The determination of source depth of the October 1981 (Ms = 7.2) outer-rise earthquake off the Chilean coast has important implications for the stress state of the subducting plate. We analyze the Chile earthquake using three different waveform inversion methods in order to check the consistency of these methods and to estimate the source depth. The methods we used are:
Honda and Seno's (1989) inversion method in which we incorporate both long-period P waves and surface waves.
Nábělek's (1984) body-wave inversion methods for long-period P and SH waves.
Seno and Honda's (1988) deconvolution method in which we assume vertically distributed point sources and use long-period P waves.
These methods give consistent results which suggest that the centroid depth and depth extent of the Chile earthquake lies between 20 to 45 km, measured from the sea surface. This conclusion is contradictory to those obtained by the previous workers who prefer a shallow (<25 km) source depth. We found that the GDSN long-period data alone cannot constrain the source depth well. The shape of deconvolved source-time function is found to be strongly dependent on the assumption of point and distributed sources and the crustal structure used in the deconvolution. The source depth obtained in this study shows that the 1981 Chile outer-rise earthquake possibly occurred because of the compression associated with the bending of the subducting plate.