V23G-2203
Do Plinian Eruptions of Mafic Magma Require Fast Ascent Rates?
Although rare, mafic magma is known to erupt explosively in Plinian fashion. Given that models for such eruptions often invoke high viscosities as pre-requisite, mafic magmas erupting explosively seems to pose a quandary. One possibility is that such magmas can erupt explosively, if they ascend towards the surface very fast, creating conditions that lead to explosive degassing and fragmentation. In order to estimate how fast mafic magma ascends in such eruptions, we are carrying out series of isothermal decompression experiments to examine groundmass textures in natural samples from such eruptions to infer ascent rates. One eruption we are examining is the 122 B.C. Plinian eruption of hawaiite from Mt Etna. Prior work suggests that this magma was stored at 1025°C and 75 MPa before erupting. We find that the groundmass has a total crystallinity of 58 vol.%, consisting of 34 vol.% pyroxene, 20 vol.% plagioclase, and 4 vol.% Fe- Ti oxides. The area number densities are on the order of 3-7 x 10-2μm-2. Thus far, we have been unable to mimic its groundmass textures with either the single-step or multi-step decompression experiments, in which pressure dropped from 75 MPa to 13 MPa at rates ranging from 0.18 to 0.001 MPa s -1. Although the slowest decompression produced the correct groundmass assemblage and total crystallinity it has twice the plagioclase and far too little pyroxene. In addition, area number densities are 1-2 orders of magnitude less. Additionally, hopper shaped plagioclase is observed in all but the slowest experiment, whereas the natural sample contains only tabular to acicular plagioclase. Our results suggest that the textures of the hawaiite scoria cannot be explained by rapid ascent alone. One possibility is that the magma stalled and slowly crystallized prior to eruption. The additional microlites would have increased the viscosity of the melt, allowing the mafic magma to erupt in a Plinian style. In order to further test the rapid ascent model for mafic Plinian eruptions, we will also perform decompression experiments on basaltic andesite from the Fontana eruption of Masyasa. Prior work suggests that this magma was colder and stored shallower than the Etna hawaiite, but interestingly the groundmass is far less crystalline with a wider range of microlite morphologies.
V23G-2204
Mathematical and Analogue Models of Fluid Filled Fracture Propagation in Layered Elastic Media
In order to overcome the practical impossibility of direct observation of dike propagation within the crust, we
develop mathematical and analogue models to describe the physical processes involved in their dynamics.
We focus our attention on what happens when a dike approaches the boundary between two media with
different rigidities.
In our 2D mathematical models, a dike opens and propagates in an infinite elastic medium, made up of 2
welded half-spaces with different elastic parameters.
Dikes are modelled as boundary element fluid-filled cracks. The pressure gradient along the crack is
proportional to the difference between the densities of the host rock and the fluid. We take into account the
compressibility of the fluid and a variable density in order to conserve the mass of the intrusion during its
motion. The mathematical model allows to set a tectonic stress field and an arbitrarily tilted boundary
separating different media. The growth, arrest and direction of propagation of the crack is governed by an
energetic criterion: the motion of the dike is driven by the minimization of the elastic deformation energy plus
the gravitational energy. Propagation is allowed when the energy release during the motion exceeds a
fracture threshold.
The output of this code gives us the path followed by the crack during the propagation, its shape and the
stresses
induced in the elastic medium.
Interestingly, the mathematical simulations provide a sort of refraction phenomenon, that is a sudden
change in direction of propagation when the crack crosses the boundary separating different rigidities.
In order to validate our mathematical results, we perform laboratory experiments of air filled cracks
propagating in gelatine. Gelatine represents well an elastic medium: it is brittle at refrigerator temperature
and varying the concentration of dry gel powder dissolved in water we can control its rigidity. By injecting air
from the bottom of a trasparent cylinder containing the gelatine, we obtain an air filled crack, tilted with
respect to the vertical and propagating upwards toward the rigidity transition surface.
