JOURNAL OF PETROLOGY VOLUME 55 NUMBER 3 PAGES 529^548 2014 doi:10.1093/petrology/egt075 Quantification of the Intrusive Magma Fluxes during Magma Chamber Growth at Soufrie're Hills Volcano (Montserrat, Lesser Antilles) C. ANNEN1*, M. PAULATTO2, R. S. J. SPARKS1, T. A. MINSHULL3 AND E. J. KIDDLEy 1 SCHOOL OF EARTH SCIENCES, UNIVERSITY OF BRISTOL, WILLS MEMORIAL BUILDING, QUEEN’S ROAD, BRISTOL BS8 1RJ, UK 2 DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF OXFORD, SOUTH PARKS ROAD, OXFORD OX1 3AN, UK 3 OCEAN AND EARTH SCIENCE, NATIONAL OCEANOGRAPHY CENTRE SOUTHAMPTON, UNIVERSITY OF SOUTHAMPTON, EUROPEAN WAY, SOUTHAMPTON SO14 3ZH, UK RECEIVED APRIL 25, 2013; ACCEPTED NOVEMBER 15, 2013 ADVANCE ACCESS PUBLICATION DECEMBER 20, 2013 Magma fluxes in the crust control the thermal viability and mechanical stability of magma chambers. We estimated the magma fluxes required to generate the negative seismic velocity anomaly observed below Soufrie' re Hills volcano, Montserrat. Growth of a magma body by accretion of andesitic sills was simulated numerically and the resulting temperatures and melt fractions were used to calculate a synthetic anomaly of seismic wave velocity, which was filtered to be comparable with the velocity anomaly obtained from a tomographic experiment. Petrology indicates that before it was reheated, remobilized and erupted, the temperature of the magma residing in the chamber was about 8508C. We ran simulations where convection is assumed to be low and heat transfer is mostly by conduction and simulations where convection is assumed to be vigorous enough to rapidly cool the magma chamber to 8508C. In both cases, magma chamber growth over the last 350 years results in tomography anomalies that are too strong, unless the magma was emplaced at an unlikely low melt fraction (50·5). Good fits between the modelled and the observed velocity anomaly were obtained with sills 2^5 km in radius emplaced over 6000^150 000 years, depending on the temperature and melt fraction of the emplaced magma. Because of a trade-off between intrusion dimensions and emplacement durations, the volumetric magma fluxes are restricted to 7 104 and 5 103 km3 a1. The velocity anomaly can be reproduced with a chamber containing high melt-fraction magma or with a mush of crystals and melt.The range of magma ages in the modelled magma chamber is much wider than the crystal residence time of the erupted *Corresponding author. Telephone: þ44117 954 54 26. Fax: þ 44 117 925 33 85. E-mail: [email protected] y Previously at School of Earth Sciences, University of Bristol, Wills Memorial Building, Queen’s Road, Bristol BS81RJ, UK andesite.This suggests that the eruption taps small pockets of recently assembled magma and that the velocity anomaly is mostly due to a non-eruptible mush. magma chamber; magma flux; Montserrat; numerical simulation; tomography KEY WORDS: I N T RO D U C T I O N The construction and differentiation of the crust involves the transfer of magma from the mantle to the lower crust and from the lower crust to the upper crust and surface. The flux of magma from depth to form and supply upper crustal magma bodies (which in turn may feed volcanic eruptions) is the main control on the thermal viability of magma chambers (Annen, 2009) and on their mechanical stability (Jellinek & DePaolo, 2003). It is now well accepted that magma bodies grow incrementally by addition of discrete magma batches, which implies that intrusive magma fluxes can vary greatly over time (e.g. de Saint Blanquat et al., 2011). Magma fluxes can be inferred from a variety of evidence. Average long-term fluxes (over tens of thousands to millions of years) are estimated by dividing intrusive volumes by the time span obtained with U-series isotopes (Matzel et al., 2006; Leuthold et al., 2012), whereas ß The Author 2013. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@ oup.com VOLUME 55 1 10 2 10 Current cycle starts Last eruption Volcano-tectonic event Volcano-tectonic event Volcano-tectonic event Current eruption starts Soufrie're Hills is the youngest of the three volcanic edifices of the island of Montserrat, which is part of the Lesser Antilles volcanic island arc. Montserrat is mostly composed of andesites with minor amounts of basalts (Rea, 1974). The oldest rocks on Montserrat are dated at 2·6 Ma (Harford et al., 2002), but volcanism might have started at 9 Ma when the northern part of the Lesser Antilles arc migrated towards the west (Briden et al., 1979). The analysis of submarine deposits indicates that Soufrie're Hills is at least 250 000 years old (Smith et al., 2007; Le Friant et al., 2008). Volcanic activity is characterized by periods of 103 to 104 years of relatively intense activity followed by longer periods of dormancy (Le Friant et al., 2008). Active periods are themselves interrupted by periods of quiescence of 103 years. The last high-activity period lasted from 31ka to 17 ka (Harford et al., 2002). Activity resumed 6000 years ago (Trofimovs et al., 2013) and was marked by the collapse of English’s Crater (Roobol & Smith, 1998). The last eruption was about 350 years ago with Castle Peak dome extruding within English’s Crater (Young et al., 1998). Three volcano-seismic crises in 1897^1898, 1933^1937 and 1966^1967 (MacGregor, 1938; Powell, 1938; Wadge & Isaacs, 1988) have been attributed to magma intrusions in the upper crust. The current eruption started in July 1995 and was preceded by 3 years of increased seismicity (Aspinall et al., 1998; Wadge et al., 2013). Key dates in Soufrie're Hills history are shown in Fig. 1. The presence of amphibole in the crystal assemblage erupted by the Soufrie're Hills magmas implies that the depth of the top of the magma chamber is no less than 5^6 km (Barclay et al., 1998). Earthquake hypocentres are located at depths shallower than 6 km (Aspinall et al., 1998) and global positioning system (GPS) data fit a point-source model (Mogi, 1958) with deformation source at 6 km depth (Mattioli et al., 1998). Modelling of deformation in 2005^2007 indicates that the deformation is related to a deeper magma chamber at 12^14 km depth (Hautmann et al., 2010). Thus petrological and geodetic data suggest that a 6 km deep magma chamber is connected to a deeper reservoir located at about 12 km depth (Mattioli et al., 1998; Elsworth et al., 2008). According to Elsworth et al. (2008), the current eruption responds to flux from below into the deeper chamber and the magma passes through the upper magma chamber without accumulating at this level. Pyroxene (Murphy et al., 2000) and Fe^Ti oxides (Devine et al., 1998) geothermometry, together with experimental studies (Barclay et al., 1998), indicate magma temperatures of 840^8708C. Plagioclase^amphibole assemblages suggest temperature variations within the magma body of up to 1008C. Complex mineral zoning, textures, mineral chemistry variations and stable isotope data imply disequilibrium related to magma mixing, convective 3 10 4 10 Time BP (yrs) 5 10 Montserrat oldest rocks CONTEXT MARCH 2014 Soufrière Hills oldest rocks short-term fluxes (over years to decades) can be estimated using ground deformation measurements (e.g. Sparks et al., 2008; Pearse & Fialko, 2010; Parks et al., 2012). Emplacement of granitoid intrusions is protracted and may last from tens of thousands of years for relatively small laccoliths (Leuthold et al., 2012) to millions of years for batholiths (Coleman et al., 2004; Matzel et al., 2006). Fluxes over hundreds or thousands of years are difficult to determine because this time span is too small to be resolved with radioactive isotopes and too long for most deformation records. However, these fluxes control much of the dynamics of shallow magma chambers and determine their ability to convect, differentiate and erupt (Annen, 2009). In this study we use seismic tomography results and numerical simulation of magma intrusion with additional constraints from petrology and geophysics to quantify the intrusive fluxes beneath Soufrie're Hills Volcano, Montserrat, by testing out different hypotheses for the age and dimensions of the current magma chamber. NUMBER 3 Last cycle ends Last cycle starts JOURNAL OF PETROLOGY 6 10 7 10 Fig. 1. Timeline showing the most relevant dates in Soufrie're Hills volcanic history (Harford et al., 2002; Trofimovs et al., 2013). 