CSMIP 00-01

CSMIP 00-01

"Prediction of Ground Motions for Thrust Earthquakes"

by P. Somerville and N. Abrahamson

February 2000, 56 pp.

Click on the link below for the full text:

PDF (Adobe PDF (SIZE 860 KB)) - Requires Adobe Acrobat Reader or other PDF viewer.


The peak accelerations recorded on alluvial sites during the Northridge earthquake were about 50% larger than the median value predicted by current empirical attenuation relations at distances less than about 30 km. This raises the question of whether the ground motions from the Northridge earthquake are anomalous for thrust events, or are representative of ground motions expected in future thrust earthquakes. Since the empirical data base contains few strong motion records close to large thrust earthquakes, it is difficult to assess whether the Northridge ground motions are anomalous based on recorded data alone. We have used a broadband strong motion simulation procedure to help assess whether the ground motions were anomalous. The ground motions from the Northridge earthquake and our simulations of these ground motions have a similar pattern of departure from empirical attenuation relations for thrust earthquakes: the peak accelerations are at about the 84th percentile level for distances within 20 to 30 km, and follow the median level for larger distances. This same pattern of departure from empirical attenuation relations was obtained in our simulations of the peak accelerations of an Elysian Park blind thrust event prior to the occurrence of the Northridge earthquake, and from twenty randomly generated rupture models of future Northridge earthquakes. Since we are able to model this pattern with broadband simulations, and had done so before the Northridge earthquake occurred, this suggests that the Northridge strong motion records are not anomalous, and are representative of ground motions close to thrust faults. Accordingly, it seems appropriate to include these recordings in strong motion data sets that are used to develop empirical ground motion attenuation relations for thrust faults, and to use this augmented data set as the basis for evaluating the need for modifications in design coefficients in the seismic provisions of building codes.

We evaluated systematic differences in ground motion on the hanging wall and foot wall during the Northridge earthquake using empirical data. An empirical model for the hanging wall effect was developed for the Northridge earthquake. This empirical model results in up to a 50% increase in peak horizontal accelerations on the hanging wall over the distance range of 10 to 20 km relative to the median attenuation for the Northridge earthquake. In contrast, the peak accelerations on the foot wall are not significantly different from the median attenuation over this distance range. Recordings from other reverse events show a similar trend of an increase in the peak accelerations on the hanging wall, indicating this systematic difference in hanging wall peak accelerations is likely to be observed in future reverse events. A modification to the near source factor in the proposed 1997 revisions to the Uniform Building Code is proposed to accommodate hanging wall effects near crustal thrust faults in California.