The experiments confirm the main characteristics of the mathematical simulations.
V23G-2205
Classifying AE spectra from deformation-decompression laboratory experiments on Etna basalt
Recent laboratory experiments (Benson et al., 2008) on Etna basalt have permitted the generation of an extensive catalogue of acoustic emissions (AE) during two key experimental phases. Firstly, AE have been generated during triaxial deformation and formation of a complex fracture damage zone. Secondly, rapid fluid decompression through the damage/shear zone after failure. Analysis of first motion and spectrograms allows to qualitatively identify high and low frequency events, which we compare to VT and LP in volcanic areas, from the two experimental stages respectively. Here, we present a new and independent analysis, aiming to quantitatively classify 'families' of AE events belonging to the same experimental stage without prior knowledge. We then test this method using the AE catalogue for verification, which is not possible with field data. FFT spectra, obtained from AE, are subdivided into equal log intervals for which a local slope is calculated. Factor analysis has been then applied to experiment results, where we use a data matrix of columns representing the variables considered (frequency range slopes) vs. rows indicating each AE data set. Factor analysis shows the method to be very effective and suitable for reducing data complexity, allowing distinct factors to be obtained. We conclude that most of the data variance (information content) can be well represented by two factors only, each one representing a well defined frequency range. Having retained two factors we can plot our AE data on a plane. Classification is then possible by identifying clusters of AE belonging to the same experimental stage. This allows us to propose a deformation/decompression interpretation based solely on the AE frequency analysis.
V23G-2206
Time-dependent Brittle Deformation in Etna Basalt
Mt Etna is the largest and most active volcano in Europe. Due to the high permeability of its volcanic rocks, the volcanic edifice hosts one of the biggest hydrogeologic reservoirs of Sicily (Ogniben, 1966). Pre-eruptive patterns of flank eruptions, closely monitored by means of ground deformation and seismicity, revealed the slow development of fracture systems at different altitudes, marked by repeated bursts of seismicity and accelerating/decelerating deformation patterns acting over the scale of months to days. The presence of a fluid phase in cracks within rock has been shown to dramatically affect both mechanical and chemical interactions. Chemically, it promotes time-dependent brittle deformation through such mechanisms as stress corrosion cracking that allows rocks to deform at stresses far below their short-term failure strength. Such crack growth is highly non-linear and accelerates towards dynamic failure over extended periods of time, even under constant applied stress; a phenomenon known as 'brittle creep'. Stress corrosion is considered to be responsible for the acceleratory cracking and seismicity prior to volcanic eruptions and is invoked as an important mechanism in forecasting models. Here we report results from a study of time-dependent brittle creep in water-saturated samples of Etna basalt (EB) under triaxial stress conditions (confining pressure of 50 MPa and pore fluid pressure of 20 MPa). Samples of EB were loaded at a constant strain rate of 10-5 s-1 to a pre-determined percentage of the short- term strength and left to deform under constant stress until failure. Crack damage evolution was monitored throughout each experiment by measuring the independent damage proxies of axial strain, pore volume change and output of acoustic emission (AE) energy, during brittle creep of creep strain rates ranging over four orders of magnitude. Our data demonstrate that the applied differential stress exerts a crucial influence on both time-to-failure and creep strain rate in EB. Stress-stepping creep experiments were then performed to allow the influence of the effective confining stress to be studied in detail. Experiments were performed under effective stress conditions of 10, 30 and 50 MPa (whilst maintaining a constant pore fluid pressure of 20 MPa). In addition to the purely mechanical influence of water, governed by the effective stress, which results in a shift of the creep strain rate curves to lower strain rates at higher effective stresses. Our results also demonstrate that the chemically-driven process of stress corrosion cracking appears to be inhibited at higher effective stress. This results in an increase in the gradient of the creep strain rate curves with increasing effective stress. We suggest that the most likely cause of this change is a decrease in water mobility due to a reduction in crack aperture and an increase in water viscosity at higher pressure. Finally, we show that a theoretical model based on mean-field damage mechanics creep laws is able to reproduce the experimental strain-time relations. Our results indicate that the local changes in the stress field and fluid circulation can have a profound impact in the time- to-failure properties of the basaltic volcanic pile.
