Broadened Modeling Approach
The WG02 report builds on previous analyses of earthquake likelihood, modifying some of the methodologies used in those studies and introducing new ones. The earthquake probabilities presented here are the product of model calculations consisting of three basic elements. The first element is the SFBR earthquake model, which forecasts the average magnitudes and long-term rates of occurrence of earthquakes on the principal faults and for the region as a whole. These average long-term rates of earthquakes lead to average, time-independent probabilities of earthquakes at or above a particular magnitude level of interest (e.g., M>=6.7).
The second element consists of a suite of time-dependent earthquake probability models that incorporate physical aspects of the causes and effects of earthquakes that vary with time. The two most important of these are the progression of faults through the "earthquake cycle" and the interactions of faults, through which the stress released by an earthquake on one fault is transferred in part to other faults or adjacent fault segments. The most significant interaction effect - that produced by the 1906 earthquake - figures prominently in the modeling. There is no consensus within the earth science community, or within this Working Group, as to whether the SFBR remains within the 1906 stress shadow (as suggested by seismicity data for the past 96 years), is now emerging from it (as suggested by the occurrence of the Loma Prieta earthquake and by calculations based on models of viscous flow in the lower crust and mantle), or has emerged from it (as suggested by simple elastic fault interaction models). The addition of a suite of probability models to represent this range of thinking represents the most substantial difference between the analysis reported by WG99 and that reported here.
The third new element introduced in our calculations is the characterization of the rate of occurrence of "background" earthquakes - earthquakes in the Bay region that do not occur on the principal faults. The probability for these events is based on seismicity rates known since 1836, extrapolated to M>=6.7 events. Background earthquakes include events such as the September 2001 M5.1 Napa earthquake, and the 1989 M6.9 Loma Prieta earthquake.
WG02 has devoted considerable effort to defining and quantifying uncertainties in all data, models, and parameters used in the analysis. In the calculations, estimates of uncertainty from all parts of the model are carried through to the end, providing an objective basis for assessing the reliability of the model calculation results and pointing to critical research needed to increase the precision and reliability of future assessments.