Cosmic Rays May Help Predict Earthquakes

A new method of dating past earthquakes can provide information that may help predict future quakes on certain types of faults, according to a paper published today (Nov. 6) in the journal Science by Zreda and his colleague, Jay S. Noller of Vanderbilt University.

Zreda, a cosmogenic isotope geochemist and assistant professor in the department of hydrology and water resources, and Noller, a seismologist, measured trace amounts of the chlorine isotope chlorine-36 (36Cl) on surfaces of bedrock exposed by faulting action at a site in Montana in order to date earthquake events there. The 36Cl results from chemical changes in surface elements that react with neutrons and muons produced by cosmic rays that enter the Earth's atmosphere.

These cosmic rays, produced in deep space and deflected into Earth's atmosphere by its magnetic field, bombard our planet at a rate that is known and predictable, Zreda says, for any given site's latitude and altitude. The longer a rock surface has been exposed to the air, the more 36Cl accumulates. By measuring the amount of 36Cl at different heights on a scarp, or vertical surface of rock exposed by a fault, and calibrating for geographic position, rock type, and erosion history, the researchers were able to calculate the time each portion of rock had been exposed.

They found that six earthquakes had struck the site in the last 24,000 years, and were able to date them to 400, 1,700, 2,600, 7,000, 20,300 and 23,800 years ago. The estimates are accurate to within 400-3,000 years.

Knowing the pattern of timing of past earthquakes at a particular fault may help in predicting when the next one will hit. "This approach has the potential for predicting large earthquakes along major faults, especially if the earthquakes are not very frequent, and many major faults fall into this category," Zreda says.

But of course it's not that simple. Predicting quakes based on past patterns will only work if the past pattern is uniform -- that is, if temblors come at evenly-spaced time intervals. And on that issue, the scientific jury is out.

Two schools of thought exist on temporal patterns of earthquakes in the Great Basin of western North America, Zreda explains. One holds that quakes come at evenly-spaced intervals; and the other, that they come in clusters, with long quiescent gaps between periods of activity.

"These two models," Zreda says, "have very different implications for our ability to predict earthquakes," with the uniform model making prediction more possible.

Unfortunately for would-be quake-forecasters, Zreda and Noller's data support the clustering model, which is the more widely held view among seismologists who study Great Basin faults.

In addition, Zreda points out that there are other limitations in the ability to make predictions using his method. Current abilities of analytical precision limit the accuracy of dating estimates to roughly 10 percent of the age, he says, even under optimal geological conditions. And the technique works only with fairly uneroded bedrock faults that experience quakes less often than about every 1,000 years. Also, faults that move horizontally rather than vertically -- such as California's famed San Andreas Fault -- cannot be studied with this technique.

However, the method could be used on many major faults, and is a marked improvement on earlier methods of dating. Previously, Zreda says, researchers using isotopes derived from cosmic rays for earthquake dating did so indirectly, for example using alluvial deposits offset by quakes. In addition, some seismologists used radiocarbon dating on fossils contained within sediments associated with faults. However, these techniques are less exact, he says, and may be compromised if the spatial arrangement of sediments is shifted. "Younger earthquakes often modify or obliterate evidence of older earthquakes," Zreda points out. "Ours is the first approach to date fault scarps directly."

Earthquake dating is the latest in a series of applications Zreda and his students have found for 36Cl isotopes derived from cosmic rays. They have previously dated other phenomena of interest, including fluvial deposits, groundwater, sand dune formations in Kansas and Nebraska, glacial moraines throughout the world, and even Meteor Crater in northern Arizona (it's 50,000 years old). They've also been able to measure the rate of "isostatic rebound" of Canadian land that was covered by glaciers. By dating rock outcrops on ancient beach heads that were left high and dry as the land rose, they determined that land freed from the pressure of glaciers 10,000 years ago at first rose 5 cm per year and now has slowed to one percent of that rate.

In the future Zreda hopes to expand his earthquake work to determine whether the timing of Great Basin quakes at various faults tends to follow the clustering model or the uniform model.