Balanced rocks can tell us about earthquake risk

Seismologists study precarious boulders to determine how hard the ground might shake.


A precariously perched granite boulder in the Mojave Desert of Southern California. Balanced rocks in these desert regions are thought to have begun standing between 12,000 and 18,000 years ago after eroding from underlying strata.
Douglas Fox

A slender monument stands in the slanted morning light: a column of granite boulders stacked like toasted marshmallows, throwing a crooked shadow down a hillside of sagebrush and scrubby juniper.

“You don’t find anything like this near the San Andreas (fault),” says James Brune, climbing up to inspect this 10-foot tower in western Nevada. He wears a trim white beard, wool sweater and wide-brimmed hat.

Brune, now 80, might have retired years ago from his post as a seismologist at the University of Nevada in Reno, but his interest in precariously balanced rocks keeps him busy. He is using them to estimate the hidden earthquake risks faced by a growing West. He hopes to learn something about the dreaded “Big One” — the kind of catastrophic shaking that occurs just once in 1,000, or even 10,000, years. These are the rare, deadly events that engineers have to plan for when they build bridges, dams, hospitals and nuclear power plants, and yet, as Brune likes to say, “How do you predict once in 10,000 years when you only have a record of 100 or 150 years?” The information simply doesn’t exist.

Historic records of earthquakes in the West go back only to the 1800s. To track older quakes, geologists trench across known faults in search of places where the silt layers are offset by several feet, marking major movements. Their studies show that Southern California’s southern San Andreas Fault has experienced 10 magnitude-7-plus quakes over the last thousand years, most recently around 1720. But the magnitude of a quake provides only a rough estimate of how much the ground shook, on average, over a very large area. “Shaking from earthquakes is not a uniform pattern, like when you toss a rock into a pond and the ripples go,” says Lisa Grant Ludwig, a seismologist at the University of California in Irvine who spent 20 years studying prehistoric earthquakes on the San Andreas. “There’s a lot of variability,” because shock waves change as they travel through different types of rocks and soils. As a result, scientists have had no reliable way to figure out how hard the ground actually shook in any particular area during an earthquake — until now.

James Brune and his son, Richard, who designs electronic doors for aircraft hangars and fire stations, have spent 25 years mapping precariously balanced rocks in California and Nevada. They have created digital 3-D models of the rocks and calculated the shaking required to topple them. They want to know how many G’s of acceleration a balanced rock can withstand before falling — not how far the ground underneath the rock must move, but rather how violent that motion has to be.

The Brunes have discovered some surprising things in the process. In some places, judging by the rocks still standing, the biggest earthquakes in the last 10,000 years weren’t quite as large as we’d thought. And that, in turn, suggests that, at least in some places, the future Big One may not be as bad as expected. If bridges or dams need less strengthening than previously thought, more resources could be freed up for the hundreds of bridges and other structures in the West that are already at risk, not from earthquakes but from old age and inadequate maintenance. “I think the (balanced) rocks are very important tools,” says Grant Ludwig. Information gained from studying them, she says, can communicate risk to the public in a concrete way that standard “2 percent risk in 50 years” seismic hazard maps don’t.


Brune’s interest in balanced rocks began by chance. In the early 1990s, he was assessing earthquake risk at Yucca Mountain, site of a proposed nuclear waste dump. He noticed volcanic rocks stacked awkwardly atop one another, painted in dark rinds of desert varnish that would have taken thousands of years to form. Using standard methods, engineers had predicted that the nuclear waste dump would experience maximum shaking up to about 0.8 G’s over a period of 10,000 years. But Brune doubted these rocks could survive more than 0.3 or 0.4 G’s — prompting him to suspect that scientists had overestimated the ground acceleration that could happen. His results suggested that the waste dump might not need to be quite so heavily- fortified.

Yucca Mountain was eventually shelved, but the two Brunes began a broader search for balanced rocks, hoping they could improve shaking estimates in other places. Richard Brune outfitted a remote-control airplane with a live-feed video camera — a rudimentary drone that they used to search in remote and rugged terrain. Later, Richard rode in a friend’s Cessna with the door removed, leaning out to shoot photos as the plane flew tight crisscross grids over the Mojave Desert. The father and son pinpointed several thousand promising rocks, which they later hiked to and measured.

When you picture a balanced rock in the desert, you may envision a ruddy sandstone spire. But these balanced sandstone rocks often form and erode away too quickly to be of much use for studying earthquakes over thousands of years. So as the Brunes comb through California’s San Bernardino, Los Angeles and Riverside counties and the western half of Nevada, they seek out balanced granite boulders instead. Geologic history has honed these columns into exquisite seismic record-keepers. Around 100 million years ago in Western North America, plumes of magma cooling miles beneath the surface formed granite monoliths. These buried blocks were alternately stretched and squished by shifting tectonic plates — splintering them with cracks that acidic groundwater widened. A few thousand years ago, when water erosion finally exposed these rocks, they were rounded and fragmented, sitting upon one another but not attached. Wind and water whisked away the last grains of sand from their joints, leaving them balanced in the air — sometimes just a single granite boulder perched precariously on a stone pedestal, sometimes a column of three or four rocks.

