No description available.

Ocean Islands Basalts provide important windows into the compositional variations of the Earth's mantle, which in tum constrains models for mantle convection and evolution. The Canary Islands show contrasting styles of eruption and evolution of magmas in an ocean island setting. U-Th-Ra disequilibrium have been used to constrain rates and timescales of melt generation and differentiation beneath ocean islands, and to estimate the buoyancy flux, potential mantle temperature and the depth and degree of melting. The Canary Islands provide a rare opportunity to observe U-Th-Ra disequilibrium, because they are underlain by a region of low buoyancy flux, and were expected to show significant disequilibrium. Tenerife is underlain by numerous magma chambers, in which magmas have time to differentiate from basanites to phonolites, erupting to form large strato-volcano complexes. The fissure and small vent eruptions of unusually primitive basanites and alkali basalts from Lanzarote show little evidence of magma chambers, unless of substantial size and longevity at depth. The U-Th results indicate that lavas underwent rapid transport from the melt region. The historic and recent pre-historic eruptions (1824, 1730-36, Corona) from Lanzarote have some of the most primitive compositions found on oceanic islands with low SiO² contents (< 51 %), Mg numbers of 67-74 and high Cr and Ni contents. The rocks are restricted in Sr, Nd and Pb isotopes, being displaced from MORB towards the HIMU om field. The major and trace elements have been modelled by mixing a deep smaller degree (1 %) melt and a shallower larger degree (4%) melt. Negative K anomalies were observed in the small degree melts indicating that melt generation may have continued at a shallow level, perhaps to within the lithospheric mantle with melting in the presence of residual phlogopite. The Lanzarote source was modelled as a mixture of HIMU and EMIl asthenospheric mantle, with a small contribution from a shallow, lithospheric source. Thermal erosion of the lithospheric mantle is required for melting at depths (58 and 73 km) modelled from the major and trace elements. The Lanzarote lavas exhibit significant (²³⁰Th/²³⁸U) disequilibrium with ²³⁰Th excesses of 6 - 81 %. This was modelled by dynamic melting giving a calculated melt rate of 0.125 x 10^-3 kg.m^-1.yr^-1, a timescale of melt generation (matrix transfer time) of 270 ka for the 1 % melt and 1,100 ka for the 4 % melt. A consistent upwelling rate of I cm.yr^-1 and an assumption that the melting process has remained consistent over tens of km at depth. The Teide-Pico Viejo complex lavas have undergone fractionation and mixing to form compositions from basanite to phonolite. Crystallising phases differ in the Pico Viejo series, where amphibole is dominant in the more evolved lavas, and Pico Teide series, where olivine in the major control. The more evolved lavas require assimilation and fractional crystallisation to explain the range in ⁸⁷Sr/⁸⁶Sr. (²³⁰Th/²³⁸U) ranges from 1.004-1.39 and gives information regarding the timescales of differentiation within the magma chambers, not least because the youngest mafic rocks have the highest (²³⁰Th/²³⁸U) and the most evolved phonolites have the lowest. The timescale of differentiation from basanite to phonolite is of the order of 150,000 years, which links to the periodicity of the eruption cycles on the island. A Ra-Th 'pseudo' whole rock isochron gave an age of fractionation for the Montafia Blanca eruption of 2.3 ka ± 80, which is a maximum of 300 years prior to eruption, indicating that fractionation of plagioclase as a possible trigger of an eruption.

