Pressure evolution during explosive caldera-forming eruptions

Abstract

Caldera-forming eruptions of silicic magmas result from a complex coupling of the mechanics and fluid dynamics of the associated magma chamber. Field studies of caldera-forming eruption products suggest that great pressure variations occur inside the magma chamber and associated conduits during these eruptions. Pressure evolution during explosive caldera-forming eruptions is investigated through a simple model that describes the first-order quantitative behaviour of the chamber. We consider a piston-like model that assumes a coherent block subsiding along circular, sub-vertical, ring faults into the magma chamber. This subsidence occurs after significant decompression of the chamber by an initial central vent eruption. We assume that the initial pressure distribution in the chamber is magmastatic. Once collapse has begun the chamber roof is supported by the magma, so that magma pressure at the chamber roof increases to lithostatic. We suggest that pressure variations during caldera-forming eruptions are mainly controlled by variations in magma volatile content. Regardless of what induces the formation of ring faults, the model suggests that the occurrence of explosive caldera-forming events depends on the strength of the chamber walls, and the depth, water content and aspect ratio of the magma chamber. No significant differences exist between model results for a cylindrical, or a more realistic elliptical magma chamber geometry of comparable aspect ratio. Assuming a constant strength of the host rock, the mass fraction of magma that must be erupted during the central vent phase in order to trigger caldera collapse ranges, for deep, gas-poor chambers, from a few percent up to 40% for shallow, gas-rich chambers. The model suggests that zoned chambers tend to collapse earlier than homogeneous chambers. Dike-shaped chambers will erupt less magma than sill-like chambers before caldera collapse initiates, although dike-like geometries are not associated with stress fields appropriate to create ring faults. The model suggests that once initiated, caldera collapse will tend to force out most or all of the volatile-rich magma from the chamber. For volatile-rich magma chambers, the total volume of erupted magma during caldera-forming event is of the same order as the chamber volume. The model also explains the variation in the erupted mass during the different phases of explosive caldera-forming eruptions, and is in good agreement with natural examples.