This article is reproduced with the permission of New Scientist for exclusive use by Nova users.

Balancing boulders with a shaky past
25 April 2007
NewScientist.com news service
Julian Smith

Enlarge

James Brune and his friends looked at the 2-metre boulder balanced on a rocky ledge south of Lake Tahoe, Nevada, and wondered if it would roll. After a lot of effort, it did. The rock finally came to rest on the highway in the valley below, miraculously missing any cars. Brune's father, a miner, was called in to blast it.

Years later, what had been just a teenage prank sowed the seed of an idea in Brune, now a geologist at the University of Nevada, Reno. If California and Nevada are the most earthquake-prone states in the US, how come the precariously balanced rocks that litter the deserts of both states are still on their perches? If only you could measure the force it would take to tip a particular boulder over, he reasoned, and estimate how long it has been standing, then surely you could infer that no quake above a certain magnitude has hit that area for at least that long. Perhaps, Brune thought, these balanced stones are more than mere geological oddities: they might represent a vast untapped data set going back tens of thousands of years - one that could help predict future quakes.

Shaky predictions

For an endeavour with such potential to save lives, earthquake prediction has so far involved a frustrating amount of guesswork. Before the theory of plate tectonics became widely accepted in the late 1960s and made plate boundaries the focus of earthquake research, prediction was based on phenomena such as shifts in the Earth's magnetic field, changing water levels in wells and strange animal behaviour.

In the late 1960s, C. Allin Cornell of Stanford University in California, proposed using data from past earthquakes to estimate the likelihood of future ones at a given point. This approach, called probabilistic seismic-hazard analysis, takes into account factors such as the distance to nearby active faults, their history of activity and the rigidity of the ground at the point in question.

Probabilistic analysis has since been incorporated into the national seismic hazard maps that are produced every three years by the US Geological Survey. These charts colour-code regions according to their chance of experiencing a sizeable quake within 50 years. The most earthquake-prone regions in the country - Puget Sound in Washington state, the California coast and southern Alaska - are coloured bright red. In these areas the authorities enforce strict building codes.

Despite these attempts to quantify risk, earthquake prediction is still far from an exact science. Mathematical models are inevitably simplifications of the real thing, and accurate quake measurements only go back a few decades, a blink of the eye in geological terms. To some extent, geologists are as much in the dark as ever, and faults are still being discovered even in well-studied areas of California.

Brune's aim was to extend the seismic record of the region back tens of thousands of years to help predict what might happen in the future. He's not the first to use toppled objects to study earthquakes. In 1862 the Irish physicist and seismologist Robert Mallet published a treatise on the 1857 earthquake in Naples in southern Italy, using overturned blocks as indicators of the direction and intensity of motion. Subsequent studies tended to focus on fallen man-made objects, such as trains and statues, until Brune and colleagues began investigating precariously balanced rocks, or PBRs, in the 1990s.

Together with Mathew Purvance and Rasool Anooshehpoor, also at the University of Nevada, Brune has mapped 1000 PBRs in southern California and Nevada. A PBR is defined as a rock at least three times as high as it is wide, and is typically a boulder ranging in size from dishwasher to school bus, resting on a stone pedestal. They form where water seeps into cracks in fractured bedrock and erodes away the surrounding stone, occasionally producing precariously balanced remnants.

The next step was to calculate how long each boulder had been in its present position. The researchers examined layers of rock varnish, a thin layer of clay, minerals and organic matter that slowly accretes on desert stones. In this particular area rock varnish is estimated to accumulate at 2.3 to 9.9 micrometres every 10,000 years.

The team also analysed levels of certain isotopes, including chlorine-36, aluminium-26 and beryllium-10, which are created by the impact of cosmic rays, high-energy particles that constantly rain down on the Earth. By measuring the levels of these isotopes, researchers can calculate how long the rocks have been exposed to cosmic rays. The findings suggest that each PBR has stood in its present position for at least 10,500 years, with some having balanced for over 30,000 years.

Dating the stones was only part of the story. To discover the maximum possible shaking that the stones could have withstood in that time, the team had to nudge a few PBRs - not tipping them all the way this time, just far enough to find out how much force would be necessary to topple each boulder. "We try very hard to do non-destructive evaluations," says Purvance. Although this wouldn't place an exact limit on the magnitude of past earthquakes in the region, it would give them an idea of how much the ground could have shaken in the life of the PBRs, and how much it is likely to shake in future.

