The dynamics and thermodynamics of large ash flows

 Ash flow deposits, containing up to 1000 km³ of material, have been produced by some of the largest volcanic eruptions known. Ash flows propagate several tens of kilometres from their source vents, produce extensive blankets of ash and are able to surmount topographic barriers hundreds of metres high. We present and test a new model of the motion of such flows as they propagate over a near horizontal surface from a collapsing fountain above a volcanic vent. The model predicts that for a given eruption rate, either a slow (10–100 m/s) and deep (1000–3000 m) subcritical flow or a fast (100–200 m/s) and shallow (500–1000 m) supercritical flow may develop. Subcritical ash flows propagate with a nearly constant volume flux, whereas supercritical flows entrain air and become progressively more voluminous. The run-out distance of such ash flows is controlled largely by the mass of air mixed into the collapsing fountain, the degree of fragmentation and the associated rate of loss of material into an underlying concentrated depositional system, and the mass eruption rate. However, in supercritical flows, the continued entrainment of air exerts a further important control on the flow evolution. Model predictions show that the run-out distance decreases with the mass of air entrained into the flow. Also, the mass of ash which may ascend from the flow into a buoyant coignimbrite cloud increases as more air is entrained into the flow. As a result, supercritical ash flows typically have shorter runout distances and more ash is elutriated into the associated coignimbrite eruption columns. We also show that one-dimensional, channellized ash flows typically propagate further than their radially spreading counterparts. As a Plinian eruption proceeds, the erupted mass flux often increases, leading to column collapse and the formation of pumiceous ash flows. Near the critical conditions for eruption column collapse, the flows are shed from high fountains which entrain large quantities of air per unit mass. Our model suggests that this will lead to relatively short ash flows with much of the erupted material being elutriated into the coignimbrite column. However, if the mass flux subseqently increases, then less air per unit mass is entrained into the collapsing fountain, and progressively larger flows, which propagate further from the vent, will develop. Our model is consistent with observations of a number of pyroclastic flow deposits, including the 1912 eruption of Katmai and the 1991 eruption of Pinatubo. The model suggests that many extensive flow sheets were emplaced from eruptions with mass fluxes of 10⁹–10¹⁰ kg/s over periods of 10³–10⁵ s, and that some indicators of flow "mobility" may need to be reinterpreted. Furthermore, in accordance with observations, the model predicts that the coignimbrite eruption columns produced from such ash flows rose between 20 and 40 km. Received: 25 August 1995 / Accepted: 3 April 1996     

Evaluating suitability of a tephra dispersal model as part of a risk assessment framework

In volcanic risk assessment it is necessary to determine the appropriate level of sophistication for a given predictive model within the contexts of multiple sources of uncertainty and coupling between models. A component of volcanic risk assessment for the proposed radioactive waste repository at Yucca Mountain (Nevada, USA) involves prediction of dispersal of contaminated tephra during violent Strombolian eruptions and the subsequent transport of that tephra toward a hypothetical individual via surface processes. We test the suitability of a simplified model for volcanic plume transport and fallout tephra deposition (ASHPLUME) coupled to a surface sediment-transport model (FAR) that calculates the redistribution of tephra, and in light of inherent uncertainties in the system. The study focuses on two simplifying assumptions in the ASHPLUME model: 1) constant eruptive column height and 2) constant wind speed and direction during an eruption. Variations in tephra dispersal resulting from unsteady column height and wind conditions produced variations up to a factor of two in the concentration of tephra in sediment transported to the control population. However, the effects of watershed geometry and terrain, which control local remobilization of tephra, overprint sensitivities to eruption parameters. Because the combination of models used here shows limited sensitivity to the actual details of ash fall, a simple fall model suffices to estimate tephra mass delivered to the hypothetical individual.     