530 ANNEN et al. SOUFRIE'RE HILLS MAGMA FLUXES processes and assimilation of older intrusions (Murphy et al., 2000; Harford & Sparks, 2001; Couch et al., 2003; Humphreys et al., 2009). These petrological data have been interpreted as evidence of injection of basalt into a shallow (5^6 km) magma chamber that reheated a cold (830^8608C), highly crystalline (60^65 wt %) andesite and initiated the current eruption (Barclay et al., 1998; Devine et al., 1998, 2003; Murphy et al., 2000). Mafic inclusions in the new andesites (since 1995) are common and represent about 1% of the rock. The proportion of cryptic mixing of mafic magma into the andesite is much higher, at several per cent or more (Humphreys et al., 2009). However, according to Zellmer et al. (2003a) these mafic magmas are not the parental magma of the andesite. The observation that U^Th isotopes are close to secular equilibrium suggests protracted residence and differentiation of andesite magma at a deep level (lasting 4250 000 years), which contrasts with the very short (5320 years) crystal residence time in the shallow magma chamber, as inferred from Sr diffusion profiles in plagioclase (Zellmer et al., 2003b). In these andesites, the crystal population is highly heterogeneous at the microscopic scale (Humphreys et al., 2009), but the whole-rock is chemically homogeneous at the macroscopic scale (Murphy et al., 2000; Couch et al., 2001). This observation has been interpreted as evidence for convective stirring and magma mixing. Oxygen and H isotopes (Harford & Sparks, 2001; Zellmer et al., 2003a) confirm that the erupted lavas contain a mixture of crystals with different histories and origins. In addition, the andesites show evidence for reheating (Devine et al., 1998; Murphy et al., 2000; Couch et al., 2001) and assimilation of older intrusions (Harford & Sparks, 2001). All these observations support a model in which magma pulses of different origins were amalgamated and mixed within the crust before erupting. Sparks & Young (2002) suggested that cycles in seismic and geodetic signals during the current eruption could be due to magma chamber overturns. M AG M A C H A M B E R MO D E L S In general, the geometry of magma chambers is not well constrained. Many physical models rely on the assumption that magma chambers are spherical or ellipsoidal. However, models of magma chambers based on petrology and geophysics range from sill-like or dyke-like chambers to a plexus of interconnected sills and dykes, for mafic and andesitic systems, and to large tabular magma chambers for large dacitic and rhyolitic eruptions [for a summary, see Zellmer & Annen (2008)]. There has been less attention on how magma chambers nucleate and grow. A magma chamber cannot start as a small sphere and grow by inflation, because for small volumes the pressure on the walls of a spherical magma chamber is much higher than the strength of crustal rocks and would result in dyke opening and draining of the chamber (Jellinek & DePaolo, 2003). Sheet intrusions are commonly observed in the field (Menand, 2011, and references therein) and it is most likely that a magma chamber starts as a sheet intrusion (Gudmundsson, 2012), with the magma pressure resulting in magma propagation at the tips. Recent studies of medium to large intrusions, several hundreds of metres to several kilometres thick, indicate that they formed by stacking of smaller sheet intrusions (Searle, 1999; Searle & Godin, 2003; de Saint-Blanquat et al., 2006; Pasquare' & Tibaldi, 2007; Michel et al., 2008; Galland, 2012). Field observations, analogue experiments and stress analysis all support a model in which magma is transported from depth through dykes, which change their orientation to form sills, either because they encounter a rheological barrier, or because of a change in the stress state of the crust (Vigneresse et al., 1999; Kavanagh et al., 2006; Menand, 2008, 2011; Gudmundsson, 2011; Maccaferri et al., 2011; Miller et al., 2011). The addition of these sheets builds up an igneous body that can result in a magma chamber if the sheet addition rate is large enough (Annen et al., 2008; Annen, 2009). Heat transfer models show that a magma chamber that grows by accretion of sheet intrusions, if it becomes thick enough, tends towards an ellipsoidal or even spherical shape. Incremental growth results in large fluctuations of temperatures and can explain the widespread evidence of magma reheating and remelting before eruptions at Soufrie're Hills and elsewhere (Devine et al., 1998; Murphy et al., 2000; Bachmann et al., 2002; Druitt et al., 2012). Alternatively, such reheating is commonly interpreted as due to mafic injection into the magma chamber (Devine et al., 1998; Murphy et al., 2000; Bachmann et al., 2002). Incremental growth can also explain the contrast between small-scale heterogeneities and large-scale homogeneity if heterogeneous magma pulses are eventually homogenized by convection before erupting (Davaille & Jaupart, 1993; Jellinek & Kerr, 1999; Couch et al., 2001). Paulatto et al. (2012) modelled the growth of the shallow Soufrie're Hills magma chamber as resulting from the vertical stacking of sills and identified a model for which the calculated seismic velocity anomaly corresponding to the modelled temperatures and melt fractions, smoothed by an appropriate filter, matched well the velocity anomaly determined from seismic tomography. Importantly, they showed that, although the inversion of tomographic data suggested low melt fractions (7^10%), direct modelling indicated that the low-velocity anomaly could also be due to higher melt fractions (430%), representative of mobile, eruptible magma. Here we comprehensively explore the parameter space for the model and also include constraints from petrology, to put limits on magma chamber dimensions and on magma fluxes. 531 JOURNAL OF PETROLOGY VOLUME 55 TOMOG R A P H Y The seismic P-wave velocity structure beneath Montserrat was constrained with first-arrival travel-time inversion of active-source seismic data collected in December 2007, during the SEA-CALIPSO experiment (Voight et al., 2010; Paulatto et al., 2012). The inversion is based on the regularized least-squares approach developed by Hobro et al. (2003), and has been explained in detail by Paulatto et al. (2012). The starting model was a one-dimensional model (1D) derived from a preliminary two-dimensional (2D) inversion of a subset of the data (Paulatto et al., 2010). The model extends over an area of 45 km 50 km and to a depth of 12 km, but is well constrained only to 7·5 km depth. The tomography revealed the presence of a low-seismicvelocity volume (LVV) extending vertically from 4 km to at least 7·5 km depth and horizontally over an area of 6 km 8 km (Fig. 2). The LVV is slightly elongated in the east^west direction. The largest negative velocity anomaly beneath Soufrie're Hills is 0·7 km s1 (Fig. 2). The LVV is attributed to the presence of partial melt and elevated temperatures corresponding to the magma chamber feeding the current eruption. Partial melt is required because the anomaly can be explained by temperature alone only if temperatures were much higher than the known eruption temperature range of Soufrie're Hills andesite (Paulatto et al., 2012). TH ERMA L MODELS The temperatures resulting from the amalgamation of horizontal magma sheets (sills) are calculated with an explicit finite-difference code that solves the heat equation rc @T @f þ rL ¼ kr2 T @t @t ð1Þ where r is density, c is specific heat, T is temperature, L is latent heat of crystallization or fusion, and