Scientists from the University of ÃÛÌÇÖ±²¥ at Boulder's Cooperative Institute for Research in Environmental Sciences accurately forecasted two earthquakes greater than magnitude 5.1 while a paper describing the forecasting method was still in press.Ìý
In a paper published in the Feb. 19, 2002 Proceedings of the National Academy of Sciences, researchers at the ÃÛÌÇÖ±²¥-Boulder headquartered ÃÛÌÇÖ±²¥, correctly identified two earthquake events since January 2001 greater than magnitude 5.1 in southern California. ÃÛÌÇÖ±²¥ is a joint institute of ÃÛÌÇÖ±²¥-Boulder and the National Atmospheric and Oceanic Administration.
"To date, earthquake forecasting has focused on analyzing the total rate of earthquake activity in a region and defining probabilities for future occurrence, without significant success," said John Rundle, the chief author on the paper.
The new method was developed by Rundle and Kristy Tiampo of ÃÛÌÇÖ±²¥, along with colleagues from the Jet Propulsion Laboratory in Pasadena and the Los Alamos National Laboratory in New Mexico. The method involves analyzing changes in activity rates of small earthquakes through time to compute the probabilities that areas are likely to be subject to future large earthquakes.Ìý
The method is a means of processing data, said Rundle. The major idea of the research is that the seismic activity in a region such as southern California can be characterized by means of what is known as a "state vector." State vectors describe space and time locations of activity patterns, he said.
Rundle and his team examined southern California earthquake patterns of magnitude 6 and larger. Since earthquakes of these magnitudes generally have fault lengths of roughly six miles and longer, the physicists used a corresponding grid of square boxes having side length of about seven miles at southern California latitudes, covering the entire study area.Ìý
The "gridding process" yielded 3,000 boxes, areas for which the scientists established individual earthquake histories from 1932 to 1991. Using these histories of activity, or "time series," they then developed a scale of seismic activity for each 0.1 degree region.Ìý
The scales served as an index adequate to define a state vector that characterizes southern California's seismic activity, said Rundle. From the state vector, Rundle and Tiampo next developed a probability index that showed the seismic potential for the occurrence of large earthquakes.
Using the probability index, they computed and displayed regional anomalies in the state vector corresponding to changes in seismic potential of large events spanning the interval from Jan. 1, 1990 through Dec. 31, 2000 for southern and central California.Ìý
The researchers submitted the article June 1, 2001 to Proceedings of the National Academy of Sciences, correctly identifying three events since January 2001, two of them subsequent to the article's submission.
The forecasting method showed considerable promise when applied retrospectively, said Rundle. Although the method was designed to be sensitive to earthquakes having magnitudes larger than 6, the researchers found that the method also seems sensitive to smaller earthquakes as well, in some cases down to magnitude 5.Ìý
Retrospective studies are questionable due to unavoidable biases. The only means of accurately evaluating a forecast methodology is a "true test," in which scientists make a real forecast using past data, and wait to see if the method actually works in the future, where the data are truly unknown.
Rundle and Tiampo's research takes place in a broader area of science called nonlinear complex systems. Their forecasting method may eventually be useful in forecasting future changes in other complex systems, including the way neurons fire in the brain, or even in anticipating stock market dynamics.
Rundle and Tiampo's real time forecast of large earthquakes anticipated from 2001 to 2010 is available. Rundle, lead author of the study, is also a distinguished visiting scientist at the JPL and is an investigator in the Southern California Earthquake Center based at the University of Southern California.