Viscoelastic crustal response to magma supply and discharge in the upper crust: Implications for the uplift of the Aira caldera before and after the 1914 eruption of the Sakurajima volcano

Abstract

Crustal viscoelasticity is expected to influence volcano deformation as the crustal viscosity at volcanoes is lowered from regional values by higher crustal temperatures and the eventual presence of magma. In this study, we examine viscoelastic crustal response to continuous magma supply into, and sudden discharge from, a sill emplaced in the upper crust, using a three-dimensional (3-D) finite element model composed of an elastic layer underlain by a linear Maxwell viscoelastic layer with spatially uniform viscosity. We find that viscoelastic response to sill deflation due to magma discharge is an important driving force for post-eruption surface displacement, influenced both by how much viscoelastic relaxation occurs in response to sill inflation due to magma supply in the pre-eruption period and by the amount of co-eruptive sill deflation. Following an eruption, the surface elevation is always recovered later than the sill volume, because the viscoelastic response to the sill inflation detracts from the uplift. We apply this viscoelastic model to geodetic observations in the Aira caldera, southern Kyushu, Japan, over a time span of more than 100 yr. It is found that a sill emplacement with constant inflation rate can well explain the observed first-order exponential-like surface recovery after the 1914 eruption until ∼1975 if the effective crustal viscosity is ∼5×1018 Pas, the sill emplacement with an equatorial radius of ∼2 km occurs at a depth of ∼11 km, and the sill is deflated by ∼0.4 km3 during the 1914 eruption, ∼4 times less than the eruptive volume. The sill emplacement is also required to have begun ∼50 yr before the 1914 eruption with a constant inflation rate of ∼0.009 km3/yr. In the post-eruption period, the accumulation of magma in the sill during the first ∼50 yr is lower than that estimated by an elastic model, but larger thereafter. It is necessary to introduce temporal variation of the inflation rate for higher-order improved fitting to the geodetic data after ∼1975. Our results suggest that geodetic signals observed before and after magma discharge due to an eruption need to be revisited, considering the effects of crustal viscoelasticity.