It has been a steep learning curve, says Purvance. The deserts are remote, and most of the boulders are too far from roads to use tow trucks or cranes to tip the rocks. Instead, the team had to rig up aircraft-grade aluminium cables or steel chains attached to soil anchors designed for circus tents, which are able to supports several tonnes in tension.

Each rock called for a slightly different approach. In some cases the team had to use winches or pistons to pull or push against nearby anchor rocks, coupled with instruments to measure the forces. Once a rock reached its tipping point, they carefully returned it to its original position. Sometimes they tipped the same rock in several directions.

It was hair-raising work, says Purvance. "We learned the hard way that if you have any elasticity in the system" - like the nylon straps they tried early on - "they can spring back. It can be kind of scary," he says.

Heading to the lab instead offered a more controlled testing environment, and a chance to topple scaled-down versions of the blocks by simulating earthquakes of varying magnitude. The team placed wooden blocks and chunks of granite on a large stainless-steel shake table that can rattle back and forth on a bed of compressed air. Although the blocks were scaled-down models, some were still so heavy the team had to use a crane to set them into place.

They found that things can get exciting even indoors, especially when the table was programmed to mimic the shaking of a large quake like the magnitude-7.6 monster that hit Chi-Chi, Taiwan, in September 1999. "It's amazing to stand there and watch how the ground would have shaken," says Purvance. "It tosses those objects like they're nothing."

The key to whether or not something topples, the team discovered, is the ratio of peak ground velocity - the fastest the ground moves during the quake - to peak ground acceleration, the maximum rate of change of its velocity. When this ratio is low, as in a short-duration, high-frequency tremor, objects will sway or vibrate but won't always fall over. When the ratio is high, as during a longer-duration, lower-frequency quake, blocks tumble more often.

"If you rattle the table with a high frequency, you can get something to start to rock," says Purvance, "but unless you really push it, with a high velocity and low frequency, it might just shake and never fall."

When they plotted their data onto existing hazard maps, the results confirmed what Brune had suspected from the start: the standing PBRs suggested that historic shaking near faults had been less intense than was previous thought. While the current hazard maps for mid-intensity, more probable earthquakes are consistent with the PBR findings, "the [future] hazard for less frequent, low-probability events may have been overestimated," Brune says. According to the lastest ground-motion models, many standing PBRs should have been knocked over within the past 10,000 years.

Ground shift

The group is now developing a revised model of seismic hazard that is more consistent with the standing rocks, but their work is unlikely to change the US Geological Survey's hazard maps any time soon. Nor do they wish it to - it's a sticky issue involving public safety, says Purvance, not to mention the billions of dollars of construction influenced by the hazard maps every year. What if the USGS lowered the structural requirements for places like hospitals and nuclear power plants and then a huge quake hit? "Unless you have very compelling, bombproof evidence, you don't want to lower the hazard from both a regulatory and a philosophical point of view," he says.

The response to the team's work has been mixed. "It seems like a clever way of getting at least a rough estimate" of earthquake hazard, says Robert Twiss, professor of geology at the University of California, Davis. "It all depends on evaluating how stable the base is these rocks are sitting on."

Another area of uncertainty is the true age of the rocks, says geophysicist Dan O'Connell of the US Bureau of Reclamation, as well as how much they have changed in shape over time. PBR data "might provide some valuable information in certain situations," he says, but "rocks evolve with time - they're only precarious for some fraction of their life cycle. You need to be able to quantify that." At the moment the work seems "a little more evangelical than scientific", O'Connell adds. "Brune thinks it's so intuitively obvious that this has to be the answer."

Brune and his colleagues acknowledge that the motion of boulders during quakes is much more complex in reality than in their mathematical and lab models. A real quake may shake stones and structures in three dimensions, not just one like on the table, and stones can bounce, slip, teeter on various balancing points and lose energy through deformation, all of which muddies the waters.

Despite the project's limitations, says Brune, "One of the nice thing about precarious rocks is that they go directly to what is important to designing buildings and reducing hazard: how much the ground will shake." Collecting and interpreting more PBR data will be important for creating more accurate models and predictions, he adds.

The PBR work "puts the onus on the USGS to critically inspect the way they have been calculating seismic hazard," says Purvance. "We're hoping that the USGS thinks this is important enough that they invest in going out into the field and assessing precarious rocks for their seismic hazard maps."

As it happens, the agency is just starting to finalise its 2012 map. And with the San Francisco Bay area considered overdue for the "Big One", California needs all the help it can get.

Julian Smith is a travel and science writer based in Portland, Oregon

From issue 2601 of New Scientist magazine, 25 April 2007, page 42-45

For the latest from New Scientiist visit www.newscientist.com