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

20  

Scientific Drilling, No. 11, March 2011

Science Reports

array was also lost as a consequence of the accident; the 

lowermost twenty-five levels were severed during the inter-

section, and the remaining seven levels were decommis- 

sioned in the spring of 2005 when an unsuccessful attempt 

was made to regain access to the Pilot Hole below the inter-

section.

During the nine-month hiatus (September 2004 to June 

2005) between the end of Phase 1 and the beginning of 

Phase 2, a number of seismometers were deployed in the 

SAFOD Main Hole as part of an instrument testing program 

for eventual deployment of the SAFOD observatory. A num-

ber of shots were set off while the seismometers were in the 

borehole to better constrain the velocity model and reduce 

uncertainty in the location of the target earthquakes. In addi-

tion, an eighty-level, 240-component seismic array was made 

available by Paulsson Geophysical Services, Inc. (PGSI) and 

recorded by Geometrics at no cost to the project. This array 

was deployed in the borehole for a period of five weeks in 

order to test its suitability for recording microearthquakes 

and to record additional shots for structural imaging 

(Chavarria and Goerrtz, 2007). In addition to recording 

microearthquakes and shots during this period, a tectonic 

(i.e., non-volcanic) tremor was recorded on this array. The 

tremor occurred in the lower crust directly below the sur-

face trace of the San Andreas Fault for at least 70 km to the 

northwest and 80 km to the southeast of SAFOD (Shelly and 

Hardebeck, 2010). The likely source of the tremor recorded 

by the PGSI array was in the vicinity of the energetic tremor 

source near Cholame (Nadeau and Dolenc, 2005) near the 

base of the crust (~25 km; Shelly and Hardebeck, 2010). 

As shown in Fig. 3, Phase 2 drilling passed from the 

arkosic sandstones and conglomerates into mudstones and  

shales at a depth of 2600 m, and at a position ~500 m 

southwest of the surface trace of the San Andreas Fault. 

Microfossil evidence from core obtained at the bottom of the 

Phase 2 hole indicates that these formations are part of the 

Cretaceous Great Valley sequence, which was deposited on 

the North American plate in a forearc environment at a  

time when subduction was occurring along the western 

margin of California (K. McDugall, pers. comm., 2005).  

In the long-term geologic sense, the contact between the 

Salinian-derived arkosic sandstones and conglomerates and 

the Great Valley formation is the boundary between the 

Pacific and North American plates. As shown by progressive 

deformation of the casing discussed below (Fig. 4), the 

south-westernmost of the active traces of the San Andreas 

Fault Zone at depth is located several tens of meters to the 

northeast of this geologic boundary.

No evidence was found that we had encountered the 

Franciscan Formation in the borehole, even though it is 

exposed at the surface about 600 m east of the San Andreas 

Fault (Fig. 3), and was predicted by several of the geophysi-

cal surveys conducted in advance of drilling. However, there 

is evidence of serpentinite directly within the fault zone asso-

ciated with either the Coast Range ophiolite or Franciscan 

formation. Hence, there is likely serpentinite in contact with 

the San Andreas along strike and/or at greater depth. A rea-

sonable conceptual model is that slivers of Great Valley and 

the Franciscan are intermixed at depth along the fault, just 

as they are found in surface exposures at several locations in 

central California. 

Rotary drilling through the San Andreas Fault during 

Phase 2 was accomplished with no small amount of diffi-

culty—some caused by the fault zone, some caused by unre-

lated operational problems (for example, the top drive, an 

extremely important component of the drill rig, broke and 

was inoperable for two weeks). We also noted a considerable 

degree of time-dependent wellbore failure (Paul and Zoback, 

2008), especially after passing through the active traces of 

the San Andreas Fault Zone. An appreciable amount of time 

was required to clean the hole through wash and ream ope-

rations. In fact, the combined result of time-dependent 

wellbore instabilities and a mistake by the drilling crew 

resulted in the drillstring being stuck in the hole for four 

days at a vertical depth of 2800 m. Despite these problems, 

drilling across the entire fault zone was successfully achie-

ved. Comprehensive cuttings and gases were sampled over 

the entire Phase 2 interval (Table 2), and a number of geo-

physical measurements were made in real-time as drilling 

across the fault zone was underway (Run 4, Table 3). After 

the hole was drilled, a comprehensive suite of geophysical 

logs was obtained, and fifty-two 19-mm-diameter side-wall 

cores were obtained in the open hole (Run 4, Table 3). After 

the hole was cased and cemented, 3.9 meters of core (mud-

stones of the Great Valley formation, mentioned above) were 

obtained from the very bottom of the hole. 

Phase 1 and 2 Real-time Sampling. Drill cuttings and for-

mation gases were collected in real time as drilling was 

taking place. Drill cuttings were collected every 3 m and pre-

served in both washed and unwashed states, and larger vol-

umes of cuttings were collected at less frequent intervals, as 

were samples of the drilling mud. Table 2 summarizes the 

cuttings samples, side-wall cores, and the three cores ob-

tained after casing was cemented into place at various 

depths. Photographs, detailed descriptions, and other infor-

mation about the extensive collection of cuttings are avail-

able online (Table 1). A summary of the lithologies encoun-

tered during Phases 1 and 2 is provided by Solum et al. 

(2006) and Bradbury et al. (2007), principally based on X-ray 

diffraction (XRD) analyses and optical analyses of mineral-

ogy and texture of the cuttings, augmented by the spot and 

sidewall cores. 

The near-continuous collection of cuttings revealed a 

number of lithologic changes along the trajectory of the hole 

that correlated very well with geophysical logs and other 

information. In addition, analysis of these cuttings revealed 

trace amounts of serpentine and a high level of clay minerals 

in the localized intervals that proved to be the active San 

---