54. Bird, P., and Z. Liu  Seismic hazard inferred from tectonics: California, Seismol. Res. Lett., 78(1), 37-48.
Abstract. We propose two related hypotheses concerning long-term-averages of shallow seismicity: (1) The long-term seismic moment rate of any tectonic fault, or any large volume of anelastically deforming lithosphere, is approximately that computed using the coupled seismogenic thickness of the most comparable type of plate boundary. (2) The long-term earthquake production of any tectonic fault, or any large volume of anelastically deforming lithosphere, is approximately that computed from its moment rate using the frequency-magnitude distribution of the most comparable type of plate boundary. Suggested rules for assigning the most comparable plate boundary type are presented in tables, and the seismicity parameters of these plate boundaries have already been estimated [Bird & Kagan, 2004]. Long-term forecasts based on these hypotheses do not use the local seismic history in a direct way, so they can alert us to regions which may be temporarily below their long-term seismicity rates. However, because of poor resolution of hypocentral and/or centroid depths in teleseismic catalogs, it may be desirable to base the depth-distribution part of the forecast on local catalogs. As an example, we compute the long-term seismicity of the California region based on a long-term-average velocity and strain rate model computed with kinematic finite-element program NeoKinema. This velocity field is a weighted least-squares solution to constraints from plate tectonics, geologic slip rates, geodesy, and stress directions. In the geographic rectangle 31.5~43°N, 113.1~125.4°W surrounding California, our forecast of 64 m>5.663 earthquakes per 25.75 years exceeds the Harvard CMT catalog count of 48 for 1977.01~2002.09. In the slightly smaller RELM test area, our forecast of 235 m>4.95 earthquakes per 21 years exceeds the ANSS catalog count of 165 for 1984.01~2004.12. In the smaller rectangle 32.5~36°N, 115~121°W in southern California, our forecast of 315 m>5 events per 72.67 years is more than twice the TriNet catalog count of 147 for 1932.01~2004.08. We suggest that California, and especially southern California, may be temporarily below long-term seismicity because of the recent lack of great earthquakes (such as those of 1700, 1857, 1872, and 1906) to stimulate aftershocks.
P.S. This paper may have continuing value as a concise statement of the SHIFT (Seismic Hazard Inferred From Tectonics) hypotheses. However, the specific kinematic model of California neotectonics presented in this paper is now out-of-date, as it has been replaced by newer NeoKinema models based on improved and expanded community datasets: see Bird [2009, J. Geophys. Res.]. P. Bird, 2010.04.09
Figure 4. Common logarithm of forecast long-term seismicity (in epicenters per square meter per second, including aftershocks) in the California region for threshold magnitude 5.663, according to the SHIFT model. Seismicity in California and surrounding regions (with shortwavelength structure) is based on kinematics from NeoKinema model GCN2004084, as described in the text. Deep-sea seismicity and southern Arizona seismicity are based on strain rates from Shells model Earth5-013. The spatial integral of the forecast rate is equivalent to 63 earthquakes per 25.75 years in the depth range 0~70 km. (To convert seismicity from earthquakes/m2/s to earthquakes/km2/year, add 13.5 to the values along the scale. To convert to earthquakes/(100 km)2/century, add 19.5.).
Figure 5. Colored background shows long-term forecast, exactly as in Figure 4. For retrospective comparison, the Harvard CMT catalog shows 48 events (“beachballs”) with m > 5.663 at 0~70 km depth in the 25.75-year interval 1977.01~2002.09. Black dots show shallow earthquakes with m > 3 in 1984~2004 from the ANSS catalog.
Appendix: Our forecast in RELM gridded format (if not available directly from RELM site, which is preferred)