V23G-2207
Anisotropy of Magnetic Susceptibility of a Magma Flow Sheet in the Trachyte Mesa Laccolith of the Henry Mountains in South-Central Utah
The Trachyte Mesa Laccolith (TML) in the Henry Mountains of south-central Utah is thought to have been formed by the coalescing of numerous horizontal magma sheets that stemmed from the nearby igneous intrusion, Mount Hillers. To understand the flow behavior within a magma sheet, cores (ranging in length from 60 cm to 1.2 m) were cut through the top sheet of the laccolith with a construction grade drill. Individual samples were then cut every 5 cm down the core to obtain a downward gradient of foliation. These samples were cut parallel to lineation and were used for fabric analysis and the collection of Anisotropy of Magnetic Susceptibility (AMS) data. To determine average crystal percentage versus matrix in the sheet, thin sections were made from two samples from the top of the sheet. Each sample was divided into rows and columns and a microphotograph was taken for each section of the grid. The area calculation and crystal analysis of the microphotographs was done using NIH software. The software allowed each plagioclase and oxide crystal to be colored in and deleted from the total area which provided the crystal percentage of each photograph. An average percentage for each row of the thin sections was calculated and determined to be approximately 35 to 40%. In both samples, one row, several cm from the top of the sheet, deviates by 22% less than the average crystal percentage. The row lying directly underneath shows a 22% increase in crystallinity. These changes in crystal content are possibly due to grain dispersive forces that cause crystals to be displaced away from the contact shear zone. This is known as the Bagnold effect and helps to define the plug zone within the sheet where no shearing in occurring. Analysis of the AMS data revealed the bulk magnetic susceptibility (Km) to range from 2.71E-3 to 1.54E-2 SI with the average being 7.69E-3 SI. The Km values appear to be changing at approximately 40 cm down. At this same level the AMS foliation begins to curve from a northwesterly dip to a southeasterly dip in each of the core samples drilled. These data suggest there is a magma sheet contact at this depth. All evidence collected suggests the magma sheet was emplaced as a plug flow. This plug flow behavior is supported by the displacement of the crystal percentages, the curvature in the AMS foliation, and a change in the bulk magnetic susceptibility that corresponds to the change in foliation curvature.
V23G-2208
Absolute healing of pyroclasts during welding of a rheomorphic ignimbrite
The architectural description of ignimbrites often shows evidence for post-deposition development of a rheomorphic, ductile shear zone - a feature which may strongly affect the progression of the flow. Rheological experiments were performed on a welded rheomorphic unit from the Grey's Landing ignimbrite in the Snake River Plain to characterize its behaviour and assess the degree of welding. The investigated sample contains 5 vol.% open pores and is made of approximately 5 vol.% crystals bathing in a relatively degassed, peraluminous glass containing 79 wt.% SiO2. Pre-eruptive temperature determination from geothermometry on pyroxenes yielded values at around 900-1050 ° C. Dilatometric measurements suggest a calorimetric glass transition temperature during deposition of approximately 845 ° C and a H2O content of approximately 0.04 wt.%. Repeated series of heating and cooling using an advanced dilatometric technique shows an increase of the glass transition temperature to 880 ° C, which is in accordance with degassing of approximately 0.02 wt.% H2O. Complementary investigation using a uniaxial press revealed an absence of strain rate dependence of the viscosity (1010.78 Pa*s) at a temperature of 900 ° C and at strain rates up to 2.5 x 10-4 s-1. Under similar conditions, a fully degassed lava with an equivalent composition would yield a comparable viscosity of 1010.89 Pa*s. Our findings may help constrain the flare up of the Grey's Landing ignimbrite. The presence of small amounts of water in the glass and the narrow temperature window between the residence in the reservoir and the transition to a glass (which would have mechanically locked this unit in place) in the flow indicates a high discharge rate and rapid post-fragmentation deposition, mass agglutination and welding. Moreover, the Newtonian character of this welded unit suggests that healing of the pyroclastic flow was absolute (that is, no thixotropic effects from the pores remain), and thus that the term 'lava-like' is adequate to rheologically describe rheomorphic pyroclastic flows.
