Cyclic Geochemical Variation in Prehistoric and Historic Lavas, Sakura-jima, Japan

Abstract:

Sakura-jima volcano, on Kyushu Island, is one of Japan’s most active volcanic centers. It has been in an eruptive phase since 1955, primarily in the Vulcanian style. Prior to its current activity, Sakura-jima had 11 prehistoric and 4 historic lava-producing events, after its initiation 12-13,000 years ago (Moriwaki, 1992). We present geochemical analyses of a suite of lavas collected from all the mapped historic eruptions and a representative suite of pre-historic eruptions. Lavas vary from andesite to dacite (60.6-66.7 wt.% SiO₂), with typical phenocryst assemblages of 20% plagioclase, 5% pyroxenes, and <5% magnetite. Major and trace element studies of historic eruptions indicate that Sakura-jima lavas have become steadily more mafic since 1475 (Yanagi et al., 1991; Arakawa et al., 1998). Consideration of prehistoric lavas, however, suggests that the variations are cyclic, shifting from more felsic to more mafic compositions at least twice during this period of activity. Previous researchers hypothesized a cumulate plug in the magma chamber that gradually sinks, allowing basaltic magma to ascend into a shallower chamber where it fractionates (Yanagi et al., 1991). Alternatively, Arakawa et al. (1998) propose a model in which magmas from two chambers, one basaltic and one dacitic, are injected into a third, shallower chamber where they mix. To test these models, we performed Sr and Pb isotopic analysis on a subset of prehistoric and historic Sakura-jima lavas. The ⁸⁷Sr/⁸⁶Sr, ²⁰⁶Pb/²⁰⁴Pb, and ²⁰⁷Pb/²⁰⁴Pb ratios decrease steadily from the oldest prehistoric lavas to the youngest historic lavas, in direct contrast to the cyclic variations of the major and trace elements. None of the isotopic ratios correlate strongly with major element compositions. Thus it appears that the simple AFC mechanisms to which the major and trace element variations were originally attributed are not the only process responsible for the cyclic behavior. One of the simplest alternative hypotheses is that early magmas experienced greater extents of assimilation than more recently erupted material, and that the sub-volcanic magma chamber has been refilled steadily by progressively less contaminated, mantle-derived material. The chamber then undergoes cooling and fractionation that is not directly related to the replenishment process.