Probabilistic Seismic Hazard Analysis for Southern California Coastal Facilities

Abstract

Listen

Translate article

The overall objective of this study was to develop probabilistic seismic hazard estimates for the coastal and offshore area of Ventura, Los Angeles and Orange counties for use as a basis for the University of Southern California (USC) to develop physical models of tsunami for the coastal regions and by the California State Lands Commission (SLC) to develop regulatory standards for seismic loading and liquefaction evaluation of marine oil terminals. The probabilistic seismic hazard analysis (PSHA) was carried out by the Lawrence Livermore National Laboratory (LLNL), in several phases over a time period of two years, following the method developed by LLNL for the estimation of seismic hazards at Department Of Energy (DOE) facilities, and for 69 locations of nuclear plants in the Eastern United States, for the Nuclear Regulatory Commission (NRC). This method consists in making maximum use of all physical data (qualitative, and quantitative) and to characterize the uncertainties by using a set of alternate spatiotemporal models of occurrence of future earthquakes, as described in the SSHAC, PSHA Guidance Document (Budnitz et al., 1997), and implemented for the NRC (Savy et al., 2002). In general, estimation of seismic hazard is based not only on our understanding of themore » regional tectonics and detailed characterization of the faults in the area but also on the analysis methods employed and the types of physical and empirical models that are deemed appropriate for the analysis. To develop this understanding, the body of knowledge in the scientific community is sampled in a series of workshops with a group of experts representative of the entire scientific community, including geologists and seismologists from the United States Geological Survey (USGS), members of the South California Earthquake Center (SCEC), and members of academic institutions (University of California Santa-Cruz, Stanford, UC Santa Barbara, and University of Southern California), and members of consulting firms. The purpose of the workshops was to analyze and evaluate existing data and formulate tectonic models that represent all the possible and physically valid models envisioned by the group. The basic input for the PSHA was a set of alternate earthquake source characterizations and a multi-model representation of ground motion attenuation, for adequate representation of the uncertainties. In the first phase, the physical modeling enabled rigorous analysis of uncertainty that arises from a lack of full knowledge in the characterization of both earthquake sources and ground motion. The set of ground motion prediction models included models that were updated to benefit from near field data from the most recent earthquakes (Taiwan and Turkey). The calculation were performed with LLNL computer software that is based on the Cornell, 1968 analytical model, and that propagates the knowledge uncertainties using a Monte-Carlo simulation approach (see, Bernreuter et al., 1989). Although the calculation were performed for rock-site conditions and generic soil sites, only the results for rock are given here. It is assumed that development of design parameters will include a correction of the spectral shape to reflect the site specificity. The results are for the average of the two horizontal components of the ground motion. The PSHA was calculated for thirteen sites, including two sites offshore. These sites are: Catalina Island site 1, Catalina Island site 2, Goleta, Offshore Santa-Monica, Offshore San-Clemente, Port Dume, Palos Verde site 1, Palos Verde site 2, Port of Long Beach, Port of Los Angeles, Port Hueneme, San Pedro Escarpment, and Redondo Canyon. For these thirteen sites, the hazard curves in terms of probability of exceedence of the peak ground acceleration (PGA), was calculated. In addition for Port of Long Beach, Port of Los Angeles, Santa Monica, and a site Offshore, east of San Clemente the (5% damping) uniform hazard response spectra were calculated for five Return Periods (100, 500, 1000, 2000, 10,000 year Return Periods). The detailed results are given in chapter 7.« less