V23G-2209
Textural evidence for strain partitioning in highly vesicular volcanic rocks
The rheology of multi-phase systems is inherently complex; these complexities make it difficult to constrain the rheology of magma (melt, +/-bubbles, +/-crystals) in a volcanic conduit. Shear textures in silicic pumice samples provide an entry point into understanding the role of shear in explosive volcanic eruptions. We show examples of shear localization in crystal-rich silicic pumice from large caldera-forming eruptions and in crystal-poor pumice from small pumice cone eruptions and discuss the rheological implications of this behavior. In crystal-rich clasts, the presence of crystals in magma has focused shear deformation into distinct shear zones. We document S-C fabrics in pumice clasts using deformed vesicle textures around crystals. Trails of elongated and aligned small vesicles are found in relatively crystal-free zones, whereas large, undeformed vesicles are present within strain shadows adjacent to phenocrysts. These shear thinning regions are similar to those described in experimental deformation of nonvesicular, highly crystalline magmas under high strain rates. In experiments, flow behavior transitions from Newtonian to non-Newtonian to Binghamian with increasing shear rate. While nonvesicular experiments show that shear thinning is more pronounced for low crystal fractions, due to the decreased likelihood of crystal-crystal interactions, our observations suggest that the expanded structure of highly vesicular magma allows for strong shear-thinning behavior in highly crystalline (>45% on a vesicle-free basis) material as well. In crystal-poor clasts, zonal variations in the degree of shear are also present. Similar to experimental deformation of liquid foams, vesicle-rich pumice records regions of rigid behavior, separated by regions of liquid behavior. These shear zones are discontinuous and are oriented in multiple directions. We suggest that jamming regions in crystal-poor pumice samples may, like their crystal-rich counterparts discussed above, represent non-Newtonian, shear-thinning behavior in the conduit. The textures preserved in crystal-rich and crystal-poor pumice samples provide evidence of strain partitioning behavior in both small and large explosive silicic eruptions.
V23G-2210
Forward solutions giving displacement and stress fields around irregularly opened rock fracture, and their extension to inverse modeling
Many rock fractures are partly or entirely driven open by fluids such as ground water, geothermal water, gas, oil, and magma. These include dykes, mineral veins, and many faults and joints. Field measurements show great variations in aperture. Such variations may result in flow channeling, and are of fundamental importance for understanding fluid transport in the crust, particularly in fractured reservoirs, seismogenic fault zones, and volcanoes. Most analytical solutions for fracture displacement and stress fields assume the loading to be either constant or with a linear variation. While these solutions have been widely used, it is clear that a fracture hosted by a heterogeneous and anisotropic rock is normally subject to loading that is neither constant nor with a linear variation. Here, we first present new general solutions as forward model for the displacement fields and stress fields around rock fractures opened by irregular overpressure variations given by the Fourier cosine series. Next, by changing form of the equation of the forward model, we show the matrix equation that can estimate Fourier coefficients giving the overpressure distribution from crack configuration. The overpressure distribution can be derived by inverse Fourier transform. In numerical examples in forward model, it was shown that the displacement and stress fields close to the crack followed the overpressure variation but tend to be more uniform far from the crack. The solutions can be used to estimate the displacement and stress fields around any fluid-driven crack. As simple examples of inverse analysis, we attempted to estimate the overpressure conditions from crack configurations given analytically by conditions that 1) overpressure in the crack is constant and that 2) overpressure variation in the crack can be described by linear function. As a result, it was found that the overpressure distributions explaining the crack configurations were correct by comparing the estimated overpressures with the assumed overpressure condition, although small noises were included a little in the estimated overpressures. The solutions add to our understanding of local stresses, displacements, and fluid transport associated with fractures in the crust.