Studies of mid-ocean ridges have provided evidence of magma chambers beneath both fast and intermediate spreading ridges. However no such features have been observed to date beneath slow spreading ridges. These contradictory observations are in direct conflict with seismic studies which reveal that the resulting crustal structures are similar and hence crustal structure is independent of the spreading rate. These latter observations in turn lead to the implication that the accretionary processes operating at all ridge types must also be similar. The aim of this study is to attempt to resolve between this discrepancy in geophysical observations of magma chambers at fast, intermediate and slow spreading ridges and investigate the nature of accretionary processes operating such that the same crustal structure is achieved. Therefore an apparently currently magmatically active section of the slow spreading Mid-Atlantic Ridge at 57 45'N on the Reykjanes Ridge, was selected as the target of a multidisciplinary geophysical experiment to be conducted aboard the RRS Charles Darwin in 1993. Wide-angle seismic data recorded using 10 digital ocean bottom seismometers were used to generate models of the crustal structure along and across-axis. These models were confirmed and further constrained by modelling of normal incidence seismic and gravity data and by comparison with the results of modelling controlled source electromagnetic data. The resultant models indicate that a magma chamber exists beneath the axial volcanic ridge studied, providing the first geophysical observation of such a feature at any slow spreading ridge. This magma chamber is similar in dimensions to those observed beneath fast and intermediate spreading ridges and consists of a thin, narrow sill-like body which appears to be continuous along-axis, and which is underlain by a region of partial melt extending almost to the Moho. This latter feature also appears to be both longer-lived and more extensive than the magma chamber. The 2.5 km depth to the top of the magma chamber is only slightly greater than that observed at fast spreading ridges, which indicates that magma chamber depth does not vary significantly with spreading rate. However, there ore insufficient data available to fully constrain and develop this relationship to its fullest. Therefore the results of this study indicate that the processes of crustal accretion occurring at all spreading ridges are similar, with the lack of observations of magma chambers being due to the fact that the periods of magmatic activity at slow spreading ridges are considerably more widely separated in both space and time than for fast and intermediate spreading ridges. The main difference however, appears to occur in the process of emplacement of layer 2A, which is observed to thicken off-axis at fast spreading ridges due to the less viscous lavas produced at these ridges being able to flow further off-axis. The results of this study, and of two other studies at slow spreading ridges, show that layer 2A is completely formed on-axis and thins off-axis due to extensional faulting. The remainder of the crust is completely emplaced, and the Moho formed, on-axis at all spreading rates.

Solidified igneous intrusions from originally liquid magma chambers display a large number of different sedimentary features. These features include the gravitational collapse of sidewalls producing slumps and the layering produced by gravitational settling of crystals. In the chamber fluid-dynamic processes such as convection are expected to occur due to cooling at the roof producing dense gravitationally unstable liquid, and the crystallisation of interstitial liquid changing the composition of the remaining liquid possibly reducing the density causing the liquid to rise up. The crystals which form in basaltic magma chambers have a high propensity to be mobilised due to convection and other fluid-dynamic processes including replenishment by a secondary intrusion. Convective mobilisation of plagioclase grains in vertical, tabular intrusions is seen from flat profiles of apparent aspect ratio as a function of dyke width. These flat profiles were formed due to scouring of gravitationally unstable sidewall mushes, and these crystals then become entrained in the convecting liquid. Convection only ceases once the volume of crystals in suspension reaches a critical volume fraction leading to an increase in viscosity, which dampens the vigour of convection. The majority of this study is performing and analysing a number of different experiments to look at the behaviour of different styles of analogue particle piles. Particle piles that are formed of inert, plastic particles are subjected to convection in the particle layer and in the bulk overlying fluid, and different styles of mobilisation depending on the heat flux driving convection and the density profile of the pile are observed. The mobilisation style goes from rolling of particles on the surface, to puffs of particles from the surface being lofted into the interior, followed by large particle fountains and then the entire particle pile being completely disaggregated and lofted into the interior of the chamber as the force driving convection is increased. The initiation of mobilisation can be explained by the fluidisation of a particle pile, whilst the high degrees of mobilisation seen in some high Rayleigh number regimes can be explaining by resuspending particles. In experiments where particle piles have a positive density profile (dense particles overlying low density particles) the underlying low density particles can break through the overlying layer in particle fountains and can be explained by a modified fluidisation parameter. These experiments lack the reactivity and cohesion that realistic crystal piles would have. To try and quantify this, I have also performed a series of experiments looking at the rheology of an ice-sucrose suspension, where ice crystals can sinter and aggregate together. Under sheared conditions the forces required to disaggregate ice aggregates can be calculated, with the viscosity of an ice-sucrose suspension being described by a power-law relationship of shear rate and crystal radius. The particle pile experiments show that mobilisation of equivalent crystal piles in magma chambers should be readily observed. As it is not observed, except in replenished magmatic systems, this suggests that the additional forces coming from cohesion and aggregation in crystal piles prevent mobilisation of magmatic crystals. The replenishment by secondary intrusions can lead to forces which overcome the strength of the pile.

