SEISMOLOGICAL ASPECTS OF THE 1989-1990 ERUPTIONS AT REDOUBT VOLCANO, ALASKA - THE SSAM PERSPECTIVE

SSAM is a simple and inexpensive tool for continuous monitoring of ave rage seismic amplitudes within selected frequency bands in near real-t ime on a PC-based data acquisition system. During the 1989-1990 erupti on sequence at Redoubt Volcano, the potential of SSAM to aid in rapid identification of precursory Long-Period (LP) event swarms was realize d, and since this time SSAM has been incorporated in routine monitorin g efforts of the Alaska Volcano Observatory. In particular, an eruptio n that occurred on April 6 was successfully forecast primarily on the basis of recognizing the precursory LP activity on SSAM. Of twenty-two significant eruptions that occurred between December 14 and April 21, eleven had precursory swarms longer than one hour in duration that co uld be detected on SSAM. For individual swarms, the patterns of relati ve spectral amplitudes are distinct at each station and remain largely stationary through time, thus indicating that one source may have bee n preferentially and repeatedly activated throughout the swarm. Typica lly, a single spectral band dominates the signal at each seismic stati on: for the vigorous one-day swarm that preceded the first eruption on December 14, signals were sharply peaked in the 1.9-2.7 Hz band at th e closest station, located 4 km from the vent, but were dominated by 1 .3-1.9 Hz energy at three more distant stations located 7.5-22 km from the vent. The tendency for the signals from different swarms recorded at the same station to be peaked in the same frequency band suggests that all of the sources are characterized by a predominant length scal e. Signals from the precursory LP swarms became weaker as the eruption sequence progressed, and swarms that occurred in March and April coul d only be detected at seismographs on the volcanic edifice. Onset time s of precursory LP swarms prior to eruptions ranged from a few hours t o about one week, but after the initial vent-clearing phase that ended December 19 these intervals tended to become progressively shorter fo r successive swarms. These trends in the relative onset times and inte nsities of successive precursory LP swarms are consistent with an over all depressurization of the magmatic system through time. In general, each of the swarms had an emergent onset, but the intensities did not always increase steadily until the eruptions. Instead, as the time of an eruption approached the intensity usually increased more rapidly be fore peaking and then declining prior to the eruption; for three of th e swarms, two distinct peaks in intensity were apparent. The time inte rvals between final peaks in swarm intensity and ensuing eruptions ran ged from about 2 hours to almost 2 days, but the peaks always occurred closer to the eruptions than to the swarm onsets. Both the onset of L P swarm activity and a decline in intensity prior to an eruption may r epresent critical points in the process of pressurization that drives the now of fluids and gas in a sealed magmatic system. A notable excep tion to this pattern is the eruption of March 9 which lacked a detecta ble precursory LP swarm, but was followed by an unusually long period of strong LP seismicity that may have been stimulated by a depressuriz ation of the magmatic system resulting from dome failure. On both Dece mber 14 and January 2, the spectra of early syn-eruptive signals have peaked signatures much like those of the spectra of precursory LP acti vity from shortly before the eruptions; these similarities may indicat e that the source of precursory seismicity continued to be active duri ng at least the early part of each eruption. In syn-eruptive signals f rom March and April recorded at stations on the volcanic edifice, the dominant spectral energy progressively shifts with time during the eru ption to lower frequencies; at least part of the energy in these signa ls may have been generated by the debris flows associated with dome fa ilures.