- Sep 14, 2004
- 2,677
- 481
- 98
The San Andreas Fault runs about one-half mile from my house in Santa Barbara, CA.
Journey to the Center of a Quake
http://www.newscientist.com/channel/earth/mg18524851.300
FOR my rendezvous, I drive hours south of San Francisco, then turn down a private road where radio reception dwindles and the only traffic signs warn drivers to keep an eye out for cattle. Pretty much all I find when I arrive is a mud-splattered hole in the ground, but to hear Bill Ellsworth describe it, this spot is a field of dreams. "It's like we've been gazing at this game from a distance for decades," he says. "Now someone has given us front row seats. Our noses will be pushed up against the glass, at the heart of the action."
The reason that Ellsworth can see so much in this unremarkable chunk of real estate is because he is a senior researcher at the US Geological Survey in Menlo Park, California. For him the action lies 3 kilometres underground, on a football field-sized patch of the San Andreas fault, where two pieces of the Earth's crusty skin have been scraping past each other for aeons.
Ellsworth, his USGS colleague Stephen Hickman and Mark Zoback of Stanford University in California have the site earmarked for what could be the ultimate extreme sport: they will use an oil drilling rig to burrow to within metres of where earthquakes are born. Then they plan to set up an electronic network and watch the tremors go off again and again.
The project, known as the San Andreas Fault Observatory at Depth (SAFOD), is part of the US National Science Foundation's $200 million EarthScope survey of North America. EarthScope researchers will install seismometers, strain meters and GPS positioning devices at hundreds of sites across the continent. But SAFOD will take that research to a whole new level.
By drilling directly into the fault, the team will be able to observe the chemistry and physics of what happens before, during and after quakes as never before. SAFOD should reveal what drives the engine of earthquakes, what keeps them going and why they stop.
Many of the questions about the San Andreas fault apply to fault lines elsewhere, including the one that wreaked havoc around the Indian Ocean at the end of last year. "We are asking the same questions about the materials, fluid pressure, how earthquakes start and what stops them," Zoback says. The results from SAFOD will help to refine the way scientists model earthquakes, and could even help determine how precisely they will be able to predict a quake's location, timing and size. (I'll believe that when I see the evidence.)
The seeds of SAFOD were sown by earthquake experts in the 1980s, when a report by the US National Resource Council argued that researchers would need to dig to the very heart of earthquake zones if they were to unravel the big questions. Yet despite this huge potential pay-off, funding agencies refused to open their pockets until a few years ago.
The three geophysicists and their colleagues were not idle while they waited for the cash. In the 1990s, they pinpointed an ideal site, an interesting section of the San Andreas near the small town of Parkfield, California, which is hit every few years by small quakes. Using data from seismometers scattered across the surrounding landscape, they traced nearly all of those quakes back to a single origin: our 100-metre-square patch of fault some 3 kilometres below the surface.
Once they had established where the earthquakes came from, they decided that the best line of attack would be to drill into one side of the fault - straight down at first, and then at an angle to puncture the fault zone close to where the earthquakes are triggered. For this purpose too, the site seemed ideal. Earthquakes can originate 10 kilometres or more below the surface, but here the rupture zone is relatively shallow. Also, the two sides of the fault, which lie in the Pacific and North American tectonic plates, have strikingly different geology at this point so it will be simple to identify when the drill pierces the divide (see Diagram). As a final bonus, a region of the fault immediately to the north-west of this zone never generates earthquakes. Instead it undergoes a continuous creep, so will provide a useful control and could help to explain why some parts of the fault generate quakes while others do not. "SAFOD is really two projects in one," Hickman says. "That's about as perfect a natural laboratory as you are going to find."
The long wait for funding meant that the project could take advantage of technological advances made in nearby Silicon Valley, which among other things saw the development of smaller and better sensors during the boom of the 1990s. At the same time there was also a revolution in drilling technology, including the development of computer-aided, gyroscopically-guided drill bits, and new imaging techniques. "We are benefiting from all the tricks the oil industry has learned in order to squeeze every drop of oil they can from the ground," Ellsworth says.
