Volcanology, Geochemistry, Petrology [V]

 CC:718B  Wednesday  1030h

When Worlds Collide: From Magma Rheology to Rock Mechanics I

Presiding:  Y Lavallee, University of Munich; L A Kennedy, University of British Colombia; P M Benson, University of Toronto


Rheology and Strain Rates of Natural Systems

* White, J C (clancy@unb.ca), Dipartimento di Scienze della Terra, Università di Pisa, via Santa Maria, 53, Pisa, 56126, Italy
* White, J C (clancy@unb.ca), Department of Geology, University of New Brunswick, 2 Bailey Dr., Fredericton, NB E3B 5A3, Canada

The improvement of tools with which to both measure and infer strain rates in natural geological systems has enabled progressively more detailed integration of observed phenomena and time. There are several aspects of classical strain rate analysis that can skew such analyses. The latter revolve around the following: (1) strain rate is a normalized parameter in both space and time; its essence is to remove the effect of spatial and temporal scales by conflating the heterogeneity of response (strain) into a single time-independent descriptor; (2) such a descriptor is dependent on a point of reference e.g. our framework (external) or that of the deforming material (internal); differences between the two result from averaging out responses in the external framework. The need to consider scale, not necessarily small vs. large, but the length scale over which displacement of material actually occurs, arises from the tendency in nature for displacement to be partitioned; there will be an corresponding partition of the work done, energy stored elastically and, in turn, dissipated by permanent deformation. The rate of energy transfer in a heterogeneous system with varying stiffness amongst different components contributes importantly to the partitioning. Entwined with the externally imposed heterogeneity is the wide range of material states that can co-exist and/or evolve within a rock system, and which produce discontinuities in the rheological response; for example, abrupt transitions during ductile flow and ductile-brittle transitions under widely varying ambient conditions. By considering geological materials in terms of the different relaxation times of the controlling deformation processes, it is possible to dissect the similarity in response, often transient, among systems encompassing igneous and frictional melts, fluidized particulate flow, cataclasis and 'normal' rocks.


When melts start behaving like rocks: the chemical vs. the physical complexity of melt rheology.

* Dingwell, D B (dingwell@lmu.de), LMU - Uni Munich, Theresienstr. 41/III, Muenchen, 80333, Germany

The rheology of silicate melts requires a description that is, for the most part, chemically very complex but physically very simple. As such, the role of melts and the melt phase, in the deformation of multiphase magmas and partially melted rocks, is essentially the challenge of describing the chemistry of the homogeneous melt phase, or, failing that, a micro-description of the chemical variations present in the melt phase distributed throughout the system. These microchemical variations can be very significant for the deformation of the system. One of the most common must certainly be water concentration gradients generated during phase separation and equilibration such as bubble growth and crystallisation. Microanalytical advances hold out the hope of great improvements in this field in the coming years. The melt phase is however, capable of complex physical behavior under the extreme conditions of strain rate to which it is typically exposed, for example, in explosive volcansim.. In this regime the chemical and thermal variation of the Newtonian viscosity is complicated by the emergence of nonNewtonian shear thinning resulting from unrelaxed melt deformation. This process has been studied in compression, tension and simple shear and is thought to be well-described or understood in an empirical sense. Viscous heating may precede or follow the shear thinning effect depending on the strain rates involved, further complicated the picture. A careful consideration of the complete range of possibilities for melt deformational respond is an essential prerequisite for extracting the contribution of melt deformation from magma or rock deformation at high strain rates.


A Laboratory Study of the 2004-2008 Mount St Helens Lava Dome: Mechanical Behaviour, Rheology, and Earthquakes.

* Smith, R (rosanna.smith@ucl.ac.uk), Aon Benfield UCL Hazard Research Centre, Department of Earth Sciences, UCL, Gower Street, London, WC1E 6BT, United Kingdom
* Smith, R (rosanna.smith@ucl.ac.uk), Rock and Ice Physics Laboratory, Department of Earth Sciences, UCL, Gower Street, London, WC1E 6BT, United Kingdom
Sammonds, P R (p.sammonds@ucl.ac.uk), Aon Benfield UCL Hazard Research Centre, Department of Earth Sciences, UCL, Gower Street, London, WC1E 6BT, United Kingdom
Sammonds, P R (p.sammonds@ucl.ac.uk), Rock and Ice Physics Laboratory, Department of Earth Sciences, UCL, Gower Street, London, WC1E 6BT, United Kingdom
Tuffen, H (h.tuffen@lancaster.ac.uk), Department of Environmental Science, Lancaster University, Lancaster, LA1 4YQ, United Kingdom
Meredith, P G (p.meredith@ucl.ac.uk), Rock and Ice Physics Laboratory, Department of Earth Sciences, UCL, Gower Street, London, WC1E 6BT, United Kingdom

