Imagine a brick resting on a board.
If you lift one end of the board, you put stress on the brick to move.
Lift it high enough and the brick will begin sliding.
The brick's passing from a state of rest to a state of motion is called the nucleation phase of the movement.
Understand the nucleation that occurs before a major earthquake, and you could help save hundreds of thousands of lives.
Bill Ellsworth, chief scientist for the U.S. Geological Survey's Earthquake Hazards Team in Menlo Park, Calif., used the brick-on-board example to explain a key puzzle piece in understanding earthquakes, a mystery that can't be solved by measurements taken at the surface.
"What we don't know is how that nucleation process takes place within the Earth," Ellsworth said.
So the most important drilling project in the United States isn't targeting oil or gas production.
It's sending a hole into the San Andreas Fault in California, in hopes of capturing vital information about the origins and processes of earthquakes.
Ellsworth is co-principal investigator for the San Andreas Fault Observatory at Depth (SAFOD) project, with Stephen Hickman, a USGS geophysicist, and AAPG member Mark Zoback, a professor of geophysics at Stanford University.
"We've collected a treasure trove of information about how large earthquakes take place and what they do," Ellsworth said. "SAFOD is going to let us study how the nucleation takes place."
And that's just one of the goals of SAFOD, a relatively small but intensely ambitious component of the National Science Foundation's $200 million EarthScope program.
By the end of 2007, SAFOD will put an array of instruments next to the San Andreas Fault two miles below the surface, to monitor earthquake origins for 20 years.
Scientists on the project talk about the wealth of deep-Earth data promised by SAFOD, but the big payoff would come from any step toward predicting earthquake occurrence.
As the Asian tsunami of December 2004 showed, advance warning of a natural disaster could save a very large number of lives.
"Are we trying to predict earthquakes in SAFOD? No," Zoback said. "Are we testing the predictability of earthquakes? Definitely."
The SAFOD drill site in central California, north of Paso Robles, sees a steadily repeating pattern of magnitude 2 earthquakes about every two years.
These repeaters were first discovered 15 years ago, according to Zoback, originating from identifiable patches along the San Andreas.
"If you were to look at the seismograms recorded at the surface coming from these patches and you laid down the seismograms on top of each other, they're indistinguishable, wiggle for wiggle," he said.
When Zoback says "patch," he means a football field-size area along the fault.
"It slips about a centimeter or a half-inch in each of these magnitude 2 earthquakes," he said. "That's the source, the seismogenic patch."
Phase 1 of SAFOD drilled just over 10,000 feet down in 2004. Now in Phase 2 drilling, which began in early June, the hole will be directionally drilled at a 55 degree angle, aiming toward the fault.
"One challenge we have is that there appears to be two active strands of the fault, separated by about 280 meters," Zoback said.
"We can miss the patch we're drilling toward slightly and still be able to core into it. We're going to try to just graze it. That makes the later multilaterals easier," he added.
The Phase 1 hole section was cased with 9-5/8 inch casing and cemented, then cleaned of drilling fluid.
"Then we pulled a wet string, which depressed the water table down to about 3,500 feet," Zoback said. "Since that time, roughly since October 1, that pore fluid had been coming up."
Phase 2 began with fluid tests, part of a long-term fluid study that's another essential part of the project.
"There are a lot of hypotheses about the role of fluids in the fault zone, and how they affect the earthquake process," Zoback noted.
"Some of the hypotheses argue that these fluids in the fault zone migrate from very great depth," he said, "or possibly even the Earth's upper mantle."
Next came a mini-frac to measure least principal stress, "what in the oil industry would be called an extended leak-off test," Zoback said.
"After that we intend to perforate casing in 10 different places," he added. "We'll sample the fluids, if possible, but we'll definitely do mini-fracs at each of those 10 points."
Drilling the 14,000-foot (measured length) hole to a total depth of two miles will take another 36 days.
