[S-GD02] Crustal Deformation

[SGD02-03] Geophysically inferred low viscosity zone controls post-eruption viscoelastic deformation of Aira caldera, southern Kyushu, Japan

*Tadashi Yamasaki¹, Freysteinn Sigmundsson², Takeshi Tameguri³, Masato Iguchi³ (1.Geological Survey of Japan, AIST, 2.Nordic Volcanological Center, Institute of Earth Sciences, University of Iceland, 3.DPRI, Kyoto University)

Volcano deformation may be controlled by more complex rheological structure of the crust than a simple rheological assumption having often been made in conventional models. Indeed, geodetic data in and around Aira caldera, southern Kyushu, Japan require a previous uni-viscous (UNV) model to have lower and higher effective viscosities, respectively, earliest and later in post-eruption crustal deformation after the 1914 eruption of Sakurajima volcano. This motivates us to implement a spatial viscosity variation inferred from recent seismic imaging beneath the caldera. In this study, using a 3-D linear Maxwell viscoelastic model, where an elastic layer underlain by a viscosity layer, we examine the influence of a low viscosity zone (LVZ) in the lower layer on ground surface displacement rate in viscoelastic responses to two different deformation source modes: (M1) an instantaneous source deflation at a major eruption and (M2) a subsequent continuous source inflation due to magma recharge. The LVZ model behaviour is quantified by the comparison with a UNV model with an apparent uniform viscosity (η_a) that best mimics the LVZ model displacement. It is found that for a given LVZ structure, η_a in viscoelastic response to the M1 mode is smaller than that to the M2 mode, i.e., the rate-controlling viscosities are the ones in the inner and outer part of LVZ for the M1 and M2 modes, respectively. Such LVZ model behaviour is applicable for the geodetic data in and around Aira caldera, where ground surface displacement is predominantly controlled by the viscoelastic responses to the 1914 eruption and subsequent magma recharge earliest and later in the post-eruption period, respectively. We compare model predictions and the geodetic data, and confirm that a LVZ model explains well the data at any stage after the 1914 eruption. This study emphasises interdisciplinary investigations that integrate geodetic and other geophysical datasets for better understanding of volcanic unrest.