A reappraisal of shear wave splitting parameters from Italian active volcanic areas through a semiautomatic algorithm

Shear wave splitting parameters represent a useful tool to detail the stress changes occurring in volcanic environments before impending eruptions. In the present paper, we display the parameter estimates obtained through implementation of a semiautomatic algorithm applied to all useful datasets of the following Italian active volcanic areas: Mt. Vesuvius, Campi Flegrei, and Mt. Etna. Most of these datasets have been the object of several studies (Bianco et al., Annali di Geofisica, XXXXIX 2:429–443, 1996, J Volcanol Geotherm Res 82:199–218, 1998a, Geophys Res Lett 25(10):1545–1548, 1998b, Phys Chem Earth 24:977–983, 1999, J Volcanol Geotherm Res 133:229–246, 2004, Geophys J Int 167(2):959–967, 2006; Del Pezzo et al., Bull Seismol Soc Am 94(2):439–452, 2004). Applying the semiautomatic algorithm, we confirmed the results obtained in previous studies, so we do not discuss in much detail each of our findings but give a general overview of the anisotropic features of the investigated Italian volcanoes. In order to make a comparison among the different volcanic areas, we present our results in terms of the main direction of the fast polarization (φ) and percentage of shear wave anisotropy (ξ).     

Quantitative models of the fallout and dispersal of tephra from volcanic eruption columns

A theoretical model of clast fallout from convective eruption columns has been developed which quantifies how the maximum clast size dispersal is determined by column height and wind strength. An eruption column consists of a buoyant convecting region which rises to a heightH _B where the column density equals that of the atmosphere. AboveH _B the column rises further to a heightH _T due to excess momentum. BetweenH _T andH _B the column is forced laterally into the atmosphere to form an upper umbrella region. Within the eruption column, the vertical and horizontal velocity fields can be calculated from exprimental and theoretical studies and consideration of mass continuity. The centreline vertical velocity falls as a nearly linear function over most of the column's height and the velocity decreases as a gaussian function radially away from the centreline. Both column height and vertical velocity are strong functions of magma discharge rate. From calculations of the velocity field and the terminal fall velocity of clasts, a series of particle support envelopes has been constructed which represents positions where the column vertical velocity and terminal velocity are equal for a clast of specific size and density. The maximum range of a clast is determined in the absence of wind by the maximum width of the clast support envelope.The trajectories of clasts leaving their relevant support envelope at its maximum width have been modelled in columns from 6 to 43 km high with no wind and in a wind field. From these calculations the shapes and areas of maximum grain size contours of the air-fall deposit have been predicted. For the no wind case the theoretical isopleths show good agreement with the Fogo A plinian deposit in the Azores. A diagram has been constructed which plots, for a particular clast size, the maximum range normal to the dispersal axis against the downward range. From the diagram the column height (and hence magma discharge rate) and wind velocity can be determined. Historic plinian eruptions of Santa Maria (1902) and Mount St. Helens (1980) give maximum heights of 34 and 19 km respectively and maximum wind speeds at the tropopause of m/s and 30 m/s respectively. Both estimates are in good agreement with observations. The model has been applied to a number of other plinian deposits, including the ultraplinian phase of theA.D. 180 Taupo eruption in New Zealand which had an estimated column height of 51 km and wind velocity of 27 m/s.     

Preliminary sensitivity study of eruption source parameters for operational volcanic ash cloud transport and dispersion models — A case study of the August 1992 eruption of the Crater Peak vent, Mount Spurr, Alaska

