In the early 1800s the central United States was hit by large earthquakes that, according to folklore, rang the Liberty Bell in Philadelphia, about 900 miles away.
Earthquakes in the center of the country. It's a great story, but whether true or not, a big question remains: Could it happen again today?
National seismic hazard maps -- maps that essentially map earthquake dangers -- have long indicated that the seismic zone centered in Missouri, around New Madrid, is in some ways more hazardous than California, despite the fact that no major earthquakes have hit the region since 1811 and 1812.
Data from a new study using the Global Positioning System satellites, however, show that the risk posed by large earthquakes in the central United States' New Madrid seismic zone may be significantly lower than previously believed -- and, in fact, there may be no risk at all.
These new results illustrate the need to reassess ideas about earthquake risk in the area.
"We've never really understood what causes earthquakes in the middle of plates," said Seth Stein, a professor of geology with the department of geological sciences at Northwestern University in Evanston, Ill. "They shouldn't occur according to simple plate tectonic theory.
"In California, for example, we can make assumptions on how often large earthquakes should occur because we know the rate of motion between the Pacific and North American plates," he said. "But in the middle of the plate historically we have had no such information.
"Geodesy or Global Positioning System technology is giving us our first chance to look at what the motions are in the middle of plates," Stein continued. "We know these motions are very small, and GPS is good enough to measure these small motions very precisely -- to fractions of an inch per year."
The study was conducted by geologists from Northwestern, the University of Missouri, the University of Miami Geodesy Lab and Grand Valley State University, with funding from NASA and technical assistance from the University NAVSTAR Consortium, a national consortium of universities supporting GPS research.
"A reassessment of ideas about the earthquake risk in this area would include the National Seismic Hazard maps prepared by the U.S. Geological Survey," Stein said, "which determine the appropriate level of earthquake-resistant construction."
Until now, it has been assumed that large earthquakes like those that struck the area in 1811 and 1812 were magnitude eight events that should occur in the New Madrid seismic zone every 500 to 1,000 years.
This hazard level is quite high, Stein said -- in some ways exceeding that in California.
"In particular, the formerly predicted peak ground motion for the New Madrid seismic zone exceeds that in Los Angeles," he said, "and the predicted highest ground motion area for the zone is larger than for Los Angeles or San Francisco."
The study's principle investigators became interested in this region several years ago, when a group from Stanford University published a paper stating they had measured very fast strain rates in the New Madrid seismic zone.
"That caught our attention because they were using GPS and because strain rates they recorded, if translated into displacement, gave displacements of four to seven millimeters a year, which is very high for a non-tectonic area," said Timothy Dixon, a professor at the Rosenstiel School of Marine and Atmospheric Science at the University of Miami.
"There is essentially no topographic relief in this area -- and four to seven millimeters of displacement a year continued over geologic time ought to result in the generation of some topographic relief," Dixon said. "That implied that either the processes that the Stanford group were measuring had just started up and should not be averaged over geologic time, or that there was a problem with the GPS measurements.
"We decided to undertake an independent GPS survey to look at the possibilities."
The study got under way in 1991 with the first set of measurements from 24 sites in the immediate area and 16 sites at greater distances in various parts of North America.
"I was suspicious that part of the signals or apparent signals obtained in the Stanford survey might be due to monument instability," Dixon said.
"When you make measurements with GPS you have to make those measurements over some highly stable mark set into the earth's crust," he continued. "Many of the marks used by the Stanford group were old marks not established for high precision geodesy -- and some of these marks were in areas where the alluvium might be unstable, which could lead to spurious motion.
"We put in a large number of our own marks, engineering them to be very stable," he said. "We also did a more rigorous analysis of the noise or error attached to the measurements, which is important for getting the highest quality and most accurate data."
The group took measurements again in 1993 and 1997, and the data indicated that movement within the New Madrid seismic zone is no more than two millimeters a year -- and there could in fact be no motion at all, according to Andrew Newman, a graduate student at Northwestern University who has been involved with the study.
