Aundance and solubility of H2O in island arc magmas

by Isoji Miyagi

ABSTRACT

Volcanoes located at subduction zones often erupt explosively. Precise understanding of the explosive eruptions mechanisms is an urgent task in order to make a prediction of volcanic activities. Violent eruptions of subduction zone magmas are caused by extensive degassing of magmatic volatiles, particularly water. The degassing of magmatic water is a consequence of water saturation and dehydration during the ascent of magmas. The important controlling factors for the water saturation and dehydration are, initial water content of the magma, water solubility of silicate melts, and dehydration process of water from hydrous melt to aqueous bubbles. The initial water content of the magmas is the most important parameter which controls subsequent dehydration history. Because of the extensive dehydration in the subaerial eruptions, however, the initial water content of magma is the most difficult variable to know. In the chapter one, analytical method by the SIMS instrument was newly established for water analysis in hydrous silicate glasses. Standard hydrous glasses were synthesized and carefully examined their homogeneities. Matrix effect on the determination of hydrogen content analysis by the SIMS is correlated using their SiO2 content (wt. %). Using the SIMS, water contents of andesitic to rhyolitic glasses can be estimated precisely (ﾅｱ 0.2 wt. % H2O, ﾅ 10ｵm in diameter).

In the chapter two, water solubility in silicate melts (basaltic andesite, andesite, dacite, rhyolite, albite) are newly determined at 2000 bar. The solubility data obtained were significantly lower than those previously reported. I found that the hydrous glass synthesized at high pressure and temperature from excess water starting materials, can be significantly hydrated by aqueous bubbles during quenching. This finding leads to conclusion that most of the previously determined water solubilities are significantly overestimated.

In the chapter three, I determined liquidus temperatures of typical rock suites of island arc magmas ranging from basaltic andesite to rhyolite as a function of melt water content at 800 to 1250 ｡C, 2000 bar, 1 wt. % step of melt H2O content up to water saturated conditions. Based on the experimentally determined phase relationships, an empirical geohygrometer was constructed (LSG diagram: Liquidus temperature - SiO2 - Geohygrometer). Using this diagram, water contents of subduction zone magmas can be roughly estimatedfrom given magma temperatures and melt SiO2 compositions. It is implied from the LSG diagram that the most of island arc magmas are nearly water saturated prior to eruption.

In the chapter four, initial water content of magma prior to eruption was determined for the historic activities at Mt. Usu volcano, and for a largest plinian eruption occurred at the Aira caldera in South Kyushu. The initial water content of the Us-b rhyolitic magma and the Aira magma are estimated to be greater than 5.3 and 4.8 wt. %, respectively. Possible vertical difference in water contents are 1 and 2 wt. % at the Usu and the Aira magma chambers, respectively.

In the chapter five, I considered kinetic dehydration process during plinian type eruption. I examined water diffusion profiles in fresh matrix glass of pumices as well as melt inclusions. From melt inclusion study shown in the chapter three, initial condition (water pressure) of the Usu and the Aira eruption has been obtained. Water solubility as a function of pressure, is based on the chapter two. Numerical calculations were conducted for various melt thickness and decompression rate. According to the numerical calculations on water diffusion in dehydrating rhyolitic melt, it is shown that "equilibrium dehydration" occurs when the thickness of the melt is small and the decompression rate is slow. On the other hand, the "non-equilibrium dehydration" often occur at the later stage of the decompression even though the decompression is constant. This is due to rapid decrease in water solubility at the later stage as well as drastically decrease in water diffusivity in rhyolitic melt. It is suggested that the "non-equilibrium dehydration" occures at the most of explosive eruptions.

From the diffusion calcuration in the chapter five, a possibility is suggested that final dehydration level (fragmentation depth) of plinian type eruptions can be estimated by analyzing the inner part of melt channels in phenocrysts. My theory is useful to understand the degassing process in natural magmas when combined with the data on water content of natural melt inclusions and matrix glasses by the SIMS.