V23A-76 1330h
Very-Long Period Seismic Pulses Associated With Small-Scale Eruptions at Suwanosejima Volcano, Japan
Suwanosejima is an andesitic volcano, located in Ryukyu Islands. The eruptive activity has been repeated at the summit crater in the caldera with Vulcanian- Strombolian style since 1957. In order to make clear process of the explosive eruptions, we installed 4 broadband seismometers, 4 tiltmeters and an infrasonic microphone around summit crater in 2003, and observed many small-scale eruptions in early November. The eruptions were accompanied with ash ejection and weak infrasonic wave. Displacement seismograms observed by the broadband seismometers typically show very long-period pulse associated with eruption. Upward movement in vertical component and horizontal outward movement from the center of crater began 50-90 s before the beginning of the eruptions. When the eruptions started, the movements turned to downward in vertical component and inward to the central crater in horizontal components, respectively, and the downward and inward movements continued for 20-30 s. The very long-period pulses appeared clearly at the closest station to the crater (460m apart from the central crater). The amplitudes ranged 8-17 microns in vertical component and 21-71 microns in horizontal component. Upward and outward movements prior to the eruptions and downward and inward movements associated with eruptions were detected at all the stations except the furthest station, 837 m apart from the central crater. Although movements before the eruptions were not so clear at the furthest station, upward and inward displacements were detected associated with the eruptions. The difference in polarity in vertical components during eruptions may be caused by difference in relative elevation. The furthest station is located at lower elevation than the crater and the other stations are located at higher positions. This fact and larger horizontal components suggest that pressure source casing deflation is located at a quite shallow depth. For simplicity, Mogi's model considering elevation differences was applied to vertical and horizontal displacement at the 4 stations during eruption deflation stages, to estimate locations of the pressure source. The sources are estimated to be located at depth <100m beneath the crater and the volume changes of the sources were in the order of 10 cubic meters. On November 2, 2003, eruptions were repeated with time intervals of 3-5 minutes. Upward-downward patterns were recognized on all the eruption. At Sakurajima volcano, inflation tilt and extensional strain changes were observed 5 minutes to several hours before explosive eruptions and deflation tilt and contraction strains were recorded associated with occurrence of the explosive eruptions. Inflations and deflations were caused by pressure increase and decrease of the source at depths of 3-6 km (Ishihara, 1990). Displacement seismograms of vertical component at Suwanosejima are quite similar to the tilt records at Sakurajima. Although the intensity and depth of the pressure and durations of pressure increase and decrease at Suwanosejima are smaller than those at Sakurajima, displacement seismograms revealed that minor inflation and deflation were repeated at quite shallow depths beneath the crater of Suwanosejima volcano. Several minutes before the eruptions, ash and gas emission suddenly suspended. Upward-downward patterns may be caused by accumulation and release of the gas trapped at the uppermost part of the conduit.
V23A-77 1330h
The Basic Study of Shock Wave Propagation in Highly Viscous Liquid
$~~~~~~$The mechanical fragmentation of magma caused by intensive decompression can be viewed in two regimes, depending upon whether the characteristic deformation times are greater or less than the viscous relaxation time. Hydrodynamic fragmentation involves the rapid acceleration of magma by pressurized fluid and gas. On the other hand, brittle fragmentation is result of 3-dimentional crack growth caused by excess strain that exceeds the elastic properties of a medium. These processes are complex and difficult to direct observation at volcanic eruption. So, it is important to conduct an indoor model experiment to grasp the mechanism of magma fragmentation using transparent materials, which have similar viscoelastic properties to magma at room temperature. For trial, to visualize the shock wave propagation and rapid gas expansion in a highly viscoelastic fluid, silver azide pellet was exploded in transparent viscoelastic fluids.$~~~~~~$ The test fluids were commercial grades of starch syrup (HS-20 of Hayashibara Shoji, Inc Group Corporation, Japan) and silicone oil (KF96 of Shin-Etsu Chemical Co., Ltd., Japan). The starch syrup was dehydrated in a vacuum heater in order to adjust the water content to 30%, 20%, 15%, 13% and 10%. The theoretical kinetic viscosity of syrup increases from 5.43_~10$^{-3}$m$^{2}$/s to 1.63 m$^{2}$/s with dehydration. The test section was composed of a stainless steel chamber (100mm in the inside diameter, 100mm in depth) and an acryl window $(140mm\times140mm\times25mm)$. The chamber had a pinhole to insert optical fiber. 10mg silver azide pelet was ignited by Nd Yag laser (Laser photonics Co., Ltd., 25mJ/pulse).$~~~~~~$ The results of experiments were visualized by two methods: high speed photography and double exposure holografhic inter ferometry. Comparing starch syrup and silicone oil with same kinetic viscosity, explosives gas in silicone oil expanded spherically as like as starch syrup with relatively low kinetic viscosity. When shock wave was loaded on milimerter-size air bubble, the clollapsing bubble created jets. On the other hand, in starch syrup with relatively high kinetic viscosity, 3-dimentional crack propagates immediately after gas expansion. Stress concentrates on the top of crack in double exposure holografhic inter ferometry. No collapsing bubbles were observed.$~~~~~~$ The distance from top of crack to the explosive center shows more rapid rate of increase than those of bubble diameter, which indicates that crack propagation is faster than bubble expansion at the same energy.
