Summary of Main Results
Note: This report has been superseded by a new probability forecast.
Figure 2. Probabilities (shown in ovals) of one or more major (M>=6.7) earthquakes on faults in the San Francisco Bay Region during the coming 30 years. The likelihood varies along the length of each fault. Color indicates the probability that each fault segment will rupture in such a quake.
Drawing on new data and new methodologies, the Working Group on California Earthquake Probabilities (WG02) has concluded that there is a 0.62 probability (i.e., a 62% probability) of a strong earthquake striking the greater San Francisco Bay Region (SFBR) over the next 30 years (2003-2032). Such an earthquake is most likely to occur on one of seven main fault systems identified in this study, but may also occur on faults that were not characterized as part of the study (i.e., in the "background") (Figure 2). The WG02 results come from a comprehensive analysis led by the USGS and involving input from a broad group of geologists, seismologists, and other earth scientists representing government, academia and the private sector. The results of this study are appropriate for use in estimating seismic hazard in the SFBR, and estimating the intensity of ground shaking expected for specified "scenario" earthquakes. In addition, they provide a basis for calculating earthquake insurance premiums, planning and prioritizing expenditures for seismic upgrades of structures, and developing building codes.
Regional earthquake probability
There is a 0.62 probability (i.e., a 62% probability) of at least one magnitude 6.7 or greater earthquake in the 3-decade interval 2003-2032 within the SFBR. Such earthquakes are most likely to occur on the seven fault systems characterized in the analysis, but may also occur on faults that were not characterized in this study (i.e., in the "background"). This result is consistent with regional 30-year probability estimates made by WG88 (0.5), WG90 (0.67), and WG99 (0.70), given the differences among these studies and their uncertainty ranges.
Distribution of probability
The earthquake likelihood is distributed broadly across the SFBR, from the San Gregorio fault on the west to the Green Valley and Greenville faults on the east (Table 1). The easternmost faults along the rapidly developing Interstate 680 corridor in central and eastern Contra Costa and Alameda Counties have a mean combined probability for M>=6.7 earthquakes of 0.19 [0.16 to 0.22]. Combining this with the contributions from the Hayward-Rodgers Creek fault, the central and southern parts of the Calaveras fault, and half the background earthquake likelihood, the probability for M>=6.7 earthquakes east of San Francisco Bay is 0.46 [0.28 to 0.63]. West of San Francisco Bay, the San Andreas and San Gregorio faults have a mean combined probability for a M>=6.7 earthquake of 0.29 [0.18 to 0.40]. With half of the background probability included, this part of the SFBR has a probability of 0.34 [0.20 to 0.48] for one or more M>=6.7 earthquakes in 2003-2032.
Consistent with previous probability estimates, the Hayward-Rodgers Creek and San Andreas fault systems have the highest probabilities of generating a M>=6.7 earthquake before 2032. The Hayward fault is of particular concern because of the dense urban development along and directly adjacent to it and the major infrastructure lines (water, electricity, gas, transportation) that cross it.
The probability of a sizeable earthquake on a fault not characterized by WG02 (i.e., an earthquake in the "background") is substantial. For events M 6.7, the likelihood is 0.14 [0.07 to 0.37], greater than that on any individual fault system other than the Hayward-Rodgers Creek and San Andreas faults. Many of the significant recent earthquakes in California, including the 1989 Loma Prieta event, have occurred on faults that were not recognized at the time of their occurrence.
Larger earthquakes (M>7.0, M>7.5)
The magnitude of an earthquake is directly related to the size of the fault rupture. Our analysis suggests a 30 year probability of an earthquake M 7.5 or larger striking the region is only 0.l0 (0.02 to 0.20). Only the San Andreas and San Gregorio faults, both lying west of San Francisco Bay, have sufficient length to generate such a large event. When the magnitude threshold is dropped to M 7, the probability is considerably larger, 0.36 (0.17 to 0.60) and is concentrated on faults adjacent to the most developed parts of the region, the San Andreas, Hayward-Rodgers Creek, and San Gregorio fault systems.
Smaller earthquakes (M>6.0)
We estimated the probability of a moderate earthquake (M 6.0 to M 6.7) over the next 30 years to be at least 0.80 (at least four times as likely to happen as not). As the recent past has demonstrated, earthquakes of this magnitude and smaller can produce significant damage over localized areas. For example, the 1984 M 6.2 Morgan Hill earthquake on the southern Calaveras fault caused $10 million damage, while a M 5.1 earthquake that occurred in September 2000 in a rural area 10 miles northwest of Napa caused $70 million damage to that community.
Probability estimates for the next 30-year interval depend critically on the degree to which the SFBR has emerged from the seismic quiescence that followed the great 1906 San Francisco earthquake. The quiescence is thought to be caused by a region-wide drop in stress produced by that earthquake. Regional seismicity rates from the last few decades of the 20th century (Figure 3) suggest that the SFBR has been emerging from this quiescence, but has not returned to the high rate of earthquakes experienced in the 1800's. Until a better understanding of the evolution of the 1906 "stress shadow" is developed, this fundamental uncertainty will continue to hamper the accuracy of probability estimations in the SFBR.
Reliability of results
Generally speaking, the larger the spatial and temporal scales, the more reliable the results. The earthquake probabilities for the SFBR as a whole, for example, are more reliable than those for any individual fault. Similarly, earthquake probabilities for several decades are more reliable than those for the next year.