The Berkeley Science Review: Articles

Uncommon Ground
Probing the underlying causes of earthquakes (view PDF)
by Susan Young

The Bay Area has its faults: the notorious San Andreas Fault, which ruptured in the great 1906 San Francisco earthquake, the Calaveras Fault near the cities of Dublin and Pleasanton, the Rodgers Creek Fault extending up to Santa Rosa, and of course, the Hayward Fault, which may be under your feet right now as it snakes directly beneath the Berkeley campus. The last five major earthquakes on the Hayward Fault, have, on average, occurred every 140 years. When was the last major quake on our neighborhood fault? October 21st, 1868. That's right; the 140-year anniversary is this year.

For most of us in the Bay Area, earthquakes are a looming possibility that we hope will never materialize, but for which we are (or should be) prepared. For the scientists of the Berkeley Seismology Laboratory (BSL), earthquakes are part of a complex story of earth science, earthquake monitoring and reporting, and hazard assessment. The BSL is the oldest research center on the Berkeley campus and has been operating earthquake-monitoring networks in Central and Northern California since 1887. As well as conducting earthquake research and educating future earth scientists, the BSL provides immediate earthquake information to a variety of agencies around the Bay Area, including emergency response operators like the California Office of Emergency Services. BSL seismologists are interested in the structure and physics of faults and other features of the earth's outer layers, real-time monitoring of seismic activity, and even early earthquake warning systems. The work done by Berkeley earthquake scientists covers an immensely broad spectrum of earth science, but a common goal of the BSL is to alleviate the damage that results from the greatest natural risk in our area.

An insider's perspective
The outermost shell of our planet, the crust, is made up of large tectonic plates that create the continents and the ocean floor. These plates are moving constantly, albeit slowly. This leisurely dance is guided by the plasticity of the warmer and weaker layer below. Plate movements can cause the brittle crust upon which we live to fracture, leading to the formation of faults. The continuous motion below the plates causes strain to build up in the land surrounding these faults. Strain can eventually be released by a "slip" of the fault—meaning that the two sides of the fault move relative to one another. If a slip has sufficient speed and strength, it can cause an earthquake and send those of us in the vicinity scrambling under our desks.

The San Andreas Fault, which runs from Southern California to the Bay Area, is one of the most notorious active faults in the world. It is a plate boundary fault, meaning it lies at the interface of two tectonic plates—the Pacific Plate to the west and the North American Plate to the east. The San Andreas Fault is home to some of the most infamous earthquake disasters, including the great 1906 San Francisco Earthquake (magnitude 7.8) and the 1989 Loma Prieta earthquake (magnitude 7.1).

The San Andreas Fault is now the site of an ambitious National Science Foundation funded project that aims to understand how earthquakes start. To this end, a team of scientists is investigating an active part of the fault at the depth at which earthquake initiation takes place. The San Andreas Fault Observatory at Depth, or SAFOD, is located near Parkfield, about four hours south of San Francisco. Scientists drilled a giant 3.2 kilometer (about two miles)-deep hole crossing right through the fault. The SAFOD observatory will allow seismologists to directly study the chemical and physical features of an active fault. "It's fundamental to furthering our understanding of earthquake nucleation, rupture dynamics, and faulting mechanics (fault motion and growth) on a much larger and longer-term scale," says the BSL's Douglas Dreger, Associate Professor of Geophysics at UC Berkeley.

Shake, rattle, and repeat
Near Parkfield, the San Andreas Fault generates repeating mini-earthquakes that occur with great regularity every 2.89 years. The repeating earthquakes are small, about magnitude 2.0. However, recent work from Dreger and colleagues finds that these unusual repeating quakes are kinematically similar to larger quakes. This means that the speed at which slip accumulates on the fault and the speed with which strain occurs is comparable between the small and large earthquakes, suggesting that smaller earthquakes, like those at Parkfield, have the same underlying mechanisms as their larger counterparts. Studying these frequent mini-earthquakes will help scientists understand the nature of larger quakes, like the Big One we all worry about in the Bay Area.

The model for the intriguing clock-like behavior of the repeater quakes is based on the geological composition of the fault and the surrounding land. Soft rock at the face of the fault would allow for slip without any earthquake behavior; the two faces of the fault slide past one another smoothly and no seismic waves are generated. However, explains Dreger, near the sliding portion of the fault there may be patches of high strength rock that are immobile. "You can think of everything around these little patches as moving, but [the patches themselves] act as pins that are holding the fault together," says Dreger. As movement continues around the stuck patches, stress increases on the patches of high strength rock. Finally, an intrinsic failure threshold will be reached at which point stress is released through a fast, seismically-active slip. "After the earthquake, it starts again. Stress starts accumulating [anew] because the surrounding area of the fault is undergoing continuous deformation. We refer to this as an earthquake cycle model." In support of the geologic aspects of this model, samples obtained by drilling at the SAFOD observatory contained a soft rock, serpentinite. This soft rock may act as a lubricant, allowing the smooth slow slip movements seen near the repeating region of the fault.