V23G-2211
Oxidation State Dependence of Viscosity in Peralkaline Lavas: Phonolite and Pantellerite.
Although the dependence of the shear viscosity of silicate melts on their oxidation state has been demonstrated for decades, the effects on melt and magma viscosity in nature remain obscure. The effect observed in very iron-rich silicates (acmite, hedenbergite and related compositions) has been observed to be moderate but measurable. Simple scaling on the basis of Fe-content from synthetic to natural liquids led to the preliminary conclusion that redox effects on the viscosity of natural liquids would not be significant for modelling porposes. Based on that premise, recent multicomponent models of melt viscosity such as the GriD model (Giordano, Russell, Dingwell, EPSL, 2008) neglect redox effects. Tests the validity of neglecting oxidation state, based on assigning oxidation states to the experimental basis of the GriD model have failed to result in an improvement in model preformance. Systematic investigation of a redox effect in natural melts provides a potential alternative to improving the model as well as providing insight into structure-property relationships in these Iron-bearing melts. Wish this in mind we have initiated such an experimental campaign. Here we report and compare the viscosity-oxidation state relationships of two natural melt compositions with ca. 8.5% iron oxide, a phonolite (ca. 50% SiO2) and a pantellerite (ca. 70% SiO2). Both melts have been investigated using concentric cylinder techniques under gas-mixing oxygen fugacity control as well as micropenetration viscometry on melts of variable oxidation state obtained from the quenched glasses. The oxidation state of the melts covers the oxidized half of the range of ferric/total iron. Viscosity decreases significantly with melt reduction. The magnitude of the effect is much larger than would be anticipated by linearly scaling the results from synthetis melts on the basis of Fe content. The magnitude of the effect increases with decreasing temperature. The results indicate clearly that improved models for melt viscosity must include a redox parameter. We will discuss the results in terms of possible structural origins of the effect.
V23G-2212
Brittleness and shear thinning: the explosive-effusive transition of lava dome experimentally investigated.
Dome-building eruptions repeatedly alternate between effusive growth and catastrophic failure, the latter of which sometimes generates pyroclastic flows. The rheology of highly crystallized dome lavas is a key to understanding their emplacement and possible failure. In this study we investigate experimentally the stress and strain-rate dependence of Mt Unzen type dome lavas. Their rheology has been determined at temperatures of 900 to 1010°C and under compressional stresses ranging from 2 to 60 MPa. Additionally, tensile tests were performed on volcanic products to characterize the effects of porosity on their tensile strength. Two kinds of non-Newtonian behavior are observed for highly crystalline melts. The first is a shear thinning effect observed through an instantaneous decrease of the viscosity with each applied stress increment which is recoverable during stress release. Shear thinning is typical for crystal-bearing melts. Under low applied stresses (e.g., 2 MPa), the apparent viscosity is high and can be estimated by the Einstein-Roscoe equations which considers the crystal fraction and the interstitial melt viscosity. However with increasing stress (or strain rate), the apparent viscosity may decrease to approximately that of the crystal-free melt. At very high applied stresses, viscous heating may lead to an additional viscosity decrease. Our observations confirm that the crystalline phases respond elastically to the applied stress and relaxes once the load is withdrawn. The second effect is time-dependent and non-recoverable. The time-weakening effect appears to be complex and exhibits hysteresis. It depends on the stress and/or strain-rate history. We distinguish four different domains: ductile with a Newtonian rheology, ductile with a non-Newtonian rheology, partially brittle with sporadic crack propagation and brittle leading to complete failure. Each of these domains expresses itself as a different regime of viscosity decrease. The presence of crystals in multi-phase melts allows stress localization, thus cracking may initiate in the phenocrysts and ground mass melts at strain rates lower than the compositionally equivalent, crystal-free melt. Through Brazil tests we find that once the tensile strength is overcome, weakening of the material ensues and leads to the onset of the time-weakening effect. Moreover, the strength decreases with porosity as is observed for threshold values of fragmentation by decompression. This study demonstrates the dominance of non-Newtonian rheology in Unzen's type lavas which may help explain the cycles of dome extrusion and destruction.