f is melt fraction. k is thermal conductivity: k ¼ Krc ð2Þ where K, the thermal diffusivity, is temperature dependent and taken from Whittington et al. (2009). The values of the physical parameters used in the simulation are reported in Table 1. The geometry of the modelled sills and magma chamber is cylindrical and simulations are run on a vertical 2D slice of a 3D system. A regular square grid and cylindrical coordinates are used (Fig. 3). The boundary conditions are a fixed temperature at the Earth’s surface (208C) and no heat flux at the right and left boundaries. The condition of no heat flux on the left boundary results from the symmetry of the system. A no heat flux condition on the right boundary is valid as long as the numerical domain is NUMBER 3 MARCH 2014 large enough so that the temperature anomaly induced by the sills does not reach the right boundary on the timescale of the simulation. The bottom boundary condition is based on the results of geodetic models for Soufrie're Hills that indicate the presence of a second reservoir deeper than 10 km (Mattioli et al., 1998; Elsworth et al., 2008; Hautmann et al., 2010). We calculated the heat flux resulting from a 6 km radius spherical body at magmatic temperature, with a top at 12·5 km depth and a centre aligned with the symmetry axis of the numerical domain (Fig. 3). Beyond the horizontal extent of this deep body (6 km), the heat flux is set to zero. This boundary condition underestimates the heat flux in the outer region, but our results are weakly dependent on this assumption. The emplacement of sills is simulated by setting the temperature of the cells corresponding to the location of the sill to a magmatic temperature. Space is made for the successive sill by shifting the cells downwards below the sills. According to Murphy et al. (2000) the melt fraction f at 8508C is 0·35. We assumed a solidus at 7008C (f ¼ 0) and a liquidus at 9508C (f ¼ 1). The exact evolution of melt fraction with temperature between these three points is not well constrained and, for the sake of simplicity, we assumed a linear relationship (Table 1). We considered that the magma was not necessarily thermally at equilibrium at the time of emplacement and might be undercooled. The liquidus phases are missing in the crystalline assemblage of Soufrie're Hills products, which suggests that the andesitic liquid came from a deeper level, where it can be liquid at lower temperature, and crystallized by decompression (Blundy & Cashman, 2001; Annen et al., 2006). The emplacement temperature is unlikely to be above 9008C considering the absence of clinopyroxene and the abundance of amphibole in the crystalline assemblage. We therefore tested the model with an emplacement temperature of 8508C and melt fractions of 0·5 and 1, and with an emplacement temperature of 9008C and melt fractions of 0·675 and 1. Latent heat (3·5 105 J kg1; Hale et al., 2007) is released by crystallization and absorbed by melting. Emplacement of undercooled magma (all the cases, except emplacement at T ¼ 9008C and f ¼ 0·675) results in rapid crystallization, which approximates the process of decompression crystallization (Blundy & Cashman, 2001), and in the release of latent heat, which causes a transient increase in temperature. We ran two sets of simulations, which correspond to two end-members in terms of heat transfer between the magma chamber and the wall-rock. In the first set, we assume that magma convection is weak enough to be neglected and heat transfer is by conduction only. In the second set, we assume that convection is so vigorous that the magma loses heat very rapidly until its melt fraction is less than 0·35, and convection comes to a halt. To simulate this process, all the cells with melt fraction in excess 532 SOUFRIE'RE HILLS MAGMA FLUXES ANNEN et al. (a) 16˚50' SH CH B’ 16˚45' SHV C C’ B 16˚40' −62˚15' (b) 20 −62˚10' (c) Section BB’ (km) 22 24 26 20 Section CC’ (km) 22 24 26 28 −2 0.5 0 −1.5 −2−1 −2.5 −3 0.5 5 −0. 2 0 0.5 Depth (km) 0 −0.5 SHV −1−2 −1.5 0 0 4 .5 −0 .5 −0 6 8 −2 −1 0 vp difference (km/s) 1 Fig. 2. (a) Elevation map of Montserrat showing the locations of seismic stations used in the seismic tomography and the sections shown in (b) and (c). (b, c) Seismic velocity anomaly beneath Soufrie're Hills calculated with respect to the average seismic velocity of the island; (b) north^south section; (c) east^west section. SHV, Soufrie're Hills Volcano; SH, Silver Hills; CH, Centre Hills. See also Paulatto et al., 2012. 533 JOURNAL OF PETROLOGY VOLUME 55 of 0·35 are instantaneously and systematically set to a temperature of 8508C and a melt fraction of 0·35 and the excess heat is transferred to the roof cells (Huppert & Sparks, 1988). Below a melt fraction of 0·35, heat transfer is by conduction only. We checked carefully that the law of heat conservation was respected during this process. The initial geothermal gradient for the simulations presented here is 508C km1. We ran a few tests with a lower geothermal gradient of 308C km1 but the results we Table 1: Values of parameters used in the simulations Density r (kg m3) 2500 Specific heat c (J K1 kg1) 1000 Latent heat L (J kg1) 2 1 Diffusivity K* (mm s ) 3·5 105 567·3/(T þ 273·15) – 0·062 T55738C 0·7320–0·00013(T þ 273·15) T45738C Geotherm (8C km1) 50 Melt fraction f 0 T57008C 0·0023T – 1·63 7008C5T58508C 0·0065T – 5·175 8508C5T59508C 1 T49508C T is temperature in 8C. *Whittington et al. (2009). NUMBER 3 MARCH 2014 obtained were not significantly different. Temperatures and melt contents are tracked during model evolution and the final temperature and melt content distributions are used to estimate expected seismic velocity anomalies. To estimate the effect of temperature we use a theoretical expression for the dependence of P-wave velocity on temperature (Karato, 1993). The effect of melt content is estimated with the self-consistent approach for the calculation of elastic properties of composite materials (Berryman, 1980). These calculations are highly sensitive to the microscopic geometry of the melt pockets (Tai Te, 1966). We used two melt models that represent end-members of the range of possible melt models, thus providing an estimate of the sensitivity of our calculations. In the ‘tube’ model the melt resides in microscopic elongated pockets at the edges of crystal grains, modelled as elongated spheroids. As the melt fraction increases the aspect ratio of the pockets decreases eventually approaching the shape of a sphere. In the ‘crack’ melt model, melt is assumed to reside in microscopic crack-shaped pockets modelled as low aspect ratio spheroids. The latter model results in higher sensitivity to melt content and hence stronger seismic velocity anomalies. The tube model is more realistic for the magma storage region in Montserrat, where pore shape is probably controlled by crystal grain geometry (Kiddle, 2012). T = 20 oC q=0 q=0 sill q = qm q=0 Fig. 3. Model setup. The numerical domain, grid and sill are not to scale. q is heat flux; qm is the heat flux from a deep spherical magma chamber located outside the numerical domain. The geometry of the system is cylindrical in the horizontal plane normal to the figure. Temperatures are computed at the grid nodes over a half domain with explicit finite differences. 