Linking Long-Lived and Transient Magma Plumbing Systems Beneath Volcanoes Using Dense Magnetotelluric Observations

Complete images of magma plumbing systems are fundamental for understanding volcanic activity. Earthquake hypocenter distributions, their migration, and geodetically inferred pressure sources provide valuable constraints, but these dynamic signals are usually spatially localized and temporally short-lived (days to tens of years). In contrast, petrological and geophysical studies often image large trans-crustal magma plumbing systems beneath volcanoes, inferred to occupy volumes of ~1000 km³ and to develop over the long lifetime of a volcano. This discrepancy highlights a key gap between short-lived, small-volume magma involved in unrest and eruptions (<0.1 km³) and long-lived, large-scale magmatic reservoirs.

To bridge this gap, we integrate recent geophysical observations at active volcanoes in Japan and propose a unified magma plumbing framework linking long-lived and short-lived magmatic processes. We present electrical resistivity structures beneath Kirishima, Sakurajima, and Hakone volcanoes derived from dense broadband magnetotelluric (MT) observations. All three volcanoes have experienced significant crustal deformation, seismicity, and eruptions within the past 15 years.

Beneath each volcano, inclined columnar-shaped conductive bodies with volumes exceeding ~1000 km³ are imaged beneath active craters, extending from depths of a few kilometers to the lower crust. Common features include: (1) tectonic earthquake hypocenters are largely distributed outside the conductive bodies, and (2) geodetically inferred pressure sources and deep low-frequency earthquakes are concentrated along their edges. At Kirishima volcano, the conductor geometry corresponds closely to a low-V_SV region imaged by surface-wave tomography. At Sakurajima volcano a magmatic dike intrusion on 15 August 2015 occurred near the top of the conductor.

We interpret the large conductive bodies as long-lived magmatic reservoirs dominated by crystal mush, within which sill complexes are developed. In contrast, small and transient magma pockets likely form along reservoir margins. We propose an edge-ascent model in which magma and volatiles preferentially migrate along conductor boundaries, feeding normal small eruptions, whereas magma stored in the large reservoirs may only be mobilized during large eruptions.