V23A-78 1330h
Variation in void fraction of pumice from exposive eruptions and fragmentation mechanisms
There is a wide variety in void fraction of pyroclastic fragments from explosive volcanic eruptions. This variety is considered to reflect diverse physical conditions of the eruptions. In order to clarify the relation between the void fraction of pyroclasts and the geological conditions, the dynamics of explosive volcanic eruptions is numerically studied using a one-dimensional steady conduit flow model coupled with bubble growth. Spherical cell model is used for bubble growth, and the magma is treated as a viscous fluid. The viscosity of magma depends on the volatile content which varies according to the equilibrium exsolution law. We introduce two types of fragmentation condition; `stress fragmentation' and `expansion fragmentation'. As the pressure of the magma decreases, the bubbles expand and the viscous stress around bubble walls caused by the pressure difference between the gas in bubbles and the liquid magma is enhanced. The stress fragmentation occurs when the viscous stress reaches the strength of magma. The void fraction at the time of the stress fragmentation is generally small because the stress fragmentation occurs before the bubbles sufficiently expand. When the stress is not enhanced enough, magma may fragment due to fluid dynamical instability of melt film as the void fraction reaches a critical value (say 0.8). This fragmentation mechanism is defined as expansion fragmentation. We found that the condition whether expansion or stress fragmentation occurs is determined by a non-dimensional number, that is, the ratio of wall friction force in conduit flow to the tensile strength of magma. As the viscosity increases, the stress fragmentation is more likely to occur. Furthermore, the fact that the wall friction is a key parameter means that the mode of fragmentation depends on not only the magma viscosity but also other geological conditions such as conduit radius and exit velocity. For instance, for a given initial volatile content and temperature, the stress fragmentation is more likely to occur as conduit radius is smaller and exit velocity is greater. We investigate how fragmentation mode and exit velocity change as a function of the geological parameters such as the pressure of magma chamber and the conduit radius by systematically searching possible steady solutions that satisfy the boundary conditions both at the vent and at the magma chamber. As the chamber pressure increases, the velocity and hence the wall friction increase for a given conduit radius. Then the stress fragmentation occurs and the void fraction at fragmentation is small. The effect of the variation of the conduit radius is more complex. For a constant velocity, the wall friction force decreases as the conduit radius increases. However, the velocity also increases as the conduit radius increases, which can results in increase of the wall friction. Numerical results show that the effect of the increase of the velocity is predominant. From the above results, we can derive a sequence of the changes in the exit velocity and the fragmentation mode during a Plinean eruption. It is considered that the pressure of the magma chamber decreases due to the depletion of magma and the conduit radius increases due to erosion with time during an eruption. If the decrease of the chamber pressure is much faster than the increase of the conduit radius, for example, the exit velocity decreases. Then the fragmentation mode changes from the stress one to the expansion one. Thus the variation of void fraction observed in deposits of explosive eruptions can be attributed to the variation of the fragmentation mechanisms and the change of the pressure of the magma chamber and the conduit radius.