Rocky relationships
Regularly occurring earthquakes like the ones in Parkfield can be thrown off schedule by other seismic activity. In September 2004, a magnitude 6.0 earthquake occurred on the San Andreas Fault near the SAFOD observatory site. Two days after the large quake, the repeating earthquakes occurred one year early, and then again just 69 days later. Dreger explains that while the large quake reduced stress at the site of the rupture, the surrounding areas of the fault, including those where the repeating earthquakes occur, experienced a stress increase. "Basically, when an earthquake occurs, you redistribute stress," he explains.

The ability of the crust to redistribute stress may be one reason that the Bay Area has had relatively light earthquake activity for the last century. "The Bay Area over the last 100 years or so has been so quiet, probably because the 1906 earthquake on the San Andreas Fault unloaded the whole system and reduced [stress in the region]," says the BSL's Roland Bürgmann, Professor of Earth and Planetary Science. An earthquake can change the forces in the surrounding crust. This can cause regions of nearby faults to become more "pinned" and thus less likely to slip, while others may become more likely to slip because increased stress is transmitted to the surrounding rock. "Fully understanding these fault interactions may be key to improving our ability to forecast earthquakes," says Bürgmann. Furthermore, if we know that the stress on a particular fault has been reduced by a rupture on another, then perhaps we can better assess real earthquake risk.

But I'm a creep...
The fault activity most of us are familiar with is the ground-shaking, vase-tossing earthquake. However, exploring a variety of types of fault activities provides seismologists with more pieces of the crust-and-mantle puzzle, allowing them to better understand the behavior of faults and aiding in making accurate earthquake risk assessments.

Creep is one of these less well-known types of fault activity. Creep refers to small, slow movements of earth at the interface between two tectonic plates, similar to the smooth slip discussed before. Although these movements are too small to cause seismic activity, they can cause slow damage to football stadium walls, or form that uneven sidewalk crack you always trip over in front of Mulford Hall. Seismologists think that in some locations, creeping movements relieve pressure at the fault. Imagine a fault that is locked, with no movement at the interface. The crust adjacent to the fault will become strained due to continuous distortion of the land around it. "This deformation will be recovered when the fault slips," says Bürgmann. In other words. when the locked fault eventually gives, the surrounding land that was pushed and pulled out of place by the continuous movements of the crust will "snap" back, leading to a potentially severe earthquake. A creeping fault, which releases the strain of crust deformation via slow and smooth movements, may be less likely to experience a severe rupture.

The Hayward Fault has creeping segments from Fremont to Point Pinole, moving about four to seven millimeters per year. Whether the Rodgers Creek Fault near Santa Rosa creeps or not has been questioned for some time, and the resolution of this issue is important for estimating earthquake probabilities in the region. In a recent paper, Bürgmann and colleagues report that like the Hayward Fault, the Rodgers Creek Fault in the North Bay also has creeping segments.

Modern methods of studying surface land movements allowed Bürgmann and colleagues to better observe the area around Rodgers Creek Fault. They combined information from GPS (Global Positioning System) with information from InSAR (Interferometric Synthetic Apeture Radar), which records radar waves sent from space and bounced off the Earth's surface back to the satellite that sent them. Both methods allow seismologists to study movements of the land surface, including the land surrounding the Rodgers Creek Fault. The researchers found that near Santa Rosa the Rodgers Creek Fault is creeping at up to 7.5 millimeters per year. "Now that we have GPS and InSAR data, we find a whole spectrum of behavior, [including] slow earthquakes or so-called 'silent earthquakes.' This is exciting because it opens up a new window of fault behavior," says Bürgmann.

Tiny temblors, ponderous plates
The dynamic activities of tectonic plates are not limited to land. Assistant Professor of Earth and Planetary Science Richard Allen and colleagues recently reported an unusual type of slow fault movement just off the northwestern coast of the North American Continent. Here, two tectonic plates meet: the Juan de Fuca Plate and the North American Plate. At this long fault, called the Cascadia Subduction Zone, the oceanic Juan de Fuca Plate is sinking below the continental North American Plate.

Initial studies at the subduction zone found that the region is experiencing very slow slip events. Whereas a ground-shaking earthquake exhibits slip in a matter of seconds, these slower movements exhibit slip over several weeks. During these events, Allen says he and his colleagues noticed "continual chatter on our seismometers" that they determined to be tremors associated with these slow slip movements. The researchers call this recurring association of slow slip with tremor "Episodic Tremor and Slip" (ETS).

Understanding ETS events provides a new window into understanding the physics of the earthquake process. Tremors like these have been found at other subduction zones, such as in Japan and Mexico, and the geology of plate subduction may be related to the tremor process. "The tremors have very low amplitude, so there's no ground shaking, but studying them can teach us about the processes close to these large faults that have the potential to generate an earthquake at some point in the future," says Allen.

