Eruption Forecasting Challenges at Large Calderas

Abstract:

Forecasting eruptions at large calderas can be more difficult than at stratovolcanoes due to the complexity and size of these systems. While some eruptions at calderas, such as Ijen and Raung volcanoes in the Ijen caldera and Lokon-Empung volcanoes in the Tondono caldera in Indonesia, do proceed as relatively typical stratovolcano-like eruptions along the caldera rim, others are far more complex. For example, the 2008 eruption of Okmok volcano had < 2 hours of clear precursory seismicity and only subtle changes in long-term inflation rates. Therefore the eruption was not forecast, despite being well monitored. In this case a combination of factors probably led to rapid eruption run-up: the system was metastable after decades of gradual inflation, frequent eruptions left it open to degassing, a small new intrusion into relatively ductile crust remained aseismic until the final stages of dike ascent, and sudden contact with water triggered an explosive phreato-magmatic eruption. Overall our experience in studying unrest and eruptions at calderas suggests a strong need for site-specific conceptual models of the magmatic and hydrological systems and an understanding of local crustal conditions based on multi-disciplinary data to forecast eruptions successfully. Although this is desirable at any volcano, it may be critical at large calderas. Recent high-resolution earthquake studies at the Long Valley Caldera and the Katmai Volcanic Cluster demonstrate that earthquakes associated with unrest at these systems occur repeatedly along pre-existing weak zones under regional tectonic stresses and are triggered by upward diffusion of high-pressure, magmatic/geothermal fluids. In combination with insights derived from geodetic and other monitoring data, this understanding of seismically active fluid pathways would be helpful in interpreting any future pre-eruptive seismicity.