Lava domes are often modelled as a fluid whose dynamics are controlled by the viscosity and pressurisation of the fluid. However, the behaviour of active domes such as the 2004-2008 Mount St Helens dome and spine complex reveals that most of the lava dome deformation occurs on shear fracture planes. Evidence from seismology and exposed magma conduits at other volcanoes also indicates that the final ascent of magma into these domes may be controlled by shear fracture zones at the conduit margins. These observations demonstrate that fracturing may exert a stronger control on lava dome dynamics than fluid mechanics does. It is therefore important to expand the limited existing data on the high temperature rock mechanics of dome lavas under eruptive conditions. Acoustic emissions (AE) recorded whilst producing such data can provide a link between laboratory experiments and seismicity recorded during lava dome eruptions. Here we present results of uniaxial and triaxial deformation of a dacite sample extruded at Mount St Helens lava dome in December 2005, which has unsurpassed age constraints. This provides the unique opportunity to compare experimental results to the geophysical signals recorded as the sample was extruded. A newly modified high temperature triaxial compression apparatus was used to deform 25 mm diameter cylindrical samples at temperatures up to 1000°C, effective pressures up to 10 MPa, and strain rates from 10-4 s-1 to 10-6 s-1. It was thus possible to deform samples at temperatures, strain rates, and effective pressures typical of the Mount St Helens lava dome system and of active andesitic and dacitic lava dome systems in general, whilst also recording AE. The experimental results show the effect of temperature, effective pressure, and strain rate on the compressive strength, failure mode, and rheology of dome lavas within the brittle ductile transition. They provide key parameters and constraints for developing numerical and analytical models of lava dome growth and stability. The amount of AE before and during sample failure did not drop significantly below that in room temperature experiments until temperatures exceeded 900°C. The characteristics of individual AE events did not change in the full range of temperatures tested where AE could be detected, which was up to 970°C. This indicates that earthquakes from fracture of the erupting lava itself ascending through the conduit may be indistinguishable from earthquakes in the surrounding rocks.


Origins of Mt. St. Helens Cataclasites: Experimental Insights

* Kennedy, L A (lkennedy@eos.ubc.ca

Russell, J K (krussell@eos.ubc.ca

Nelles, E (edward_nelles@hotmail.com) AB: Thick (1-3 m) layers of fault rocks that formed during the 2004-2006 dome building events are interpreted to be linked to the 'drumbeat' microseismicity associated with dome building (Iverson et al. 2006). We experimentally deformed Mt. St. Helen's dacite with the intent of reproducing the fault textures observed in nature and placing further constraints on their conditions of formation. Experiments were run at confining pressures of 0, 25, 50, and 75MPa, at room temperature and strain rate of ∼ 1 x 10-4 S-1. The dacite starting material has low porosity (7.29 and 7.73%), a uniform bulk composition (65 wt. percent SiO2), is highly crystalline, containing 41-45% euhedral to subhedral phenocrsyts and microphenocrysts of plagioclase, hypersthene and amphibole set in a microcrystalline matrix. The experimental run products show a progressive increase in peak strength with increasing confining pressure and all show brittle behavior, characterized by a rapid stress drop. Run products contain macroscopic fractures with deformation extremely localized around the shear fractures. Natural and experimentally deformed dacites show extreme grain size reduction. Total grain size distribution plots (wt%) show that natural gouge sieved to < 63 μm contains a sharp peak at 1.9 μm and has a second, broad peak, at ∼ 4 μm. The experimentally derived gouge has a peak 1.5 μm. Our experiments were performed dry and at room temperature, whereas the temperature measured at fissures in the dacite domes at Mt. St. Helens, was ∼ 730oC (Iverson et al., 2006). Despite the lack of temperature, the microstructures from our run products are remarkably similar to those developed at Mt. St. Helens. Mode I fragmentation and shear fracture of grains and matrix, with subsequent cataclastic flow and grain size comminution occurred in both natural (Cashman et al. in press) and experimental fault rocks. Following Iverson et al (2006), we propose that the fault rocks generated at Mt. St. Helen's formed incrementally by rapid, small displacement faults, and that temperature did not greatly affect the mode of failure. We compare our experimental data (e.g. displacement, shear zone thickness, strain rate) to that obtained from Mt. St. Helen's to place constraints on the mechanisms of formation of the fault rocks. To evaluate the role of porosity on failure mechanisms, Augustine dacite from the 2006 eruption (Power et al. 2006) was also deformed. The Augustine starting material is porous (~20 - 24 %) with a glassy matrix. In contrast to Mt. St. Helen's dacite, the Augustine dacites deformed by ductile strain. The mechanism of deformation for Augustine dacites is a combination of localized shear fracturing and cataclastic flow. We suggest that porosity is the most important factor in promoting distributed cataclastic flow (Augustine) over wholesale localized faulting (Mt. St. Helen's).
Cashman, et al., in press. USGS Prof. Paper. Iverson, et al. 2006; Nature, 444. P. 439-444. Power, et al., 2006. EOS Trans, AGU, vol. 87, no. 37.


