HR: 1340h
AN: S43B-1078    [Abstracts]
TI: Magnitude - Rupture Area Scaling of Large Strike-Slip Earthquakes
AU: * Somerville, P G
EM: Paul_Somerville@urscorp.com
AF: URS Corporation, 566 El Dorado Street - 2nd Floor, Pasadena, CA 91101 United States
AU: Graves, R W
EM: Robert_Graves@urscorp.com
AF: URS Corporation, 566 El Dorado Street - 2nd Floor, Pasadena, CA 91101 United States
AB: There is unresolved debate, reviewed most recently by the Working Group on California Earthquake Probabilities (WGCEP, 2002), about the way in which the source parameters of large crustal earthquakes, especially strike-slip earthquakes, scale with earthquake magnitude (M_w). WGCEP (2002) considered both a self-similar model where average fault displacement D, fault length L and fault width W all increase uniformly together (M_w ~ log Area), and a L model where the displacement D grows in proportion to the fault length L with increasing magnitude once W reaches its maximum value at the base of the seismogenic zone (M_w ~ 4/3 log Area). Like many other scaling relations, the WGCEP models are largely based on data in which the earthquake rupture dimensions, and in some cases, the seismic moment, are either inferred from indirect data, such as the aftershock zone and surface rupture length, or are obtained from very preliminary analyses of the events. This may lead to a systematic bias to underestimate the fault rupture area. In this study, we have used recent crustal earthquakes whose rupture models have been derived from strong motion, teleseismic, and geodetic data, providing detailed images of the distribution of slip on the fault. Since these rupture models are derived directly from the seismic radiation from the fault, they are ideally suited for scaling relations that are used to characterize rupture models for the prediction of strong ground motions from future earthquakes. Our analysis includes rupture models of several recent large strike-slip earthquakes, including the M_w 7.9 Denali, Alaska earthquake of 2002, the M_w 7.8 Kunlunshan, Tibet earthquake of 2001, and the M_w 7.4 Kocaeli, and M_w 7.2 Duzce, Turkey earthquakes of 1999. We find these data to follow a self-similar model in which M_w = log A + 4.0, consistent with Wells and Coppersmith (1994). This relationship estimates a seismic moment one-half that of the preferred relation of WGCEP (2002), and a magnitude M_w 0.2 units lower. For empirical ground motion models, this 0.2 unit magnitude difference has minor significance, resulting in about 15% variability in ground motion level at long periods. This is because the empirical ground motion models implicitly account for the change in rupture area associated with the magnitude difference. However, for numerical waveform simulations where the fault area is fixed, the 0.2 units in magnitude (factor of two in moment) scales almost directly into ground motion level. The factor of two in ground motion variability is about six times greater than that predicted by the empirical relations. This large difference has important implications for seismic hazard analysis.
DE: 7209 Earthquake dynamics (1242)
DE: 7212 Earthquake ground motions and engineering seismology
DE: 7215 Earthquake source observations (1240)
SC: Seismology [S]
MN: Fall Meeting 2005