A magnitude 7.75 earthquake along the San Andreas fault in southern California would shake the Los Angeles basin far more severely than expected, say three scientists, including a seismologist at MIT, in an article published in the December 8 issue of Science.
The study was undertaken to explore the risks of living in large, sediment-filled bowls like the Los Angeles basin, which are some distance from a major fault. Earlier studies have shown that the compliant sediments filling such basins can greatly amplify ground motion, prolonging shaking and increasing its intensity.
The scientists' new evidence comes from a detailed computer simulation, which they describe as the most ambitious ever attempted for a major earthquake-one they give a 27 percent chance of occurring in the next 30 years. The work was supported by the National Science Foundation's Southern California Earthquake Center.
The MIT link to the study is Dr. Joseph R. Matarese, until last month a postdoctoral fellow at the Earth Resources Laboratory (ERL) in the Department of Earth, Atmospheric and Planetary Sciences. Dr. Matarese, who is now a scientist with nCUBE, Inc., the California-based manufacturer of massively parallel supercomputers, contributed his expertise in performing large-scale wave propagation simulations on such machines. Since 1990, nCUBE has had a research and education agreement with MIT, which is leasing an nCUBE machine. Under the agreement, the company provides an on-site analyst-most recently Dr. Matarese-and funds graduate fellowships at the ERL headed by Professor M. Nafi Toksoz.
The other authors of the paper were Ralph Archuleta and Kim Olsen, both of the University of California at Santa Barbara.
The simulation showed that long-period ground motion-the motion created by successive seismic waves arriving at intervals greater than 2.5 seconds-in the Los Angeles area would be 4 to 10 times greater than that of the 1992 magnitude-7.3 Landers earthquake, California's most powerful in 40 years.
The study reported in Science is the first detailed look at the likely shaking in the 8,500-square-mile greater Los Angeles area, which has a population of more than 10 million people. The three-dimensional nature of the problem presented an enormous computing challenge which the scientists overcame by developing special computer techniques to simulate an earthquake originating on the section of the San Andreas fault closest to the basin. These computational techniques allowed the 120-second simulation to be conducted in 23 hours on the MIT nCUBE 2, a machine that is 500 times faster than, and has 500 times the memory of, a typical home computer.
Massively parallel supercomputers are capable of performing numerous calculations simultaneously, as opposed to sequentially. This technology provides much faster performance and reliability than conventional computers in handling large amounts of data. The ERL uses the nCUBE computer to solve large-scale physical modeling and inversion problems relating to nuclear monitoring, petroleum exploration and environmental remediation, in addition to earthquake seismology.
The San Andreas fault was chosen because it has been the source of the area's largest quakes in the past, and is likely to be in the future. The 700-mile San Andreas fault marks the boundary where two tectonic plates slide past each other.
Dr. Matarese and his colleagues postulated an event that began with a rupture six miles deep under Quail Lake, near Tejon Pass. It continued southeastward through the Mojave Desert and San Bernardino Mountains until it reached Mill Creek, east of San Bernardino, 104 miles away. Total ground displacement on opposite sides of the fault was just under 16 feet. The rupture itself lasted 68 seconds, though the ground vibrated for nearly another minute.
The simulation indicated that the area from downtown Los Angeles to Anaheim and Santa Ana would experience strong swaying motion for about 60 seconds. Even so, the authors point out, such long-period motion is generally not associated with damage to structures less than 10 stories high. Nor would it probably pose a serious threat to most family dwellings in the Los Angeles basin, they add. However, the simulation suggested that cities closest to the fault, such as Ontario, Pomona, San Bernardino and Redlands might feel even more intense ground motion.
In spite of the "what-if" nature of their exercise, the scientists emphasized that it was firmly grounded in reality: on December 12, 1812, the same stretch of the San Andreas fault was hit by an earthquake of similar size. Moreover, because large earthquakes along the fault's Mojave and San Bernardino segments recur at similar intervals-about once every 150 years-it was "reasonable," they felt, to consider a scenario that had both segments rupturing together.
They concluded that the simulation showed that a major earthquake on the San Andreas fault would trigger still larger long-period ground motions. Many scientists have tended to believe that Los Angeles Basin earthquakes between magnitude 6 and 7 present a greater threat than similar-size events on the San Andreas.
The authors plan additional studies to address the effects of seismic wave scattering and attenuation by loosely compacted sediments and of shorter-period ground motion.
In a related projected, sponsored by the Idaho National Energy Laboratory, the ERL is collaborating with Professor Eduardo Kausel of MIT's Department of Civil and Environmental Engineering and researchers at the Plasma Fusion Center to study the more localized effects of strong ground motion on building structures.
A version of this article appeared in MIT Tech Talk on January 24, 1996.