Estimated ground motions in Los Angeles due to MW = 7 earthquake on the Elysian thrust fault

Abstract

Abstract The Elysian thrust fault has been identified as a blind thrust fault (Davis et al., 1989; Hauksson and Jones, 1989) representing a potentially serious seismic hazard to the metropolitan Los Angeles and its neighboring areas. We have simulated time histories, peak ground accelerations and their uncertainties using a semi-empirical method for a M W = 7.0 earthquake on the Elysian thrust fault using a flat-layered crustal structure. The accelerograms from the 4 October 1987 Whittier-Narrows aftershock (10h:50m; M L = 5.3) are used to represent the source functions of each subfault on the fault surface. To account for the velocity variation in the surface sediments, we compared simulated ground motions using two separate shearwave velocities, 0.6 km/sec and 0.9 km/sec, respectively, in the top layer of the crustal model. The use of such a simple crustal model is validated by modeling the accelerograms recorded during the 1 October 1987 Whittier-Narrows mainshock ( M L = 6.0). The duration and the relative frequency content observed on these accelerograms are successfully modeled. At some stations, the simulated peak accelerations agree well with the observed values; however, at some sites, the simulated and observed values differ by a factor of 2 to 3. The variation is attributed to a laterally varying crustal structure that is significant for small earthquakes. Additional site specific study is needed for an improved prediction of the observed peak accelerations. For large magnitude earthquakes, the near-source peak ground acceleration appears to be controlled by the source-receiver geometry relative to the fault, as well as the location of asperities on the fault surface. This is validated by simulating the peak acceleration data of the 1989 Loma Prieta earthquake recorded within 30 km of the source. For our simulations on the Elysian thrust fault, an additional component is added to the uncertainty by analyzing several asperity models. Finally, an analytical representation is given to the simulated average peak horizontal ground accelerations, Y ( R ), which is expressed by ln ( Y ( R ) ) = ( 5.38 ± 0.085 ) + ( − 2.09 ± 0.0268 ) ln ( R + 8.0 ) ± 0.343 in the range of 7.5 ≤ R ≤ 35 km ( R is the closest distance to the seismogenic rupture zone). This functional form predicts a rapid fall-off rate for the simulated ground motions compared with the fall-off rate predicted by the other published empirical attenuation relations. We use the finite-difference method to investigate the effects of irregular structure on ground motions resulting from point sources on the Elysian Park fault. The response computed at several depths beneath the basin suggests that the response is dominated by the initial direct arrivals for sources located interior to the basin. When the receivers are located at one end of the basin, and the seismic sources are located at the opposite edge, the characteristic features on the seismograms are the long durations caused by the trapped energy within the basin. The level of trapped energy decreases as the depth of the source increases. Thus, the fault model becomes important in determining the level of peak ground motions and shaking duration.