| ||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 8 | Pages 1345-1347 | 2003
© Oxford University Press 2003
Introduction to the Thematic Issue on Montserrat
Environment Centre, Lancaster University
The 1995–present (February 2003) eruption of the Soufrière Hills volcano, Montserrat, has been monitored and studied in exceptional detail. The volcanic hazard monitoring programme of the Montserrat Volcano Observatory has been complemented by, inter alia, ground deformation studies, earthquake seismology, measurements of dome volume and gas emission measurements using both COSPEC and FTIR. Petrographic examination and geochemical analyses of eruptive products have proceeded continuously from the earliest stages of the eruption. The papers in this thematic issue have drawn on a wide range of modern petrological techniques to provide important new insights into the processes occurring in the magma chamber prior to and during the eruptive activity, in the upper-crustal conduit and in the dome itself.
The petrology of the eruptive rocks is relatively simple. The dominant lithology is andesite in the range 
57–61 wt % SiO2, with the phenocryst assemblage (45–55 wt % modally) plagioclase (An93–48), amphibole, orthopyroxene and titanomagnetite, with minor quartz and ilmenite. This deceptive simplicity masks, however, an extremely complex evolutionary history for these rocks. For example, variations in plagioclase morphology and composition are ascribed in the papers here to combinations of reheating in the storage chamber (as a result of convection related to injection of mafic magma), mixing between basalt and andesite magmas, amalgamation of batches of partially crystallized andesite, the variable effects of textural coarsening, and varying amounts of decompression-induced crystallization resulting from differing ascent and degassing rates. Similarly detailed study of any arc volcano will surely reveal equally complex magma histories.
In the opening paper on the geochemical evolution of the Soufrière Hills volcano, Zellmer et al. note that the most mafic magma recorded in the Soufrière Hills eruption occurs as relatively rare basaltic inclusions in the dominant andesite of the suite. These basalts are clearly not primary (mg-number <60). On the basis of isotope ratios (e.g. high 143Nd/144Nd) and trace element characteristics (e.g. low Nb/Th), Zellmer et al. speculate that the mantle source has been enriched by small (
1·2%) amounts of partial melt of sediments. High U/Th and light rare earth element abundances point to modification of the source by fluids released from the dehydrating slab. The primary magmas underwent closed-system fractionation of amphibole–plagioclase assemblages at deep crustal levels, before being intruded into a chamber, or chambers, filled with andesite. The andesite was formed by fractionation of mafic magma of different composition to the basalt found as mafic inclusions. Zellmer et al. interpret within-sample oxygen isotope heterogeneities to suggest that the andesites represent a mixture of crystals with different growth histories, implying reheating and amalgamation of a number of discrete intrusive bodies of andesite prior to eruption.
Devine et al. compare Ti zoning patterns in titanomagnetite crystals in the andesite with those found in the controlled phase equilibria experiments of Rutherford & Devine and find (1) evidence in some rocks of heating by basaltic magmas throughout the course of the eruption and (2) that heating of some (but not all) batches of andesite occurred just (4 weeks) prior to eruption. They envisage that the basaltic magma ponded at the base of a pre-existing magma storage region (‘chamber’ of other workers) at 5–6 km depth. They also propose that an internal conduit within the storage region carries magma from the heating zone to an overlying conduit, which connects to the surface. Devine et al. ascribe the current activity to triggering by basalt injection and thus to processes acting in the magma storage region of the basalt. Such processes may include increasing buoyancy as a result of increase in melt water content of the fractionating basalt, or injection into the lower storage region (10 km depth?) of more primitive mafic magma.
Higgins & Roberge, using textural analyses and crystal size distributions (CSDs) of plagioclase and amphibole, studied some of the mineralogical effects of the injection of mafic magma into the andesite chamber(s). The occurrence in the andesites of plagioclase as separate crystals and as chadocrysts in amphibole has permitted the first study of volcanic rocks in which the CSDs of plagioclase have been measured at two different stages of development. In their model, nucleation and crystallization of plagioclase followed emplacement of andesite at 
5 km depth. Injection of mafic magma or convective overturn of hotter magma then resulted in a rise in magma temperature, which was eventually buffered by plagioclase solution and then crystallization. Simultaneous dissolution of small crystals and growth of larger crystals resulted in textural coarsening.
Higgins & Roberge appeal to the ‘Communicating Neighbours’ model, where each crystal ‘sees’ only its neighbours during textural coarsening. Thus, each crystal has a unique growth and solution history; complex textural and zoning patterns in plagioclase and amphibole are not necessarily, therefore, the result of mixing of magmas crystallized in different environments. The complexity is increased during repeated heating cycles, as a result of either convective overturn or injection of mafic magma.