"With maximum movements of two millimeters per year it would take at least 2,500 years to generate enough strain energy to produce a magnitude eight earthquake," Newman said. "Magnitude eight earthquakes release about five to 10 meters of slip. That's the worse case scenario based on our measurements.
"Another earthquake may never happen again in this region."
The New Madrid seismic zone overlays a Paleozoic rift system that is buried under the thick sediments of the Mississippi Embayment, according to Stein.
No faults are visible on the surface, but scientists believe that earthquakes occur through reactivation of faults left over from the Paleozoic rifting. The faults are reactivated by stresses within the North American plate.
"What we know is that there were quite large earthquakes in this region in 1811 and 1812, and since that time there has been a low level of activity," Stein said. "There were some magnitude six earthquakes just before the turn of the century and some magnitude fives in this century.
"Based on our measurements of no more than two millimeters of movement per year, this region could accumulate enough strain to cause a magnitude seven earthquake every 500 to 1,000 years -- but a magnitude eight would take much longer, 2,500 years or more," he continued.
"What that indicates to us is that the earthquakes of 1811 and 1812 were either quite a bit smaller than previously thought or that similar earthquakes will be far less frequent."
Magnitude seven earthquakes are 10 times smaller than magnitude eight. The GPS results agree with analysis of the earthquake history of the area. In the 1950s seismologists B. Gutenberg and C. Richter noted that in a given area, the time between earthquakes of a certain size is approximately ten times longer than that for earthquakes one magnitude unit smaller.
Since 1816, the New Madrid zone has had earthquakes with magnitude greater than five about every 10 years, and earthquakes with magnitude greater than six about every 100 years.
Based on those calculations, magnitude seven earthquakes should occur about every 1,000 years and magnitude eight earthquakes about every 10,000 years.
Earlier geological studies show that earthquakes similar to those in 1811 and 1812 occurred in about 1300 AD and 900 AD, Stein said.
Although these earthquakes have been thought to be magnitude eight, the GPS data and earthquake history -- plus the fact that these were about 500 years apart -- make it likely that the earlier earthquakes were much smaller, probably magnitude seven.
So why do earthquakes occur in the middle of a plate?
Stein said there is no definitive answer, but the study group and others think it's likely that the stresses within the plate are released on some set of old faults for a period of time and then the locus of that seismicity migrates somewhere else.
"We are thinking in terms of these transient features, and the intraplate strain release migrates from one weak zone in the crust to another over time," Stein said. "We know of a couple of other examples of intraplate earthquakes -- one near Charleston, South Carolina, in the 1800s and along the continental margins in Massachusetts in the 1700s.
"Currently, New Madrid is the most active of these zones, and we don't know yet whether this area is special in some way or -- what we think is more likely -- that these areas of seismicity move around the plate to weak zones."
This theory does imply that the current New Madrid seismicity is relatively recent -- it hasn't been going on very long in geologic time.
"And if we continue to see no motion over the next 10 to 12 years it may very well be that the system is shutting down," he continued. "One of these systems likely turns on and is active for a few thousand years and then shuts down and moves to another weak zone."
The geologic record provides evidence that supports this theory.
"We know from the geologic record that the New Madrid seismic zone could not have been active for 10,000 years," Stein said. "You can go back in the geologic record and look for disturbances due to strong shaking, or paleo seismology, and those don't extend back more than a few thousand years -- so it looks like this zone turned on relatively recently, while the plate motions that presumably provide the stress that makes the system work haven't changed in millions of years.
"From that," he said, "we can infer that these systems can turn on and turn off anywhere there is a fault zone."
The GPS study continues, and scientists currently are modeling the data to gain a greater understanding of its implications -- particularly how it relates to seismic hazard estimates in the area, such as the Seismic Hazard maps.
"Our study certainly indicates there is substantial evidence to support reducing the seismic hazard in this area," Stein said. "This change would have a major impact on construction costs, since higher expected hazards require more expensive earthquake-resistant construction.
"Lower expected hazards encourage construction to meet realistic and less expensive seismic safety standards."