The latest assessment of earthquake risk for the San Francisco Bay region has not changed the odds much, but a lot has changed behind the numbers. Working Group 2002 (WG02), a committee of about 100 earthquake experts, issued its report for maximum impact during the same week as the 1906 San Francisco earthquake's 97th anniversary. The report is on the Web in two versions, one for a general audience and another for more technical readers with all the supporting detail.
The ten-second soundbite version of this report? Between 2003 and 2032, there's a 62 percent chance of an earthquake of magnitude 6.7 or greater somewhere in the Bay region, between Monterey and Mendocino. There's an 80 percent chance of one between 6.0 and 6.7. The map includes lots more detail. This is not an earthquake prediction, but rather a statement of the odds.
Growing Confidence
It's all the detail that makes this report so exciting. The first time the group went through this exercise, in 1990, many of the Bay area's geologic faults had to be ignored for lack of data about their earthquake histories before 1850, when decent written records began. And even estimating the uncertainties in the numbers felt like a poorly educated guess.
The geoscience community, coordinated by the U.S. Geological Survey (USGS), really dug into the evidence during the 1990s. Trenching studies, like those I described for the Hayward fault, were carried out on many faults to learn more about their past earthquakes. Data for fault movement was improved with a decade of precise GPS-based measurements. Historical research corrected some long-standing misconceptions—most notably, the earthquake of 1836 was placed outside the Bay area instead of on the southern part of the Hayward fault. You can see on the map that most of the probability numbers are new since the 1990 forecast.
Zeroing in on Uncertainty
The scientists also grew much more sophisticated in their use of theoretical tools. One of the unsung strengths of scientists is their approach to uncertainty—they may not know something for sure, but they can tell you exactly how unsure they are. For example, WG02 used four different models of earthquake behavior (five for the San Andreas fault) to estimate probabilities rather than one "favorite" model. They went beyond the uncertainty of the data and allowed for the uncertainty of science itself!
They also added a factor to account for quakes on previously unknown faults—after all, the Loma Prieta event of 1989 was one of those. And new in the WG02 process was a consideration of the stress shadow from the 1906 San Francisco earthquake, which clearly affected the whole region's seismic pattern over the following century.
If you get off on following the thinking behind a complex intellectual project like the WG02 assessment, then the industrial-strength summary report will be a treat. And much related material is also on the WG02 main Web site, so you can delve to your heart's content. That page also links to the USGS's preparedness site, so you can turn this new knowledge into effective action.
