Feasibility study for the materials science approach to volcanic eruption prediction
Pennsylvania State University, Altoona
The materials science approach to eruption prediction defines the time of eruption by the relationship $\ddot\Omega$ = $A\dot\Omega\sp\alpha$, where $\Omega$ represents deformation or energy release. The eruption time approximately equals the time of terminal rock failure following accelerating deformation and is gained from the graphical or numerical extrapolation of inverse rate curves. A linear inverse rate extrapolation (implying $\alpha$ = 2) yields mostly conservative predictions, since slightly concave upward inverse rate trends ($\alpha < 2$) are typical. Precursory data (EDM-derived slope distance, tilt, fault movement, seismicity) for twenty dome building events at Mount St. Helens 1980/1986 give mean $\alpha$ values ranging from 1.38 to 1.62. Simulated "forecast" analyses (conducted in hindsight) for three eruptions (1985/1986) show how a comprehensive picture of event predictions can be developed using various techniques, including a graphical approach, nonlinear curve fits, and calculations of eruption windows based on data statistics. The success ratio also reflects a comprehensive and intensive monitoring effort. Hindsight analyses for the 1989/90 eruption of Redoubt volcano, Alaska, were conducted with seismicity in the form of Real-time Seismic Amplitude Measurement (RSAM). Precursory rate changes to the eruptions of January 2 were apparently sufficient in consistency, duration, and intensity for quantitative predictive analyses. Precursory trends for the spring of 1990 succession of events are more problematic to evaluate and illustrate the difficulty of having to rely on one data type only. The materials science theory is tested against experimental data for notched bar constant-load tests on siltstone. Creep fracture is assumed to be related to stress corrosion (Charles' equation). The stress corrosion index n relates crack velocity to stress intensity during subcritical crack growth. A relationship between this index, n, and the materials science parameter $\alpha$ is established: $\alpha$ = (2n-2)/n. This study indicates that the materials science approach is a promising tool which can provide valuable information for volcanic emergency management. As with all such methods, success is not guaranteed for every application. The analyses should never stand alone, but should be incorporated within a comprehensive evaluation scheme that considers all pertinent geologic evidence and the results of other forecasting methodologies.
Minerals Data and Information Rescue in Alaska (MDIRA)