Even so, precision can be hard to achieve when it comes to drilling a hole deep in the ground - as the SAFOD team was soon to discover. A pilot project in 2002 involved drilling a 2.2-kilometre hole straight down at the site. This passed without a hitch, and the team implanted an array of 32 monitoring stations, each with a set of three sensors, along a 1-kilometre section of the hole. The array was to serve as another control for the project, taking measurements near - but not in - an active fault. But plans to take a core sample from the bottom of the hole using a razor-edged pipe had to be abandoned when the drilling tool became irretrievably lodged near the bottom.
Another surprise was awaiting the researchers when drilling of the main SAFOD hole began during last year's drilling season. This hole will eventually be 3 kilometres deep and travel 2 kilometres horizontally. But after just 1 kilometre, the drill bit cut through the pilot hole despite the team's efforts to deflect it, and severed communication with all but seven of the monitoring stations. Three days later the team was still struggling to direct the new hole away from the pilot hole using a cement plug. As the project's web journal notes: "Convincing the bit to drill though granite instead of cement is not as easy as it sounds."
Setting off a Quake
After that the digging went more smoothly. By the end of the drilling season in October 2004, the hole was 3 kilometres deep. And this time, the sharp cutting pipes were able to sample more than 10 metres of rock from the tunnel's bottom. The hole was rigged with sensors and capped for the season.
Even then, the work did not stop. In November, researchers carried out a "virtual quake" experiment to get information on rock structures deep underground. During a natural quake, seismic waves created by movement at a fault are detected at the surface and analysed to estimate their point and time of origin. In their virtual quake the researchers turned this on its head by firing off seven explosions at the surface. They then used the newly installed sensors to detect these seismic waves at depth. Since the timing and location of the explosions were known, the team was able to map the structure of rock even more precisely and refine SAFOD's computer models. This data will help to guide next season's drilling, when the hole will pierce the fault zone. Finally, side holes will be drilled so that different regions in the creeping and quake zones can be examined.
At every step, scientists at the surface follow the action, eager to learn everything they can about the fault zone. They examine the fragments of rock belched up by the pressurised mud that is used to power the drill bit, looking for traces of microbes in rock samples. And they sniff the gases that waft from the bore for clues to the region's geology and biology. One project is even monitoring the vibrations generated by the drill bit itself to build up images of nearby rock.
Ultimately, all this activity is aimed at helping geophysicists understand the setting in which earthquakes develop and the factors that control them. Earthquakes occur because the tectonic plates that make up the Earth's surface are in constant motion. The plates that meet along the San Andreas fault, for example, should glide a few centimetres relative to each other each year. But at most places along the fault the two plates are held in place by friction for decades or even centuries, building up huge stress in the surrounding crust.
Earthquakes are the result of the sudden release of that stress. Each quake begins at a small so-called nucleation patch, the place where the fault first begins to slide. If conditions are right, that displacement will grow suddenly and travel along the fault at supersonic speeds like a crack running through glass, creating seismic waves as it goes. These are felt at the surface as an earthquake.
The longer the rupture the bigger the quake. The rupture in last December's massive quake in the Indian Ocean, for example, travelled over 1000 kilometres (New Scientist, 15 January, p 17). But why does one earthquake grow while another peters out? Geophysicists have learned a lot by analysing the shaking of earthquakes, the stress in fault zones and the material involved in sections of faults that have been exposed at the surface by uplift or natural erosion. This has allowed them to develop many models of the earthquake cycle.
In fact, they have too many models. Lab experiments and computer analysis have spawned a plethora of plausible theories of how earthquakes might nucleate, spread and halt. The most popular of these involve three factors: frictional properties of the rocks in the fault zone, fluids that lubricate the fault, and feedback between vibrations created by a quake and the fault itself.
