Statistical analysis of crystal populations and links to volcano deformation for more robust estimates of magma replenishment volumes

Forecasting the timing and size of volcanic eruptions requires a proper interpretation of multiparametric monitoring signals. Studies of the erupted rocks can provide critical information on the processes and volcano plumbing system that is needed to decode the monitoring signals. Here we present the results of a petrological study of plagioclase phenocrysts using a new statistical approach that allows us to estimate the amount of intruded magma before eruption. Our crystal population analysis of the 2006 and 2010 CE Merapi volcano (Indonesia) eruptions shows that ∼60 ± 20 vol% of the 2010 magma was left over from the 2006 magma, and thus ∼40 ± 20 vol% was newly intruded magma. Using the published values of the 2010 erupted magma volume, this corresponds to >8 to 20 (±7) × 106 m3 of new magma. This is a minimum estimate and is similar to the inferred pre-eruptive deformation volume (18 × 106 m3), although given the uncertainties, several million cubic meters of magma intruded in 2010 could still be in the Merapi plumbing system. Our approach could be used at other volcanoes to quantify the volume of intruded magma and thus help in better understanding the unrest signals that anticipate eruptions. INTRODUCTION Many volcanoes erupt magmas of the same composition for decades or centuries, e.g., Mayon in the Philippines (Newhall, 1979), Merapi in Indonesia (Costa et al., 2013), Arenal in Costa Rica (Reagan et al., 1987), or Soufrière Hills in Montserrat (Murphy et al., 2000). What drives these frequent eruptions of the same magma? How much of the magma produced by each eruption is new replenishment, and how much was already stored in the system? Being able to answer these questions is important for understanding how volcanos work, and thus for being able to properly interpret monitoring data and