Re-analysis of the Normal Mode Spectra of the 1960 Chile Earthquake

Abstract:

After the great 1960 Chile earthquake, Press et al. (1961) created a unilateral source rupture model with a length of 1000 km at a speed of 3- 4km/s. Kanamori and Cipar (1974), looked at long-period motion before the main shock in the strain-meter record at ISA and hinted the existence of a slow precursor. Kanamori and Anderson (1975) interpreted that the anomalous decrease of the Fourier spectral amplitude near 1.8 and 3.0mHz of the UCLA gravimeter record and the Press-Ewing seismometer record at PAS are caused by the interference between the main-shock and an un-modeled precursor. Cifuentes and Silver (1989) collected IGY seismographic data and confirmed the existence of the spectral holes at 4 stations all situated in the continental U.S. and modeled the holes with a large precursor or a post-cursor with magnitude comparable to the main shock. CS89 reported no such normal mode spectral holes outside North America. If the spectrum holes are due to destructive interference of a large pre- or post-cursor, the spectrum pattern will be a global feature. Seismograms outside North America are well explained by the unilateral rupture model. KA75 and CS89 synthetic spectra did not consider the spheroidal-toroidal modal coupling effect.

We compared the amplitude spectra computed for earthquake models of a line source and a point source, and for the PREM and a rotating elliptic Earth model with laterally heterogeneity inside. The spectra are computed for the same time series presented in CS89 and directly compared with their amplitude pattern. The observed amplitude spectrum patterns are re-produced at 8 globally distributed stations including the amplitude holes at 4 stations in the U.S by a 920km-long line source (same to the CS89 source) for a rotating elliptic earth model with lateral heterogeneity inside. The synthetic spectrum tests indicate that the long-period precursor or post-cursor proposed by Cifuentes and Silver (1989) and Kanamori and Anderson (1975) is likely unnecessary. This is consistent with absence of the report of a long-period precursor in the strain-meter records at Nana (Peru) and Ogdensburg (New Jersey). Even a possible post or precursor is excluded, magnitude Mw9.6 estimate by seismological methods is still larger than geodetic (Mw9.3) and tsunami (Mw9.2) estimates.