Plagioclase phenocrysts in andesites erupted from the Soufrière Hills between 151 ± 4 ka and AD 1999 are the focus of the study by Zellmer et al. on magma emplacement and remobilization timescales beneath Montserrat. Multiple crystal resorption events and significant variations in the anorthite content are ascribed to temperature variations. Using the degree of intracrystalline disequilibrium for Sr and Ba in the plagioclases, they show that the timescale for crystal growth is considerably shorter than the residence time at magmatic temperatures. Bulk crystal residence times at 850°C range from 10 to 1200 years. Crystal residence times in the current andesite range from 15 to 320 years. Zellmer et al. reconcile the complex crystal growth histories, rapid crystal growth and variable crustal residence times by suggesting that small volumes of andesite, periodically intruded into the upper crust, crystallize rapidly by degassing. Later injection of basalt amalgamates the andesites by convection, leading to eruption. Because magma residence times are short compared with the age of the volcano, they suggest that the Soufrière magmas spend most of their crustal residence deep in the crust.
Rutherford & Devine interpret complex compositional zonation in large hornblende phenocrysts in the Soufrière Hills andesites as indicating temperature rises. The heating occurred shortly (30 days) before eruption, judging from the width of Ti-rich rims on titanomagnetite phenocrysts. The occurrence of pargasite and An-rich plagioclase phenocrysts in the andesites is taken to require mingling and hybridization with mafic magma. Some hornblende crystals have rims of Ca-rich pyroxene, consistent with heating above the hornblende stability limit. The rims were reproduced experimentally in andesites held for 2 days at 10–15°C above the stability limit at 130–200 MPa. The ascent rate of magma during the eruptive activity is estimated at >0·02 m/s, on the basis of the general absence of decompression-induced reaction rims on the hornblendes. Variations in ascent rate are ascribed to breaks in the flow of mafic magma into the andesite storage zone.
Using a synthetic melt composition representative of the groundmass of the Soufrière Hills andesite, Couch et al. provide experimental constraints on the formation of highly calcic plagioclase microlites based on phase equilibria at 825–1100°C, 5–225 MPa and fO2 NNO + 1·3 (where NNO is nickel–nickel oxide). The natural plagioclase microlites have size-dependent compositions. Large crystals (>60 µm2) have cores of An75–60; smaller crystals (<60 µm2) have cores of An60–40. Couch et al. explain the high An contents as the result of convective self-mixing within the andesite, the convection being triggered by intrusion of mafic magma. This conclusion contributes to the continuing debate as to whether magma chambers can support vigorous convection.
The smaller microlites and sodic overgrowth rims on some dome samples are thought to result from decompression-induced crystallization, the extent of which varies with ascent history. Rapidly decompressed samples, such as pumices, lack smaller grains and more sodic plagioclase compositions; dense, slowly decompressed dome samples contain both. There is significant groundmass crystallization (up to 70%) during decompression. The latent heat released by crystallization of microlites is estimated to raise magma temperature by up to 45°C.
Couch et al., in their second paper, examine the kinetics of decompression-induced crystallization of the same bulk composition. A series of experiments was held at 875°C and pH2O = 160 MPa for 16 h, and then decompressed to final pressures between 125 and 30 MPa. Judging from comparisons with the equilibrium experiments reported in their first paper, the decompression resulted in undercoolings (
T) of 40–150°C for plagioclase. There were two types of experiment. Single decompression experiments resulted mainly in the nucleation and growth of new plagioclase crystals, whose morphology varied systematically with T. Multiple decompression experiments took place over eight equal instalments and resulted dominantly in growth on existing crystal sites.
Different microlite compositions and types of groundmass textures formed between leaving the magma chamber and eruption, depending on magma ascent rate in an open-system degassing situation. Pumice magmas ascend in <4 h, prohibiting crystallization. Dome lava magmas ascend more slowly, possibly stalling for periods of hours to days and then stagnating at shallow depths for several days; they are thus crystal rich.
Harford et al. take up the theme of decompression-driven crystallization and degassing, using the composition of matrix glasses in the andesites, which, they estimate, are the products of 20–70 wt % crystallization during eruption and variable amounts of degassing. Lava blocks and pumice clasts show major differences in crystallinity and Cl contents. Chlorine contents are related to variations in extrusion rates and in residence locations and times within the lava dome. Residual water contents in pumice clasts (0·2–0·6 wt %) indicate fragmentation processes in the 1997 Vulcanian explosion of 1·1–3·7 MPa, whereas a ballistic block from the sub-Plinian explosive eruption of 17 September 1996 records larger pressure drops (20 MPa). Water contents in glass from lava dome rocks imply pressures 9 MPa.
The low melt Cl contents of many samples cannot be accounted for by models of either open- or closed-system degassing. Taken in conjunction with 
D values for matrix glasses that are isotopically heavier than would be predicted by either model, the data suggest that there has been heterogeneous leaching of Cl by groundwater circulation within the lava dome and upper conduit. HCl emission during the eruption is adequately explained by Cl loss from the melt during magma ascent and crystallization.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||