No description available.

Les grandes éruptions caldériques sont parmi les phénomènes les plus destructeurs de la Terre, mais les processus à l’origine des grands réservoirs de magma siliceux et pauvre en cristaux qui alimentent ces éruptions ne sont pas bien compris. Le temps de stockage de ces réservoirs dans la croûte supérieure a un intérêt particulier. De longs temps de stockage—jusqu’à 105 ans—ont été estimés en utilisant les temps de repos entre les éruptions et les âges radiométriques des cristaux qui se trouvent dans les produits éruptifs. Par contre, des travaux récents sur la diffusion dans des cristaux suggèrent que les réservoirs qui alimentent même les plus grandes éruptions peuvent se mettre en place pendant une période beaucoup plus courte—101–102 ans. Afin de répondre à cette question, j’ai étudié l’éruption dacitique de Cape Riva de Santorin, Grèce (>10km3, 22 ka). Pendant les 18.000 ans précédant cette éruption, une série de dômes et de coulées dacitiques a été émise, alternant avec des dépôts de ponce dacitique (le complexe de dômes de Therasia). Ces dacites ont des compositions similaires à celle qui a été émise pendant l’éruption de Cape Riva, et ont été décrites précédemment comme des « fuites » provenant du réservoir de Cape Riva pendant sa croissance. Cependant, le magma de Cape Riva est appauvri en éléments incompatibles (tels que K, Zr, La, Ce) par rapport au magma de Therasia, une différence qui apparaît également dans les cristaux de plagioclase. Cette différence ne peut pas être expliquée par des processus peu profonds, tels que la cristallisation fractionnée ou l’assimilation de la croûte, ce qui suggère que les magmas de Cape Riva et Therasia ont des origines différentes. En outre, il existe des arguments tendant à montrer que les dacites de Therasia n’ont pas été alimentées par un réservoir majoritairement liquide ayant eu une longue durée de vie. Il y a des variations non systématiques dans la composition du magma, les compostions des bords ainsi que les caractéristiques des cristaux de plagioclase tout au long de la séquence. De plus, les temps de résidence à haute température des cristaux de plagioclase et d’orthopyroxène estimés par des modèles de diffusion sont 101–102 ans. Ces temps sont courts par rapport au temps moyen entre éruptions (1.500 ans), ce qui suggère que les cristaux observés dans chaque coulée ne se sont formés que peu de temps avant l’éruption. Les différentes teneurs en éléments incompatibles indiquent qu’un nouveau magma s’est mis en place dans le système volcanique superficiel peu de temps avant l’éruption de Cape Riva. Cet apport de magma a eu lieu après la dernière éruption de Therasia, qui s’est produite <2.800±1.400 ans avant l’éruption de Cape Riva selon les âges 40Ar/39Ar. Les périphéries des cristaux de plagioclase présents dans la dacite de Cape Riva sont en équilibre avec une rhyodacite, avec une composition similaire à celui du verre de l’éruption. Cependant, les zonations dans les éléments majeurs et traces enregistrent des changements dans la composition du liquide magmatique pendant la croissance des cristaux. La composition du centre de la plupart des cristaux de plagioclase est la même que celle des bords ; toutefois ces cristaux sont souvent partiellement résorbés, et la croissance a repris avec du plagioclase plus calcique. Ces cycles se répètent jusqu’à trois fois. La relation étroite entre la teneur en anorthite, Sr et Ti des différentes zones suggère que la composition des plagioclases est corrélé avec la composition du liquide, allant de liquides dacitiques à rhyodacitiques. Des cristaux d’orthopyroxène révèlent une séquence similaire. Les motifs de zonation sont interprétés comme un témoin de la formation du réservoir de Cape Riva dans la croûte supérieure par le mélange de plusieurs magmas ayant des compositions diverses. Des modèles de diffusion de Mg dans le plagioclase et de Fe–Mg dans l’orthopyroxène suggèrent que ce mélange a eu lieu 101–102 