JordanJ[1991]Jdeveloped a formalism for detecting slow precursors to earthquakes based on observations of the Earth's free oscillations. He derived an inequality that states that an earthquake must be precursive (non-zero moment release before the high-frequency origin time) if the characteristic duration of the moment-rate function is greater than a simple expression involving the skewness of the moment-rate function and the centroid time shift. Jordan [1991] found no evidence for a precursor to the 1989 Loma Prieta, California earthquake; however he did find evidence for a slow precursor to the April 12, 1983 Peru-Bolivia earthquake and speculated that slow precursors might be related to the silent earthquakes proposed by Beroza and Jordan [1990].
Ihmlé et al. [1993] found evidence in the free oscillation spectra of the 1989 Macquarie Ridge earthquake for a slow precursor with a time scale greater than 100s and a moment magnitude of 7.6. The observation that lead to this conclusion was a large characteristic duration and small values of the time shift spectrum at frequencies less than about 6 milliHz. Ihmlé et al. [1993] also demonstrated that it was possible that a precursor could have been smooth enough not to be detected at higher frequencies. Kedar et al. [1994] studied isolated normal-mode peaks for this earthquake that were specifically selected to be free of strong coupling effects. They found large amplitude anomalies suggesting that the earthquake was about 50% larger than indicated by surface-wave analysis [ Satake and Kanamori, 1990]. Kedar et al. [1994] argued that the phase and possibly the amplitude anomalies could be due to finite source effects, rather than a slow precursory event.