Bulletin of the Seismological Society of America. Vol. 64, No. I, pp. 149-172. February1974 A THREE-DIMENSIONAL DISLOCATION MODEL FOR THE SAN FERNANDO, CALIFORNIA, EARTHQUAKE OF FEBRUARY 9, 1971 BY M. D. TRIFUNAC ABSTRACT The data from five strong-motion accelerograph stations centered above and surrounding the fault are used to develop an approximate three-dimensiunal dislocation model for the San Fernando earthquake. In the resulting model, the dislocation originates near the instrumentally determined epicenter at a depth of 9.2 km and then propagates southward and upward with a velocity of 2 kin/see. Calculated dislocation amplitudes of about 10 m in the hypucentral region have been found to decay to about 1 m toward the center of the fault and then build up again to about 6 m just before the fault intersects the ground surface in the San Fernando Valley. The assumed fault area of 130 km 2 and the assumed rigidity p = 3 × 1011 dyne/cm 2 give a moment M o = 1.53 × 1026 dyne-cm. This study indicates that, with several strong-motion accelerographs suitably located in the epicentrai region, it is possible to find a kinematic faulting process associated with the periods of ground motion which are longer than about I sec. INTRODUCTION One of the fundamental problems in strong-motion seismology is the determination of the spatial and temporal characteristics of seismic energy release from several measure- ments of ground motion close to the source. This is clearly a difficult inverse problem since, knowing the output of the system, one is faced with the task of finding the input. The description of the black box, which models the effects of the transmission path, is at best only approximately known (Hudson, 1972). Despite these difficulties, however, the inverse problem in strong-motion seismology will remain the focal point of research for many years, since it is only through a thorough analysis involving high-frequency measurements that it will be possible to understand the physical nature of the processes going on at the source of energy release. The purpose of this paper is to explore the extent to which an earthquake source can be modeled by a simple dislocation model (Haskell, 1969) and to determine the frequency (or wavelength) bandwidth up to which an adequate representation by such a simple model is possible. To do this, we chose the San Fernando, California, earthquake of February 9, 1971, since it is one of the most completely recorded and documented earth- quakes to date (e.g., Trifunac and Hudson, 1971 ; Jennings, 1971 ; USGS Profess. Paper 733, 1971). SUMMARY OF FIELD OBSERVATIONS, AFTERSHOCK STUDIES, AND PREVIOUS SOURCE MECHANISM WORK It has already been pointed out (e.g., Trifunac, 1972b) that the geometry of faulting during the San Fernando earthquake was more complicated than that for many other shocks in Southern California. Preliminary studies by Whitcomb (1971) and Wesson et al. (1971) indicated a tentative fault surface striking N64°W-N72°W and dipping about 50° toward the north. The extent of this dislocation was outlined by the aftershocks 149