What, for instance, decides where a quake nucleates? One idea is that a fluid "soda pop" of water and carbon dioxide - with a dash of helium, radon and methane gas - percolates out of the rocks and into the fault zone, where it pressurises pores and reduces friction by lubricating the spaces between rocks. And if the fluid's pressure exceeds a critical threshold, one side of the fault can start to slide across the other. Alternatively nucleation might occur at points where the junction between two plates contains weak material such as slippery clay.
Data from SAFOD could help eliminate some of these models, validate others or even produce new ones. Perhaps sensors will show that fluid pressure in a small section of the fault suddenly changes before an earthquake, which would make fluid the prime suspect for nucleation. Similarly, readings taken in the regions where earthquakes start, travel and stop could reveal the role different minerals play in each step of the process.
Part of the advantage of SAFOD's subterranean lair is that it will allow project scientists to test these hypotheses on many more earthquakes than has been possible so far, and examine them at a much higher degree of sensitivity. While earthquakes at the site must reach a magnitude of 0.5 to be detectable at the surface, the SAFOD team will be able to measure truly tiny quakes that are 10,000 times less energetic - with a magnitude of -2 or less on the logarithmic Richter scale - which they suspect occur several times a day.
One question the researchers are eager to tackle is whether they can predict exactly where and when the next quake will strike the SAFOD region and, even more importantly, how big it will be. "Size is really the goal of early warning earthquake systems," Hickman says. "If we can't say how large the next one will be, that isn't going to be very useful." He argues that if SAFOD proves size prediction is too complex to be practical, it would be valuable to know that as soon as possible. Efforts could then focus on making long-term forecasts of quake hazards and better estimates of shaking to guide building regulations near fault lines.
The mixed success of earthquake prediction is already evident in a section of the San Andreas immediately south-east of the drill site. This area is home of the Parkfield experiment where, on average, one magnitude 6 earthquake occurs every 22 years. In September 2004, a quake of this size rocked the region (I felt it in Santa Barbara), just as researchers had predicted - only the quake arrived 16 years later than expected. The reason for the quake's tardiness is the subject of intense study. The frequency of smaller quakes at the SAFOD site increased immediately after the big quake and is now beginning to tail off.
This pattern has been seen after previous earthquakes but no one can explain it. The SAFOD team hope their work might determine the underlying physical laws that govern this phenomenon. But it will probably take years of study, which is one reason why they plan to keep SAFOD online for at least 15 years. "You can try to prepare quickly," Ellsworth says, "but ultimately you need to stop, be patient and let the Earth do its work."
To the Seabed and Beyond
If drilling into a fault on dry land is a challenge, how about doing the same thing in the deep ocean? That is the objective of the Integrated Ocean Drilling Program (IODP), which will use gear mounted on a massive ship specially equipped to drill into the deep ocean floor to unprecedented depths.
The 54,000-tonne, 210-metre Japanese vessel Chikyu (meaning "Earth") now in testing will ultimately be able to drill in 4 kilometres of water to a depth of 8 km below the ocean floor. Its equipment will include an outer pipe, called a riser, to contain the drill pipe. This will allow the drilling mud to be pumped down to power the drill bit and will stabilise any soft sediments it drills through. The ship's drilling tools also include a monitor to check the release of potentially explosive gases and avert a disastrous blow-out should the drill hit pressurised pockets of oil or gas.
These capabilities will allow Chikyu to go where no geological expedition has gone before: the Nankai trough, a seismically active subduction zone off Japan's main island, where two tectonic plates ride over each other. The Nankai trough has a history of generating destructive quakes dating back at least to the 15th century, when a quake generated by it killed tens of thousands of people.
Unlike the SAFOD team, Chikyu's crew does not expect to pinpoint the nucleation region of earthquakes or observe the action close-up. But the team will learn a great deal about the deep geology of the area that spawns these massive quakes. "Scientists have been studying this region and wondering what lies far beneath the ocean floor for a long time," says Manik Talwani, president of the IODP's management team. "But in the end you don't know what's there unless you get a piece of it."
