If you search a satellite image of upstate New York -- or just about anywhere in the United States -- you can pick out a maze of lines, winding paths and other patterns.

Robert Jacobi, a geologist who hunts for earthquake faults, is not interested in most of these lines, which represent roads, fences, ditches, pipelines or other things built by people.

Jacobi is interested in natural lines, or lineaments, as he calls them: simple streams; a long row of dead, dying or stressed plants; maybe a hint of out-of-place autumn color skipping along the top of a forest canopy.

Yellow, green, dark blue and red lines show suspected fracture systems near the Finger Lakes region of upstate New York.

Faults on major cities?

Jacobi and colleague John Fountain, along with a host of students, studied a small, remote portion of a long, flat plateau that stretches from Ohio to upstate New York and into Canada. From the ground, the plateau appears relatively featureless. Geologists call it pancake geology.

From space, though, the studied patch of the rocky region is revealed to be riddled with faults, and Jacobi says the finding may challenge assumptions of seismic safety for Cleveland, Buffalo and Toronto -- all of which sit on the same plateau.

"All those cities were assumed to have no faults," Jacobi said in a telephone interview. "And what we're suggesting is that there are more faults than we suspected."

It is not known whether any of the faults are active.

High earthquake activity occurs in places where plates of Earth's crust collide, as in California and Japan. High, steep mountain ranges are evidence of the young and violent process. The eastern United States, on the other hand, sits in the middle of a continental plate, where less frequent movement indicates a mature process, manifested with less drama in the mountain ranges.

"In the East, most of these intracontinental faults have long periods of quiescence, and then they let go," Jacobi said. Knowing when is impossible, however, because geologists have no record of past events on most faults.

"The recurrence rate of big earthquakes on intracontinental faults is around 1,000 to 2,000 years," Jacobi points out. "Our historical record only goes back 150 to 200 years."

Boston, New York, Washington and Philadelphia lie outside the plateau Jacobi studied. These cities have their own infrequent earthquake risks, however.

A 5.0 temblor in 1737, for example, knocked down chimneys in New York City and was felt from Boston to Philadelphia. A magnitude-5.5 quake in 1884 did similar damage in a wider region around New York.

But it's a major earthquake that geologists like Jacobi fear. And major earthquakes are not confined to the West Coast.

In fact, the most widely felt temblors in North America were a series of four that hit near New Madrid, Missouri from December 16, 1811 through February 7, 1812. They ranged from magnitude 7.6 to 8.2. The largest was felt from the East Coast to the Rocky Mountains and from Canada to the Gulf of Mexico. New Madrid also represents an intracontinental fault system.

"Most other intracontinental faults do seem to be seismically capable of large-magnitude events," said Jacobi. "Because these faults are also intracontinental in nature, they too could be capable of large-magnitude events."

John Everett, a geologist at Earth Satellite Corp., worked on the original Landsat data that Jacobi's team ultimately incorporated into their study. Everett said there is plenty of evidence to indicate that the East Coast has active faults. Because faults can go so long between major movements, Everett said, Jacobi's technique is a valuable to way to help spot potentially active faults.

"You can't guarantee that you'll know the faults are active," Everett said. "But the biggest problem is finding these things in the first place."

Landsat images show natural landforms and also shapes created by humans.

Once potential faults have been catalogued, geologists can go into the field and look for signs they still are active. A leaning tree, for example, can indicate that a fault is slipping quietly over time, possibly building stress that might give way in a sudden snap. Or, related rock outcroppings on two sides of a valley can be compared, showing height differences that reflect movement going back millions of years.

There are many clues to look for, the most obvious being streams.

"Faults chew up the ground, and so the ground is more susceptible to erosion -- so many streams follow faults and fractures," Jacobi said.

Variations in plant species or health can also be indicators: Stressed vegetation blooms later in the spring and turns color earlier in the fall.

"Some of these lineaments are there because faults conduct water and soil gas," Jacobi said "They conduct fluids and gases that will make trees and grass respond differently."

Even within a forest canopy, scientists can pick out differences by flying over a region at various times of the year. In drier areas, where there is little vegetation, what greenery there is typically follows fault lines, where water reaches the surface.

From airplanes, readings of magnetic fields emanating from rock below helped Jacobi's team calculate that some of the faults extended several miles (kilometers) below the surface. Matching their results with known faults in the far north of the study area helped show the method was reliable.

Using satellites to hunt for gas and oil, as well as mineral deposits, is nothing new, said John Martin of the New York State Research and Energy Development Authority, which produced the 1997 study from which Jacobi pulled his satellite data.

"Sometimes if you look at something real close, you can't see anything," Martin said. "But if you step back, a pattern takes shape."

Martin said using lineaments to search for fault lines that might indicate oil and gas reserves is a proven technique that has been refined for a quarter century.

"I think the technique is widely accepted, though there are still detractors," he said, adding that Jacobi's method of combining ground, airborne and satellite data was unique and, like anything new, might not be gain favor right away.