Midwest's Earthquake Hazard Lower Than Thought, Satellite Data Show

The results, published in the April 23 issue of the journal Science, suggest the National Seismic Hazard maps, prepared by the U.S. Geological Survey to determine the appropriate level of earthquake-resistant construction, should be revised to reflect lower estimates of earthquake risk in the area. High hazard estimates could in the future lead to higher costs for new construction.

"There is an earthquake hazard there -- it's not zero -- but it's much lower than the maps say," says Seth Stein, professor of geological sciences at Northwestern University, who led the study. "The current hazard map says parts of the Midwest are more dangerous than L.A. or San Francisco. Most geologists would say that's unlikely, and that's exactly what our new data show."

Until now, it has been assumed that large earthquakes like those that struck the New Madrid area in 1811 and 1812 were magnitude 8 events, which should recur every 500 to 1,000 years. Because the hazard maps used this model, Stein said, the assumed seismic hazard in the mid-continent is quite high, in some ways higher than in California. The predicted peak ground motion for the NMSZ exceeds that in Los Angeles, and the area of peak ground motion predicted in the NMSZ is larger than for either Los Angeles or San Francisco.

A different picture emerges from the new study, which was conducted by geologists from Northwestern University, the University of Missouri, the University of Miami 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. Using GPS, the tiny movements of a network of geodetic markers in Missouri, Tennessee, Illinois, Arkansas and Kentucky were tracked since 1991 to accuracies of less than an inch.

"The results show little or no motion across the seismic zone," said Timothy Dixon, professor of marine geology and geophysics at the University of Miami, whose group processed the GPS data. "GPS sites all over eastern North America also show little if any motion across the seismic zone."

The small -- or zero -- motion observed implies that it would take a very long time to accumulate the stress needed to generate large earthquakes.

"More than 2,500 years would be needed for a future magnitude 8 earthquake, and at least 1,000 years would be needed before a future magnitude 7 earthquake," said Stein. Because magnitude 7 earthquakes are 10 times smaller than magnitude 8 earthquakes, the largest earthquakes in the New Madrid Zone are either 10 times smaller than assumed in the hazard maps or will occur much less frequently, he said.

The study also includes a new analysis of the earthquake history of the area, which agrees with the GPS data. In the 1950s, seismologists noted that in a given area, the time between earthquakes of a given size is about 10 times longer than the interval between earthquakes one magnitude smaller.

"Since 1816, the New Madrid zone has had earthquakes with magnitude greater than 5 about every 10 years, and earthquakes with magnitude greater than 6 about every 100 years," said Andrew Newman, a Northwestern graduate student and lead author on the new study. "So magnitude 7 earthquakes should occur about every 1,000 years, and magnitude 8 earthquakes should be about 10,000 years apart."

Earlier studies showed that earthquakes similar to those in 1811 and 1812 occurred in about 900 A.D. and 1300 A.D., said Joseph Engeln, professor of geological sciences at the University of Missouri.

Although these earthquakes have been thought to be magnitude 8 earthquakes, said Engeln, who was also an author on the study, "the GPS data and the earthquake history, together with fact that these were about 500 years apart, make it likely that the earlier earthquakes were much smaller, probably magnitude 7."

The new results illustrate that estimates of seismic hazard in the area, such as the Seismic Hazard map, should be reduced, said Stein. "It will be important to assess how these new ideas might modify estimates of the expected building damage and financial loss from future earthquakes." Such estimates, he said, depend not only on the size of expected earthquakes, but also on the specifics of building construction and local geology.

The GPS results give important new insight not only into earthquake hazards in the Midwest, but into the fundamental mystery of why earthquakes occur in this region at all, Stein said.

"The New Madrid seismic zone is in the center of the North American plate," Stein said. "It's easy to see how earthquakes occur along the San Andreas fault; that's the boundary between the moving Pacific and North American plates. But it's hard to see why they occur within the plate." The GPS data, he said, show that the North American plate is amazingly rigid. "On average, parts of eastern North America move relative to each other by less than 1/25 of an inch per year. Somehow, this tiny motion adds up over geologic time to cause large earthquakes."

One intriguing possibility is that large New Madrid earthquakes may never happen again. This possibility is suggested, Stein said, by the GPS observations that little if any motion occurs today, and the fact that the New Madrid area is very flat.

"Over time, fault motions typically build up impressive topography, as seen in the western U.S.," Stein said. "The New Madrid Seismic Zone may be a short-lived feature that turned on less than 10,000 years ago, and is now shutting off with a series of small earthquakes. We've got a lot to learn about New Madrid, but this is certainly a possibility."

Other authors on the study published in Science are John Weber of Grand Valley State University and Ailin Mao of the University of Miami.