Allen and his team are interested in many aspects of the Cascadia subduction zone and the interaction of the two major plates there. In addition to the phenomenon of ETS, the group is examining the structure of the crust and mantle that comprises the whole of western North America. "At the moment, those two studies are at two ends of a spectrum. One's very large scale—structural—and one is relatively fine scale—the details of earthquake processes," Allen explains. "What we are trying to do is fill in that spectrum between the two so we understand how earthquake processes relate to large-scale structure."

Straight from the source
One of the BSL's most critical jobs is to monitor, analyze, and report seismic data from stations across Northern and Central California. The Berkeley Digital Seismology Network is a network of seismic instrumentation that continuously reports seismic information from remote sites in Northern and Central California back to the BSL here on campus. "What's great about it is, with the modern technology, it only takes a few seconds for the information to come back to Berkeley," says Dreger. This allows BSL researchers to perform real-time analysis of earthquakes. Groups ranging from the California Office of Emergency Services to PG&E need to know how big an earthquake was and how strong the resulting ground motions were in order to best respond to a disaster, and they need this information quickly. Institutions like PG&E can make decisions about whether it's safe to keep operations running or not; emergency services can decide which areas need the most help and attention.

An early warning
But what about assessing the impacts of an earthquake before the shaking begins? This may sound like earthquake prediction, but it is not. Researchers at the BSL have developed a methodology that can provide tens of seconds of warning before the ground-shaking movements of an earthquake hit by catching and evaluating its earliest signals. When an earthquake occurs, seismic waves called P-waves, or primary waves, are emitted. (P-waves are generally too weak to cause ground shaking.) The P-waves are then followed by potentially damaging S-waves, or secondary waves. S-waves are the forces that cause the earth to shudder, buildings to sway, and books to fly off their shelves.

In 2003, Allen first published his idea for an early earthquake warning methodology (see BSR Spring 2006). An ElarmS (Earthquake Alarms System) warning describes when, where, and how strong the shaking will be in an AlertMap, a map predicting the scale of shaking across the region. Since the initial 2003 report, Allen's group has been busy testing and improving the system. Allen applied the ElarmS methodology to a set of 32 past earthquakes in Southern California to test the accuracy and timeliness of the system. "[The results] were very encouraging, so we decided to move forward and test it in real-time," says Allen.

In 2007, Allen reported the results of running ElarmS in an automated test-mode in Northern California. From February 2006 to September 2006, the system monitored and predicted the effects of 75 earthquakes in the area. The system turned out to be quite accurate. Compared to the shaking intensities that were observed, the ground motion predictions were correct to within an average of 0.1 units of the modified Mercalli intensity scale, a measure of an earthquake's ability to cause structural damage on a scale of 1 (no movement) to 12 (widespread damage to buildings)—as opposed to the Richter scale, which measures the energy released. Among the 75 events monitored, two earthquakes represented likely hazardous scenarios for the Bay Area. The warning times for these cases ranged from 3–30 seconds, depending on the Bay Area city. For example, for an August 2006 magnitude 4.7 quake near Santa Rosa, ElarmS generated an accurate prediction of ground shaking with six seconds of warning for San Francisco and Oakland and 24 seconds of warning for San Jose.

An ideal first user of ElarmS would be an organization like the Bay Area Rapid Transit (BART) public transportation system. BART is completely computer-controlled, enabling it to instantly respond to the warning. In the event of a severe earthquake, trains could be stopped to reduce the risk of passenger injury, and in the event of a false alarm, they could simply be restarted. Integrated into school and work buildings, an ElarmS warning system would give us time to duck under a table or desk.

Bringing ElarmS online however, is a major undertaking. More real-time testing needs to be done and a lot of infrastructure needs to be developed and deployed, including more seismic sensors, which are required to improve the ground-shaking intensity predictions. Currently, Allen is in the process of getting the system running in an automated test-mode in the rest of the state, with a goal of having it running on a statewide basis sometime this year.

It's our fault!
This year is the 140th anniversary of the last major Hayward Fault earthquake. While most of us are comforted to know the BSL will continue its history of cutting edge seismology research, earthquake monitoring, and earth-science education, we also have our own important work to do to prepare. I've been living in earthquake country for almost four years now, and I have to admit I haven't yet prepared for the Big One. Taking the advice from Dreger, "the most important thing is to have a plan of self-sufficiency for three days." Allen agrees: "As individuals, we need to be ready. That means food, water, shelter, and medical supplies. You also need to be able to look after the people around you, and have a plan of how to contact the people who are going to be concerned." With this advice in mind I'll be ready—come what may—during this anniversary of our very own neighborhood fault.

Susan Young is a graduate student in molecular and cell biology.

Want to know more? Check out:
The Berkeley Seismological Laboratory: seismo.berkeley.edu
The 1868 Earthquake Alliance: 1868alliance.org
The US Geological Survey page on San Andreas Fault Observatory at Depth:
earthquake.usgs.gov/research/parkfield/safod_pbo.php
ElarmS: www.elarms.org

Comments on this article? Drop us a line at with 'letter to the editor' in the subject!