
as a sequence of slipping high-strength asperities?
What is the effective (dynamic) stress during seismic
faulting? How important are processes such as shear
heating, transient increases in fluid pressure, and
fault-normal opening modes in lowering the dynamic
frictional resistance to rupture propagation?
• How do earthquake source parameters scale with magni-
tude and depth? What is the minimum size earthquake
that occurs on faults? How is long-term energy release
rate partitioned between creep dissipation, seismic
radiation, dynamic frictional resistance, and grain
size reduction (determined by integrating fault zone
monitoring with laboratory observations on core)?
• What are the physical properties of fault-zone materials
and country rock (seismic velocities, electrical resistiv-
ity, density, porosity)? How do physical properties from
core samples and downhole measurements compare
with properties inferred from surface geophysical
observations? What are the dilational, thermoelastic,
and fluid-transport properties of fault and country
rocks, and how might they interact to promote either
slip stabilization or transient over-pressurization dur-
ing faulting?
• What processes control the localization of slip and strain?
Are fault surfaces defined by background microearth-
quakes and creep the same? Would active slip surfaces
be recognizable through core analysis and downhole
measurements in the absence of seismicity and/or
creep?
In addition, a substantial body of evidence indicates that
slip along major plate-bounding faults like the San Andreas
occurs at much lower levels of shear stress than expected,
based upon laboratory friction measurements on standard
rock types and assuming hydrostatic pore fluid pressures
(i.e., it is a weak fault). Yet, the cause of this weakness has
remained elusive (Hickman, 1991). In the context of the San
Andreas, two principal lines of evidence indicate that the