Crystallization of microlites and degassing during magma ascent: Constraints on the fluid mechanical behavior of magma during the Tenjo Eruption on Kozu Island, Japan

Abstract

Bubble and crystal textures provide information with regard to the kinetics of the vesiculation and crystallization processes. They also provide insights into the fluid mechanical behavior of magma in a conduit. We performed textural (bubble and crystal) and compositional analyses of pyroclasts that were obtained from the Tenjo pyroclastic flow, which resulted on account of the eruption in 838 A.D. on Kozu Island, about 200 km south of Tokyo, Japan. Pyroclasts in one flow unit (300∼2,060 kg/m³; average density 1330 kg/m³) can be classified into three types on the basis of vesicle textures. Type I pyroclasts have small isolated spherical bubbles with higher vesicularities (67–77 vol.%) and number density (10.8–11.7 log m⁻³). Type II pyroclasts have vesicularities similar to type I (61–69 vol.%), but most bubbles exhibit evidences of bubble coalescence, and lower number densities than type I (8.9–9.5 log m⁻³). Type III pyroclasts contain highly deformed bubbles with lower vesicularities (16–34 vol.%) and number densities (8.2–9.0 log m⁻³). The microlite volume fraction (DRE converted) also changes consistently across type I, type II, and type III as 0.06, 0.08, and 0.10–0.15, respectively. However, the number density of the microlites remains nearly invariant in all the pyroclast types. These facts indicate that the variation in the microlite volume fraction is controlled not by the number density (i.e., nucleation process), but by the size (i.e., growth process); the growth history of each type of microlite was different. Water content determinations show that the three types of pumices have similar H₂O contents (2.6±0.2 wt%). This fact implies that all three types were quenched at nearly the same depth (35±5 MPa, assuming that the magma was water-saturated) in the conduit. If the crystal sizes are limited only by growth time, a variation in this parameter can be related to the residence time, which is attributed to the flow heterogeneity in the conduit. By assuming a laminar Poiseuille-type flow, these textural observations can be explained by the difference in ascent velocity and shearing motion across the conduit, which in turn results in the differences in growth times of crystals, degrees of deformation, and bubble coalescence. Consequently, for crystals in the inner part of the conduit, the crystal growth time from nucleation to quenching is shorter than that near the conduit wall. The vesicle texture variation of bubbles in types I, II, and III results from the difference in the deformation history, implying that the effect of degassing occurred primarily towards the conduit wall.

Appendix A

Crystal growth induced by cooling at the conduit wall

The crystal growth induced by cooling at the conduit wall may produce varying degrees of crystallization processes. The cooling rate of magma is a function of the distance from the conduit wall, assuming a simple one-dimensional conductive cooling without vertical flow. In such a simple case, the solution for the thermal conduction equation is controlled by a dimensionless quantity κ∼tx ², where κ is the thermal diffusivity, t is time, and x is the distance from the conduit wall (e.g., Carslaw and Jaeger 1959). Here, we consider the relationship between the ascent timescale of magma and the thermal diffusion time scale (Fig. A1); the characteristic time for the ascending magma along the conduit (t _flow) is represented as a function of the vertical distance H and the ascent velocity v:

$$t_{{\rm flow}} \sim \frac{H}{v}.$$

(A.1)

Fig. A1

Photograph representing the schematic images of the relationship between the ascent timescale of magma and the thermal diffusion time scale

In general, the ascent velocity during typical explosive eruptions is expected to be high, at least 1 m/s. When we assume that the ascent velocity v is 1 m/s and the vertical distance H is 1 km along the conduit, the calculated ascent time t _flow is 1.0×10³ s. On the other hand, during this ascent of magma, the propagation distance of conductive cooling l from the conduit wall can be represented as a function of magma ascent time and thermal diffusivity:

$$l_{{\rm th}} \sim \sqrt {\kappa t_{{\rm flow}}}.$$

(A.2)

After obtaining the heat capacity, melt density, and specific heat proposed by Murase and Mcbirney (1973) and Richet and Bottinga (1986), the thermal diffusivity κ is calculated as 5.5×10⁻⁷ m²/s. Thus, the distance of the conductive cooling l is approximately 0.02 m while the magma traverses a vertical distance of 1 km. This implies that the conductive cooling may influence a negligible spatial scale from the conduit wall. Therefore, we believe that crystal growth induced by the cooling at the conduit wall does not play a significant role in the microlite growth process that results in different microlite sizes in various vesicular types.