Forecasting the failure of heterogeneous magmas

Abstract:

Eruption prediction is a long-sought-after goal of volcanology. Yet applying existing techniques retrospectively (hindcasting), we fail to predict events more often than we success. As much of the seismicity associated with intermediate to silicic volcanic eruptions comes from the brittle response of the ascending magma itself, we clearly require a good understanding of the parameters that control the ability to forecast magma failure itself. Here, we present suites of controlled experiments at magmatic temperatures using a range of synthetic magmas to investigate the control of microstructures on the efficacy of forecast models for material failure. We find that the failure of magmas with very little microstructural heterogeneity – such as melts – is very challenging to predict; whereas, the failure of very heterogeneous magmas is always well-predicted. To shed further light on this issue, we provide a scaling law based on the relationship between the microstructural heterogeneity in a magma and the error in the prediction of its failure time. We propose this method be used to elucidate the variable success rate of predicting volcanic predictions. We discuss this scaling in the context of the birth, life and death of structural heterogeneity during magma ascent with specific emphasis on obsidian-forming eruptions such as Chaitèn, 2008. During such eruptions, the repetitive creation and destruction of fractures filled with granular magma, which are thought to be the in situ remnants of seismogenic fracturing itself, are expressions of the life-cycle of heterogeneity in an otherwise coherent, melt-rich magma. We conclude that the next generation of failure forecast tools available to monitoring teams should incorporate some acknowledgment of the magma microstructure and not be solely based on the geophysical signals prior to eruption.