Ash clouds are one of the major hazards that result from volcanic eruptions. Once an eruption is reported, volcanic ash transport and dispersion (VATD) models are used to forecast the location of the ash cloud. These models require source parameters to describe the ash column for initialization. These parameters include: eruption cloud height and vertical distribution, particle size distribution, and start and end time of the eruption. Further, if downwind concentrations are needed, the eruption mass rate and/or volume of ash need to be known. Upon notification of an eruption, few constraints are typically available on many of these source parameters. Recently, scientists have defined classes of eruption types, each with a set of pre-defined eruption source parameters (ESP). We analyze the August 18, 1992 eruption of the Crater Peak vent at Mount Spurr, Alaska, which is the example case for the Medium Silicic eruption type. We have evaluated the sensitivity of two of the ESP – the grain size distribution (GSD) and the vertical distribution of ash – on the modeled ash cloud. HYSPLIT and Puff VATD models are used to simulate the ash clouds from the different sets of source parameters. We use satellite data, processed through the reverse absorption method, as reference for computing statistics that describe the modeled-to-observed comparison. With the grain size distribution, the three options chosen, (1) an estimated distribution based on past eruption studies, (2) a distribution with finer particles and (3) the National Oceanic and Atmospheric Administration HYSPLIT GSD, have little effect on the modeled ash cloud. For the initial vertical distribution, both linear (uniform concentration throughout the vertical column) and umbrella shapes were chosen. For HYSPLIT, the defined umbrella distribution (no ash below the umbrella), apparently underestimates the lower altitude portions of the ash cloud and as a result has a worse agreement with the satellite detected ash cloud compared to that with the linear vertical distribution for this particular eruption. The Puff model, with a Poisson function to represent the umbrella cloud, gave similar results as for a linear distribution, both having reasonable agreement with the satellite detected cloud. Further sensitivity studies of this eruption, as well as studies using the other source parameters, are needed.     

The fluid dynamics and thermodynamics of eruption columns

A new model for Plinian eruption columns is derived from first principles and investigated numerically. The dynamics particular to the momentum-driven basal gas-thrust region and the upper buoyancy-driven convective region are treated separately. The thermal interactions in the column are modelled by the steady-flow-energy equation. The main results of the present paper are that: (1) the basal gas-thrust region model predicts a very rapid initial expansion of the material on leaving the vent; (2) the gas-thrust region height decreases with initial temperature, inital gas content of the erupted material and initial velocity, but increases with vent radius; (3) the total column height increases with initial temperature, initial velocity and vent radius, but decreases with initial gas content; (4) column collapse occurs for initial velocities of the order of 100 m/s; the precise value increases as the initial gas content in the erupted material decreases; (5) for large vent radii or low initial gas content of the erupted material, the velocity in the column can increase with height once in the buoyancy-driven region instead of decaying to zero monotonically; (6) the interaction of the potential energy with the enthalpy is found to be the dominant thermal interaction in the upper part of the column. Previous models of eruption columns involve inconsistencies and simplifications; these are shown to lead to significant differences in the results in comparison to the present model.     

BET_VH: exploring the influence of natural uncertainties on long-term hazard from tephra fallout at Campi Flegrei (Italy)

In this paper, we explore the effects of the intrinsic uncertainties upon long-term volcanic hazard by analyzing tephra fall hazard at Campi Flegrei, Italy, using the BET_VH model described in Marzocchi et al. (Bull Volcanol, 2010). The results obtained show that volcanic hazard based on the weighted average of all possible eruptive settings (i.e. size classes and vent locations) is significantly different from an analysis based on a single reference setting, as commonly used in volcanic hazard practice. The long-term hazard map for tephra fall at Campi Flegrei obtained here accounts for a wide spectrum of uncertainties which are usually neglected, largely reducing the bias intrinsically introduced by the choice of a specific reference setting. We formally develop and apply a general method to recursively integrate simulations from different models which have different characteristics in terms of spatial coverage, resolution and physical details. This outcome of simulations will be eventually merged with field data through the use of the BET_VH model.     

Tephra fallout hazard assessment at the Campi Flegrei caldera (Italy)

Tephra fallout associated with renewal of volcanism at the Campi Flegrei caldera is a serious threat to the Neapolitan area. In order to assess the hazards related with tephra loading, we have considered three different eruption scenarios representative of past activity: a high-magnitude event similar to the 4.1 ka Agnano-Monte Spina eruption, a medium-magnitude event, similar to the ∼3.8 ka Astroni 6 eruption, and a low-magnitude event similar to the Averno 2 eruption. The fallout deposits were reconstructed using the HAZMAP computational model, which is based on a semi-analytical solution of the two-dimensional advection–diffusion–sedimentation equation for volcanic tephra. The input parameters into the model, such as total erupted mass, eruption column height, and bulk grain-size and components distribution, were obtained by best-fitting field data. We carried out tens of thousands simulations using a statistical set of wind profiles, obtained from NOAA re-analysis. Probability maps, relative to the considered scenarios, were constructed for several tephra loads, such as 200, 300 and 400 kg/m². These provide a hazard assessment for roof collapses due to tephra loading that can be used for risk mitigation plans in the area.     

