Building a geochemically-constrained time-varying eruption hazard forecasting model for Mt Taranaki

Summary

In this report we present our latest work on the development of a time-varying eruption hazard forecasting model for Mt. Taranaki. The goals of this work are twofold: to develop a probabilistic forecast model that takes into account the underlying magmatic system; and to achieve integration between geochemical-based interpretations and statistical approaches based on high-precision temporal records of eruption events.

The success of this project relied on the development of a most-complete volcanic history record of Mt. Taranaki. This was achieved by using pumice-bearing tephra layers (indicative of fast- ascending magmas which led to the production of high eruption columns of sub-plinian eruptions). Sites along only one dispersal direction were sampled, assuming a long-term even probability of ashfall in this direction. From this record, an annual eruption rate was calculated. This rate revealed that a regular periodicity in eruption frequency has occurred over the Holocene. A hypothesis proposed before this study was that magma recharge events were behind eruption frequency cycles at Mt Taranaki. These recharges are events that move fresh magma from the base of the crust (25 km) to mid crustal (7-10 km) reservoirs. Identified recharge events (evidenced within crystals of plagioclase and titanomagnetite within tephras), however, show no relationship to eruption cycles. Hence this hypothesis was rejected.

This study discovered that another magmatic feature appears to control eruption frequency. Geochemical evidence with titanomagnetite and glass chemistry, as well as whole-rock trace-element chemistry, showed that over the Holocene period, eight distinctive “batches” of magma were erupted. Each of these batches produced magmas of progressively evolving compositions over a period of c. 1500 years. The timing of magmatic batches clearly matches eruption frequency, with highest eruption rates near the mid-point of each batch “life”. With the aid of petrological, geochemical and geophysical data, a model consisting of a lower crustal zone is used to explain the geochemical and subsequent eruption cycles. Each cycle is the result of mantle-sourced intrusions into a lower crustal sill/dyke complex. The intruded melt undergoes 1500 years of assimilation, fractionation and crystallisation processes and residual melt from these cooling bodies periodically rises to feed the upper magma storage region (producing the recharge signature of the phenocryst assemblage). Superimposed on this 1500-yr cyclic trend is that fact that from ~3000 yrs BP Taranaki’s magma composition has included a much greater mafic component, associated with development of the satellite cone of Fanthams Peak.

In addition, the largest known Holocene eruptions from Mt Taranaki occur regularly in the trough of the eruption frequency curve, coinciding with the end of a compositional magma batch. These largest events are uniformly the most evolved (silicic) andesites. These datasets illustrate the relationship between the magma system and eruption frequency, and are therefore important for constraining the timescales and nature of the magmatic processes. Using Hidden Markov statistical models we can now quantify the eruption variability and therefore provide a more realistic probabilistic eruption forecast.