Variations of Strong Earthquake Ground Motion In

Accelerograms recorded at six stations in the metropolitan Los Angeles area during the Borrego Mountain, 1968, the Lytle Creek, 1970, and the San Fernando, 1971, earthquakes in southern California have been studied. In comparing the ground motions recorded during different earthquakes at each of the six stations and in correlations of these motions recorded at different stations during the same earthquake, those aspects of the analysis which emerge from this study and are relevant for seismic zoning have been emphasized. It has been found that the patterns of strong ground shaking in this area depend predominantly on the mechanism and the distance of an earthquake source from a recording station and that the local soil conditions played only a minor role in modifying the ground motion at this particular area. It has been shown that the gross spectral characteristics of ground motion recorded at various stations can be approximately related by the seismic moment at the low-frequency end and by the stress drop at the high-frequency end. I N T R O D U C T I O N Numerous reports on the distribution of earthquake damage indicate that it can vary widely from one point to another and that the areas of intense damage may be highly localized (e.g., Jennings, 1971; Sozen et al., 1968; Hudson, 1972; Gutenberg, 1957; Richter, 1958). In addition to real differences of ground shaking a few significant factors contributing to such a distribution of damage are the overall quality of construction in the damaged area and the poor soil conditions leading to foundation failures, surface faulting, water inundation, and landsliding (e.g., Lee and Monge, 1966; Waseda University Publication, 1966; Alaska Earthquake, 1964). The fact that the distribution of damage seems to follow some specific pattern has led many investigators to advance theories in an attempt to explain the observed patterns of damage in terms of the local amplification of ground motion, which are assumed to be related to the subsoil conditions. The simple and most frequently used theory based on vertically ascending S waves through a set of parallel horizontal layers has been developed by Kanai and his co-workers in Japan (e.g., Kanai, 1957; Kanai and Tanaka, 1961; Kanai et al., 1953). This approach has been further developed and extended in the United States (e.g., Duke, 1958; Idriss and Seed, 1967; Schnabel et al., 1972) during the last 15 years. Although in some special cases such a simple theory can explain the gross effects of local subsoil conditions, especially at sites having exceptionally high impedance ratios (for example, in the metropolitan areas of Tokyo and Mexico City), no one has so far demonstrated that it can portray the gross features of the observed ground motions in areas characterized by less pronounced material contrasts. Nevertheless, during the last decade or so this method has gained considerable popularity, presumably for the following two reasons: (1) because of the relative ease with which local subsoil conditions 1429 1430 M . D . TRIFUNAC AND F. E. UDWADIA can be obtained through seismic prospecting, bore holes, and trenching and (2) because the modeling of ground motion in parallel isotropic and homogeneous layers requires only a moderate theoretical and computational effort (e.g., Kanai et al., 1953; Herrera and Rosenblueth, 1965; Tsai, 1969; Sakurai and Takahashi, 1971). It is well known that the spectral characteristics of strong earthquake ground motion are strongly dependent on the earthquake source mechanism (e.g., Brune, 1970; Aki, 1957; Aki, 1968 ; Trifunac, 1972a; Trifunac, 1972b), the earthquake radiation pattern (e.g., Suzuki, 1932; Udwadia, 1972; Haskell, 1969), the nature and the configuration of different geological discontinuities along the wave propagation path (e.g., Haskell, 1962; Aki and Larner, 1970; Gupta, 1966; Sato, 1963; Hudson, 1962; Abubakar, 1962; Tsai, 1969; Trifunac, 1971a), and the surface topography (e.g., Boore, 1972; Trifunac, 1973a) in addition to possible local subsoil effects. Nevertheless, there is a significant effort, currently under way at many institutions engaged in earthquake engineering research, aimed at introducing the local soil effects as an important factor into the characterization of the amplitudes and the frequency content of strong ground shaking. Since it is clearly of great importance to predict whether particular regions may experience high levels of heavy shaking and whether those can be correlated with the local subsoil conditions, it is timely to reexamine this question in the light of the most recently acquired strong-motion data for the Los Angeles metropolitan area. A similar study was carried out for the E1 Centro strong-motion accelerograph site in Imperial Valley, California (Udwadia and Trifunac, 1973) by comparing the spectra of 15 strong-motion earthquakes, all recorded at the same station. For the E1 Centro site we found that the local subsoil conditions play an insignificant role in determining the overall spectral trends of the strong earthquake ground motion and that the amplitudes and the shape of the spectra strongly depend on the source mechanism and the distance from the earthquake source. In most studies of damage distribution patterns, when the explanation is sought in terms of assumed predominant effects of local subsoil conditions, a correlation of the spatial distribution of some index of damage with a salient feature (typically the depth of alluvium) of the proposed subsoil model is made. There is, however, no basis for the assumption that the pattern of damage will be the same when the next earthquake occurs. Thus, it is necessary to examine several recordings of strong ground motion at the same point, caused by different earthquakes, before the possible dominant effects of the local subsoil conditions can be established. The purpose of this paper is to present such a comparison of strong earthquake ground motion recorded at six selected accelerograph stations in the metropolitan Los Angeles area during three moderate earthquakes in Southern California. The analysis in this paper focuses on the question of whether the local subsoil conditions can be detected from repeated measurements of strong shaking. Having shown that there are no prevalent "site periods" at the stations selected, we proceed to examine the coherence of ground displacements in the Los Angeles and Hollywood areas by comparing the radial, transverse, and vertical components of ground motion at four stations during the same earthquake. This analysis shows that seismic waves whose frequencies are less than one to several Hertz propagate from Hollywood to Los Angeles and vice versa with only a minor distortion. Although the three earthquake recordings studied at the six stations are certainly not adequate to reach any definite conclusions, our results indicate, so far, a lack of strong local site periods, in agreement with several previous investigators (Hudson, 1972; Byerly, 1947; Wiggins, 1964; Udwadia and Trifunac, 1973). Such findings raise the question of the usefulness of seismic zoning and the preparation of seismic risk maps for VARIATIONS OF STRONG EARTHQUAKE GROUND MOTION IN LOS ANGELES AREA 1431 the Los Angeles area based on local amplifications related to subsoil conditions alone and reflected in predominant site periods. STATION AND EARTHQUAKE CHARACTERISTICS AND THE STRONG-MOTION DATA USED The present work is restricted by the availability of usable records from all three earthquakes at the selected stations. Since Los Angeles is one of the best instrumented metropolitan areas of the world (currently instrumented with about 600 accelerographs), it is worthwhile to point out that it will be necessary to wait for many years before a