V23A-79 1330h
Transition between Plinian Activity and Dome Growth at Novarupta: the Dacite Dome
During the eruption of Novarupta on June 6-8, 1912, Plinian volcanism was followed by dome extrusion ultimately leading to the present-day rhyolitic dome. The fine-ash deposited at the close of Plinian activity underlies a dacite block bed covering an elliptical 4 km$^{2}$ area around the vent. The blocks appear to have mostly followed ballistic trajectories; often blocks broke into pieces upon impact. The largest clasts are up to 12 m$^{3}$ in volume or 10$^{4}$ kg in mass. The blocks have a wide range of textures and lithologies and include juvenile dacites and nonjuvenile material, e.g., agglutinates of recycled lava and wall rock. Many of the blocks were apparently ejected at high temperatures; breadcrust textures are pervasive among all lithologies. Over 700 blocks have been mapped, measured, and classified into four main lithologic categories: 1) pumiceous dacite, 2) dense dacite, 3) welded breccia, and 4) flow-banded dacite. Characteristic densities range from 860-2400 kg-m$^{-3}$, and point counting of 200 blocks suggests the ratio between lithologies is 6:1:9:4, respectively. Textural analyses of high, modal, and low density pieces of pumiceous dacite blocks along with a piece of a dense dacite block show portions of the melt underwent variable degrees of outgassing. Vesicle size distributions and qualitative observations include markedly coarse bubbles, thick glass walls, and irregularly shaped vesicles, all characteristic of bubble coalescence and collapse. Compared to textural data from the current rhyolitic dome, the blocks have a wider range of vesicularities and bubble sizes and a smaller dominant bubble size. While both compositions show melt maturation, the dacite magma might have had less time for free-escape of volatiles. The block bed seemingly formed in a series of Vulcanian explosions that followed the close of Plinian volcanism and disrupted a body of dacitic magma which had undergone a very heterogeneous pattern of degassing and, to a much lesser extent, wall rock. Block morphology and textures strongly suggest the magma was first erupted as a relatively gas-rich lava dome but incomplete outgassing led to explosive disruption.
V23A-80 1330h
Visible Microspectroscopy For Characterizing Color Changes Of Volcanic Materials
Volcanic materials often show reddish colors and this red coloring is considered to be due to the high temperature oxidation process. In order to study quantitatively the origins of the red coloring of volcanic materials, we studied here a scoria sample from Takatsukayama volcano having varying colors from black to red. Spectro-colorimetry of the powdered samples was first carried out to describe quantitatively their colors using CIE L*a*b* color space. The reddish samples show increasing a* (redness) and b* (yellowness) values. This red coloring of the scoria was reproduced by the laboratory heating experiments under an atmospheric condition. X-ray fluorescence spectrometry (XRF) of the sample powders showed constant total Fe contents, while the phenanthroline method indicated decreasing Fe2+ contents for reddish scoria (Yamanoi et al. 2004). In order to elucidate changes in chemical forms of iron in the scoria, we analyzed the scoria thin sections by visible and Raman microspectroscopy. These methods indicate the presence of hematite-like materials in the red parts of the olivine phenocryst and groundmass both in the natural samples and heated products. These imply that the red coloring of Takatsukayama scoria is induced by the high temperature oxidation of Fe2+, resulting in the formation of hematite-like materials in the olivine phenocryst and the matrix glass. The kinetic aspects of the high temperature oxidation of Fe2+ to hematite-like materials are now being studied by developing in-situ high temperature visible microspectroscopy for tracing this process directly on olivine and glass thin sections. A heating stage is set under the visible microspectrometer and the sample thin sections are put on a Pt foil on the stage. Olivine sections are heated on this stage at 500-600 C and changes in visible absorption spectra are monitored every 60 seconds. Quantitative treatments of these spectral changes and their kinetic analyses will be discussed for evaluating temperature-time scales of the red coloring.
V23A-81 1330h
A new Experimental Apparatus for Hydrous Alteration Reactions With Super-Critical Fluid Flow
A new super-critical fluid flow apparatus is developed to reproduce hydration and alteration processes of rocks and minerals with fluid related to volcanic eruptions. The behavior of high-pressure fluid flow and permeability of rocks have close relation to volcanic eruptions, especially, phreatomagmatic eruptions caused by excess pressures of the fluid in volcanoes. Alteration, dissolution and / or precipitation processes of rocks and minerals can change permeability of rocks by spreading of the fluid path or obstruction with precipitated minerals. Kinetics of flowing super-critical fluid - rock interaction is essential to understand rise and fall of excess pressure of the fluid. The pressure vessel is made from hastelloy-C alloy. Inner diameter and length of the pressure vessel is 45mm and 600mm, respectively. The maximum temperature and pressure of the pressure vessel is $600\deg$C and 80MPa. Fluid flow rate of the can be set from 0.01ml to 30ml / minute. Pressure of the reactor vessel is maintained with a high-pressure regulator after cooling of the outflow fluid. Starting materials of the experiments are placed in a SUS sample basket sealed to an outlet of the pressure vessel with pure Ti gaskets. Inner diameter of the sample basket is 9.4mm for powdered samples and 25mm for drilled rock core samples. Pressure difference between the inlet and the outlet of the pressure vessel is monitored to watch permeability of the sample with the super-critical flow. Temperature gradient of the pressure vessel is controlled by triple electric furnaces. Run products of the powdered samples are retrieved by cutting off the sample basket. Volcanic glass is easily hydrated by the super-critical water. Alteration experiments of powdered obsidian are carried out at $450\deg$C and 50MPa with 0.2ml / minute. Alteration products of the volcanic glass including kaolinite occur in grain boundaries of the obsidian powder and cemented grains within a few days.