534 ANNEN et al. SOUFRIE'RE HILLS MAGMA FLUXES Seismic velocities are also affected by fluid-related velocity dispersion mechanisms, such as the flow of melt between interconnected pockets (squirt flow). If seismic frequencies are sufficiently low (low-frequency approximation) squirt flow equalizes pressure imbalances and dissipates energy, resulting in reduced seismic velocities compared with the case where squirt flow cannot happen (high-frequency approximation) (Budiansky & O’Connell, 1976). The difference between the two cases is significant only for the crack model. We adopt the low-frequency approximation, which leads to lower estimates of melt volume and is more likely for Montserrat upper crustal rocks (Kiddle, 2012). R E S U LT S Tomography results suggest the presence of melt between about 5^6 km depth and at least 7·5 km depth (Paulatto et al., 2012). We ran a series of magma body growth simulations with the first sill emplaced at 5, 5·5 or 6 km depth and with sill radius varying from 1 to 5 km. Sills are emplaced by under-accretion down to 8, 8·5 or 9·0 km. The exact position of the first and last sill does not affect significantly the results, so here we show only the simulation with sills emplaced from 5·5 down to 8 km. Based on the geological record (Fig. 1), we tested the hypotheses that the LVV is related to magma injection that started: (1) at 250 ka (the magma body is as old as Soufrie're Hills); (2) at 30^40 ka (the magma body started to grow at the beginning of the last cycle of high activity); (3) at 17ka (the magma reservoir started to (re)fill at the end of the last cycle of high activity); (4) 6000 years ago, when activity resumed with the formation of English’s Crater; (5) 350 years ago, after the last eruption before the current activity. We also ran models over 100 and 150 kyr to refine our estimates of fluxes and durations. In comparison with simulations where heat transfer is by conduction only, the volumes of molten materials that are produced with simulations that integrate convection are larger and their temperatures are lower (Fig. 4). This effect is noticeable only if magma is injected at high temperatures and high melt fractions, and it is never strong enough to affect our conclusions regarding the fit with the tomography model. Final temperatures and melt fractions in the magma body depend on how fast it was assembled (Fig. 5), on the total heat content of the injected magma [i.e. on its temperature (sensible heat) and melt fraction (latent heat) (Fig. 6)], and on the intrusion radius (Fig. 7). The log of magma intrusive fluxes is shown in Figs 8^10 as contour lines. This average flux corresponds to the total volume of magma injected during the simulation divided by the duration of the simulation. Figure 8 shows which combination of parameter values (magma heat content, simulation duration, intrusion radius) result in magma temperatures at the end of the simulation that are hotter, colder or in the same range of temperature (830^8808C) compared with Soufrie're Hills magmas as determined by experimental petrology (Barclay et al., 1998; Devine et al., 1998; Murphy et al., 2000). At temperatures below 8508C, melt fractions are less than 0·5 and the magma is unlikely to behave as a fluid and be eruptible. We note that our modelled magma chambers can be hotter than 8508C only in conductive models. For emplacement durations of more than 100 kyr and flux of less 103 km a1, the model temperatures are always colder than the petrology-inferred temperatures, independently of the effect of convection. Emplacement over 250 kyr does not produce any melt, because each magma pulse completely solidifies before the injection of the next one. Figure 9 shows where, in the parameter space, the modelled velocity anomaly fits the field tomography (Fig. 2); that is, where its intensity is between 0·5 and 0·9 km s1 and the diameter of the 0·5 km s1 isoline is between 2 and 4 km. For protracted chamber growth duration and/or magma emplaced with low heat content, the seismic velocity anomaly is too weak to fit observations, whereas short durations and high heat content produce an anomaly that is too strong (Figs 5 and 6). A fit is dependent on the chamber radius (Fig. 7) and on growth duration (Fig. 6) (i.e. on the average magma flux), with longer duration requiring larger chamber dimensions (Fig. 9). A chamber radius of 1km always produces anomalies that are too small (Figs 7 and 9). Fits were found for chamber radii from 2 to 5 km and a large range of durations from 6 to 150 kyr (Fig. 9), but all fits involve fluxes that are in a relatively restricted range between 7 104 and 5 103 km3 a1. DISCUSSION The current eruptive flux of Soufrie're Hills volcano is about 7·5 102 km3 a1 (Sparks et al., 1998; Elsworth et al., 2008; Wadge et al., 2013), but the average eruptive flux over its lifetime is estimated as (1·5^ 1·7) 104 km3 a1 (Harford et al., 2002; Le Friant et al., 2004). Our estimated intrusive flux (Fig. 9) of 7 104 to 5 103 km3 a1 lies between these two values. The model of incremental igneous body growth that is now widely accepted implies that several magma fluxes characterize the growth of igneous bodies (de Saint Blanquat et al., 2011): a ‘long-term’ average flux, which corresponds to the total volume of an igneous body divided by the total emplacement time, a ‘quasi-instantaneous’ flux, which corresponds to the emplacement of the smallest unit, and one or a series of ‘intermediate fluxes’ corresponding to periods of high activity. In comparison with plutonic fluxes based on radiogenic isotopes, the flux at the origin of Soufrie're Hills LVV is higher than the fluxes determined by Matzel et al. (2006) for Tenpeak intrusion 535 Depth (km) 536 0 400 800 26 Temperature (oC) 200 400 800 SHV 400 800 200 SHV 24 Y distance (km) 22 1200 28 20 20 30 26 Melt fraction (vol)% 10 10 30 SHV 10 30 SHV 24 Y distance (km) 22 40 −2 28 20 0 26 28 20 −0.5 −1 −1.5 SHV 24 1 −2 −1 0 vp difference (km/s) −0.5 −1 SHV 26 Y distance (km) 22 (b) Strong convection vp difference (km/s) −1 −0.5 −2 SHV −0.5 −1 −3.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −3 −0.5 SHV −3.5 −0.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 −1 −1.5 SHV −0.5 −1 −1.5 SHV 24 Y distance (km) 22 1 28 NUMBER 3 8 6 4 2 0 −2 8 6 4 2 0 −2 20 (a) Weak convection VOLUME 55 Fig. 4. Simulation results. (a) Heat transfer is by conduction only. (b) Strong magma convection results in rapid heat transfer to the roof. Emplacement duration is 6000 years. The intrusion radius is 2 km. Magma is injected at 9008C with a melt fraction of unity. First column, temperature; second column, melt fractions; third column, unsmoothed seismic velocity anomalies assuming tube-shaped melt pockets; fourth column, smoothed seismic velocity anomalies assuming tube-shaped melt pockets; fifth column, unsmoothed seismic velocity anomalies assuming crackshaped melt pockets; sixth column, smoothed seismic velocity anomalies assuming crack-shaped melt pockets. Depth (km) JOURNAL OF PETROLOGY MARCH 2014 Depth (km) 537 400 26 Temperature (o C) 800 600 800 6 8 200 400 0 SHV 400 800 200 SHV 24 Y distance (km) 22 4 2 0 −2 8 6 4 2 0 −2 20 1200 28 20 10 20 10 30 30 26 Melt fraction (vol)% 20 SHV 10 30 SHV 24 Y distance (km) 22 40 −2 28 20 24 0 28 20 −0.5 −1 SHV 24 1 −2 −1 0 vp difference (km/s) −0.5 SHV 26 Y distance (km) 22 (b) Duration = 6000 yrs 26 vp difference (km/s) −2 −0.5 SHV −0.5 −2 SHV −1 22 Y distance (km) (a) Duration = 350 yrs 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 − −3 SHV −3 −0.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 −1 SHV −0.5 −1 −1.5 SHV 24 Y distance (km) 22 1 28 Fig. 5. Simulation results for duration of magma emplacement of (a) 350, (b) 6000, (c) 17 000, and (d) 30 000 years. The intrusion radius is 2 km. The temperature and melt fraction of the injected magma are 8508C and 0·5, respectively. Heat transfer is by conduction only. The columns are as in Fig. 4. (continued) Depth (km) ANNEN et al. SOUFRIE'RE HILLS MAGMA FLUXES Depth (km) 538 0 400 800 26 Temperature ( oC) 400 600 200 SHV 400 600 800 200 SHV 24 Y distance (km) 22 1200 28 20 20 10 30 26 Melt fraction (vol)% 10 10 SHV 20 SHV 24 Y distance (km) 22 40 −2 28 20 −1 28 20 −0.5 SHV 24 26 Y distance (km) 22 0 1 −2 0 vp difference (km/s) −1 SHV (d) Duration = 30,000 yrs 26 vp difference (km/s) −0.5 −1 SHV −0.5 −1 −1.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −2.5 −0.5 SHV −3 −0.