Broadband displacement and velocity records of P waves recorded at teleseismic distances are analyzed to determine the static and dynamic source parameters of the Chilean earthquake of March 3, 1985 (Ms 7.8) and seven large (mb>5.6) aftershocks. Besides the usual parameters of depth, moment, and focal mechanism, the analysis provides estimates of radiated energy, associated stresses, and source dimension for seven of the eight modeled earthquakes. To insure that the parameters are describing the same physical aspect of the rupture process, the parameters for each shock are computed from the time window in which most of the seismic energy is radiated. The hypocenters of an additional 149 small foreshocks are aftershocks of 4.5≤mb≤5.6 are computed with the method of joint hypocenter determination. The spatial and temporal variation of source parameters and earthquake locations are used to infer a description of the rupture process before, during, and after the main shock. The main shock and modeled aftershocks occurred as thrust faulting on the interface between the Nazca and South American plates. A cluster of foreshocks in the 10 days preceding the main shock also involved thrust faulting on the plate interface.

The main shock itself was a complex rupture consisting of three events. The first two events, denoted as ms1 and ms2, released minor amounts of energy, and they occurred on the periphery of the small zone defined by the foreshocks. The major release of energy occurred with the third event, denoted by MS, which nucleated downdip of ms1 and ms2. Within the time window of major energy release, rupture extended approximately 90 km to the south from the point at which MS nucleated. The size of the early aftershock zone, which far exceeds our inferred dimensions for the major shock, and the strong frequency dependence of scalar moment imply that substantial slow slip occurred on the plate interface that was not associated with major energy release. The dip of the seismically active interface increases landward from about 15¿, at 20 km depth, to about 35¿, at 40 km depth, where the seismogenic interface reaches its maximum observed depth. The vast majority of small and moderate aftershocks occurred in the shallower part of the interface, as did the foreshocks ms1 and ms2. It appears, however, that the asperities that control the rupture of the largest earthquakes are the asperities on the deeper interface. MS and the largest 1985 aftershock, that of April 9 (Ms 7.2) occurred on the deeper interface.

The Chilean earthquake of July 9, 1971 (Ms 7.5), which occurred just north of the 1985 earthquake, also involved main shock rupture on the deeper interface and was followed by aftershocks on the shallower interface. Asperities on the deeper interface appear on average to be stronger, larger, and more uniform in size than asperities on the shallow interface. A greater average strength for deeper asperities is suggested by the observation that stress drops and apparent stresses of the main shock and the aftershock of April 9 (Ms 7.2) are higher than those of the large shallow aftershocks and is consistent with arguments that the maximum friction on a fault zone. should be near the base of the seismogenic zone

One deep interface aftershock with low-stress drop and apparent stress may have been situated below the point of maximum friction on the interface. The inferred larger and more uniform sizes of the deeper asperities are suggested by the higher proportion of large to small shocks on the deeper interface. The downdip change in the asperity size distribution may reflect increased friction on the plate interace or the coalescence of small asperities into large asperities with increased subduction. The radiated energies of the main shock and large aftershocks that are computed directly from the broad bandwidth data are smaller than those predicted by standard empirical formulas by factors ranging from 4 to 20.