Petrologic constraints on the decompression history of magma prior to Vulcanian explosions at the Soufrière Hills volcano, Montserrat

Abstract

A series of 88 Vulcanian explosions occurred at the Soufrière Hills volcano, Montserrat, between August and October, 1997. Conduit conditions conducive to creating these and other Vulcanian explosions were explored via analysis of eruptive products and one-dimensional numerical modeling of magma ascent through a cylindrical conduit. The number densities and textures of plagioclase microlites were documented for twenty-three samples from the events. The natural samples all show very high number densities of microlites, and > 50% by number of microlites have areas < 20 μm². Pre-explosion conduit conditions and decompression history have been inferred from these data by comparison with experimental decompressions of similar groundmass compositions. Our comparisons suggest quench pressures < 30 MPa (origin depths < 2 km) and multiple rapid decompressions of > 13.75 MPa each during ascent from chamber to surface. Values are consistent with field studies of the same events and statistical analysis of explosion time-series data. The microlite volume number density trend with depth reveals an apparent transition from growth-dominated crystallization to nucleation-dominated crystallization at pressures of ∼ 7 MPa and lower. A concurrent sharp increase in bulk density marks the onset of significant open-system degassing, apparently due to a large increase in system permeability above ∼ 70% vesicularity. This open-system degassing results in a dense plug which eventually seals the conduit and forms conditions favorable to Vulcanian explosions. The corresponding inferred depth of overpressure at 250–700 m, near the base of the dense plug, is consistent with depth to center of pressure estimated from deformation measurements. Here we also illustrate that one-dimensional models representing ascent of a degassing, crystal-rich magma are broadly consistent with conduit profiles constructed via our petrologic analysis. The comparison between models and petrologic data suggests that the dense conduit plug forms as a result of high overpressure and open-system degassing through conduit walls.

Introduction

The current eruption of the Soufrière Hills volcano (SHV) is characterized primarily by effusive dome building with intermittent periods of explosive activity (Young et al., 1998). Between August and October 1997, 88 pyroclastic-flow-producing Vulcanian explosions occurred with a mean repose of just 9.5 h and mean high-flux-phase duration of tens of seconds (Druitt et al., 2002). Up to 10⁶ m³ of andesite magma (dense rock equivalent, DRE) were ejected at speeds up to 170 m s^− 1, and plumes reached as high as 15 km above sea level (Druitt et al., 2002). The average inter-explosion magma flux rate was 3–13 m³ s^− 1 (based on erupted volumes and time between eruptions, Druitt et al., 2002). A statistical study of the time series of explosions in September and October 1997 indicates that each batch of magma experienced 6 to 8 episodic decompressions between leaving the chamber and erupting at the surface, recording 60–87 h of eruptive history (Jaquet et al., 2006).

Several studies have suggested that the transition from effusive to explosive eruptive behavior and variations in explosion scale can be explained by changes in conduit dynamics (Jaupart and Allegre, 1991, Woods and Koyaguchi, 1994, Melnik and Sparks, 2002, Melnik and Sparks, 2005), including changes caused by exsolution of volatiles, degassing of the system via permeability (Taylor et al., 1983, Eichelberger et al., 1986, Jaupart, 1998), groundmass crystallization and corresponding changes in magma rheology (Sparks, 1997, Melnik and Sparks, 1999, Massol and Jaupart, 1999, Cashman and McConnell, 2005, Mastin, 2005). Numerical models have investigated relationships between complex conduit processes and eruptive behavior (Melnik and Sparks, 1999, Papale, 1999, Barmin et al., 2002, Melnik and Sparks, 2005, Mastin, 2005). However, boundary and initial conditions, along with many simplifying assumptions, must be applied in order to solve equations for conduit flow, making empirical constraints and validation imperative. Unfortunately, subsurface data are difficult to obtain directly and parameters need to be inferred indirectly (Voight et al., 1998, Voight et al., 1999). Eruptive products are potential sources of data which can constrain subsurface processes.

The groundmass of pyroclastic material commonly contains tiny crystals, less than 100 μm² in cross-sectional area (microlites) that develop during magma ascent from chamber to quench due to degassing (e.g., Cashman, 1992, Hammer et al., 1999). Crystal textures reflect pre-explosion conduit conditions, decompression rate and style, as well as time between decompression and quench (residence time, T_r), allowing estimation of magma storage and ascent conditions (e.g., Swanson et al., 1989, Cashman, 1992, Geschwind and Rutherford, 1995, Cashman and Blundy, 2000, Hammer et al., 2000, Hammer and Rutherford, 2002, Couch et al., 2003a, Couch et al., 2003b, Cashman and McConnell, 2005). Plagioclase and potassium feldspar are the dominant microlite minerals because the feldspar liquidus is strongly affected by decreasing water content in the melt (Burnham, 1979, Sisson and Grove, 1993, Moore and Carmichael, 1998, Cashman and Blundy, 2000). Plagioclase feldspar is abundant in the SHV andesite (Murphy et al., 1998, Devine et al., 1998).

Two recent experimental studies quantify relationships between groundmass crystal textures and dynamic conduit parameters, where groundmass compositions representative of both the Pinatubo (Hammer and Rutherford, 2002) and SHV systems have been explored (Couch et al., 2003a, Couch et al., 2003b). Two styles of decompression were simulated; the first is decompression to final pressure (P_f) in a single step (SDEs); the second is decompression by several staggered steps of equal decompression magnitude, ΔP (MDEs). These experimental data provide a valuable framework relating feldspar crystal morphology, aspect ratio, crystal size distribution, and volume fraction to decompression (ascent) rate and style (MDE vs. SDE), quench pressure and residence time, within which natural samples can be interpreted.

This study has four parts. First, we characterize explosion samples from the 1997 SHV events by measuring clast density and groundmass plagioclase microlite morphology, volume fraction and size distribution. Second, we compare these data to experimental results (Hammer and Rutherford, 2002, Couch et al., 2003a, Couch et al., 2003b) in order to constrain decompression style and quench pressure. Third, we compare our results to estimates made by field observations of the Vulcanian explosions (Druitt et al., 2002) and by a statistical study of time-series data of the same events (Jaquet et al., 2006). Fourth, we compare findings against independent calculations of a one-dimensional numerical model of magma ascent.