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 SHV −1 −0.5 SHV 24 Y distance (km) 22 1 28 NUMBER 3 8 6 4 2 0 −2 8 6 4 2 0 −2 20 (c) Duration = 17,000 yrs VOLUME 55 Fig. 5. Continued. Depth (km) JOURNAL OF PETROLOGY MARCH 2014 Depth (km) 539 8 6 4 2 0 −2 8 6 4 2 0 −2 0 20 400 800 200 200 26 Temperature (oC) 400 600 800 SHV 400 600 SHV 24 Y distance (km) 22 1200 28 20 10 20 10 30 30 26 Melt fraction (vol)% 20 SHV 10 SHV 24 Y distance (km) 22 40 −2 28 20 −1 −0.5 −1.5 SHV −0.5 −1 SHV 24 28 20 SHV 24 0 1 −2 −1 0 vp difference (km/s) −0.5 SHV 26 Y distance (km) 22 (b) T = 900 C, f = 0.675 26 vp difference (km/s) 22 Y distance (km) (a) T = 850 C, f = 0.5 1 −2 28 20 −1 0 26 vp difference (km/s) −3 −0.5 SHV −2.5 0.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 SHV −0.5 SHV 24 Y distance (km) 22 1 28 Fig. 6. Simulation results for different heat content of the magma that is repeatedly injected: (a) Temperature (T) is 8508C, melt fraction (f) is 0·5. (b) T is 9008C, f is 0·675. (c) T is 8508C, f is unity. (d) T is 9008C, f is unity. The emplacement duration is 30 kyr and the intrusions radius is 2 km. Heat transfer is by conduction only. The columns are as in Fig. 4. (continued) Depth (km) ANNEN et al. SOUFRIE'RE HILLS MAGMA FLUXES Depth (km) 540 0 400 800 26 Temperature (oC) 400 600 800 200 SHV 400 600 800 200 SHV 24 Y distance (km) 22 1200 28 20 20 20 10 30 30 26 Melt fraction (vol)% 10 10 30 SHV 20 SHV 24 Y distance (km) 22 40 −2 28 20 −1 0 28 20 −0.5 SHV 1 −2 24 −1 0 vp difference (km/s) −0.5 SHV 26 Y distance (km) 22 (d) T = 900 C, f = 1 26 vp difference (km/s) −2.5 −0.5 SHV −2.5 −0.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −3.5 −0.5 SHV −3 −0.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 −1 SHV −1 −0.5 SHV 24 Y distance (km) 22 1 28 NUMBER 3 8 6 4 2 0 −2 8 6 4 2 0 −2 20 (c) T = 850 C, f = 1 VOLUME 55 Fig. 6. Continued. Depth (km) JOURNAL OF PETROLOGY MARCH 2014 Depth (km) 541 8 6 4 2 0 −2 8 6 4 2 0 −2 0 20 400 800 26 Temperature (oC) 600 400 200 SHV 600 400 200 SHV 24 Y distance (km) 22 1200 28 20 20 30 26 Melt fraction (vol) % 10 SHV SHV 24 Y distance (km) 22 40 −2 28 20 24 0 26 28 20 SHV 1 −2 24 −1 0 vp difference (km/s) −0.5 SHV 26 Y distance (km) 22 (b) Radius = 3 km vp difference (km/s) −0.5 SHV SHV −1 22 Y distance (km) (a) Radius = 2 km 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 −2.5 SHV −1.5 −0.5 SHV 24 Y distance (km) 22 1 −2 28 20 −1 0 26 vp difference (km/s) −0.5 SHV SHV 24 Y distance (km) 22 1 28 Fig. 7. Simulation results for different intrusion radii of 2 km (a), 3 km (b), and 4 km (c). Emplacement duration is 100 kyr. Temperature and melt fraction of the injected magma are 9008C and unity, respectively. Heat transfer is by conduction only. The columns are as in Fig. 4. (continued) Depth (km) ANNEN et al. SOUFRIE'RE HILLS MAGMA FLUXES 542 8 6 4 2 0 −2 0 20 26 400 800 o Temperature ( C) 600 400 200 SHV 24 Y distance (km) 22 1200 28 20 26 10 20 30 Melt fraction (vol %) 10 SHV 24 Y distance (km) 22 40 −2 28 20 26 −1 0 vp difference (km/s) −0.5 −1 SHV 24 Y distance (km) 22 1 −2 28 20 26 −1 0 vp difference (km/s) −0.5 SHV 24 Y distance (km) 22 1 −2 28 20 26 −1 0 vp difference (km/s) −3 −0.5 −1 SHV 24 Y distance (km) 22 1 −2 28 20 26 −1 0 vp difference (km/s) −0.5 −1 SHV 24 Y distance (km) 22 1 28 VOLUME 55 NUMBER 3 Fig. 7. Continued. Depth (km) (c) Radius = 4 km JOURNAL OF PETROLOGY MARCH 2014 ANNEN et al. SOUFRIE'RE HILLS MAGMA FLUXES Radius (m) (a) T = 850 C, f = 0.5 5.5 0 5 4.5 4 3.5 0 3 2.5 2 1.5 1 1 0.5 2 10 (b) T = 900 C, f = 0.675 −1 5.5 0cold −1 5 hot fit 4.5 4 3.5 0 −1 3 2.5 2 1.5 −2 1 −1 0.5 2 3 10 10 −3 −2 −3 −1 −2 −3 −4 −3 −2 −4 −5 10 3 10 4 10 5 Radius (m) (c) T = 850 C, f = 1 5.5 0 5 4.5 4 3.5 0 3 2.5 2 1.5 1 −1 0.5 2 10 5.5 0 5 4.5 4 3.5 0 3 2.5 2 1.5 1 −1 0.5 2 10 −3 −2 −3 −2 −3 −4 −2 −3 −4 −5 3 10 −3 −2 −3 −4 −3 −4 −5 10 4 10 5 (d) T = 900 C, f = 1 −1 −1 −3 −2 4 10 Duration (yrs) 5 10 −1 −3 −2 −3 −1 −2 −3 −4 −2 −3 −4 −5 3 10 4 10 Duration (yrs) 5 10 Fig. 8. Comparison between simulations and petrological data plotted in radius^duration space for different magma emplacement temperatures T and melt fractions f. Grey squares indicate magma chamber temperatures at the end of the simulation similar to those determined by petrology (830^8808C). Open circles indicate lower temperatures and open diamonds indicate higher temperatures. Contour lines are log of fluxes in km3 a1. (1·5 104 km3 a1) and Mount Stuart Batholith (2·2 104 to 5·8 104 km a1) but it is comparable with the 8 104 km3 a1 intrusive flux of the Torres del Paine granite (Leuthold et al., 2012). However, it is about one order of magnitude lower than the fluxes estimated on the basis of ground surface deformation at Uturuncu (Sparks et al., 2008) and at Santorini (Parks et al., 2012) and on the basis of petrological study of the Santorini Minoan eruption products, which indicate that magma was injected in the shallow reservoir shortly before the eruption with a flux 45 102 km3 a1. These comparisons suggest that the flux that built the Soufrie're Hills shallow reservoir is intermediate between the long-term average fluxes that form batholiths and the short-term fluxes that replenish volcanic reservoirs. From the perspective of plutonic^volcanic relationships, the shape of intrusions exerts a strong control on heat loss, so that with an equivalent flux, intrusive bodies with the highest aspect ratio might not accumulate melt, whereas bodies with lower aspect ratio would evolve into eruptible magma chambers. For a magma body that grows by addition of sills, temperature and melt fractions are controlled by the sill addition rate. For example, if an average addition rate of 3 cm a1 is needed to accumulate eruptible magma (Hanson & Glazner, 1995; Yoshinobu et al., 1998; Annen et al., 2008), the minimum flux required to form a magma chamber is about 103 km3 a1 if the magma body diameter is 4^8 km as is predicted for Soufrie're Hills, whereas for larger systems, which can feed a caldera-forming eruption and with a 20^40 km diameter, 543 JOURNAL OF PETROLOGY VOLUME 55 (a) T = 850 C, f = 0.5 5.5 Radius (m) 5.5 5 4.5 4 3.5 0 3 2.5 2 1.5 1 −1 0.5 2 10 −3 −2 4.5 −3 4 3.5 −2 −1 0 3 −3 2.5 2 −4 1.5 1 −1 −2 −3 3 4 −4 −5 0.5 2 10 10 5 10 10 (c) T = 850 C, f = 1 5.5 0 5 −1 Radius (m) −3 3.5 −2 −1 −3 2.5 2 1.5 1 −1 0.5 −2 −3 −4 −4 −5 2 10 3 10 weak −1 too small too strong Fit tomo 5.5 0 5 4.5 4 3.5 0 3 2.5 2 1.5 1 −1 0.5 2 10 −3 −2 4 0 0too −3 −2 −3 −2 −1 −3 −4 −3 −2 −4 −5 3 4 10 5 10 10 (d) T = 900 C, f = 1 4.5 3 MARCH 2014 (b) T = 900 C, f = 0.675 −1 0 5 NUMBER 3 4 10 Duration (yrs) 5 10 −1 −3 −2 −3 −2 −1 −3 −2 −3 −4 −4 −5 3 10 4 10 Duration (yrs) 5 10 Fig. 9. Comparison between simulations and field-recorded tomography plotted in radius^duration space for different magma emplacement temperaturesTand melt fractions f. Grey squares indicate synthetic velocity anomalies similar to field velocity anomalies. Upward-pointing triangles indicate synthetic velocity anomalies that are too weak. Downward-pointing triangles indicate synthetic velocity anomalies that are too strong and diamonds indicate synthetic velocity anomalies that are too small. Contour lines are log of fluxes in km3 a1. this minimum flux is 101^102 km3 a1 (Annen, 2009). In other words, the flux needed to maintain a typical andesitic arc volcano magma chamber is at least one order of magnitude lower than the flux required to assemble the magma for a supereruption. The residence times of crystals in the Soufrie're Hills eruptive products are very short (5320 years; Zellmer et al., 2003b). According to our calculations, the LVV cannot be due to the growth of a magma chamber over the last 350 years, unless emplacement temperatures and melt fractions were very close to the pre-eruptive ones that we are trying to match, because on such a short timescale the magma does not have time to cool and crystallize. This would imply emplacement with a high crystal fraction of about 0·65^0·7. Magma viscosity strongly depends on its crystal content and increases dramatically beyond a critical crystal fraction, which usually falls between 0·4 and 0·6 (Marsh, 1981; Lejeune & Richet, 1995; Caricchi