Drilling San Andreas

Despite decades of study, why earthquakes happen and when they do remains one of the great mysteries of science. A two-mile-deep tunnel boring into America's most infamous fault may soon change that

By Brad Lemley|Thursday, March 31, 2005

A team of roughnecks pull giant hydraulic wrenches from a freshly joined section of pipe. “These guys are amazing,” says geophysicist Naomi Boness. “They can tell you what kind of rock they are going through just by the way the drill behaves.”

Photograph by Vicky Samburnaris

The odds against a major earthquake happening on the day I arrive in central California to write about earthquakes are approximately 8,030 to 1.

If only I were a betting man. It hits at 10:15 a.m. on September 28, 2004, as I am biting into a breakfast bar, waiting in a trailer at the San Andreas Fault Observatory at Depth. The project is using an oil-drilling rig to bore deep into the infamous fault, allowing geophysicists to plant instruments at the precise spot where earthquakes begin. At first I think the rig, roughly 100 yards from the trailer, is making the ground vibrate, but soon the trailer’s floor is juking and jiving beyond all reason.

“Oh my God, it’s an earthquake!” shouts geophysicist Naomi Boness, clutching her desk. I appreciate the professional confirmation.

The quake lasts 10 seconds—10 really, really long seconds. It is shaky, rumbly, scary, and bad enough to have flattened bad housing and killed some people if it had struck a city. Here in rural, earthquake-ready central California, it just knocks out power for a few hours, topples a couple of chimneys, and breaks some bottles at a local winery. Nobody is hurt, but the quake is a once-in-a-lifetime thrill ride for roughneck Tim Camargo, who is working atop the drilling rig when it hits. “What did I feel? I felt a whole lotta shakin,’ ” he says after quickly climbing down. “It was magnified about 10 times up there.”

Everything becomes a bit nutty after that. The quake is felt over a 500-mile-long area, so a whole slew of local news media rapidly converge on the tiny town of Parkfield, population 18, which is about eight miles from both the earthquake’s epicenter and the drill site. I drive there and count 13 live-TV trucks on the dusty street. Television reporters try to put tough, in-your-face questions to Stephen Hickman, the U.S. Geological Survey geophysicist who is one of the observatory’s chief investigators.

TV REPORTER: So, did you guys cause this with your drill?


Meanwhile, I am stunned and weirdly jumpy all day. With the trailer rolling and yawing dramatically even in this modest temblor, the fact that earthquakes have killed more than 1.8 million people since 1900 is suddenly very easy to believe. The indirect effects of a temblor can be equally deadly, as demonstrated by the devastating Indian Ocean tsunami that struck last December. One has to wonder how many of those people would have lived if they had known in advance that an earthquake was coming.

“Earthquake prediction is the bugaboo of geophysics. People don’t even like to talk about it,” says Mark Zoback, a professor of geophysics at Stanford University and another principal investigator at the drill site. Right now scientists understand far too little about Earth’s inner motions to make reliable earthquake forecasts, and the cost of an unreliable forecast could be huge: Millions of dollars of business would be lost by shutting down San Francisco for a week awaiting a temblor that might never arrive. On the other hand, even a few minutes of advance warning would allow enough time to evacuate people, shut down gas and water lines, and save a city from ruin.

Zoback picks his words carefully as he explains why he is drilling into the San Andreas Fault. “With this project, we are not trying to learn how to predict earthquakes, but we are trying to see if they are predictable. Whatever we find will be interesting, because it’s never been done before.” Hickman echoes the sentiment: “What happened today is a reminder of what this is all about.”

The San Andreas Fault Observatory at Depth is located 25 miles northwest of the spot, near the intersection of Highways 41 and 46, where James Dean wrecked his Porsche Spyder and died on September 30, 1955, an event commemorated by an unattractive, stainless-steel roadside shrine erected by a Japanese businessman. All around is remote, desolate, sun-blasted cattle country dotted with live oaks, brown grass, and weather-beaten barns. The nearest gas station is 20 miles away, cell phone service is virtually nil, and Alice, the desk clerk at the four-room Parkfield Inn, asks me to close the gate behind me to keep the wild pigs off the porch.