V23G-2213
High Temperature Strain-Rate Dependent Rheology of Strombolian Magmas
Volcanism at Stromboli generates a variety of products that need to be fully characterized in terms of their
rheological behaviour mainly according to phase distribution, volatile content and stress-strain regimes.
These eruptive products, similar in overall chemical composition (HK-shoshonitic basalts), strongly differ in
terms of crystallinity and vesicularity (from dark holocrystalline scoria to aphiric golden pumice).
In order to quantify the effect of crystals on the rheology of basaltic magma from Stromboli, we have
investigated the pure liquid viscosity and the strain-rate dependent liquid+crystal rheology of remelted
products of nearly aphiric pumices of the 15th March 2007 paroxysmal event.
The high T viscosities of fully molten (1225° C
V23G-2214
Permeable Gas Flow Influences Magma Fragmentation Speed.
Highly viscous magmas undergo fragmentation in order to produce the pyroclastic deposits that we observe, but the mechanisms involved remain unclear. The overpressure required to initiate fragmentation depends on a number of physical parameters, such as the magma's vesicularity, permeability, tensile strength and textural properties. It is clear that these same parameters control also the speed at which a fragmentation front travels through magma when fragmentation occurs. Recent mathematical models of fragmentation processes consider most of these factors, but permeable gas flow has not yet been included in these models. However, it has been shown that permeable gas flow through a porous rock during a sudden decompression event increases the fragmentation threshold. Fragmentation experiments on natural samples from Bezymianny (Russia), Colima (Mexico), Krakatau (Indonesia) and Augustine (USA) volcanoes confirm these results and suggest in addition that high permeable flow rates may increase the speed of fragmentation. Permeability from the investigated samples ranges from as low as 5 x 10-14 to higher than 9 x 10- 12 m2 and open porosity ranges from 16 % to 48 %. Experiments were performed for each sample series at applied pressures up to 35 MPa. Our results indicate that the rate of increase of fragmentation speed is higher when the permeability is above 10-12 m2. We confirm that it is necessary to include the influence of permeable flow on fragmentation dynamics.
V23G-2215
Field And Experimental Constraints on the Rheology of Silicic Magma
The microphysics governing the rheology of crystal-rich silicic magma remains an open question. We conduct detailed field measurements of the Pleasant Bay layered gabbro-diorite intrusion (Coastal Maine, USA), which records the temporal evolution of a silicic crystal-liquid mush periodically injected with mafic material (Wiebe 1993). Comparison of field measurements of kinematic indicators including crystal alignments and textural gradients around the margins of stoped basaltic blocks, as well as at interfaces between basaltic and silicic layers, with scaling analyses drawn from thin viscous sheet theory suggest that the silicic magma had a viscoplastic rheology with a yield stress. A key outstanding issue, however, is the relationship of this macroscopic rheology to the microphysics governing the deformation of the crystal-rich magma. In particular, how the microtextures we observe in plutonic rocks can be related quantitatively to the coupled crystal-liquid dynamics that ultimately govern the rheological response of the magma is unclear. To understand how a magma's microstructure might evolve to produce the textures we observe in the field, we perform a series of quasi-two-dimensional lab experiments on the responses of mono- and poly-disperse mixtures of plastic particles (analogue crystals) and viscous liquid to an imposed simple shear using a rotating cylinder-Couette setup under laminar flow conditions. We aim to understand phenomena such as particle alignment and particle migration in terms of the three-way (solid-liquid, liquid-solid, solid-solid) solid-fluid coupling in our experiments. We use particle image velocimetry and video analysis to characterize the microstructural response to imposed shear for a broad range of strain rates, particle concentrations and shape distributions. Stress-strain rate curves are estimated and the corresponding rheologies are applied in a simple numerical model to understand measured velocity profiles.
