Shortcut Navigation:

Forecasting Earthquakes Using Paleoseismology


Thomas Rockwell in Mongolia digging trenches by hand along the Bogh fault, Mongolia. Photo courtesy of Thomas Rockwell. 

“My mother first got me interested in geology,” says Tom Rockwell. “We used to go out into the desert and collect rocks, and I started a rock collection when I was about six, along with my shell collection and my coin collection and all my other collections…” Now a paleoseismologist at San Diego State University, Rockwell studies the recent history of faults. A fault he defines as “a fracture in the Earth’s crust across which movement occurs,” and it is this movement—the friction between the two sides of the fault—that produces an earthquake. Rockwell became interested in faults and plate tectonic theory in his first geology class, and, in particular, “how faults formed in these large, worldwide systems. That eventually evolved into the work that I do now, where we go out and directly study the activity of faults and how they generate earthquakes over a period of time.”

More specifically, he and his collaborators head for active fault zones all over the world, dig trenches across the faults, and examine the sedimentary layers for evidence of quakes large enough to have broken through to the ground’s surface. “Our job is to try to put all the pieces of the puzzle back together, to reconstruct the earthquake history at that site,” Rockwell explains. “Then we want to date the sediment as precisely as we can, so that we can then reconstruct the actual timing of past quakes.” The clearest record is found in sites where fine-grained sediments have been deposited more or less continuously, right up to the present day. “If the sediments are deposited much more frequently than the quakes occur, then they should show every quake that has broken the surface at that point along the fault. It’s really the sediment that provides the record,” he confirms. It’s a cycle. Sediments accumulate; an   earthquake blasts through the sediments; more sediment is deposited across the fault, capping the layers that were deformed by the quake; another quake occurs; and on and on. “After five or six cycles, the story can get pretty messy,” Rockwell acknowledges. It’s a subtle story to begin with. Paleoseismologists may have to rely on delicate changes in grain size or color. Some layers are defined by nothing more than a faint line of individual grains.

The first step is to open a trench across the fault, usually with the help of a backhoe that digs a deep slot across the fault. “We’ll keep the slot open with hydraulic shores—big aluminum planks held open with a piston to keep the trench from collapsing,” explains Rockwell. “Then we try to cut the face of the trench perfectly flat, commonly with thin-bladed, spatula-like tools called taping knives. If it’s dry and sandy, we can also use a brush.”

The next step is to map out the exposure in detail, specifically its contacts. Contacts are the places where the nature of the sediment changes, for example from a sandy stream deposit to a silty flood plain deposit. The scientists place a grid of string across the exposure and document its surface as precisely as possible, using several different methods. They may photograph it and draw the contacts directly on the photograph. They may take measurements from the grid and map onto paper. Or they may skip the gridding altogether, rely on precise surveying equipment, and shoot the contacts directly into a computer which generates a picture of the exposure. According to Rockwell, “that technique is great if you don’t have too many contacts.”


Aerial image of the San Andreas transform fault in the Southern California desert. Photo courtesy of the United States Geologic Survey. 

Choosing where to dig is part technology, part research, and part intuition. To start with, Rockwell and his colleagues study aerial photography and satellite imagery for features that indicate the presence of a fault, such as areas lifted by fault movement. “Something that you wouldn’t necessary pick up on the surface stands out like a straight line on the aerial photography or satellite image,” says Rockwell. He avoids old, weathered surfaces because they won’t be up-to-date, and looks for fine-grained material with lots of organic matter. The sediments need to contain something like charcoal or organic peat deposits in order to be dateable by carbon-14 methods. Still, he says, “to some degree it’s a shot in the dark. It depends on the geomorphology—the lay of the land—and a lot of it comes with experience.” A number of new methods have been developed to study faults, including drilling and ground-penetrating radar, but according to Rockwell, “there really is no substitute for going out and digging a trench.”

The only thing better than one trench is lots of them, and it’s usually necessary to dig a number along any fault to characterize its behavior. “That usually also means doing a number of trenches at a single site, sometimes in three dimensions to characterize its faulting history in all directions. Ideally, we’d like to know the history every few kilometers along any given fault, but that’s obviously a lot of work.” Cities pose a real challenge. “Oh, it’s tough,” says Rockwell. “In San Diego, we did the standard look at aerial photography to locate the fault, and then we had to find a parking lot we could trench in. Sometimes in urban environments it’s almost impossible, because they’re almost completely built over.”

Rockwell’s work has taken him to more exotic places than downtown San Diego, including the Dead Sea fault zone in Israel, the northern Anatolia fault in Turkey, the Himalayan frontal thrust fault in Nepal, and the northern Gobi Desert in Mongolia. The “paleo” in paleoseismology means ancient, but in Rockwell’s field it refers to any earthquake that occurred prior to the instrumental record (approximately at the turn of the twentieth century). Instrumentation is a relatively recent development, but in the Dead Sea zone a long written record exists in ancient diaries and church and mosque records. “The work in the Middle East in particular is driven by these long historical records,” explains Rockwell. “This enables us to make much stronger statements about long-term fault behavior in those areas. And from this information we improve our chances of correctly forecasting future earthquakes in the U.S.”

“One thing we suffer from in the U.S., in California in particular, where we have the main fault, is a short written historical record,” he continues. “There are some older earthquakes that we know about from mission records, for example, but we don’t know which fault produced which quakes.” Paleoseismology is one way to find out. “In essence, what we’re trying to do is establish a several-thousand-year record of large quakes in California and all over the western U.S. That’s where I study, because that’s where most of the action is.”

Rockwell and his colleagues can’t predict when an earthquake will occur, but they can make a statement about probability based on past history. For example, “prior to the 1989 Loma Prieta quake, the Santa Cruz mountain segment of the San Andreas fault had been forecast to have a high probability of an earthquake with the potential to do real damage. And it certainly occurred within the prescribed zone in the prescribed time frame. So you can call that a success, at least in terms of reducing hazard and making people more aware of it,” Rockwell points out. “Ultimately we’d like to be able to forecast the likelihood of an earthquake in the fault in your backyard, say for the next thirty- to fifty-year period, so that you can design your building accordingly and make sure you don’t build it on an active fault.” On the other hand, the paleoseismologist ruefully admits, “it’s a rather amazing thing, but most people do not want you to trench in their backyard and show that a fault goes through their living room. They simply don’t want to know.”

American Museum of Natural History

Central Park West at 79th Street
New York, NY 10024-5192
Phone: 212-769-5100

Open daily from 10 am - 5:45 pm
except on Thanksgiving and Christmas
Maps and Directions