The fault behaves oddly here. Formed by the Pacific and North American tectonic plates grinding against each other, the San Andreas Fault curves up 800 miles from the Gulf of Mexico, slicing a shallow arc across western California. The north and south sections are locked together, slipping rarely but violently, leading to quakes like the 1906 monster that killed 3,000 people in San Francisco. The section just north of Parkfield is different. It moves all the time at about the speed human fingernails grow, roughly 1.3 inches a year. Build a house there and it would be ripped in half in slow motion.

The five-acre drill site, a 12-minute drive from Parkfield, is on the transition between the southern, locked section and the creeping section. Zoback, along with Hickman and William Ellsworth of the U.S. Geological Survey Earthquake Hazards Team, decided to drill here because this transition zone cranks out little magnitude 2 quakes, too small to feel on the surface, with metronomic regularity every couple of years. These are repeating earthquakes, meaning they produce identical seismograph patterns, “wiggle for wiggle,” says Zoback. Monitoring how the ground moves repeatedly, at many different depths, will allow the team to piece together the world’s first three-dimensional map of a fault in action, like peering straight into the rock. The “observatory” part of the project’s name is more than just metaphor.

Putting Earth Under the Microscope
The San Andreas Fault Observatory at Depth is one of several elements of EarthScope, an ambitious $200 million initiative by the National Science Foundation that is investigating the geophysical forces that shape the North American continent. The other projects within EarthScope include:

The Plate Boundary Observatory—Global Positioning System receivers at about 1,000 sites, combined with strainmeters at about 200 locations, are being installed along the western edge of North America from Alaska to Mexico. The aim is to monitor the continent’s ongoing shrinking, bulging, and general deformation, watching changes over timescales ranging from days to decades.

United States Seismic Array—A network of seismic stations, including portable ones that will be moved across the entire United States over a 10-year period, will map subtle differences in the seismic energy traveling through our planet, yielding an improved understanding of deep-earth structures.

Interferometric Synthetic Aperture Radar—A proposed satellite-based monitoring system that, if funded, will track the movement and deformation of the North American and Pacific plates by bouncing radar waves off the surfaces. This technique could identify ground motion as small as one millimeter.


The earthquake of September 28, 2004, struck at Parkfield, California, 15 miles from the San Andreas Fault observatory. “I wish the quake had waited until we were done drilling,” laments Mark Zoback.

Courtesy of M.J. Rymer/U.S. Geological Survey

The observatory itself has an unexpected paramilitary vibe. Though it is on a sprawling cattle ranch on a remote, lonely plain, it is nonetheless protected by a barbed-wire-topped fence and a gate guard, a determined woman known as Toad. If you are not on Toad’s checklist, you can’t get in. When I drive back there after the media frenzy in Parkfield, Toad (a.k.a. Torie Osborn-Heilmann) says she’d had a funny feeling an earthquake was coming: “The squirrels have been gone for two days. That should have told me something.”

The 182-foot-tall drilling tower, brilliant white in the blazing morning light, dominates the tract, which is filled with lab trailers, residence trailers, oceans of dried muck pumped out of the drill hole, and piles of gritty rigging equipment. Right after the quake, the drilling crew clambers down from the work platform and sits in the shade while safety inspectors climb the rig to look for cracks. By noon, the roughnecks are back at it. After the quake, the work seems even more important.

The rig, owned and operated by the Nabors Drilling company and previously used to find oil or gas, first bit into the reddish soil here in July 2002, drilling a pilot hole 1.4 miles deep, which was outfitted with a string of seismometers. Using information gleaned from that effort, the observatory team began boring the steel-jacketed main hole in June 2004. It will go down vertically for the first 1.1 miles, then bend eastward toward the fault at a 50 degree angle for another 1.8 miles, using the directional drilling techniques invented to allow more complete extraction of tough-to-reach oil deposits.

By the end of this summer, researchers hope to reach a football-field-size section of fault plates roughly two miles underground, the stress point where the little magnitude 2 quakes begin. About 50 instruments—a mix of seismometers, strainmeters, temperature sensors, and fluid-pressure transducers—will monitor every change in the fault, sending readings to the surface through fiber-optic cable. “We will have instruments within tens of meters of the source, so over the next 20 years we can watch what is essentially the same earthquake over and over,” Zoback says.