V23A-82 1330h
Banded tremor at Miyakejima volcano, Japan: Implication for the periodic instability of hydrothermal system
Miyakejima volcano eruption started in 2000 and its activity is still going on in 2004. In these three years, the main activity is the steam emission and the geodetic observation records slow shrinkage of volcanic body and the seismicity is not so high as that in the dike intrusion stage in 2000. Under such a situation, we record some typical "banded tremor" in some periods of the activity. The banded tremor has some characteristics as: (1) their intervals are almost constant as long as 20 - 40 minutes, (2) One banded tremor continues about 5 minutes, (3) the amplitudes of the tremor are decreasing as a long-term point of view, (4) before Oct. 2002, the characteristic frequencies are about 5-7 Hz and the spectra has about three distinct peaks, (5) after Oct. 2002, the continuous tremor with 4Hz characteristic frequencies was observed and the spectral peak of banded tremor has laid on it with broad peaks. The source of this banded tremor is inferred to be shallow (about 1-2 km or less deep) beneath the volcano, since the nearest station could detect. Associated with this banded tremor, the borehole tiltmeter recorded jerky crustal deformation, suggesting that the expansion of summit area. There is no relation between this banded tremor and some apparent phenomena. The source of this banded tremor seems to be related to some instability of the hydrothermal system beneath the summit caldera at the depth less than 2-3 kilometers. The source model has to explain some distinctive features of the phenomena, their intervals (20-40 minutes), the duration time (5 minutes) and characteristic frequencies with a few spectral peaks. In this presentation, we propose two cases of hydrothermal pipe flow system. (A): Instability and periodicity of the heated water - steam system (two-phase flow): the density wave oscillation, (B): the open and shut of a valve in a pipe flow. In the model A, the cycle of the heating and vesiculation controls the periodicity of the banded tremor. In addition, the length of the two-phase flow area determines the characteristic frequencies. In the model B, the instability between the flux and the differential pressure is caused by the non-elasticity of the valve. The transition between the stable and instable condition corresponds to the periodicity of the banded tremor.
V23A-83 1330h
SPH Simulations of Field Explosive Experiments: Relations of Explosive Energy and Depth to Explosive Features
Volcanic explosions cause sometimes catastrophic destruction of mountains. However, it is difficult to predict how the mountain side collapses and how much disaster is caused due to volcanic explosions, in detail. Taniguchi et al. 2003 conducted field explosion experiments in order to relate explosive conditions to the resultant surface phenomena. Their results indicate that explosive features, such as, the crater diameter, and the aspect ratio of explosive clouds, can be scaled by the amount of the energy and its depth. The results can be scaled to the real scale of volcanic eruptions, although it is difficult to conduct explosive experiments of the same order of the energy of volcanic eruptions. In order to verify their predictions, and to simulate the mountain collapse due to volcanic explosion, numerical codes to realize volcanic eruptions have been developed. The code utilizes fully Lagrangian method called SPH, Smoothed Particle Hydrodynamics. It was used to simulate impact phenomena of meteoroids on planetary surfaces and is modified to apply to mountain collapse. It is suitable to realize large deformation. Two types of EOS of fluid and elastic-plastic deformation are taken into consideration. As the first step of simulations, field explosive experiments are simulated to test our numerical code. The energy source is allocated in some depth of 2-dimensional semi-infinite target. The high energy source region develops, and thereafter it can be traced as a function of time. Various conditions of total amount of energies and depths of the energy source are applied and then, crater diameters and explosive feature are examined. As a result, the correlations of scaled depth to explosive feature, such as, the crater diameter and aspect ratio estimated by field explosive experiments are confirmed on conditions of the yield strength of the order of 1 MPa and the shear modulus of the order of 100 MPa in the case of elastic-plastic EOS. Field experiments were conducted in soil farm, and these conditions can be applied. However, strength of materials of fields or volcanoes may become one of the new key parameter of scaling law of volcanic explosions. Hereafter, considerations of EOS and parallelization of the code are necessary to expand it to 3 dimensional real volcanic eruptions.
