All of the nation's 104 nuclear reactors will need to undergo analysis using cutting-edge technology and the most recent data to assess how well they can withstand earthquakes, the Nuclear Regulatory Commission says.

Plant operators will be required to study the safety of their facilities using a new seismic risk model created by the NRC, Electric Power Research Institute (EPRI) and U.S. Geological Survey, which should be available later this year.

The modeling is expected to give clearer indications of the risks facing each of the plants, providing details on the ground shaking that plant operators can expect at any given site, NRC spokesman Scott Burnell said. Plant operators must then show the commission their facilities are equipped to handle the worst-case scenarios the model generates.

The NRC will likely start with 27 reactors in the eastern and central United States. Data in past USGS reports for those facilities have shown the "largest increase in seismic risk," Burnell said, while acknowledging the risk is slight and still covered by the plants' designs. Using the new model for those facilities is expected to show "areas where the plants can improve what is already an acceptable response to seismic events," he said.

"There's been some talk about these being the first 27," NRC spokeswoman Beth Hayden said. "But we may just look at all of them."

But even as it analyzes risks at nuclear plants, the NRC acknowledges it has challenges in identifying the risks that reactors face from seismic activity, which cannot always be pinpointed to a fault line or seismic region. It's particularly difficult in the eastern and central United States, the NRC and geologists say, because quakes are less frequent there than on the West Coast.

"One of the questions which has come up repeatedly is which of the plants are near faults or how many plants are in moderate or high seismicity regions," Annie Kammerer, senior seismologist and earthquake engineer in the NRC's Office of Nuclear Regulatory Research, told the commission at a meeting Monday.

"That's a very challenging question to answer because these seismic zones are not well-defined boundaries."

The review was proceeding before safety concerns were piqued in the wake of the March 11 earthquake and tsunami that crippled a nuclear plant in northeast Japan. Burnell said the review is "in no shape or form a response" to events in Japan.

27 reactors

The NRC will first review the following plants: Farley 1 and 2 in Alabama; Crystal River 3 and St. Lucie 1 and 2 in Florida; Dresden 2 and 3 in Illinois; Duane Arnold in Iowa; Wolf Creek in Kansas; River Bend in Louisiana; Seabrook in New Hampshire; Indian Point 2 and 3 in New York; Perry 1 in Ohio; Limerick 1 and 2 and Peach Bottom 2 and 3 in Pennsylvania; Oconee 1, 2 and 3 and Summer in South Carolina; Sequoyah 1 and 2 and Watts Bar 1 in Tennessee; and North Anna 1 and 2 in Virginia.

The NRC has been reviewing the strength of plants since 2005, and in 2008 the commission began applying new seismic information from EPRI into the design of new nuclear power plants, as well as USGS findings for existing eastern and central reactor sites. Western reactors, the NRC said in 2008, had already taken into account the greater seismic activity within that region.

The USGS in its 2008 report, which updated a 2002 report, presented updated information on how ground shaking is likely to be as a result of earthquakes. Because temblors of different magnitudes generate different amounts of force, the USGS presents the potential movement as g force, or acceleration relative to free fall.

The USGS report estimates the likelihood that a particular amount of force will happen over a certain time period. In a region, for example, it might warn that there is a 10 percent chance of getting a force equal to 20 percent of g or larger over the next 50 years, said Arthur Frankel, a USGS research seismologist.

The 2008 report included new information on faults and earthquakes developed since the USGS's 2002 analysis. Because there are not many quakes in the eastern and central United States, USGS also used models, Frankel said. There were new models created between 2002 and 2008, he said.

Calculating risks

Ninety percent of all the earthquakes occur at the boundaries of the Earth's tectonic plates, said Christopher Scholtz, a professor of geophysics at Columbia University. One of the best-known of those areas is between the West Coast of the United States and the east coast of Asia.

"Most of the earthquakes occur in the places where we expect them to occur," which are the areas with active faults, said Larry Ruff, a professor in University of Michigan's geological sciences department.

But three major earthquakes in the range of magnitude 7 in the early 1800s struck near the town of New Madrid, Mo. The epicenter of that quake has never been located, the NRC said.

A fault line responsible for a magnitude 7.3 quake in Charleston, S.C., in 1886 also has never been located, several geology experts said.

There also is the risk posed by undiscovered faults in earthquake-prone areas. In California, the 1994 Northridge and 1987 Whittier Narrows quakes both happened on fault lines that were mostly undiscovered. They occurred on "blind thrust" faults, which are buried beneath the top layers of rock in the Earth's crust, so there is no evidence on the surface that they exist.

Because there are fewer quakes in the eastern and central United States than on the West Coast, there is less opportunity to gather information about faults, experts said. Quakes help scientists study the potential for new temblors.

"There's probably places where there's faults lurking ... where there are large faults we don't know about yet," said Frankel of USGS.

Experts disagreed about the chances for a major quake on an unknown fault in the country's central-eastern region.

A major quake would need to happen on a large fault line, and those would be visible, Ruff said.

The magnitude 9.5 earthquake in Chile in 1960 happened on a fault equal in size to the area of California, Ruff said.

"That's not something you hide in the San Fernando Valley," Riff said, referring to a region of Southern California known for earthquakes.

"It's easy to hide, and therefore have an unknown fault, something that's small," Ruff said. "It's hard to hide a fault that's as large as you need to have a magnitude 9."

Scholtz disagreed. The New Madrid quake of 1811, he said, happened on a fault line unknown at that time. And even though today's technology is far more advanced, "we wouldn't have any special reason to make a study of them to know there was a fault there unless there was an earthquake," Scholtz said. "It's an unlikely place to look."

USGS in its seismic hazards report tries to account for the uncertainty posed by unknown faults, Frankel said. It also studies potential evidence of past earthquakes like sand deposits in the ground. Those indicate there were New Madrid quakes in A.D. 1450 and 900, he said.