V23G-2216
Field and Experimental Constraints on the Dynamics of Replenished Silicic Magma Chambers
The underlying causes of catastrophic caldera-forming volcanic eruptions remain poorly understood. However, the occurrence of magma mixing within bimodal systems has become increasingly linked with such eruptions. In particular, buoyancy effects related to unstable density contrasts arising as a result of silicic- basaltic magma interactions may play an important role in the growth, differentiation and catastrophic eruption of silicic magma chambers. Evidence of such magmatic interactions can be found in layered intrusions from the Coastal Maine Magmatic Province (USA), where well-exposed cross-sections reveal hundreds of laterally-extensive basaltic sheets, apparently injected as intrusive lava flows onto the growing floors of silicic magma chambers. Interfaces between mafic and silicic layers are commonly sharply defined and exhibit deformation parallel to the inferred direction of palaeo-gravity. Our field observations suggest that the cooling, settling and buckling of gravitationally-unstable mafic replenishments may have driven large-scale (basalt layer depth) and small- scale (crystal diameter) upwelling and/or overturning of underlying buoyant silicic cumulate material. In order to characterize the full range of buoyancy effects, we carried out extensive spectral analysis of high- resolution digital field measurements from the Pleasant Bay and Mount Desert Island intrusions. In many cases, Rayleigh-Taylor theory and the longest measured wavelength of deformation indicate that a large and potentially-quantifiable fraction of the original, pre-replenishment silicic cumulate thickness may be missing, implying that vertical mass transfer has occurred. In addition, the shortest wavelengths of deformation are generally consistent with observed length-scales of crystals and clumps of crystals at these localities. With the aim of understanding the initial conditions that gave rise to these field observations, we conduct a series of laboratory experiments in which we observe the development of a Rayleigh-Taylor instability between a buoyant basal fluid layer overlain by a denser fluid layer. In order to identify the important parameters in the problem, we perform these experiments for a wide range of density contrasts, layer thicknesses and fluid rheologies (i.e. we vary particle concentration from the dilute to highly-crystalline limits to simulate freezing basalt and re-heated silicic cumulate rheologies). Regimes in which the entire silicic layer becomes unstable are potentially responsible for overturning the system and iniating large volcanic eruptions.
V23G-2217
The GRD Model for Silicate Melt Viscosity: Volcanological Applications
We recently published a model for predicting the non-Arrhenian Newtonian viscosity of silicate melts as a
function of temperature (T) and melt composition (X), including the volatile constituents H2O and F
(Giordano et al. 2008). The non-Arrhenian T-dependence is accounted for by the VFT equation [log η =
A + B/(T(K) -C)] and the model is calibrated on > 1750 measurements of melt viscosity. All compositional
dependence is accommodated by 17 model coefficients embedded in the parameters B and C. The
optimization assumes a common, high-T limit (A) for silicate melt viscosity and returns a value for this limit of -
4.55 (± 0.2) (e.g., log η ~ 10-4.6 Pa s) making for a total of 18 model coefficients. The
effects of pressure on the silicate melt viscosity are not accounted for in this model, however, the model has
the following attributes: a) it covers over fifteen log units of viscosity [10-1 to 1014 Pa s], b) it spans
most of the compositional range found in naturally-occurring volcanic rocks, c) it is computationally
continuous across the entire compositional and temperature spectrum of the database, and d) it is capable of
accommodating both strong (near-Arrhenian T-dependence) and fragile (non-Arrhenian T-dependence)
behaviour of silicate melts. Lastly, the model for melt viscosity can be used to predict other transport
properties including glass transition temperatures (Tg) and melt fragility (m). Volcanic regimes feature