et al., 2007), although, depending on the grain shape and aspect ratio, the critical crystal fraction might be lower (Mueller et al., 2010; Cimarelli et al., 2011; Schmeling et al., 2012). Experiments show that crystal size bimodality can increase critical crystal fraction to 0·66 (Castruccio et al., 2010), but it is unlikely that a magma with a crystal fraction of 0·65^0·7 can be transported through a dyke. Emplacement with crystal fraction of 0·5 or less is much more plausible. In this case, matches can be found only if the chamber grew over 6000 years or more. However, for those simulations, the magmas in the chamber cover a much wider age range than the 320 years suggested by the plagioclase study of Zellmer et al. (2003b). Residence times and tomography can be reconciled if most of the 544 SOUFRIE'RE HILLS MAGMA FLUXES 5 No fit fit −3 −2 0 5.5 −1 ANNEN et al. 4.5 −3 4 −1 3 −2 3.5 0 Radius (m) deformation that the crust is unlikely to be able to accommodate. We expect early injections to be accommodated by brittle deformation but as the system heats up and the crust become more ductile, new magma volume might be accommodated viscously, which might have consequences on the behaviour of the ground surface deformation signal (Pearse & Fialko, 2010). −3 2.5 2 −4 −2 1 −3 1.5 CONC LUSIONS −4 0.5 102 103 104 Duration (yrs) 105 Fig. 10. Comparison between simulation results and tomography, integrating the rheological state of the magma. The grey squares show the simulations that fit the field-recorded tomography and where the simulated magma body melt fractions are less than 0·5; that is, most of the magma is not eruptible. The open circles are for simulations that either do not fit the field-recorded tomography or where the magma is at a melt fraction of 0·5 or more. Contour lines are log of fluxes in km3 a1. low-velocity volume is a non-eruptible mush of crystals and melt and the current eruption taps smaller bodies of magma that have been assembled at shallow level during the last few centuries. If we take into account this additional constraint (i.e. that the LVV is related to a reservoir that is largely crystalline with a relatively small amounts of melt), we find magma emplacement durations between 30 and 150 kyr and fluxes between 7 104 and 2·4 103 km3 a1 (Fig. 10). Our results combined with data from geophysics, petrology and geochemistry support the following model. Magma differentiates and resides over timescales of more that 350 kyr (Zellmer et al., 2003a) as partial melt in a lower crustal hot zone (Sparks & Young, 2002; Annen et al., 2006; Hodge et al., 2012; Svensen et al., 2012). This melt is episodically released to an intermediate magma reservoir at a depth of 10^12 km, which in turn feeds a shallow reservoir at 5^6 km depth, where most of the magma partially crystallizes and forms a non-eruptible mush. Occasionally, eruptions tap the magma that has been recently injected and is still mobile. The long time scale of deep melt storage, segregation and episodic ascent might be related to the physics of melt separation in a slowly accumulating region of basalt underplating (Solano et al., 2012). Our simulations do not include emplacement mechanics, which is beyond the scope of this study. As discussed above, magma chamber growth over less than 6000 years requires emplacement at a very low melt fraction and high viscosity. Another difficulty with emplacement over short durations is that the fluxes and sill addition rates are very high (9 102 km3 a1 and 7 m a1, respectively for emplacement over 350 years) and would result in The temperatures and melt fractions in a growing magma body strongly depend on the heat content of the injected magma and on the duration of the magma body growth. We found that the low-velocity volume imaged by seismic tomography that underlies Soufrie're Hills volcano can be reproduced with intrusive magma fluxes in the range of 7 104 to 5 103 km3 a1 over durations that range from 6 to 150 kyr depending on the intrusion radius (2^ 5 km). Crystal residence times that are much shorter than the age of our modelled magma reservoir might indicate that the eruption taps small pockets of recently assembled magma and that the low-velocity volume detected below Soufrie're Hills is a non-eruptible crystalline mush. This additional constraint limits the intrusive fluxes to a maximum of 2·4 103 km3 a1. AC K N O W L E D G E M E N T S We gratefully thank Bruce Marsh for his positive and encouraging comments, and an anonymous reviewer for helping us to clarify our argument. FU N DI NG This work was supported by a European Research Council Advanced Grant (project VOLDIES). The tomography experiment on which this work is based was supported by the National Science Foundation, the National Environment Research Council, the British Geological Survey (BGS), Discovery Channel TV and the British Foreign and Commonwealth Office. R EF ER ENC ES Annen, C. (2009). From plutons to magma chambers: Thermal constraints on the accumulation of eruptible silicic magma in the upper crust. Earth and Planetary Science Letters 284, 409^416. Annen, C., Blundy, J. D. & Sparks, R. S. J. (2006). The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology 47, 505^539. Annen, C., Pichavant, M., Bachmann, O. & Burgisser, A. (2008). Conditions for the growth of a long-lived shallow crustal magma chamber below Mount Pelee volcano (Martinique, Lesser Antilles Arc). Journal of Geophysical Research 113, B07209, doi: 10.1029/ 2007JB005049. Aspinall, W. P., Miller, A. D., Lynch, L. L., Latchman, J. L., Stewart, R. C., White, R. A. & Power, J. A. (1998). Soufrie're Hills 545 JOURNAL OF PETROLOGY VOLUME 55 eruption, Montserrat, 1995^1997: volcanic earthquake locations and fault plane solutions. Geophysical Research Letters 25, 3397^3400. Bachmann, O., Dungan, M. A. & Lipman, P. W. (2002). The Fish Canyon magma body, San Juan volcanic field, Colorado: Rejuvenation and eruption of an upper-crustal batholith. Journal of Petrology 43, 1469^1503. Barclay, J., Rutherford, M. J., Carroll, M. R., Murphy, M. D., Devine, J. D., Gardner, J. & Sparks, R. S. J. (1998). Experimental phase equilibria constraints on pre-eruptive storage conditions of the Soufrie're Hills magma. Geophysical Research Letters 25, 3437^3440. Berryman, J. G. (1980). Long wavelength propagation in composite elastic media. 2. Ellipsoidal inclusions. Journal of the Acoustical Society of America 68, 1820^1831. Blundy, J. & Cashman, K. (2001). Ascent-driven crystallisation of dacite magmas at Mount St Helens, 1980^1986. Contributions to Mineralogy and Petrology 140, 631^650. Briden, J. C., Rex, D. C., Faller, A. M. & Tomblin, J. F. (1979). K^Ar geochronology and paleomagnetism of volcanic rocks in the Lesser Antilles Island Arc. Philosophical Transactions of the Royal Society of London 291, 485^528. Budiansky, B. & O’Connell, R. J. (1976). Elastic moduli of a cracked solid. International Journal of Solids and Structures 12, 81^97. Caricchi, L., Burlini, L., Ulmer, P., Gerya, T., Vassalli, M. & Papale, P. (2007). Non-Newtonian rheology of crystal-bearing magmas and implications for magma ascent dynamics. Earth and Planetary Science Letters 264, 402^419. Castruccio, A., Rust, A. C. & Sparks, R. S. J. (2010). Rheology and flow of crystal-bearing lavas: Insights from analogue gravity currents. Earth and Planetary Science Letters 297, 471^480. Cimarelli, C., Costa, A., Mueller, S. & Mader, H. M. (2011). Rheology of magmas with bimodal crystal size and shape distributions: Insights from analog experiments. Geochemistry, Geophysics, Geosystems 12, Q07024, doi: 10.1029/2011GC003606. Coleman, D. S., Gray, W. & Glazner, A. F. (2004). Rethinking the emplacement and evolution of zoned plutons: Geochronologic evidence for incremental assembly of the Tuolumne Intrusive Suite, California. Geology 32, 433^436. Couch, S., Sparks, R. S. J. & Carroll, M. R. (2001). Mineral disequilibrium in lavas explained by convective self-mixing