Starting today, Californians will be able to check the Internet for their daily earthquake forecast.
A map, which can be viewed at pasadena.wr.usgs.gov/step/, shows, in color coding, the probability that a location will be hit by severe shaking over the next 24 hours.
Large earthquakes are notoriously difficult to predict. They don't occur often enough for scientists to gather data and there is an inherently random aspect as to how quakes occur. The new model primarily forecasts the additional ground shaking that follows an earthquake.
To build their model, the scientists combined long-established earthquake probability models with information about California's faults, such as the San Andreas.
"The breakthrough is getting something like this to work in an automated sense, in real time, and to have it out there running on its own," said Matt Gerstenberger, a geophysicist with the U.S. Geological Survey in Pasadena, California.
Gerstenberger led the research, which will be published tomorrow in the academic journal Nature.
Background Check
Earthquake forecasts are usually very vague. The simplest model assumes randomness—it says the probability of an earthquake of a particular size occurring in any given location is always the same.
One model, known as the Gutenberg-Richter law, states that small earthquakes occur much more often than large ones. Another model, known as the Omori law, shows that big earthquakes are followed by smaller aftershocks, which then decrease with time.
"Everything that we do is based on these two ideas that are well established," Gerstenberger said.
He and his colleagues used so-called background hazard maps that show California's earthquake risk over 30 years. These maps are based on the behavior of the region's fault lines. The scientists then translated those 30-year probability levels into 24-hour probability levels.
The next step involved factoring in the anticipated add-on effects of any medium or large earthquakes that have occurred in the region in the past few years.
|
SOURCES AND RELATED WEB SITES
|
