Linking Seismicity at Depth to the Mechanics of a Lava Dome Failure - a Forecasting Approach

Abstract:

Soufriere Hills volcano (SHV), Montserrat has been in a state of ongoing unrest since 1995. Prior to eruptions, an increase in the number of seismic events has been observed. We use the Material Failure Law (MFL) (Voight, 1988) to investigate how an accelerating number of low frequency seismic events are related to the timing of a large scale dome collapse in June 1997. We show that although the forecasted timing of a dome collapse may coincide with the known timing, the accuracy of the application of the MFL to the data is poor. Using a cross correlation technique we show how characterising seismicity into similar waveform "families'' allows us to focus on a single process at depth and improve the reliability of our forecast. A number of families are investigated to assess their relative importance. We show that despite the timing of a forecasted dome collapse ranging between several hours of the known timing of collapse, each of the families produces a better forecast in terms of fit to the seismic acceleration data than when using all low frequency seismicity. In addition, we investigate the stability of such families between major dome collapses (1997 and 2003), assessing their potential for use in real-time forecasting.

Initial application of Grey's Incidence Analysis suggests that a key parameter influencing the potential for a large scale slumping on the dome of SHV is the rate of low frequency seismicity associated with magma movement and dome growth. We undertook numerical modelling of an andesitic dome with a hydrothermally altered layer down to 800m. The geometry of the dome is based on SHV prior to the collapse of 2003. We show that a critical instability is reached once slope angles exceed 25°, corresponding to a summit height of just over 1100m a.s.l.. The geometry of failure is in close agreement with the identified failure plane suggesting that the input mechanical properties are broadly consistent with reality. We are therefore able to compare different failure geometries based on edifice geomorphology and determine a Factor of Safety associated with such scenarios. This modelling would be extremely useful in a holistic forecasting approach within a volcanic environment.

Reference:

Voight, B. (1988). A method for prediction of volcanic eruptions. Nature, 332: 125-130.