Scaling Relation of Mega-fault Systems for Strong Ground Motion Prediction

We are studying scaling relations between fault length (L) and seismic moment (Mo) for mega-fault systems such as the 1906 San Francisco earthquake and the 2002 Denali earthquake. Scholz (2002) proposed an idea that the scaling relation changes from L-model to W-model for extra large earthquakes. According to the idea, Ikikura et al. (2004) proposed three-stage scaling relations between fault rupture area (S) and seismic moment; i.e., Mo is in proportion with S^{2/3} for Mo<7.5x1018Nm, S^{1/2} for Mo <7.5x1020, S for extra large earthquakes. Meanwhile, Somerville et al. (1999) proposed an uniform scaling, Mo is in proportion with S^{2/3}, even for larger earthquakes. We consider that the difference of the two scaling relations causes from their treatments of fault width (W). Irikura et al. (2004) introduces saturation of fault width considering thickness of seismogenic zone. However, fault widths are not observed values but estimated from aftershock distributions. We have to carefully evaluate S-Mo relationship considering uncertainty of fault widths. We try to estimate fault width from two empirical relationships, fault length vs. fault displacement (D) and seismic moment Mo vs fault length (L). The relationship L vs D is from certain data based on geological and geomorphological investigations. As for fault displacement, Awata (2006) proposes liner relationship between behavioral fault segment length and its maximum displacement. He also proposed saturation of fault displacement in case of mega-fault system. We here propose and demonstrate new scaling relations between fault length and seismic moment for mega-fault systems compiling past earthquake data (introducing field data about behavioral fault segments). We compare the derived fault width with carefully validated fault width from waveform and geodetic inversions to develop scaling relation for strong ground motion estimation. ACKNOWLEDGEMENT: A portion of this study was supported by the JNES research topic 'Ground Motion Prediction for Mega-fault Systems Based on Dynamic Source Modeling' (P.I.: Kojiro Irikura). REFERENCES: Somerville et al. (1999), SRL. Irikura et al. (2004), 13WCEE. Scholz(2002),Cambridge University Press. Awata (2006), AGU 2006 Fall Meeting.