The project team will core four areas and run a number of tests as drilling progresses, including a detailed study of gas from the hole.
"In addition to a standard type of gas chromatograph, we're also putting it through a mass spectrometer and radon detector and so on, and even sampling for more sophisticated isotopic analysis," Zoback said.
"All of this, Phase 2, will end with a seismometer and tilt meter operating at the bottom of the hole in early September," he added.
SAFOD's drilling will penetrate and pass through the San Andreas. That part of the hole will eventually be lost to movement along the fault.
"One of the questions we have is, 'What is the width of the deforming zone at depth?' It's part of the experiment plan to watch the hole deform over time," Zoback said.
Phase 3 of the SAFOD project, the final phase, will acquire cores and emplace instruments up to the fault zone.
Zoback said SAFOD scientists originally planned continuous coring for the hole, but that idea wasn't workable. A petroleum industry specialist suggested drilling the hole and then using multilateral drilling for coring.
"This was 10 years ago, and I had not heard of multilaterals at the time. Frankly, I thought it sounded like a pretty dumb idea to drill our well twice," Zoback said.
He quickly decided it was a pretty great idea, however, given the new technologies available.
"You use a hollow-stem top drive and a higher rotation speed so you can do continuous wireline coring and actually retrieve core barrels through the top drive, much as is done in the mining industry," he said.
Without advances in drilling and instrumentation from oil and gas exploration, SAFOD could never have happened, Zoback acknowledged.
"What's making this experiment possible is our being able to modify and extend technologies that are routinely used in the petroleum industry," he said. "The monitoring instrumentation that's going into the observatory are extensions of what's being done in oil and gas."
Ellsworth said project plans call for a series of renewed and improved instrumentation for the downhole observatory as time goes by.
"We're deep. We're hot. So advances in sensor technology will be essential to the project," he explained.
SAFOD's drilling location ensures a look at repeated earthquakes at a feasible depth, according to Ellsworth.
Hunting bigger earthquakes would require much deeper drilling. Magnitude 6 earthquakes originate about six miles below the surface, he noted.
But rapid advances in deep-hole exploration could put that depth within reach of ongoing scientific study in the near future.
The EarthScope program also includes huge networks of seismic stations, GPS receivers, strainmeters and surface monitors. Zoback called it "the biggest thing that's ever happened in solid Earth science.
"For people in the petroleum industry, they have to realize that there are three components of EarthScope, and SAFOD is the smallest," he said. "We're only about 10 percent of the budget."
SAFOD will make its collected data available to the industry, primarily through the Web, and all cores will be kept at Texas A&M University's Ocean Drilling Core Repository.
Another payoff for the industry will come from EarthScope's ability to attract and train young scientists, Zoback observed.
"Some of those are Earth scientists who will find their way into the oil industry," he said. "It's been a real stimulus for bright young people to come back into Earth science."
Will the program be a major step toward predicting earthquake occurrence?
The truth is, no one can be sure, because so little is known about the origins of earthquakes.
"One of the objectives of SAFOD is to determine whether or not earthquakes are predictable, no doubt about it," Zoback said.
"We're going to put our instruments within tens or certainly hundreds of meters of the earthquake process, so we will see what the fault does leading up to an earthquake," he added.
Ellsworth, who's studied earthquake nucleation extensively, won't make any predictions, either.
If nucleation begins in an area the size of a coin before spreading violently, scientists have little hope of finding and identifying nascent earthquakes, he said.
But if nucleation takes place over a larger area, slowly building and giving off definitive signals, earthquake prediction may become reality.
"That slow process does not emit seismic waves -- it's what we'd call creep," Ellsworth said. "We'd like to understand how that occurs in the Earth.
"We can learn a lot about earthquakes at the surface," he added. "What we don't see are the nonlinear parts of the process. What we don't see is the breaking."
For the first time, SAFOD will give scientists a window to earthquakes at birth.