osedly "glitched" is not installed throughout the US and the only false read stations were in California.

As far as possible 'funny business' goes, it's interesting how two USGS maps for the same location, same time frame can post such a huge difference in number of earthquake events. One map shows 116 more earthquakes than the other.

MENLO PARK -- A deep and powerful earthquake Friday beneath the Pacific in the Northern Mariana Islands sent a seismic wave coursing more than 5,200 miles through the Earth, triggering an automated quake-detection system in Menlo Park that swiftly reported six smaller temblors in California.

David Oppenheimer, a seismologist and project chief of the U.S. Geological Survey's Northern California Seismic Network, said all 430 seismometers in the region caught the temblor's first seismic wave 12 minutes after the Pacific quake struck at 5:39 a.m.

The network's highly sophisticated central computer software instantly reported that six local quakes had struck at California locations ranging from Emeryville to Yosemite and with magnitudes ranging from 3.8 to 4.2.

Almost immediately - when software analysis indicated that two quakes with the same magnitude had apparently hit at virtually the exact same location within five seconds - the computer caught the problem and the local earthquake reports were instantly deleted, Oppenheimer said.

The computer software system, known as an "associator" at the local USGS headquarters in Menlo Park, is designed to assemble all the seismometer signals generated from quakes that strike throughout the San Andreas Fault zone and then combine them swiftly into a single analysis that determines a quake's precise location, magnitude and depth.

The midocean quake in the Marianas struck 212 miles south-southeast of Iwo Jima and 509 miles north-northeast of Saipan. It registered a magnitude of 7.4, and at 162 miles beneath the ocean floor, it was unusually deep within Earth's crust, Oppenheimer said.

And although the high-speed primary wave of seismic energy that the quake generated was greatly attenuated as it raced through the Earth from its distant epicenter to California, the wave was still strong enough to register on the network's seismometers, he said.

Then, as each instrument in the network sent its seismic signal to the associator's computer, the software speedily linked the signals, analyzed them, and indicated the variety of quake epicenters and magnitudes in California as if they were real.

"The magnitudes were legitimate, and the system performed exactly as it was supposed to do," Oppenheimer said, "except that it was designed for the San Andreas and not for such an unusual event as the deep quake in the Marianas."

The software at the heart of the "associator" is extremely complex, he said, and will now require adjustments to do its job without responding to powerful but irrelevant signals from far away.

E-mail David Perlman at dperlman@sfchronicle.com.

This article appeared on page B - 3 of the San Francisco Chronicle

http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2007/09/29/BA30SGARH.DTL