in open magma chambers. Nature 411, 1037^1039. Couch, S., Harford, C. L., Sparks, R. S. J. & Carroll, M. R. (2003). Experimental constraints on the conditions of formation of highly calcic plagioclase microlites at the Soufrie're Hills Volcano, Montserrat. Journal of Petrology 44, 1455^1475. Davaille, A. & Jaupart, C. (1993). Transient high-Rayleigh-number thermal-convection with large viscosity variations. Journal of Fluid Mechanics 253, 141^166. de Saint-Blanquat, M., Habert, G., Horsman, E., Morgan, S. S., Tikoff, B., Launeau, P. & Gleizes, G. (2006). Mechanisms and duration of non-tectonically assisted magma emplacement in the upper crust: The Black Mesa pluton, Henry Mountains, Utah. Tectonophysics 428, 1^31. de Saint Blanquat, M., Horsman, E., Habert, G., Morgan, S., Vanderhaeghe, O., Law, R. & Tikoff, B. (2011). Multiscale magmatic cyclicity, duration of pluton construction, and the paradoxical relationship between tectonism and plutonism in continental arcs. Tectonophysics 500, 20^33. Devine, J. D., Murphy, M. D., Rutherford, M. J., Barclay, J., Sparks, R. S. J., Carroll, M. R., Young, S. R. & Gardner, J. E. (1998). Petrologic evidence for pre-eruptive pressure^temperature conditions, and recent re-heating, of andesite magma at Soufrie're NUMBER 3 MARCH 2014 Hills Volcano, Montserrat, W.I. Geophysical Research Letter 25, 3669^3672. Devine, J. D., Rutherford, M. J., Norton, G. E. & Young, S. R. (2003). Magma storage region processes inferred from geochemistry of Fe^Ti oxides in andesitic magma, Soufrie're Hills Volcano, Montserrat, W.I. Journal of Petrology 44, 1375^1400. Druitt, T. H., Costa, F., Deloule, E., Dungan, M. & Scaillet, B. (2012). Decadal to monthly timescales of magma transfer and reservoir growth at a caldera volcano. Nature 482, 77^80. Elsworth, D., Mattioli, G., Taron, J., Voight, B. & Herd, R. (2008). Implications of magma transfer between multiple reservoirs on eruption cycling. Science 322, 246^248. Galland, O. (2012). Experimental modelling of ground deformation associated with shallow magma intrusions. Earth and Planetary Science Letters 317^318, 145^156. Gudmundsson, A. (2011). Deflection of dykes into sills at discontinuities and magma-chamber formation. Tectonophysics 500, 50^64. Gudmundsson, A. (2012). Magma chambers: Formation, local stresses, excess pressures, and compartments. Journal of Volcanology and Geothermal Research 237^238, 19^41. Hale, A. J., Wadge, G. & Muhlhaus, H. B. (2007). The influence of viscous and latent heating on crystal-rich magma flow in a conduit. Geophysical Journal International 171, 1406^1429. Hanson, R. B. & Glazner, A. F. (1995). Thermal requirements for extensional emplacement of granitoids. Geology 23, 213^216. Harford, C. L. & Sparks, R. S. J. (2001). Recent remobilisation of shallow-level intrusions on Montserrat revealed by hydrogen isotope composition of amphiboles. Earth and Planetary Science Letters 185, 285^297. Harford, C. L., Pringle, M. S., Sparks, R. S. J. & Young, S. R. (2002). The volcanic evolution of Montserrat using 40Ar/39Ar geochronology. In: Druitt, T. H. & Kokelaar, B. P. (eds) The Rruption of Soufrie' re Hills Volcano, Montserrat, from 1995 to 1999. Geological Society, London, Memoirs 21, 93^113. Hautmann, S., Gottsmann, J., Sparks, R. S. J., Mattioli, G. S., Sacks, I. S. & Strutt, M. H. (2010). Effect of mechanical heterogeneity in arc crust on volcano deformation with application to Soufrie're Hills Volcano, Montserrat, West Indies. Journal of Geophysical Research 115, doi:10.1029/2009JB006909. Hobro, J. W. D., Singh, S. C. & Minshull, T. A. (2003). Threedimensional tomographic inversion of combined reflection and refraction seismic traveltime data. Geophysical Journal Intenational 152, 79^93. Hodge, K. F., Carazzo, G. & Jellinek, A. M. (2012). Experimental constraints on the deformation and breakup of injected magma. Earth and Planetary Science Letters 325^326, 52^62. Humphreys, M. C. S., Edmonds, M., Christopher, T. & Hards, V. (2009). Chlorine variations in the magma of Soufrie're Hills Volcano, Montserrat: Insights from Cl in hornblende and melt inclusions. Geochimica et Cosmochimica Acta 73, 5693^5708. Huppert, H. E. & Sparks, R. S. J. (1988). Melting of the roof of a chamber containing a hot, turbulently convecting fluid. Journal of Fluid Mechanics 188, 107^131. Jellinek, A. M. & DePaolo, D. J. (2003). A model for the origin of large silicic magma chambers: precursors of caldera-forming eruptions. Bulletin of Volcanology 65, 363^381. Jellinek, A. M. & Kerr, R. C. (1999). Mixing and compositional stratification produced by natural convection 2. Applications to the differentiation of basaltic and silicic magma chambers and komatiite lava flows. Journal of Geophysical Research 104, 7203^7218. Karato, S. (1993). Importance of anelasticity in the interpretation of seismic tomography. Geophysical Research Letters 20, 1623^1626. 546 ANNEN et al. SOUFRIE'RE HILLS MAGMA FLUXES Kavanagh, J. L., Menand, T. & Sparks, R. S. J. (2006). An experimental investigation of sill formation and propagation in layered elastic media. Earth and Planetary Science Letters 245, 799^813. Kiddle, E. J. (2012). The structure of the crust and magmatic system at Montserrat, Lesser Antilles. PhD thesis, Bristol University. Le Friant, A., Harford, C. L., Deplus, C., Boudon, G., Sparks, R. S. J., Herd, R. A. & Komorowski, J. C. (2004). Geomorphological evolution of Montserrat (West Indies): importance of flank collapse and erosional processes. Journal of the Geological Society, London 161, 147^160. Le Friant, A., Lock, E. J., Hart, M. B., Boudon, G., Sparks, R. S. J., Leng, M. J., Smart, C. W., Komorowski, J. C., Deplus, C. & Fisher, J. K. (2008). Late Pleistocene tephrochronology of marine sediments adjacent to Montserrat, Lesser Antilles volcanic arc. Journal of the Geological Society, London 165, 279^289. Lejeune, A. M. & Richet, P. (1995). Rheology of crystal-bearing silicate meltsçan experimental study at high viscosities. Journal of Geophysical Research 100, 4215^4229. Leuthold, J., Mu«ntener, O., Baumgartner, L. P., Putlitz, B., Ovtcharova, M. & Schaltegger, U. (2012). Time resolved construction of a bimodal laccolith (Torres del Paine, Patagonia). Earth and Planetary Science Letters 325^326, 85^92. Maccaferri, F., Bonafede, M. & Rivalta, E. (2011). A quantitative study of the mechanisms governing dike propagation, dike arrest and sill formation. Journal of Volcanology and Geothermal Research 208, 39^50. MacGregor, A. G. (1938). The Royal Society Expedition to Montserrat, B. W. I. The volcanic history and petrology of Montserrat with observations on Mt Pele¤e, in Martinique. Philosophical Transactions of the Royal Society of London, Series B 229, 1^90. Marsh, B. D. (1981). On the crystallinity, probability of occurrence, and rheology of lava and magma. Contributions to Mineralogy and Petrology 78, 85^98. Mattioli, G. S., Herd, R. A., Strutt, M. H., Ryan, G., Widiwijayanti, C. & Voight, B. (1998). Long term surface deformation of Soufrie're Hills Volcano, Montserrat from GPS geodesy: Inferences from simple elastic inverse models. Geophysical Research Letters 37, L00E13. Matzel, J. E. P., Bowring, S. A. & Miller, R. B. (2006). Time scales of pluton construction at differing crustal levels: Examples from the Mount Stuart and Tenpeak intrusions, North Cascades, Washington. Geological Society of America Bulletin 118, 1412^1430. Menand, T. (2008). The mechanics and dynamics of sills in layered elastic rocks and their implications for the growth of laccoliths and other igneous complexes. Earth and Planetary Science Letters 267, 93^99. Menand, T. (2011). Physical controls and depth of emplacement of igneous bodies: A review. Tectonophysics 500, 11^19. Michel, J., Baumgartner, L. P., Putlitz, B., Schaltegger, U. & Ovtcharova, M. (2008). Incremental growth of the Patagonian Torres del Paine laccolith over 90 k.y. Geology 36, 459^462. Miller, C. F., Furbish, D. J., Walker, B. A., Clairborne, L. L., Cotheas, G. C., Bleick, H. A. & Miller, J. S. (2011). Growth