Particle recycling in volcanic plumes

We have developed a new theoretical model of an eruption column that accounts for the re-entrainment of particles as they fall out of the laterally spreading umbrella cloud. The model illustrates how the mass flux of particles in the plume may increase with height in the plume, by a factor as large as 2.5 because of this recycling. Three important consequences are that (1) the critical velocity required to generate a buoyant eruption column for a given mass flux increases, (2) the total height of rise of the column may decrease, and (3) we infer that in relatively wind-free environments, for eruption columns near the conditions for collapse, the recycling of particles may lead to an unsteady oscillating motion of the plume, which, in time, may lead to the formation of interleaved fall and flow deposits.     

Sedimentation of tephra by volcanic plumes: I. Theory and its comparison with a study of the Fogo A plinian deposit,Sao Miguel (Azores)

Sedimentation of ejecta from volcanic plumes has been studied as a function of distance from the source in the Fogo A plinian deposit, Sao Miguel, Azores. The Fogo A trachytic pumice deposit is reversely graded and can be divided into two parts on the basis of pumice colour, abundance of syenite accessory lithic clasts and distribution. The lower syenite-poor part was dispersed to the south and was clearly influenced by wind. The upper syenite-rich part is coarsegrained and has a nearly symmetrical distribution around the vent. Elongation of isopachs to the east indicate a weak wind influence. The grain-size variations of lithic and crystal components in the upper coarse part were studied. Total accumulation and accumulation per unit area (expressed in kg/m²) show good fits to a gaussian function at distances greater than 7 km for grain diameters less than 2 cm. These results agree with a theoretical model for a radially spreading turbulent current moving over a quiescent fluid. The gaussian coefficient is shown to be a function of grain size and the flow rate of material into the umbrella region of the eruption column. The coefficient is therefore also a function of column height. The column height deduced from these data is 21 km, which is in broad agrrement with the column height of 27 km deduced from maximum clast dispersal using the method of Carey and Sparks (1986). The accumulation of clasts larger than 2 cm agrees with a theory for the fallout of clasts from the margins of the ascending eruption column, which treats the plume as a succession of large eddies that decrease their mass of particles as an exponential function of time. Calculations are also presented for the influence of the radial inflow of surrounding air into the column on the deposition of clasts. These calculations constrain the wind speed during the later part of the Fogo A eruption to be at most a few metres per second. The study has allowed four different dynamic categories of clast behaviour to be recognised in eruption columns.     

钢筋混凝土预制拼装柱扭转力学性能数值模拟与参数分析

采用ABAQUS大型有限元分析软件开展预制拼装柱扭转力学性能数值模拟及参数分析,研究了轴压比、灌浆套筒位置及长度、预制构件拼接缝界面黏结强度对灌浆套筒连接中柱抗扭性能的影响。研究结果表明:轴压比会显著影响预制拼装柱抗扭承载力和变形,而灌浆套筒位置和长度对预制拼装柱抗扭力学性能的影响不明显;预制构件拼接缝界面黏结强度显著影响预制拼装柱抗扭性能。基于预制拼装构件和现浇构件力学性能的对比分析,提出了轴向荷载和扭矩共同作用下灌浆套筒连接预制拼装柱抗扭承载力设计方法。     

Volcanoes, the stratosphere, and climate

A list of volcanic eruption plumes observed to ascend into or near the stratosphere since 1883 shows that the volcanoes divide readily into two groups, one at low and one at higher latitudes. A model for the rise of a buoyant volcanic plume rise as applied to volcanic eruptions is corrected for realistic temperature profiles and for the finite vertical extent of the resultant debris clouds. The utility of the model can be questioned, however, owing to the highly uncertain and variable nature of the efficiency of use of heat energy of buoyant rise. The observed correlation of stratospheric plumes with climatic effects indicates that those plumes nearer the equator have the largest impact on surface temperatures. Analysis of the observations also suggests that injection of debris into the stratosphere is more important in determining the effect on climate than either the total volcanic explosivity of the eruption or the actual height reached within the stratosphere.     

