Uniform California Earthquake Rupture Forecast, Version 2 (UCERF 2) by E. H. Field, T. E. Dawson, K. R. Felzer, A. D. Frankel, V. Gupta, T. H. Jordan, T. Parsons, M. D. Petersen, R. S. Stein, R. J. Weldon II, and C. J. Wills Abstract The 2007 Working Group on California Earthquake Probabilities (WGCEP , 2007) presents the Uniform California Earthquake Rupture Forecast, Version 2(UCERF 2). This model comprises a time-independent (Poisson-process) earthquake rate model, developed jointly with the National Seismic Hazard Mapping Program and a time-dependent earthquake-probability model, based on recent earthquake rates and stress-renewal statistics conditioned on the date of last event. The models were devel- oped from updated statewide earthquake catalogs and fault deformation databases using a uniform methodology across all regions and implemented in the modular, extensible Open Seismic Hazard Analysis framework. The rate model satisfies integrating mea- sures of deformation across the plate-boundary zone and is consistent with historical seismicity data. An overprediction of earthquake rates found at intermediate magni- tudes (6:5 ≤ M ≤ 7:0) in previous models has been reduced to within the 95% confi- dence bounds of the historical earthquake catalog. A logic tree with 480 branches represents the epistemic uncertainties of the full time-dependent model. The mean UCERF 2 time-dependent probability of one or more M ≥ 6:7 earthquakes in the California region during the next 30 yr is 99.7%; this probability decreases to 46% for M ≥ 7:5 and to 4.5% for M ≥ 8:0 . These probabilities do not include the Cascadia subduction zone, largely north of California, for which the estimated 30 yr, M ≥ 8:0 time-dependent probability is 10%. The M ≥ 6:7 probabilities on major strike-slip faults are consistent with the WGCEP (2003) study in the San Francisco Bay Area and the WGCEP (1995) study in southern California, except for significantly lower estimates along the San Jacinto and Elsinore faults, owing to provisions for larger multi- segment ruptures. Important model limitations are discussed. Introduction California’ s 35 million people live among some of the most active earthquake faults in the United States. Public safety demands credible assessments of the earthquake hazard to maintain appropriate building codes for safe con- struction and earthquake insurance for loss protection. Seismic hazard analysis begins with an earthquake-rupture forecast, a model of probabilities that earthquakes of speci- fied magnitudes, locations, and faulting types will occur during a specified time interval. This paper describes Version 2 of the Uniform California Earthquake Rupture Forecast (UCERF 2; see Table 1 for list of acronyms), which estimates the long-term rate of earthquakes with magnitudes greater than five (M ≥ 5:0) and the conditional time-dependent probability of large earthquakes in California and its bound- ary zones. Figure 1 shows a representative earthquake rate calculation for the study region, annotated with some of the model elements and subregions used in the analysis. Two types of studies have traditionally developed offi- cial earthquake forecast models for California: the National Seismic Hazard Mapping Program (NSHMP), and the Work- ing Group on California Earthquake Probabilities (WGCEP). Through the NSHMP , the U.S. Geological Survey (USGS) has cooperated with the California Geological Survey (CGS) and academia to map seismic hazard, which specifies the likeli- hood that levels of shaking will be exceeded at sites through- out the state (Algermissen and Perkins, 1982; Frankel et al., 1996; Petersen et al., 1996; Frankel et al., 2002). The NSHMP framework is based on long-term estimates using a time-independent (Poisson) probability model for earth- quake ruptures. The resulting hazard maps are used to estab- lish building codes and promote mitigation efforts. Time-dependent rupture models have been the focus of four previous working groups on California earthquake prob- abilities (WGCEP , 1988, 1990, 1995, 2003). In these studies, event probabilities were conditioned on the dates of previous earthquakes using stress-renewal models in which probabil- ities drop immediately after a large earthquake releases tec- tonic stress on a fault and rise as the stress reaccumulates. Such models are motivated by the elastic-rebound theory of the earthquake cycle (Reid, 1911; National Research 2053 Bulletin of the Seismological Society of America, Vol. 99, No. 4, pp. 2053–2107, August 2009, doi: 10.1785/0120080049