Timescales of Compaction in Volcanic Systems

* Quane, S (steve.quane@gmail.com), Colorado College, 14 E. Cache La Poudre St., Colorado Cprings, CO 80903, United States
Russell, K (krussell@eos.ubc.ca), University of British Columbia, Dept. of Earth and Ocean Sciences 3669 Stores Rd., Vancouver, BC V6T 1Z4, Canada
Friedlander, B (betsyfriedlander@gmail.com), Colorado College, 14 E. Cache La Poudre St., Colorado Cprings, CO 80903, United States

Twelve high-temperature deformation experiments on fabricated cores of rhyolite ash are used to explore the effect of porosity on the rheology of pyroclastic deposits. During deformation, the cores of ash accommodate strain mainly by shortening and reduction of porosity. Porosity loss causes a strain dependent rheology and is responsible for a marked and continuous increase in effective viscosity during deformation. The effective viscosity (η) of pyroclastic materials is predicted as a function of porosity (φ) and melt viscosity (ηo) by: logη=logηo - (α * φ)/(1-φ). The optimal value for the parameter α, based on the experimental data, is 0.78 +/- 0.15. These experiments provide constraints on the timescales of compaction and flow processes in volcanic systems under variable conditions (i.e., T, load stress). Our results indicate that welding and compaction processes in pyroclastic deposits (e.g., ignimbrite sheets, conduits, etc.) can occur on timescales of tens of minutes to hours. At these timescales, welding is fully decoupled from cooling of the deposits and may be coupled to depositional processes. Within active volcanic conduits, welding processes operate on fragmented magma rapidly enough to eliminate porosity in a matter of hours, thereby, sealing off permeability and contributing to subsequent vulcanian-style explosive eruptions.


The brittle failure of volcanoes as a precursor to eruptions

* Kilburn, C (c.kilburn@ucl.ac.uk), Aon Benfield UCL Hazard Research Centre, Dept of Earth Sciences, University College London, Gower Street, London, WC1E 6BT, United Kingdom

Recurring patterns of behaviour occur among crustal precursors to eruptions after long repose, supporting the view that reliable forecasts of eruption are a realistic objective. In particular, volcano-tectonic events (or VT events, due to rock fracturing) tend to occur at an accelerating rate towards eruption. The behaviour is consistent with the activation and interaction of a population of faults within the deforming crust and volcanic edifice. At the start of unrest, the VT event rate at volcanoes in several tectonic settings commonly increases exponentially with time over intervals of months or more. In subduction-zone settings, the exponential increase evolves into faster-than-exponential trends about 1-2 weeks before eruption. In extensional settings, the faster- than-exponential stage may emerge only hours before eruption. The time series are determined by how the deformation induced in the crust by a magmatic pressure source is partitioned between elastic and brittle behaviour. Quantitative relations among stress, deformation and event rate from field observations at andesitic-dacitic and basaltic volcanoes, as well as large calderas, are consistent with fracture-mechanics measurements in the laboratory. The results identify new constraints on the rate constants that define the precursory accelerations and indicate that forecasts of eruption may be feasible as much as weeks ahead of time.