The rocks have often fulfilled expectations: The farther away they were from major faults, the more precarious they were — some could be toppled with the nudge of a finger. But there were conspicuous exceptions, and these have led to important discoveries.

The Brunes found surprisingly tippy rocks near the San Jacinto Fault in Southern California. “Current hazard maps say those rocks shouldn’t be there,” says James Brune. The rocks clustered around a so-called “step-over,” where the fault jags four miles west before continuing south. Brune concluded that such step-overs can effectively bracket an earthquake by preventing shock waves that begin in one segment of a fault from propagating strongly into other segments. It’s a discovery he’s proud of, he says, “because the precariously balanced rocks said something that nobody realized.”

Glenn Biasi, a younger seismologist at UNR, is now converting the Brunes’ copious field notes into a database — some 790 rocks, so far. Analysis of that big dataset is revealing some new and unexpected insights into the nature of seismic risk.

 In some cases, patterns of risk are actually turning out to be simpler than previously assumed. For example, even within 10 miles of major faults like the San Andreas, Biasi sometimes sees rocks that are surprisingly precarious, requiring only about 0.35 G’s of ground acceleration to topple — about what you’d feel in a modern sports sedan going 0 to 60 in 8 seconds. Despite having experienced 50 magnitude-7 to -8 quakes in their lifetimes, a couple dozen of these rocks near the San Andreas Fault are still standing. The severity of shaking depends on many complex factors, including how deep in the earth the fault rupture occurred and the type of bedrock underlying the area. But these rocks have revealed something important: Having hard bedrock at ground level dampens the shaking, so that even in really big quakes, it still doesn’t exceed about 0.35 G’s. A magnitude-8 quake may unleash far more energy than a magnitude-7 one — but not because its maximum shaking is any harder. It simply occurs over a broader area. “Nobody’s really seen that before,” says Biasi.

The San Jacinto step-over results have already found their way into California’s 2014 seismic hazard maps, which are used to decide how strongly houses, bridges and other structures should be designed, or when retrofitting is needed. The next hazard maps, due out in several years, will include more of their results.

The balanced rocks are also relevant to broader areas of the West. Geologists now understand that some faults in Nevada and Utah can cause earthquakes as severe as those in California, but because they accumulate stress more slowly, their major quakes often have repeat times of 1,000 years — this is the case for a major fault that passes through Reno. This makes the 150-year historical record even less useful than it is in California, where many faults experience major quakes every 200 years or so. The information that balanced rocks provide in these inland zones could prove even more critical.

Richard Brune prepares a rock for photogrammetry by taping the rock to aid in merging photographs to construct a 3-D digital model. This boulder, east of the San Andreas Fault, may have withstood 50 or more major earthquakes over the past 10,000 years.
Douglas Fox

One hundred and fifty yards from the offices of Biasi and Brune, engineers are studying the nuts-and-bolts implications of their findings. Inside a metal hangar sit three massive “shake tables,” resembling metal dance floors. Controlled by an array of stout hydraulic jacks, these tables can be programmed to reproduce the shake patterns of any recorded earthquake. The Brunes have balanced rocks on the tables and shaken them down, testing their calculations of how precarious they are.

These tables are typically used on real structures — in one case, a concrete bridge loaded with F-250 pickup trucks. Outside the hanger stands a macabre sculpture garden of wreckage from these experiments. Brune walks up to one such pillar. It is bent 90 degrees at the bottom, like a forlorn human figure knocked to his knees in battle. As the pillar bent, the rebar flexed inside it, causing its brittle concrete armor to pop off in chunks. “This,” he says, “is one of the commonest ways that buildings fail” during earthquakes.

Brune’s contribution to these experiments involves simply telling the engineers how hard they need to shake their structures to simulate once-in-1,000-years or once-in-10,000-years ground acceleration. But the most important consequence of his work may lie elsewhere — buried in obscure mathematics.

For decades, seismologists relied on complex statistical methods to estimate the potential ground motion at any given site. They collected the handful of available ground-shake measurements, then extrapolated that sample into a standard random distribution, similar to a bell curve. The average shake events clustered in the tall part of the curve, but the curve also included a thin tail stretching to the right, representing rare, extreme events — events that had never happened, but were predicted to happen based on the standard shape of the statistical curve.  “The width of that” tail, Biasi says, “is the most expensive thing in earthquake engineering.”

Engineers use these shake severity curves to decide when and how to retrofit a bridge, building or dam. This is crucial for safety, but enormously expensive if the severity of future shaking is overestimated and the structures are over-engineered. Retrofitting a bridge to withstand 0.5 G’s rather than 0.3 G’s can sometimes cost as much as building an entirely new structure.-   

Biasi and Brune now believe that these standard statistical methods have caused scientists to overestimate the width of the shaking curves, leading them to overstate the amount of random variation in shaking, and the severity of the rarest and worst events. This finding, says Biasi, could end up being the most significant result of their work. It could alter seismologists’ basic understanding of how all faults function and lead to further major revisions of seismic risk estimates. “We always study faults,” Grant Ludwig says. “But the rocks let us study shaking — which is what actually does the damage.”