CONTACT: Stanford University News Service (415) 723-2558
Earthquake probabilities: a never-ending quest for better numbers
STANFORD -- People who live in earthquake-prone areas such as California have become adjusted to hearing predictions like those Anne Kiremidjian is to present Monday, August 26:
If you're under 45, there's a 2 in 5 chance of a major earthquake along the San Francisco Peninsula section of the San Andreas fault during your lifetime, says the civil engineering professor. If you're under 25, the chances rise to 2 in 3.
However, the chances are only about 1 percent for a major quake in the next three decades on the South Santa Cruz section of the fault, where a major quake occurred in October of 1989.
The probabilities for the Peninsula segment of the fault are slightly higher than those reported by the United States Geological Survey in 1988, but that is not surprising to earthquake probability scientists, Kiremidjian said, because "we are constantly refining our methods to take into account new data and to more accurately simulate the physical process of how earthquakes occur."
"The differences in the forecasts of the two groups are small, and are well within expected margins of error," she added.
Both Kiremidjian's and the Survey's studies relied upon the same basic model, but they differed in the statistical methods by which they combined the limited data available and in the interarrival time distribution -- the average time between earthquake events -- that they used. Consequently, the earlier study may result in risk forecasts that decrease over time.
"That contradicts the basic premise of the model, which is that after a major earthquake event the strain is reduced significantly, and it will take a considerable amount of time to accumulate enough energy to produce another large earthquake," Kiremidjian said.
Two years ago, Kiremidjian improved earthquake models developed in the 1970s at Stanford and the Massachusetts Institute of Technology by incorporating the strain accumulation and release mechanism.
In the more recent work, Kiremidjian and graduate students Kimberly Lutz and Hjortur Thrainsson and scientists from the Electric Power Research Institute and the Geological Survey applied the model to five segments of the San Andreas fault considered most likely to produce major earthquakes -- the South Santa Cruz, San Francisco Peninsula, Mojave, San Bernardino and Coachella Valley segments.
They concluded that over the next 30 years, the likelihood of an earthquake of magnitude greater than 6.5 ranged from less than 1 percent in the South Santa Cruz area to 85 percent on the San Bernardino and Coachella Valley zones.
The low probability of an earthquake occurring in Santa Cruz can be explained by the 1989 Loma Prieta quake, which released a great deal of accumulated stress along that segment of the San Andreas fault. Similarly, the higher probabilities for the southern regions reflect the longer time since the last major earthquakes there.
The researchers are continuing to refine the model to consider ruptures that occur along only a portion of a fault segment or that span several segments.
Geologists divide faults into segments with similar quake- causing qualities. In some areas along the San Andreas fault, the Pacific and North American plates slide more or less continuously relative to one another. In other areas movement is halting. Tension builds over time and then is suddenly released. When such releases produce large, abrupt shifts, major earthquakes result.
The research team uses the amount of time since the last earthquake, the average time between events, and the expected movement of the plates each year, called the "slip rate," to forecast future earthquakes for each segment of the fault.
Limited data make earthquake forecasting an imprecise science.
"Written records of quakes cover only a sliver of geological time," Kiremidjian said, "and instruments to measure slip rates have only recently been installed along the San Andreas fault by geologists at the United States Geological Survey."
Geologists can dig trenches across a fault and use clues like the presence of crushed rock to infer past movement. However, trenches are typically only 10 to 15 feet deep and can be of little use in assessing ruptures at greater depths. The 1989 Loma Prieta quake, for example, was triggered by a break that stopped 3.7 miles below the Earth's surface.
Long-term earthquake risk assessment is only the first part of a larger study headed by Kiremidjian at Stanford's John A. Blume Earthquake Engineering Center. After determining the likelihood of quakes along the other segments of the San Andreas, as well as along the Calaveras and Hayward faults, Kiremidjian and her colleagues plan to estimate potential groundshaking for earthquakes of specified magnitude and location.
Then, the team plans to estimate damage to different types of structures, and use this information to calculate dollar losses for various regions in the event of a major quake.
This economic analysis will aid state planners in determining how much money to set aside for earthquake damage repair and other kinds of victim assistance, Kiremidjian said.
This is an archived release.
This release is not available in any other form.
Images mentioned in this release are not available online.