Seismicity and magma supply rate of the 1998 failed eruption at Iwate volcano,Japan

Iwate volcano, Japan, showed significant volcanic activity including earthquake swarms and volcano inflation from the beginning of 1998. A large earthquake of magnitude 6.1 hit the south-west of the volcano on September 3. Although a 1 km² fumarole field formed, blighting plants on the ridge in the western part of the volcano in the spring of 1999, no magmatic eruptions occurred. We reconcile the spatio-temporal distributions of volcanic pressure sources determined by previously reported studies in which GPS, strain and tilt data from dense geodetic station networks are analyzed (Miura et al. Earth Planet Space 52:1003–1008, 2000; Sato and Hamaguchi J Volcanol Geotherm Res 155:244–262, 2006). We calculate the magma supply rates from their results and compare them with the occurrence rates of volcanic earthquakes. The results show that the magma supply rates are almost constant or even decrease with time while the earthquake occurrence rate increases with time. This contrast in their temporal changes is interpreted to result from stress accumulation in the volcanic edifice caused by constant magma supply without effusion of magma to the surface. We further show that data showing slight acceleration in strain can be best explained by magma ascent at a constant velocity, and that there is no evidence for increased magma buoyancy resulting from gas bubble growth. This consideration supports the interpretation that the magma stayed at 2 km depth and horizontally migrated. These findings relating magma supply rate and seismicity to magma ascent process are clues to understanding why no magmatic eruption occurred at Iwate volcano in 1998.     

The 1996 eruption at Gjálp,Vatnajökull ice cap,Iceland: efficiency of heat transfer,ice deformation and subglacial water pressure

The 13-day-long Gjálp eruption within the Vatnajökull ice cap in October 1996 provided important data on ice–volcano interaction in a thick temperate glacier. The eruption produced 0.8 km³ of mainly volcanic glass with a basaltic icelandite composition (equivalent to 0.45 km³ of magma). Ice thickness above the 6-km-long volcanic fissure was initially 550–750 m. The eruption was mainly subglacial forming a 150–500 m high ridge; only 2–4% of the volcanic material was erupted subaerially. Monitoring of the formation of ice cauldrons above the vents provided data on ice melting, heat flux and indirectly on eruption rate. The heat flux was 5–6×10⁵ W m^-2 in the first 4 days. This high heat flux can only be explained by fragmentation of magma into volcanic glass. The pattern of ice melting during and after the eruption indicates that the efficiency of instantaneous heat exchange between magma and ice at the eruption site was 50–60%. If this is characteristic for magma fragmentation in subglacial eruptions, volcanic material and meltwater will in most cases take up more space than the ice melted in the eruption. Water accumulation would therefore cause buildup of basal water pressure and lead to rapid release of the meltwater. Continuous drainage of meltwater is therefore the most likely scenario in subglacial eruptions under temperate glaciers. Deformation and fracturing of ice played a significant role in the eruption and modified the subglacial water pressure. It is found that water pressure at a vent under a subsiding cauldron is substantially less than it would be during static loading by the overlying ice, since the load is partly compensated for by shear forces in the rapidly deforming ice. In addition to intensive crevassing due to subsidence at Gjálp, a long and straight crevasse formed over the southernmost part of the volcanic fissure on the first day of the eruption. It is suggested that the feeder dyke may have overshot the bedrock–ice interface, caused high deformation rates and fractured the ice up to the surface. The crevasse later modified the flow of meltwater, explaining surface flow of water past the highest part of the edifice. The dominance of magma fragmentation in the Gjálp eruption suggests that initial ice thickness greater than 600–700 m is required if effusive eruption of pillow lava is to be the main style of activity, at least in similar eruptions of high initial magma discharge.Editorial responsibility: J. Donnelly-Nolan     

