Tiny California town is the focus of geologist’s effort to predict quakes

"Ultimately we are looking for a fundamental understanding of how quakes work. It’s like the Holy Grail: some kind of a reliable quake predictor," says Terry Tullis.

by Kristen Cole

Papering the wall across from the entrance to Terry Tullis’ office in the Geo-Chem Building is a map of California on which the splay of colors resembling a child’s paint-by-number illustration indicates seismic activity.

Running a finger north to south, the professor of geological sciences points out the reason for earth movements in California: the San Andreas fault. And about halfway down the jagged line, his finger stops at a tiny town that may yield answers about forecasting earthquakes.

Using high-performance computers, Tullis will attempt to approximate the Earth’s behavior by simulating the evolution of a quake on that single, unstable midpoint of the fault over time.

Tullis is studying only a small portion of the San Andreas – about 20 of its 800-plus miles. But when he begins to slice and dice the area into grids to determine how land is behaving, he’ll need 1,024 computer processors running in concert to perform the calculations. That is where NASA figures in.

NASA’s computer capability enables Tullis, in collaboration with other investigators from NASA and various universities nationwide, to work toward developing an advanced earthquake modeling system called QuakeSim.

“Ultimately we are looking for a fundamental understanding of how quakes work,” said Tullis. “It’s like the Holy Grail: some kind of a reliable quake predictor.”

Tullis’ portion of the project will use information about the rate of movement – or slip – between the Earth’s plates, and information about friction, to determine the indicators for a quake in the making. He will look at how stress is distributed over the fault and how it is redistributed by quakes.

Tullis started this work years ago, but was unable to create a truly detailed picture because of the size of the computations the work would require. In this project, he hopes to create a much finer picture by creating smaller and smaller grids, each block representing a block of earth.

The ability to determine whether a pattern of earth movement is building toward a major quake holds obvious potential to save lives. For this reason, Tullis’ project – which centers on the town of Parkfield and for which his simulation tool will be named “Park” – is crucial.

Although Parkfield itself has a population of 18, the Los Angeles-area megalopolis lies a few hundred miles south. The last large earthquake on the southern San Andreas, and one of the greatest every recorded in the United States, occurred in 1857. Its foreshocks are thought to have started in Parkfield. But in 1857, Southern California was less populated and only two people died.

Quakes of a significant magnitude regularly shook Parkfield throughout the past century. All the Parkfield quakes have foreshocks, and experts speculate that one of those might set off a cascading process similar to the 1857 quake. The U.S. Geological Survey has placed an array of instruments in the Parkfield area because of the possibility that a foreshock might be the trigger for the next “Big One,” said Tullis.

“What is it about some of them that allows the small ones to stay small and others to cascade to bigger ones?” asked Tullis. “I’m looking for the patterns of microsesmicity that lead up to it.”

Like the snake oil salesmen of old, there are dubious characters whose earthquake predictions have negative consequences, economic and otherwise. However, said Tullis, “if we could [predict quakes] in a rational way then people could respond in a rational way.”

QuakeSim’s principal investigator, Andrea Donnellan of NASA’s Jet Propulsion Laboratory, said, “New quake models developed under QuakeSim are expected to yield future earthquake forecasts that will be used by a variety of federal and state agencies to develop decision support tools that will help mitigate losses from future large earthquakes.”

Tullis last set foot in Parkfield about a decade ago, yet he will learn about any major happenings on the opposite coast as soon as they occur. He is on a computer list to find out the moment the earth begins moving significantly in Parkfield.

In addition to Tullis’ work on “Park,” the QuakeSim team is working to create two other earthquake simulation tools. The work involves faculty at the Davis and Irvine campuses of the University of California, Indiana University, and the University of Southern California, along with scientists at NASA’s Jet Propulsion Laboratory, Goddard Space Flight Center, and the Ames Research Center.

“The deformation of Earth’s crust and the interaction between quake faults is a complex 3-D process happening on timescales of minutes to thousands of years,” said Donnellan. “Studying it requires sophisticated simulation models and high-performance supercomputers. The availability of space-based data and our current limited understanding of quake processes make this an ideal time to develop a system for studying deformation processes such as tectonics, quakes and volcanoes.”