01/12/1995 08:00:00

Southern California Scientists Mark Northridge Earthquake Anniversary with Suite of Papers in Science

PASADENA—Scientists affiliated with the Southern California Earthquake Center (SCEC) will publish three sobering papers in the January 13 issue of the prestigious journal Science. The studies examine the nature and likelihood of future earthquakes in the L.A. region, and the expected response of certain types of buildings to a major temblor. The scientists conclude that both the likelihood of major earthquakes in the region and the potential for damage to certain types of large buildings are greater than was previously thought. SCEC is funded by the National Science Foundation and the United States Geological Survey (USGS).

A paper by James Dolan, formerly a geologist at Caltech who is now with USC, Kerry Sieh, a Caltech geology professor, and six coauthors makes the most thorough numerical assessment to date of the potential for future destructive earthquakes in the L.A. area. In research supported by the National Science Foundation and the USGS, Dolan, Sieh, and their colleagues examine two reasonable scenarios for earthquakes on six major fault systems in the greater L.A. area in a paper titled "Prospects for Larger or More Frequent Earthquakes in the Los Angeles Region, California." The scientists, all eight of whom participate in SCEC, conclude that much of the seismic strain in the L.A. region is probably released in large—magnitude 7.2 to 7.6—earthquakes that occur, on average, about every 140 years. This research was also supported by the City of Los Angeles, the County of Los Angeles, and the California Department of Transportation.

Starting with their knowledge of the locations of six major fault systems underlying the metropolitan area, and with the measured geologic slip rates, the geologists asked how often earthquakes would occur if all the strain were released in moderate earthquakes the size of the 1994 Northridge and 1971 San Fernando earthquakes, about magnitude 6.7. Dividing the six known major fault systems into 51 segments, each capable of creating magnitude 6.7 earthquakes, and combining the expected recurrence intervals, the scientists found that we should be jolted by a Northridge-sized earthquake about every 11 years, on average. In other words, in the nearly 200 years of historic record, Los Angeles should have felt about 17 temblors of this size. In fact, we have had only two, Northridge in 1994 and Sylmar in 1971, which leaves a deficit of about 15 magnitude 6.7 earthquakes. Although they don't rule out the possibility that the strain accumulated over the past two centuries could be released in a series of future moderate earthquakes clustered closely in time, the scientists consider this scenario to be unlikely.

In another scenario, the authors asked how often an earthquake would strike if strain were released in large—magnitude 7.2 to 7.6—earthquakes that would break the entire length of one of the six major fault systems. Since big earthquakes relieve more strain than small ones, this scenario yields less frequent, but more severe events. Summing the calculated recurrence intervals for each of the six major fault systems, which ranged from about every 500 years for a magnitude 7.2 on the San Cayetano fault to roughly every 2,200 to 3,300 years for a magnitude 7.3 on the Hollywood–Santa Monica–Malibu Coast fault system, the SCEC geologists found that large earthquakes would occur, on average, every 140 years. Yet the 200-year history of the L.A. region, which is probably complete for very large temblors as far back as the 1780s, is devoid of such large events on faults within the metropolitan region. The geologists note that these scenarios are useful as minimum and maximum estimates of earthquake size, but that the actual method of strain release is certainly more complex, involving a combination of large and small events.

The scientists end by noting that while information from excavations across faults is still too sparse to reveal when the most recent large earthquakes have struck on faults in the L.A. metropolitan area, recent data from several faults suggest that individual faults slip very infrequently. This, they caution, suggests that very large earthquakes, not clustered moderate-sized ones, may be the principal type of seismic activity in the L.A. area. As to why greater Los Angeles hasn't felt anything in the mid-7 magnitude range for more than 200 years, they cite examples of earthquake clustering that indicate we could well be within a centuries-long quiescent period.

In a second Science paper, "Earthquakes in the Los Angeles Metropolitan Region: A Fractal Distribution of Rupture Size?", seismologist Susan Hough of the USGS in Pasadena also attempts to determine the frequency and size of large earthquakes on L.A.'s faults, but using a different method. She uses the measured deformation rate of the L.A. metropolitan region, approximately 1 centimeter per year, to estimate the recurrence time for various sizes of earthquakes. Her theoretical work is based not only on slip measurements, but also on a distribution pattern of earthquake sizes that is well established from decades of observations: for every magnitude 7 in a certain area and time period, there will also be roughly 10 magnitude 6s and 100 magnitude 5s.

In Hough's model, the recurrence interval is tied to the size of the earthquake, which makes sense: bigger earthquakes would relieve more crustal strain, so fewer of them would be needed. But smaller temblors would provide less relief, so more would be required to account for the measured slip. Specifically, if Hough assumes the longest possible rupture is 100 kilometers (62 miles), which would create a magnitude 7.35 quake, then the average recurrence time is 203 years. And the repeat time for smaller—magnitude 6.5—events would average about 33 years. Given the historical record of seismic activity in the L.A. region, this predicted repeat time is implausibly short; we would expect to have had many more moderately large earthquakes in the last two centuries than have actually been recorded.