of plutons by incremental emplacement of sheets in crystal-rich host: Evidence from Miocene intrusions of the Colorado River Region, Nevada, USA. Tectonophysics 500, 65^77. Mogi, K. (1958). Relations between the eruption of various volcanoes and the deformations of the ground surface around them. Bulletin of the Earthquake Research Institute 36, 99^134. Mueller, S., Llewellin, E. W. & Mader, H. M. (2010). The rheology of suspensions of solid particles. Proceedings of the Royal Society of London 466, 1201^1228. Murphy, M. D., Sparks, R. S. J., Barclay, J., Carroll, M. R. & Brewer, T. S. (2000). Remobilization of andesite magma by intrusion of mafic magma at the Soufrie're Hills Volcano, Montserrat, West Indies. Journal of Petrology 41, 21^42. Parks, M. M., Biggs, J., England, P., Mather, T. A., Nomikou, P., Palamartchouk, K., Papanikolaou, X., Parassidis, D., Parsons, B., Pyle, D. M., Raptakis, C. & Zacharis, V. (2012). Evolution of Santorini Volcano dominated by episodic and rapid fluxes of melt from depth. Nature Geoscience 5, 749^754. Pasquare', F. & Tibaldi, A. (2007). Structure of a sheet^laccolith system revealing the interplay between tectonic and magma stresses at Stardalur Volcano, Iceland. Journal of Volcanology and Geothermal Research 161, 131^150. Paulatto, M., Minshull, T. A., Baptie, B., Dean, S., Hammond, J. O. S., Henstock, T. H., Kenedi, C. L., Kiddle, E. J., Malin, P., Peirce, C., Ryan, G., Shalev, E., Sparks, R. S. J. & Voight, B. (2010). Upper crustal structure of an active volcano from refraction/reflection tomography, Montserrat, Lesser Antilles. Geophysical Journal International 180, 685^696. Paulatto, M., Annen, C., Henstock, T. J., Kiddle, E., Minshull, T. A., Sparks, R. S. J. & Voight, B. (2012). Magma chamber properties from integrated seismic tomography and thermal modeling at Montserrat. Geochemistry, Geophysics, Geosystems 13, Q01014, doi: 10.1029/2011GC003892. Pearse, J. & Fialko, Y. (2010). Mechanics of active magmatic intraplating in the Rio Grande Rift near Socorro, New Mexico. Journal of Geophysical ResearchçSolid Earth 115, B07413, doi: 10.1029/ 2009JB006592. Powell, C. F. (1938). The Royal Society expedition to Montserrat, B. W. I. Final report. Philosophical Transactions of the Royal Society, Series A 237, 1^34. Rea, W. J. (1974). The volcanic geology and petrology of Montserrat, West Indies. Journal of the Geological Society, London 130, 341^366. Roobol, M. J. & Smith, A. L. (1998). Pyroclastic stratigraphy of the Soufrie're Hills volcano, MontserratçImplications for the present eruption. Geophysical Research Letters 25, 3393^3396. Schmeling, H., Kruse, J. P. & Richard, G. (2012). Effective shear and bulk viscosity of partially molten rock based on elastic moduli theory of a fluid filled poroelastic medium. Geophysical Journal International 190, 1571^1578. Searle, M. P. (1999). Emplacement of Himalayan leucogranites by magma injection along giant sill complexes: examples from the Cho Oyu, Gyachung Kang and Everest leucogranites (Nepal Himalaya). Journal of Asian Earth Sciences 17, 773^783. Searle, M. P. & Godin, L. (2003). The SouthTibetan Detachment and the Manaslu Leucogranite: A structural reinterpretation and restoration of the Annapurna^Manaslu Himalaya, Nepal. Journal of Geology 111, 505^523. Smith, A. L., Roobol, M. J., Schellekens, J. H. & Mattioli, G. S. (2007). Prehistoric stratigraphy of the Soufrie're Hills^South Soufrie're Hills volcanic complex, Montserrat, West Indies. Journal of Geology 115, 115^127. Solano, J. M. S., Jackson, M. D., Sparks, R. S. J., Blundy, J. D. & Annen, C. (2012). Melt segregation in deep crustal hot zones: a mechanism for chemical differentiation, crustal assimilation and the formation of evolved magmas. Journal of Petrology 53, 1999^2026. Sparks, R. S. J. & Young, S. R. (2002). The eruption of Soufrie're Hills Volcano, Montserrat (1995^1999): overview of scientific results. In: Druitt, T. H. & Kokelaar, B. P. (eds) The Eruption of Soufrie' re Hills Volcano, Montserrat from 1995 to 1999. Geological Society, London, Memoirs 21, 45^70. Sparks, R. S. J., Young, S. R., Barclay, J., Calder, E. S., Cole, P., Darroux, B., Davies, M., Druitt, T. H., Harford, C., Herd, R., 547 JOURNAL OF PETROLOGY VOLUME 55 James, M., Lejeune, A. M., Loughlin, S., Norton, G., Skerrit, G., Stasiuk, M. V., Stevens, N. S., Toothill, J., Wadge, G. & Watts, R. (1998). Magma production and growth of the lava dome of the Soufrie're Hills volcano, Montserrat, West Indies: November 1995 to December 1997. Geophysical Research Letters 25, 3421^3424. Sparks, R. S. J., Folkes, C. B., Humphreys, M. C. S., Barfod, D. N., Clavero, J., Sunagua, M. C., McNutt, S. R. & Pritchard, M. E. (2008). Uturuncu volcano, Bolivia: Volcanic unrest due to mid-crustal magma intrusion. AmericanJournal of Science 308, 727^769. Svensen, H., Corfu, F., Polteau, S., Hammer, . & Planke, S. (2012). Rapid magma emplacement in the Karoo Large Igneous Province. Earth and Planetary Science Letters 325^326, 1^9. Tai Te, W. (1966). The effect of inclusion shape on the elastic moduli of a two-phase material. International Journal of Solids and Structures 2, 1^8. Trofimovs, J., Talling, P. J., Fisher, J. K., Sparks, R. S. J., Watt, S. F. L., Hart, M. B., Smart, C. W., Le Friant, A., Cassidy, M., Moreton, S. G. & Leng, M. J. (2013). Timing, origin and emplacement dynamics of mass flows offshore of SE Montserrat in the last 110 ka: Implications for landslide and tsunami hazards, eruption history, and volcanic island evolution. Geochemistry, Geophysics, Geosystems 14, 385^406. Vigneresse, J. L., Tikoff, B. & Ameglio, L. (1999). Modification of the regional stress field by magma intrusion and formation of tabular granitic plutons. Tectonophysics 302, 203^224. Voight, B., Widiwijayanti, C., Mattioli, G., Elsworth, D., Hidayat, D. & Strutt, M. (2010). Magma-sponge hypothesis and stratovolcanoes: Case for a compressible reservoir and quasi-steady deep influx at Soufrie're Hills Volcano, Montserrat. Geophysical Research Letters 37, doi:10.1029/2009gl041732. NUMBER 3 MARCH 2014 Wadge, G. & Isaacs, M. C. (1988). Mapping the volcanic hazards from the Soufrie're Hills volcano, Montserrat, West Indies, using an image processor. Journal of the Geological Society, London 145, 541^551. Wadge, G., Voight, B., Sparks, R. S. J., Cole, P. & Loughlin, S. C. (eds) (2013). An Overview of the Eruption of Soufrie' re Hills Volcano from 2000 to 2010. Geological Society, London, Memoirs (in press). Whittington, A. G., Hofmeister, A. M. & Nabelek, P. I. (2009). Temperature-dependent thermal diffusivity of the Earth’s crust and implications for magmatism. Nature 458, 319^321. Yoshinobu, A. S., Okaya, D. A. & Paterson, S. R. (1998). Modeling the thermal evolution of fault-controlled magma emplacement models: implications for the solidification of granitoid plutons. Journal of Structural Geology 20, 1205^1218. Young, S. R., Sparks, R. S. J., Aspinall, W. P., Lynch, L. L., Miller, A. D., Robertson, R. E. A. & Shepherd, J. B. (1998). Overview of the eruption of Soufrie're Hills volcano, Montserrat, 18 July 1995 to December 1997. Geophysical Research Letters 25, 3389^3392. Zellmer, G. F. & Annen, C. (2008). An introduction to magma dynamics. In: Annen, C. & Zellmer, G. F. (eds) Dynamics of Crustal Magma Transfer, Storage and Differentiation. Geological Society, London, Special Publications 304, 1^13. Zellmer, G. F., Hawkesworth, C. J., Sparks, R. S. J., Thomas, L. E., Harford, C. L., Brewer, T. S. & Loughlin, S. C. (2003a). Geochemical evolution of the Soufrie're Hills volcano, Montserrat, Lesser Antilles volcanic arc. Journal of Petrology 44, 1349^1374. Zellmer, G. F., Sparks, R. S. J., Hawkesworth, C. J. & Wiedenbeck, M. (2003b). Magma emplacement and remobilization timescales beneath Montserrat: insights from Sr and Ba zonation in plagioclase phenocrysts. Journal of Petrology 44, 1413^1431. 548
© Copyright 2026 Paperzz