V23A-84 1330h
Geodetic and seismological evidence for underground cavity collapse preceding the caldera formation of Miyakejima, Central Japan, in July 2000.
The caldera formation of the 2000 Miyakejima eruptive activity began with the small summit eruption at 6:41 p.m. on July 8 (JST). After then, the intermittent subsidence in the summit area lasting until middle of August formed the caldera with 1.6 km diameter and 400 m depth. Gravity and electro-magnetic data suggested a cavity formation at the depth of 1-2 km under the summit area during the period from the beginning of magma discharge in the late June to the eruption on July 8 (Furuya et al., 2003; Sasai et al., 2001). The eruption of July 8 preceded tremor-like seismic signals with large amplitudes and crustal deformation indicating contraction of the summit area, both of which had started in the mid-night of July 7. Our present analyses succeed to show that these signals evidently indicate the fabric of the cavity that had been formed in the early July collapsed and led to the first eruption. About one day before the eruption, tilt change toward the summit began and accelerated as the eruption approached, baseline length between GPS stations shortened about 10 cm, and amplitudes of tremor-like seismic signals increased. The crustal deformation can be explained by an accelerated contraction of a spherical source at the depth of 1 km beneath the summit. The volumetric change is -0.8x10$^{6}$m$^{3}$. The crustal deformation hardly shows occurrence of a magma ascent before the eruption. The tremor-like seismic signals are composed of the successive occurrence of small earthquakes with P and S phases. The spatial amplitude distribution indicates the source location is shallow part of the summit, by comparing the well located events. These observations highly suggest that the collapse of cavity had begun about one day before the eruption, and the collapse gradually accelerated. When the tilt change was observed, magnetic field intensity on the summit acceleratedly decreased about 30 nT, which was interpreted qualitatively by successive collapses of magnetized pillars that sustained the cavity (Sasai et al., 2001). Based on their conceptual model, we constructed a cavity collapse model that interprets quantitatively the accelerated crustal deformation as well as the magnetic intensity change. The model is a spherical cavity whose internal pressure is sustained by restitution of elastic pillars with magnetization. When stress of a pillar exceeds its failure stress, the pillar collapses. The collapse increases stress of the other pillars, and the collapse and deflate of the cavity are promoted. The accelerated collapse of magnetized pillars decrease the magnetic field intensity on the ground surface. The crustal deformation and magnetic field change can be interpreted by an accelerated collapse of the cavity. Therefore, we speculate that the eruption is a consequence of the collapse of the cavity that had been formed beneath the summit until July 6.
V23A-85 1330h
Complex Proximal Geometry of Fall Deposits From the Tarawera 1886 Basaltic Plinian Eruption: Implications for Eruption Dynamics
The 1886 eruption of Tarawera, New Zealand, is unusual for a Plinian eruption because (a) it involved entirely basaltic magma and (b) it produced abundant proximal deposits with a complex geometry not predicted by standard models of Plinian eruption columns. The distribution of the widespread tephra fall indicates that the Plinian activity was restricted to the 8 km Mt Tarawera portion of the 17 km eruption fissure. During the 5 hour eruption, over 50 point source vents were active along this segment with a variety of styles and dispersals. Most vents primarily produced localized, cone-building tephra fall, but some vents also contributed spasmodically to the Plinian plume. The proximal deposits comprise a series of lensoid packages of beds visible in continuous exposures along both sides of the 1886 fissure. Each package includes contributions from local weak vents as well as fallout from the high plume. The dispersal of a package and the nature of the clasts indicate the location, style, and intensity of the dominant vent(s) that produced that package. We have defined facies that represent products of endmember styles: facies a is a dry Strombolian style that formed very localized deposits, and facies b is a highly explosive, possibly partially phreatomagmatic style that contributed most to the high plume. A combination of observations, including package dispersal, clast morphology, and distribution of clast densities, is necessary to determine the dominant facies of a particular package. We have produced detailed cross-sectional maps of the packages in eight of the thirteen craters along the Mt Tarawera fissure in order to compare style, intensity, and relative timing of activity at different vents. We identify the vents that contributed most to the Plinian column and consider possible mechanisms for their unusually intense behavior, such as early incorporation of rhyolitic wall rock into the basaltic magma. The 1886 dispersal data require much more complex models for the velocity distribution in the lower portion of the plume than are afforded by existing numerical models.