Earthquakes: the deadly side of geometry.

In a discovery bound to raise the ire of real estate agents, a seismologist has discovered that living near certain faults poses significantly more danger than previously thought. Other faults, however, may not be as fearsome to their neighbors as they had once seemed.

Earthquakes often appear to strike capriciously, sparing some regions while devastating others nearby. By studying foam rubber models of faults and computer simulations of quakes, James N. Brune of the University of Nevada in Reno found that the geometry of faults may explain some differences among earthquakes. He reported on his research at last month's meeting of the American Geophysical Union in San Francisco.

Brune focused on two types of earthquakes, both stemming from faults that dive into the ground at an angle. In thrust fault earthquakes, land on one side of a fault gets driven up and over land on the other side, much like a block pushed up an inclined ramp. The contrasting situation is a normal fault earthquake, during which rock on one side of the fault gets pulled away from and down relative to the rock on the other side, like a block sliding down a ramp.

According to standard seismological theory, it should make little difference whether a house sits next to a thrust fault or a normal fault. Brune's simulations, however, revealed much stronger shaking near thrust faults, especially on the so-called hanging wall, the land that rises during a quake.

"If this model applies to the real earth, it's very dangerous for people living on the hanging wall of the thrust," says Brune.

The reason stems from differences in the seismic waves generated when a fracture develops and starts to grow. During a thrust earthquake, the fracture sends out a pulse of compression, which squeezes rock. When this pulse reaches Earth's surface, it reflects downward as a pulse of dilatation, which reduces stress on the rock. As these reflected waves intersect the fault, they relieve frictional pressure on the growing quake, allowing the rock on either side of the fault to slide more freely.

This amplification enhances shaking when the fracture reaches the surface. Normal fault quakes have the opposite geometry, which weakens the tremors.

Seismologists lack the critical measurements needed to test Brune's hypothesis for actual quakes. But circumstantial evidence of unusual damage patterns exists, Brune says. For instance, after the 1971 thrust fault earthquake in California's San Fernando Valley, geologists found signs of shattered ground-soil overturned by intense shaking-only on the hanging wall side of the fault.

In contrast, a strong normal fault earthquake in Nevada in 1954 failed to knock over ketchup bottles in a shack, even though the quake ruptured the ground only a few meters from the building. What's more, Brune has also discovered many precariously balanced rocks near normal faults, suggesting that the shaking immediately next to these faults is less intense than expected.

Thomas H. Heaton of the California Institute of Technology in Pasadena says that Brune's findings could explain the mysterious case of a strong earthquake that hit Japan in 1945. "The thrust fault went right through town. The story was that the houses on the hanging wall were demolished, while houses on the foot wall were much less damaged," says Heaton.

"There is little doubt that the phenomenon happens," says Heaton. "But it's not yet clear what its widespread importance is yet." The reflected waves that Brune studied would have little effect on earthquakes that fail to reach the surface, like the 1994 Northridge, Calif., earthquake. What's more, the amplified shaking only threatens regions close to thrust faults. At this point, however, seismologists cannot tell how close is too close. The affected regions could extend just a kilometer from the fault or much farther.