Behind schedule. The ruptures of the San Andreas fault near Parkfield (the southern end of the Parkfield segment is seen here) struck with fair regularity until recently.

Bad News for Quake Forecasting

A new study of how a 25-kilometer segment of the San Andreas fault accumulates stress and releases it in moderate earthquakes shows that the next quake in the series is now even longer overdue than thought. The problem seems to lie with the fuzzy picture geophysicists have of the spot kilometers beneath the surface where quakes get started.

Back in the 1980s, the San Andreas segment passing by the tiny central California town of Parkfield looked to be the ideal place to test whether seismologists could forecast earthquakes. The seeming 22-year regularity of moderate Parkfield quakes had prompted a formally endorsed forecast that building stress would trigger the next one in 1988, give or take 5 years. This forecast was based on the assumption that the next Parkfield quake would occur when stress built up to the same level that triggered the 1966 Parkfield quake. The forecasted quake still hasn't struck.

Yet the fault has accumulated far more stress than it took to trigger the 1966 quake, according to geophysicists Jessica Murray and Paul Segall of Stanford University. Murray and Segall used both traditional land surveys and Global Positioning System measurements to gauge how changing stress levels around Parkfield had deformed the ground surface. Their analysis in the 19 September issue of Nature indicates that the next quake should have struck no later than 1987, if the assumption used to forecast a 1988 repeat were correct. Even when several small quakes struck near the starting point of Parkfield quakes in the early 1990s, nothing happened. “The gun was loaded and cocked,” says Segall. “Someone was banging on the trigger, yet the gun didn't go off. It means there's some fundamental things we don't understand” about how earthquakes get started.

Seismologists are the first to agree with Segall. The new accounting of stress at Parkfield “points out we don't understand earthquake recurrence, even at Parkfield, where we have the most information,” says Gregory Beroza of Stanford. That's because "we have such a fuzzy picture” of the spot 8 kilometers beneath the surface where Parkfield quakes begin, he says.

Related sites
Segall's site
Murray's site
The USGS Parkfield experiment