California Quakes: 99.7% Sure

At a press conference the morning of 14 April, a government task force put the official 30-year odds of an earthquake somewhere in California of at least 6.7 magnitude—the size of the 1994 Northridge quake—at certainty, or close enough: 99.7 percent. The odds of a magnitude 7.5 event, which approaches the size of the historic shocks of 1857, 1872 and 1906, are 46 percent or an even chance. That's about as big as quakes can get in most of California. But a truly giant magnitude 8 or 9 shock could strike the northwestern corner in the Cascadia subduction zone—odds of that in the next 30 years are put at 10 percent. See more details of the Uniform California Earthquake Rupture Forecast (UCERF) at the UCERF home page, or download a handy fact sheet from the US Geological Survey.

This is news for several reasons. The headline is that Southern California has more than twice the risk of Northern California for that big 7.5 event, 37 percent compared to 15 percent. The two main culprits are (1) the San Andreas fault's southern end, including the central and southern segments, and (2) in the north the Hayward fault plus its northern neighbor the Rodgers Creek fault. The south state was last assessed in 1995, a full generation ago in terms of data and theory, and its odds are not just higher than last time, but also more accurately known. (For barroom arguments over where the two Californias are divided, the seismologists say the line goes from Cambria to Benton, through Fresno.)

A new wrinkle in the UCERF results is that the traditional "time-independent" earthquake odds are compared with a more innovative "time-dependent" model to estimate which faults are "getting close." Time-independent probabilities have to do only with the long-term tectonic motions on a fault without caring when the last quake struck it, or what its neighbor faults are doing. Time-dependent odds consider that recent earthquake history and those "earthquake conversations," in USGS researcher Ross Stein's phrase. The comparison of the two models shows five places with elevated probabilities: the southern San Andreas, the Hayward-Rodgers Creek (and a bit of the Calaveras), the San Jacinto fault (which runs west of the San Andreas south of San Bernardino), the Elsinore fault (east of Los Angeles) and the eastern end of the Garlock fault (in the desert northeast of Baker). Stein said that the recent evidence tying the Hayward fault to its neighbor, the Calaveras fault, came too late to be used in the new forecast.

The headline behind the headlines is that the earthquake modelers have really been getting their act together. Not only is this the first rupture forecast for the whole state, but the model used dovetails with other important initiatives including the RELM prediction tests, the upcoming national seismic risk map and the California Earthquake Authority's effort to update insurance rates this summer. Plans are to keep doing this forecast every five years.