Dynamical analysis of volcanic explosion

Visible phenomena accompanied by volcanic explosions at Sakurajima Volcano in Kyushu, Japan, were recorded by means of a TV camera and still cameras to make clear the process of explosive eruption of a Vulcanian type by image analysis and to enable a discussion of the process of explosive eruption. The most interesting phenomenon observed by the TV camera was visible shock waves passing through the atmosphere above the crater. The instant disappearance of thin clouds and the condensation of dense clouds were induced by the passage of shock waves. Explosion-quakes, which occurred at a depth of 1–2 km beneath the active crater, clearly preceded the explosion at the crater bottom. The atmospheric shock waves were generated in the crater 1.1–1.5 seconds later than the occurrence of the explosion-quake and propagated with the velocity of Mach 1.3–1.5 in a height range from 300 m to 600 m above the crater. Eruption clouds expanded subspherically for several seconds after the ejection and then the eruption column developed upwards at a certain velocity. The maximum ejection velocity of volcanic blocks, which was obtained from the analysis of photo-trajectories, was 112–157 m/sec. The internal pressure which ejected the volcanic blocks was estimated to be 138–271 bars in the case of the explosive eruptions analyzed. The results of analysis suggest that a high-pressure gas chamber was formed just beneath the crater bottom before the explosive eruption and that pressure waves caused by the explosion-quake acted as the trigger for the explosive eruption.     

Volcaniclastic deposits from the North Arch volcanic field, Hawaii: explosive fragmentation of alkalic lava at abyssal depths

Submarine explosive eruptions are generally considered to become less likely with increasing depth due to the increasing hydrostatic pressure of the overlying water column. Volcaniclastic deposits from the North Arch volcanic field, north of Oahu, have textural characteristics of explosive fragmentation yet were erupted in water depths greater than 4,200 m. The most abundant volcaniclastic samples from North Arch are clast-supported with highly vesicular, angular pyroclasts. They are most likely near-vent pyroclastic fall deposits formed in eruption columns of limited height. Interbedded with highly vesicular pillow lava, they form low (50 to 200 m), steep-sided cones around the vents. Less common are stratified samples with graded bedding; one such sample includes a layer of roughly aligned, platy, bubble-wall glass fragments (resembling littoral limu o Pele) that may have been deposited by density currents. In addition to bubble-wall glass shards, numerous glass fragments with spherical, delicate spindle and ribbon shapes, and Pele's hair-like glass strands occur in the finer size fraction (<0.5 mm) of some samples. They are probably more distal fallout. Another sample, consisting of glass fragments dispersed in a marine clay matrix, was apparently reworked and deposited farther from the vents by bottom currents. Glass compositions include low-(∼0.4-0.6 wt%) and medium-K₂O (>0.6 wt%) alkalic basalt, basanite, and nephelinite. Sulfur and chlorine abundances are high, reaching a maximum of 1,800 and 1,300 ppm, respectively. The ubiquitous presence of limu o Pele fragments, regardless of glass composition, suggests that bursts of Strombolian-like activity accompanied most eruptions. Coalescing vesicles observed in larger pyroclasts and some pillow lava suggests accumulation of volatiles. Since the great hydrostatic pressure makes steam expansion impossible, a volatile-rich, supercritical magmatic fluid probably drove the eruptions. If these volatile-rich magmas had erupted in shallow water or subaerially, tall fountains would most likely have resulted. The great hydrostatic pressure (>40 MPa) limited fountain and eruption column heights.     

