We proposed a new procedure for evaluating the parameters of the asperity models for predicting strong ground motions from inland earthquakes caused by long strike-slip faults. The evaluation formula of averaged dynamic stress drops obtained by Irie et al. (2010) 1) for surface ruptured faults 15 to 100 km long was adopted in our procedure, because the evaluation formula of the averaged static stress drops for circular cracks, commonly adopted in existing procedures, could not be applied to surface ruptured faults or long faults. The dynamic stress drop averaged on the entire fault was estimated to be 34 bars from the data of the rupture areas and the seismic moments of actual earthquakes, and the dynamic stress drop on the asperity was estimated to be 122 bars from the data of the short-period levels and the seismic moments of actual earthquakes. These results led to the conclusions that the asperity areas were 23 % of the entire ruptured areas.

Irie et al.

Short active faults observed on the ground surface have been supposed to indicate longer subsurface faults in the seismogenic layer. However, no concrete procedures have been proposed so far for modeling the subsurface faults for predicting strong ground motions. Hence, we surveyed several important research papers, and proposed a procedure for modeling the subsurface faults. Here, the seismic moment M 0 of 7.5�10 25 dyne・cm was assumed as the minimum size of the source for the case that any short active faults were observed on the surface. The modeling procedure was based on the asperity model, that was adopted by the Headquaters of Earthquake Research Promotion (2005) in Japan. Five main fault parameters of the area of the entire fault S, the averaged stress drop ��, the area of the asperities S asp , the stress drop on the asperities �� asp , and the short-period level A were determined by the two theoretical relationships and the three empirical relationships among the six main fault parameters including the seismic moment M 0 .

Two new procedures for evaluating fault parameters of asperity models for predicting strong ground motions from crustal earthquakes had been proposed: one is for long strike-slip faults by Dan et al. (2011) and the other is for long reverse faults by Dan et al. (2015). The procedures are based on important statistical findings that the average dynamic stress drop is constant, 3.4 MPa on strike-slip faults with the seismic moment larger than 7.5×10 18 N-m and 2.4 MPa on reverse faults with the seismic moment larger than 2×10 19 N-m, and that the dynamic stress drop on the asperities is also constant, 12.2 MPa on the strike-slip faults and 18.7 MPa on the reverse faults. In order to validate the new procedure for long strike-slip faults, we made an asperity model for a strike-slip fault 141 km long by this procedure, predicted ground motions, and showed that the predicted velocity motions, velocity response spectra, PGA's, and PGV's agreed well with the observed ones in the 1999 Kocaeli, Turkey, earthquake (M W 7.6). We also made an asperity model for a reverse fault 279 km long by the new procedure for long reverse faults, and showed that the predicted velocity motions, PGA's, and PGV's agreed well with the observed ones in the 2008 Wenchuan, China, earthquake (M W 7.9).

We evaluated the fault parameters of asperity models for inland earthquakes caused by long strike-slip faults based on the procedure proposed by Dan et al. (2011) 1) , and simulated strong ground motions by these asperity models. The models had a constant dynamic stress drop averaged over the entire fault of 34 bars and that on the asperities of 122 bars. We presented  ve models with the fault length of 25 km, 50 km, 100 km, 200 km, and 400 km, and simulated strong ground motions from the three models with the fault length of 50 km, 100 km, and 400 km. The simulated motions were consistent with the records of the 2000 Tottori-Ken Seibu, Japan, earthquake, and the 2002 Denali, Alaska, earthquake as well as the attenuation model of peak accelerations and peak velocities proposed by Si and Midorikawa (1999) 2). These results veri ed the procedure proposed by Dan et al. (2011) 1) for evaluating fault parameters of asperity models.