One of the few ways to account for a longer repeat time for both large and moderate events is to assume a longer maximum possible rupture length. According to Hough's model, a rupture 130 kilometers (81 miles) long would produce a magnitude 7.45 earthquake on average every 284 years, and an added six or seven events of magnitude 6.6—about the size of the Northridge and Sylmar earthquakes—in the same time span, one on average every 46 years. (For comparison, recall that the magnitude 7.3 Landers earthquake rupture was about 80 kilometers long, and the magnitude 6.7 Northridge event about 10 kilometers long.)

These numbers, from the model with a maximum fault length of 130 kilometers, are more consistent with the historical record. This model predicts that most of the seismic stress in the L.A. region will be released by infrequent but extremely large events, while moderate events, similar in size to the Northridge earthquake, are also expected every 40 to 52 years. These results are consistent with the findings of Dolan, Sieh, and colleagues, although Hough's model suggests a mixed distribution of ruptures with a variety of sizes. But even allowing for occasional large events, as this model does, the historical record of magnitude 6.5 to 6.7 events in the L.A. area is still well under the predicted long-term rate.

Hough used earthquakes of magnitude 6.5 to 6.7 as a simplification in her study, and because millions of people are familiar with this size from the Northridge and Sylmar events. There is nothing special about this particular size. In reality, scientists expect a range of different magnitudes smaller than the largest, in this study a magnitude 7.5, earthquake. For example, a magnitude 6.9, a 7.1, or a 6.4 earthquake would be possible, and smaller events in the magnitude 5s and low 6s would certainly be expected as well. A third paper examines how well two types of buildings can be expected to hold up during an imagined large earthquake directly under Los Angeles. This study, "Response of High-Rise and Base-Isolated Buildings to a Hypothetical MW 7.0 Blind-Thrust Earthquake," was written by seismologists Thomas Heaton and David Wald of the USGS in Pasadena, John Hall, an associate professor of civil engineering at Caltech, and Marvin Halling, a former Caltech graduate student who is now at Utah State University.

To test the expected behavior of buildings, the seismologists combined a theoretical model of an earthquake with actual seismic recordings to simulate the ground motions at 121 locations spread over 3025 square kilometers (1170 square miles). The engineers then calculated what would happen both to a high-rise and to a base-isolated building, if these structures were located at each of the 121 sites. A base-isolated building is one whose foundation rests on shock-absorbing rubber pads.

Using a mathematical model of a 20-story steel-frame building, the scientists found that the amount of "drift"—the amount of sway in a story of the building as a percentage of its height—exceeded 6 percent in the first story at three of the 121 locations. Drift exceeded 2 percent, which itself is considered severe, at 40 sites over a region of 1,000 square kilometers (390 square miles). An important limitation of the model is that it did not include the possibility of fractures in the welded beam-to-column connections. This means that the drift in a real building, where welds could fracture, would be larger than the calculated drifts cited above, and could be large enough to cause collapse.

The mathematical high-rise also showed that "the ground motion most damaging to a multi-story building is a rapid pulse motion, forward and back, which is timed so that the backward phase strikes the building just as it has acquired a large forward velocity." This rapid pulse creates a damaging whiplike effect as the building suddenly changes direction. Such a pulse develops tension in the exterior columns, raising concern over tensile fracture at the point where two columns are spliced together. A sample analysis in the paper demonstrates that the building could collapse if column splices were to fracture.

Base-isolated buildings are designed to sway a certain distance side-to-side in an earthquake, and have precautionary stops, such as concrete walls, around the base of the building to prevent it from exceeding the design limit and coming completely off its pads. The scientists considered an imaginary three-story building with clearances of 40, 50, and 60 centimeters (16, 20, and 24 inches) around its base, and found that the shaking at 26 of the sites (covering a region of 650 square kilometers, or 250 square miles) would be strong enough to cause the building with 40 centimeters of clearance to hit the stops, with a maximum striking velocity of 165 centimeters per second (about 3.7 miles per hour). Such a high-velocity impact would cause significant damage to the building.

The designs with 50 and 60 centimeters of clearance fared much better of course, but where the peak hypothetical ground motions were expected, even the 60-centimeter design collided with the stops. The scientists conclude that a 40-centimeter clearance, which is greater than that at several existing Southern California base-isolated buildings, is much too small for the expected ground motions near the fault in a magnitude 7.0 earthquake. The three papers taken together carry a cautionary message for the residents of Los Angeles. Scientists will discuss the papers in more detail at 3 p.m. on Thursday, January 12, when they will be available for interviews in the Caltech Media Center.