The dimensions and dynamics of volcanic eruption columns

Eruption columns can be divided into three regimes of physical behaviour. The basal gas thrust region is characterized by large velocities and decelerations and is dominated by momentum. This region is typically a few hundred metres in height and passes upwards into a much higher convective region where buoyancy is dominant. The top of the convective region is defined by the level of neutral density (heightH _B ) where the column has a bulk density equal to the surrounding atmosphere. Above this level the column continues to ascend to a heightH _T due to its momentum. The column spreads horizontally and radially outwards between heightH _T andH _B to form an umbrella cloud. Numerical calculations are presented on the shape of eruption columns and on the relationships between the heightH _B and the mass discharge rate of magma, magma temperature and atmospheric temperature gradients. Spreading rate of the column margins increases with height principally due to the decrease in the atmospheric pressure. The relationship between column height and mass discharge rate shows good agreement with observations. The temperature inversion above the tropopause is found to only have a small influence on column height and, eruptions with large discharge rates can inject material to substantially greater heights than the inversion level. Approximate calculations on the variation of convective velocities with height are consistent with field data and indicate that columns typically ascend at velocities from a few tens to over 200 m/s. In very large columns (greater than 30 km) the calculated convective velocities approach the speed of sound in air, suggesting that compressibility effects may become important in giant columns. Radial velocities in the umbrella region where the column is forced laterally into the atmosphere can be substantial and exceed 55 m/s in the case of the May 18th Mount St. Helens eruption. Calculations on motions in this region imply that it plays a major role in the transport of coarse pyroclastic fragments.     

Simulation of the 1980 eruption of Mount St. Helens using the ash-tracking model PUFF

The dispersal of volcanic ash from the May 18, 1980 eruption of Mount St. Helens (MSH) has been simulated using the Lagrangian ash-tracking model PUFF. Previous applications of the model were limited to smaller, short-lived eruptions with ash dispersal occurring mainly within the troposphere. Two high-resolution atmospheric reanalysis datasets (ERA-40 and NCEP/NCAR-40) allowed MSH ash cloud dispersal to be simulated up to 30 km elevation. The 1980 eruption was divided into two distinct eruptive phases, (1) an initial, relatively short-lived blast/surge phase that injected ash up to 30 km and (2) a subsequent nine-hour plinian phase that maintained an average eruption column height of 16 km. Using PUFF, the two phases of the MSH eruption were modeled separately based on a range of individual input parameters and then combined to produce an integrated simulation of the entire eruption. The trajectory and areal extent of the modeled atmospheric ash cloud best match the actual distribution of MSH ash when input parameters are set to values inferred from satellite and radar data collected on May 18, 1980. The prevailing wind field exerts the strongest control on the advection and ultimate position of the modeled ash cloud, making the maximum column height and the vertical distribution of ash the most sensitive of the PUFF input parameters for this event. The results indicate that the PUFF model works well at simulating the dispersal of ash injected well into the lower stratosphere from a moderate, relatively long-lived eruption, such as MSH. However, attempts to use PUFF to recreate some granulometric aspects of the MSH fallout deposit, such as the maximum particle size as a function of distance from source, were not successful. PUFF consistently predicts much greater fallout distances for small ash particles (< 500 µm) than actually observed in the MSH deposit. The effective settling velocities used by the PUFF model appear to be too slow to accurately predict fallout distances of small ash particles. As a consequence the PUFF model may overestimate the duration of ash loading in the atmosphere associated with the distal fine ash component of explosive eruptions.     

Caldera formation by magma withdrawal from a reservoir beneath a volcanic edifice

Caldera formation has been explained by magma withdrawal from a crustal reservoir, but little is known about the conditions that lead to the critical reservoir pressure for collapse. During an eruption, the reservoir pressure is constrained to lie within a finite range: it cannot exceed the threshold value for eruption, and cannot decrease below another threshold value such that feeder dykes get shut by the confining pressure, which stops the eruption. For caldera collapse to occur, the critical reservoir pressure for roof failure must therefore be within this operating range. We use an analytical elastic model to evaluate the changes of reservoir pressure that are required for failure of roof rocks above the reservoir with and without a volcanic edifice at Earth's surface. With no edifice at Earth's surface, faulting in the roof region can only occur in the initial phase of reservoir inflation and affects a very small part of the focal area. Such conditions do not allow caldera collapse. With a volcanic edifice, large tensile stresses develop in the roof region, whose magnitude increase as the reservoir deflates during an eruption. The edifice size must exceed a threshold value for failure of the roof region before the end of eruption. The largest tensile stresses are reached at Earth's surface, indicating that faulting starts there. Failure affects an area whose horizontal dimensions depend on edifice and chamber dimensions. For small and deep reservoirs, failure conditions cannot be achieved even if the edifice is very large. Quantitative predictions are consistent with observations on a number of volcanoes.