The two magmatic end-members crystallized at different depths, requiring the presence of two separate magma reservoirs; shallower SA reservoir and deeper BA reservoir. An experimental study reveals that the SA magma had been stored at a pressure of 125 MPa, corresponding to a depth of 5 km. The textures and forms of phenocrysts from the BA magma indicate rapid crystallization directly related to the 2011 eruptive activity. The wide range of H2O contents of olivine melt inclusions (5.5-1.6 wt. % ) indicates that rapid crystallization was induced by decompression, with olivine crystallization first (¡Ü 250 MPa), followed by plagioclase addition. The limited occurrence of olivine melt inclusions trapped at depths of < 5 km is consistent with the proposed magma system model, because olivine crystallization ceased after magma mixing. Our petrological model is consistent with a geophysical model that explains whole crustal deformation as being due to a single source located 7-8 km northwest of the Shinmoe-dake summit. However, even the shallowest estimated source of this deformation (7.5-6.2 km) is deeper than the SA reservoir, which thus requires a contribution of deeper BA magmas to the observed deformation.
Remobilization of mush-like SA magma occurred in two stages before the early sub-Plinian event. Firstly, precursor mixing with BA magma and associated heating occurred (925-871 ¡ãC; stage-1 of ¡Ý 350 h), followed by final mixing with BA magma (stage-2). MgO profiles of magnetite phenocrysts define timescales of 0.7-15.2 h from this final mixing to eruption. The mixed and heated magmas, and stagnant mush that existed in the SA reservoir in the precursor stage, were finally erupted together. Magnetite phenocrysts in the Feb 18 ash reveal the occurrence of continuous erosion of the stagnant mush during the course of the 2011 eruptive activity.