ans avant l’éruption. / Large, explosive, caldera-forming eruptions are amongst the most destructive phenomena on the planet, but the processes that allow the large bodies of crystal-poor silicic magma that feed them to assemble in the shallow crust are still poorly understood. Of particular interest is the timescales over which these reservoirs exist prior to eruption. Long storage times—up to 105 y—have previously been estimated using the repose times between eruptions and radiometric dating of crystals found within the eruptive products. However, more recent work modelling diffusion within single crystals has been used to argue that the reservoirs that feed even the largest eruptions are assembled over much shorter periods—101–102 y. In order to address this question, I studied the >10km3, 22-ka, dacitic Cape Riva eruption of Santorini, Greece. Over the 18 ky preceding the Cape Riva eruption a series of dacitic lava dome and coulées were erupted, and these lavas are interspersed with occasional dacitic pumice fall deposits (the Therasia dome complex). These dacites have similar major element contents to the dacite that was erupted during the Cape Riva eruption, and have previously been described as “precursory leaks” from the growing Cape Riva magma reservoir. However, the Cape Riva magma is depleted in incompatible elements (such as K, Zr, La, Ce) relative to the Therasia magma, as are the plagioclase crystals in the respective magmas. This difference cannot be explained using shallow processes such as fractional crystallisation or crustal assimilation, which suggests that the Cape Riva and Therasia magmas are separate batches. Furthermore, there is evidence that the Therasia dacites were not fed from a long-lived, melt-dominated reservoir. There are non-systematic variations in melt composition, plagioclase rim compositions, and plagioclase textures throughout the sequence. In addition, high-temperature residence times of plagioclase and orthopyroxene crystals from the Therasia dacites estimated using diffusion chronometry are 101–102 y. This is short compared to the average time between eruptions (1,500 y), which suggests the crystals in each lava grew only shortly before eruption. The different incompatible element contents of the Cape Riva and Therasia magmas and plagioclase crystals suggest that a new batch of incompatible-depleted silicic magma arrived in the shallow volcanic plumbing system shortly before the Cape Riva eruption. This influx must have taken place after the last Therasia eruption, which 40Ar/39Ar dates show occurred less than 2,800±1,400 years before the Cape Riva eruption. The rims of the plagioclase crystals found in the Cape Riva dacite are in equilibrium with a rhyodacite, with a similar composition to the Cape Riva glass. However, the major and trace element zoning patterns of the crystals record variations in the melt composition during their growth. The compositions at the centre of most crystals are the same as the rims; however, these crystals are often partially resorbed and overgrown by more calcic plagioclase. The plagioclase then grades normally back to rim compositions. This cycle is repeated up to three times. The tight relationships between the anorthite, Sr and Ti contents of the different zones suggests that the composition of the plagioclase crystals correlates with the composition of the melt from which theygrew. The different plagioclase compositions correspond to dacitic and rhyodacitic melt compositions. The orthopyroxene crystals reveal a similar sequence, although they only record one cycle. These zoning patterns are interpreted to document the assembly of the Cape Riva reservoir in the shallow crust through the amalgamation of multiple batches of compositionally diverse magma. Models of magnesium diffusion in plagioclase and Fe–Mg interdiffusion in orthopyroxene suggest that this amalgamation took place within 101–102 y of the Cape Riva eruption.

No description available.