Cracking Magmas

* Lavallee, Y (yanlavallee@hotmail.com), LMU-Munich, Theresienstrasse 41, Munich, BAY 80333, Germany
Benson, P (phil.benson@utoronto.ca), Univeroty of Toronto, 170 College Street, Toronto, ONT M5S 3E3, Canada
Hess, K (hess@min.uni-muenchen.de), LMU-Munich, Theresienstrasse 41, Munich, BAY 80333, Germany
Meredith, P G (p.meredith@ucl.ac.uk), UC-London, Gower Street, London, WC1E 6BT, United Kingdom
Dingwell, D B (dingwell@lmu.de), LMU-Munich, Theresienstrasse 41, Munich, BAY 80333, Germany

Explosive eruptions are commonly preceded by accelerations in discharge rate and seismic energy released by fracturing. Understanding the propagation of cracks in rapidly deforming magmas is thus needed to adequately describe the physics of explosive eruptions. In this study dome lavas from volcán de Colima (Mexico) are deformed at 900-980°C under high stresses (up to 50 MPa) in a uniaxial press, and crack propagation is monitored in situ by a couple of fast-acquisition, acoustic emission (AE) sensors. The obtained 1-D AE profile associated with crack propagation is then compared to a series of strain-step experiments as well as to the developed, 3-D crack network imaged after some experiments via high-resolution (30 micrometers) neutron tomography. We observe that uniaxial deformation is initially characterized by extensional cracks propagating inward from the margin of the cylindrical samples. As microscopic cracks grow and link up, a set of transversal, conjugate shear fractures developed from the inner tips of the extensional cracks. Alike the transition to explosive eruptions, the experimental transition of microscopic to macroscopic failure is accompanied by accelerations in strain rate and released AE energy. Inversion of the acceleration rate of released AE energy following the failure forecast method is used to track and predict macroscopic failure of magma. We estimate that accurate failure prediction can be made within ∼40 % of the critical strain at failure. Our findings suggest that shear fractures dominate as an acoustic precursor when macrocopic failure is approached.


Constraining Volcano Source Rheology and Mechanisms: 3D Full Wavefield Simulations and Very High Resolution Observations From Mt Etna.

* Bean, C J (chris.bean@ucd.ie), School of Geological Sciences, University College Dublin, Belfield, Dublin, 4, Ireland
O'Brien, G
EM: , School of Geological Sciences, University College Dublin, Belfield, Dublin, 4, Ireland
de Barros, L
EM: , School of Geological Sciences, University College Dublin, Belfield, Dublin, 4, Ireland
Murphy, S
EM: , School of Geological Sciences, University College Dublin, Belfield, Dublin, 4, Ireland
Lokmer, I
EM: , School of Geological Sciences, University College Dublin, Belfield, Dublin, 4, Ireland
Saccorotti, G
EM: , Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Pisa, Via U. della Faggiola, 32,, Pisa, 56126,
Patane, D
EM: , Istituto Nazionale di Geofisica e Vulcanologia-Sezione di Catania,, Piazza Roma 2, Catania, 95123, Italy
Metaxian, J
EM: , Laboratoire de Géophysique Interne et Tectonophysique,, Université de Savoie, Le Bourget-du Lac, France

Recent field observation and laboratory experiments have demonstrated a broad range of deformation mechanisms in volcanic rocks, and a juxtaposition of brittle and ductile deformation in both space and time. On the other hand seismological observations of transient deformation at volcanoes yield an equally wide variety of signal types including Volcano Tectonic (VT), Long Period (LP), Very Long Period (VLP) and tremor. A clear goal is to find robust connections between these independent sets of observations, linking detailed field studies, well controlled laboratory experiments and volcano seismology. In volcano seismology VT events are usually interpreted as the brittle response of the edifice to stressing whereas LP and VLP events are thought to result from fluid-filled conduit dynamics. However, strong wave propagation path effects and a large number of possible source mechanisms make it difficult to find a quantitative interpretation of mechanism/rheology. Numerical simulations have a key role to play in making the connection between well-controlled laboratory experiments and the field. Furthermore, many of the features seen in real volcano seismograms can be reproduced in 3D full wavefield simulations of both wet (coupled multi phase fluids and solids) and dry (rupture propagation) models. Even in simulated data the underlying rheology/source mechanisms are difficult to determine from an inversion of the synthetic seismograms, especially for sparse data with poor velocity control. With this in mind a detailed field experiment was undertaken on Mt Etna in June 2008, comprising 30+ stations in the summit area. Aided by simulated data in realistic velocity models, this has given us an unprecedented picture of shallow LP activity on Etna. These high resolution observations will be compared with recent results from laboratory experiments and with numerical simulations in an effort to better constrain the rheology/mechanism of the sources.