V24B-06
Forecasting the Beginning, Middle, and End of Eruptions

Tuesday, 15 December 2015: 17:30
308 (Moscone South)
John S Pallister1, Wendy A McCausland2, Sarah E Ogburn2, Randall A White3, Heather Michelle Nicholson Wright4 and Volcano Disaster Assistance Program (VDAP), (1)USGS Cascades Volcano Observ, Vancouver, WA, United States, (2)USGS Cascades Volcano Observatory, Vancouver, WA, United States, (3)USGS California Water Science Center Menlo Park, Menlo Park, CA, United States, (4)USGS, Baltimore, MD, United States
Abstract:
Volcanic eruptions are triggered either by “bottom-up” processes such as magmatic intrusion and recharge or by “top-down” processes such as unloading by flank failure. Eruptions end when conduit pressure drops below lithostatic pressure because of depletion of magmatic gases or cessation of magmatic replenishment, or alternatively, as a consequence of plugging of the conduit by crystallization. Examples from the Volcano Disaster Assistance Program (VDAP) show that it is possible to forecast the beginning, changes during, and the end of eruptions using a combination of multi-parametric monitoring, geologic constraints and applicable information from global databases. 

Beginning: Magmatic intrusions can be detected from patterns of precursory seismicity, CO2 emissions, and inflation. The probability that a particular intrusive episode leads to eruption can be estimated from global data modified by the local history of past eruptions and by characteristic progressions in monitoring parameters. 

Middle: Increased probability of a more explosive phase during a long-lived dome-forming eruption may be forecast on the basis of high extrusion rates and recurrence of deep or distal VT earthquakes; both indicate more rapid magma ascent and increased gas pressure. Alternatively, increased seismicity coincident with a rapid decrease in gas emission and extrusion rate may signal conduit plugging, which can also lead to an explosive phase. 

End: The end of long-lived eruptions may be forecast using a combination of: 1) global data on duration of similar eruptions, 2) comparison of eruptive volumes to those of past eruptions, 3) projection of effusion rate trends to zero, 4) reversal of regional deflation to inflation and near-vent inflation to deflation, and 5) change in morphology or composition indicative of more viscous crystalline magma. 

We find that forecasting using the criteria such as described above is best conducted by multidisciplinary teams using probabilistic event trees.