His team will search for any kinds of meaningful patterns. “If we have our instruments right in the fault zone, and we literally don’t see anything predictive, that will say something about predictability,” he says. But his hope is that quakes do give off warning signals. If so, underground monitoring stations like this one could someday save millions of lives. Even in the worst case, seismologists will gain tremendous insight into what happens when a fault slips and what kind of damage the resulting motion can cause.

What about the magnitude 6 quake we felt that morning? That came from a fault section a few miles to the south that has churned out a similarly intense tremor once every 22 years, on average, since the mid-19th century. (Twenty-two years equals roughly 8,030 days, which is how I arrived at my very rough odds calculation.)

 “When I felt it, I got out of the shower and ran outside. Didn’t even turn off the water,” Cindy Van Horn, a cook at the Parkfield Café tells me that afternoon. “I take earthquakes seriously. When I was 9 years old during the big one in 1966, the house fell down around me.”

Zoback, meanwhile, is elated. He is an intense, fast-talking researcher who loves his work, and the morning quake only amplifies his zeal. He was off-site at the time, but drove there at breakneck speed when he heard the news. “Some people in earthquake science have stopped asking the big questions. That’s unfortunate. I think this will revitalize the field,” he says. He cannot wait until the drill finally bites into the spot where the two plates scrape past each other. “Penetrating the fault will be the big deal. So far, it’s just foreplay.”

A companion observatory for tsunamis
The enormous tsunami that killed more than a hundred thousand people in Africa and southern Asia last December 26 was spawned by a subduction fault, formed when an oceanic plate dives beneath a continental plate. To study this process, researchers have initiated a drilling project, remarkably similar to the one at the San Andreas Fault Observatory at Depth.

Subduction faults, angled at 10 to 15 degrees from horizontal, are significantly different from vertical faults such as the San Andreas. “With a subduction fault, you get a much larger locked area, so the quakes can be much bigger,” says geophysics professor Mark Zoback of Stanford University. While vertical-fault quakes, such as the one that hit San Francisco in 1906, peak at roughly magnitude 8, “subduction quakes can be 9s. And remember, between 8 and 9 is a factor of 30 in energy.”

As part of the Integrated Ocean Drilling Program, funded primarily by Japan and the United States, seismologists plan to use the drilling ship Chikyu to bore into the Nankai Trough, a subduction zone paralleling the southern Japanese coastline. They will then fill the borehole with seismometers, thermal sensors, and other instruments similar to the array planned for the San Andreas project. Launched in January 2002, the Chikyu is undergoing shakedown cruises, with drilling expected to begin by 2007.

Zoback emphasizes that safety measures need not wait for results from that effort, because tsunamis are much simpler to predict than the earthquakes that spawn them. “Usually, you have hours of warning between the detection of the earthquake and the tsunami striking,” he says. “All you need to do to save lives is have the infrastructure in place.” In the aftermath of December’s devastation, the Indian, Thai, and Indonesian governments say they will move to create such an infrastructure. —B. L.


Excess ground-up rock from the well collects in a trench next to the drilling tower. Analysis of gases trapped in mud pumped up from the borehole has shown that radon, carbon dioxide, and helium are more abundant in fault areas than in solid crust.

Photograph by Vicky Sambunaris

Earthquakes are context-free natural disasters. In hurricane-force winds, nobody wonders, “Gee, is this a strong wind?” During floods, no one ponders the rising waters and asks, “Has the river overflowed?” But earthquakes start without preamble in the invisible depths, and at any one location major ones are rare in relation to the human life span. These two facts decree that everyone caught in a temblor will spend the first 10 seconds wondering what the heck is going on and the next 10 shouting “It’s an earthquake!” to each other. It is over at almost the precise moment that everyone has figured out what it is.

The brevity and inaccessibility of earthquakes also makes them devilish to study and, so far, impossible to predict. Even the fanciest seismic arrays read only an external signal, distorted by the scattering of energy waves, from an event that is fundamentally internal. That’s why researchers have toyed with the idea of drilling into the San Andreas Fault to see what makes it tick at least since the 1970s. “The surface of Earth is a mess,” Zoback says. “All faults are weak at the surface because there is no confining pressure, so we have to make inferences” about each quake’s deep-earth origin. Placing instruments underground, right where the plates start slipping past each other, will provide crucial information about the nucleation zone, the place where an earthquake begins.