constantly changing T-X melt conditions and, in many instances, these small changes generate strong non-
linear variations in melt viscosity. The GRD model allows for accurate, continuous prediction of melt
properties as a function of temperature and melt composition and, thus, is ideal for modelling transport
properties in dynamic natural systems. Below we demonstrate the utility of this model with three
volcanological applications: (A) We track variations in viscosity along liquid lines of descent predicted by
MELTS (Ghiorso et al. 1995) and discuss the overall implications for differentiation processes, (B) We
explore the rheological (η, m, and Tg) consequences of mixing between hot, dry mafic melts and cooler,
hydrous, felsic melts, whilst also accounting for issues of thermal equilibration and volatile saturation, and (C)
We couple the GRD model to that of Caricchi et al.'s (2007) model for viscosity of crystal-rich melts and
explore changes in magma rheology attending the 2007 eruption of Stromboli. These eruptions featured a
variety of eruption styles and produced nearly isochemical eruptive products that varied in crystallinity and
vesicularity (e.g., from dark holocrystalline scoria to aphyric pumice). Caricchi, L., Burlini, L., Ulmer, P.,
Gerya, T., Vassalli, M., Papale, P. (2007) Earth and Planetary Science Letters 264 : 402-419. Giordano, D.
Russell, J.K. and Dingwell, D.B. (2008) Earth and Planetary Science Letters 271: 123-134. Ghiorso, M.S.,
and Sack, R.O. (1995) Contributions to Mineralogy and Petrology 119: 197-212.
http://www.eos.ubc.ca/~krussell/VISCOSITY/grdViscosity.html
V23G-2218
Experimental constraints on the interaction between hydrous and anhydrous magmas at Stromboli volcano.
Triggers of eruptions or important changes in eruptive style have been often associated to the arrival of batches of new undegassed magma in the volcanic plumbing system. This process has been proved to occur in basaltic and in more evolved magma. What happen in a magmatic reservoir and how do melts interact when this occur has been theoretical inferred or modeled using silicate analogues, thermodynamics or numerically. But only few laboratory experiments on molten silicates exist. Analogues of magmas, as syrups or waxes allow scaling of some important parameters like density contrast, rheology, and relative proportion between volumes of the interacting melts, but some other important process like diffusion, crystallization and gas exsolution is overlooked. These processes can significantly affect the mobility of magmas and the extent of mixing. This paper reports main results of laboratory experiments carried at high pressure and temperature on natural melts aimed to reproduce interaction between hydrous and dry magmas with same compositions. A grounded basalt from Stromboli is used as starting material in order gain insights on the dynamics of magmas within the plumbing system. In fact the interaction between hydrous crystal-poor and dry crystal-rich magmas is recognized to trigger violent paroxysmal eruptions in this volcano. A dry glassy (or slightly crystallized) cylinder was superimposed to an hydrous grounded glass in a single Au- Pd capsule. Charges were held for between 15 minutes and 13 hours at about 1130°C , 50 Mpa in a EHPV and then rapidly quenched isobarically. Capsules were cut longitudinally and interactions between melts were observed by SEM, EMP and FTIR. Results evidence that in runs with shorter duration (<1 hour) water diffusion dominates at the interface and strongly controls liquidus temperature, crystallization and rheology. Experiments of longer duration show buoyant plumes or blobs of water-rich melts that rise and mix with the upstanding dry crystal-rich portion. Plumes generate convection, enhances mixing and control crystals distribution along the whole capsule. Experimental results demonstrate that interaction between magmas with different water and crystal content can be successfully reproduced using natural materials. Obtained information can be fully generalized to understand dynamics of magmas within a large number of volcanoes.