By Mark Zoback, Stephen Hickman, William Ellsworth, and the safod science Team doi: 10. 2204/iodp sd. 11. 02. 2011

16  

Scientific Drilling, No. 11, March 2011

Science Reports

repeating microearthquakes, M3 and smaller, occurring on 

the fault at depths of 2–12 km (Waldhauser et al., 2004). The 

Parkfield segment of the fault hosts the well-studied seven 

M6 earthquakes that have ruptured since 1857 (Bakun and 

McEvilly, 1984). Slip distributions for the last two Parkfield 

earthquakes—on 28 June 1966 and 28 September 2004—

determined using geodetic measurements, indicate that the 

ruptures terminated a few kilometers southeast of SAFOD 

(Murray and Langbein, 2006; Harris and Arrowsmith, 2006 

and papers therein).

Beginning at the Asilomar meeting, site selection com-

mittees winnowed down eighteen potential sites to four, and 

eventually the northwest end of Parkfield segment was se-

lected. The geology seemed ideal since Salinian granite on 

the west side of the fault was expected to be juxtaposed 

against Franciscan melange on the east side, so a major geo-

logic discontinuity was expected when crossing the fault at 

depth. Also, the San Andreas Fault is quite active in the area, 

exhibiting a combination of aseismic creep and frequent 

microearthquakes that would help define the exact location 

of the active fault trace at depth. In addition, more is known 

about this section of the San Andreas than any other, due to 

the intense interest in capturing a M~6 earthquake within a 

dense network of instrumentation.

After selecting the Parkfield segment of the San Andreas 

for the SAFOD experiment, the next question of particular 

importance was where 

exactly to site the borehole. 

The site chosen (Fig. 1) 

was selected near Middle 

Mountain because re- 

peating microearthquakes 

could be reached at the 

shallowest depth possible 

close to the fault to limit  

the horizontal reach of the 

borehole. As shown in the 

photo inset of Fig. 1, the 

selected site is a broad, 

relatively flat area where 

 

a 5-acre drill pad could 

 

be constructed 1.8 km 

southwest of the surface 

trace of the fault. Once this 

area was identified, a num-

ber of detailed geophysical 

and geologic site studies 

were carried out to allow 

results from SAFOD to be 

placed in the appropriate 

geological and geophysical 

context and to assure 

 

that the drill site selected 

would not encounter any 

large-scale faults or struc-

fault has low frictional strength: the absence of 

frictionally-generated heat, and the orientation of the 

maximum principal stress in the crust adjacent to the fault.  

A large number of heat flow measurements show no 

 

evidence of frictionally generated heat adjacent to the San 

Andreas Fault (Lachenbruch and Sass, 1980, 1992; Williams 

et al., 2004), which implies that shear motion along the fault 

is resisted by shear stresses approximately a factor of five 

less than fric-tional strength of the adjacent crust. This 

observation is sometimes referred to as the San Andreas 

stress/heat flow paradox. Saffer et al. (2003) showed that it 

is highly unlikely that topographically driven fluid flow has 

an appreciable effect on these heat flow measurements, 

indicating that the lack of frictionally-generated heat in the  

vicinity of the San Andreas Fault is indeed indicative of low 

aver-age shear stress levels acting on the fault at depth.  

In addition to the heat flow data, the orientation of principal 

stresses in the vicinity of the fault also indicates that 

right-lateral strike slip motion on the fault occurs in 

 

response to low levels of shear stress (Zoback et al., 1987; 

Mount and Suppe, 1987; Oppenheimer et al., 1988). 

Why Parkfield? SAFOD is located in central California 

(Fig. 1) near the town of Parkfield, at the transition between 

the locked (i.e., seismogenic) portion of the fault to the 

southeast and the segment of the fault to the northwest 

where slip dominantly occurs by aseismic creep. The fault is 

seismically active around SAFOD with numerous sites of 

Figure 2.

 Microearthquakes selected for targeting with SAFOD. [A] 3-D perspective view of the seismicity with 

respect to the path of the SAFOD borehole, with north pointing up, east to the right, and depth down (all axes 

in km). [B] View of the plane of the San Andreas Fault at about 2.7-km depth looking to the northeast. The red, 

blue, and green circles represent seismogenic patches of the San Andreas Fault that produce nearly identical, 

regularly repeating microearthquakes termed the San Francisco (SF), Los Angeles (LA), and Hawaii (HI) 

clusters, respectively. The point at which the SAFOD borehole passes through the central deforming zone (CDZ) 

is shown by the asterisk. [C] Cross-sectional view of these earthquakes looking to the northwest, parallel to the 

San Andreas Fault, including the trajectory of the SAFOD borehole and the principal faults associated with the 

damage zone shown in Figs. 3 and 4. Note that the HI events occur about 100 m below the fault intersection 

at 3192 m (measured depth), indicating that the HI microearthquakes occur on the southwest deforming zone 

(SDZ).  The SF and LA sequences occur on the northwest bounding fault (NBF), as discussed in the text.

D

C

B

A

SF

LA

1

1

1

2

2

2

3

3

3

4

1985

1990

1995

2000

2005

2010

SF

Year

Magnitude

1985

1990

1995

2000

2005

2010

LA

Year

Magnitude

1985

1990

1995

2000

2005

2010

HI

Year

Magnitude

0

0

0

1

1

1

2

2

2

3

3

3

M 6

   400

   300

   200

   100

     0

−100m

−2900

−2800

−2700

−2600

−2500

−2400

Fault Plane View

Depth (m)

SF

LA

SF

LA

HI

Distance from Fault Intersection (N 45 W)

*

CDZ

Intersection

Point

1100

1200

1300

1400

1500

1600m

−2900

−2800

−2700

−2600

−2500

−2400

Distance from SAFOD Wellhead (N 45 E)

Depth (m)

SF

LA

HI

Fault Normal View

3192

SDZ

3302

CDZ

3413

SAF

OD

Obse

rvato

ry

SAFOD Main 

Borehole

Damage 

Zone

NBF

---