Zoback and Hickman worked up a formal proposal for this research in the early 1990s. Their goal was to read all the vital signs of an earthquake—not just fault motions but also the precise pressures, temperatures, strains, and chemical compositions of minerals in the nucleation zone, along with an analysis of rocks and fluids extracted through the drill hole. The expense of such an undertaking kept it from happening until 2002, when the National Science Foundation secured more than $200 million from Congress to fund EarthScope, an elaborate earth-science umbrella project that aims to illuminate the geologic forces that shape North America. Through EarthScope, the San Andreas observatory will receive about $20 million through 2008. “After 12 years of meetings, to see it happening is exciting as hell,” Zoback says, grinning.

In the late afternoon with the sun pounding down, I climb the steel stairs up to the drilling rig’s platform. At the surface, the steel-jacketed well is 36 inches wide. It grows narrower as it gets deeper, because each successive lining must slide down inside the previous one; at the depth of the fault, it will be only 8 1/2 inches wide. Today the well is 10,100 feet deep and 9 5/8 inches wide—more than half finished. The four-man hard-hatted crew works in perfect synchrony, joining 90-foot sections of drilling pipe with massive orange wrenches. The pipe snakes through the installed lining and drills the hole deeper. Then pressurized fluid forces the bored earth to the surface, making space for another batch of ever-narrower steel sleeves. The labor is grueling. “It kind of rips you up,” says roughneck Joe Mason during a break, rolling up a sleeve to show a black pressure bandage. “We’ve all been smashed, cut, and beat up. I tore my biceps last week.”

Zoback and his colleagues will make sure the sacrifice is not in vain. One of the big questions this observatory should help answer is why faults like the San Andreas, where two vast plates of Earth’s crust scrape past each other, behave so strangely.

“If you look at a fault within a tectonic plate, you’ll find that it obeys the laws of the laboratory very well,” Zoback says. If a researcher extracts a one-inch sample of rock from the fault zone located inside a single plate and subjects it to compression and shearing forces in a laboratory until it slips, it will behave in much the same way that rocks in the ground do.

But for some reason, the San Andreas Fault and other interplate faults slip under very low shear stress, much lower than is required to make rock samples slip in the laboratory: The rock here slides when lab tests say it should remain immobile. “It seems crazy to say it—you know, the mighty San Andreas and all that—but it is actually a weak fault. Why?” Zoback asks.

In his view, there could be at least three reasons. One might be some oddity in the composition or structure of the rock at the nucleation site that makes it exceptionally slippery. (No one has extracted material from the actual nucleation zone before; researchers have had to assume that the rock there was similar to rock closer to the surface.) Another might be higher-than-expected pressure from underground water that pushes the plates apart, allowing them to slip easily, just as an engine with high oil pressure works more efficiently than one with low pressure. Finally, there could be some unknown dynamic mechanism that makes the plates skip against each other as they slide, much as a car’s skidding tires make microskips across a road’s surface.

“The fact is, nobody knows,” Zoback confesses. “What we have now is what Steve Hickman calls ‘an expanding plethora of hypotheses.’ ”

The San Andreas observatory should help root out the best explanation. When the fault-weakness issue is settled, many more questions await: What’s the origin of the fluid at that depth—is it rainwater or something else? What controls the origin and propagation of earthquake waves? Above all, what is the structure of the plates that mark the ground-zero point where quakes occur? “At the moment, every theory is as good as another, because we just don’t know,” Zoback says.

By 6:30 p.m., the sun is retreating behind the rapidly cooling brown hills. Zoback and Hickman have, between them, done about 50 interviews that day, and both look beat. I suddenly realize that all I have eaten was that meal-replacement bar.

Driving back to the Parkfield Inn, munching another bar, I think about how quakes have been having their way with innocent people for a long time. The biggest natural disaster in human history took place on January 23, 1556, in Shanxi Province, China, when a quake killed 830,000 people. Buildings have improved since then, but the planet is also a lot more crowded than it was in 1556. There is no reason why an equally deadly quake could  not strike tomorrow morning.

I drive past the inn, get out of my rental car, and spend a moment standing in a dry wash that runs along a creeping section of the San Andreas Fault. Now, in the dark, I can almost feel it move.

Next Page
1 of 3
Comment on this article

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar

Log in to your account

Email address:
Remember me
Forgot your password?
No problem. Click here to have it emailed to you.

Not registered yet?

Register now for FREE. It takes only a few seconds to complete. Register now »