Dome growth at the Soufriere Hills volcano (1996 to 1998) was frequently accompanied by repetitive cycles of earthquakes, ground deformation, degassing, and explosive eruptions. The cycles reflected unsteady conduit flow of volatile-charged magma resulting from gas exsolution, rheological stiffening, and pressurization. The cycles, over hours to days, initiated when degassed stiff magma retarded flow in the upper conduit. Conduit pressure built with gas exsolution, causing shallow seismicity and edifice inflation. Magma and gas were then expelled and the edifice deflated. The repeat time-scale is controlled by magma ascent rates, degassing, and microlite crystallization kinetics. Cyclic behavior allows short-term forecasting of timing, and of eruption style related to explosivity potential.

[1] The 7.5 ka Socompa sector collapse emplaced 25 km 3 of fragmented rock as a thin, but widespread (500 km 2 ), avalanche deposit, followed by late stage sliding of 11 km 3 as Toreva blocks. Most of the avalanche mass was emplaced dry, although saturation of a basal shear layer cannot be excluded. Modeling was carried out using the depth-averaged granular flow equations in order to provide information on the flow behavior of this well-preserved, long run-out avalanche. Results were constrained using structures preserved on the surface of the deposit, as well as by deposit outline and run-up (a proxy for velocity). Models assuming constant dynamic friction fail to produce realistic results because the low basal friction angles (1 to 3.5°) necessary to generate observed run-out permit neither adequate deposition on slopes nor preservation of significant morphology on the deposit surface. A reasonable fit is obtained, however, if the avalanche is assumed simply to experience a constant retarding stress of 50-100 kPa during flow. This permits long run-out as well as deposition on slopes and preservation of realistic depositional morphology. In particular the model explains a prominent topographic escarpment on the deposit surface as the frozen front of a huge wave of debris reflected off surrounding hills. The result that Socompa avalanche experienced a small, approximately constant retarding stress during emplacement is consistent with a previously published analysis of avalanche data.Citation: Kelfoun, K., and T. H. Druitt (2005), Numerical modeling of the emplacement of Socompa rock avalanche, Chile,

Crystals in volcanic rocks provide records of magma-reservoir processes and timescales prior to eruptions 7. A crystal growing from a magmatic melt incorporates trace elements in quantities governed by thermodynamic and kinetic laws 8,9 . If the crystal is subsequently mixed into another melt, trace elements that 2 diffuse sufficiently fast will begin to re-equilibrate with the new host melt, generating intra-crystalline diffusion gradients than can be used to obtain time information [10][11][12] (Supplementary Fig. S1). (Fig. 1g). In some crystals, the core also contains a euhedral to anhedral inner core of An 88-58 ('calcic inner core' ; Fig 1 a, c). A spectrum of type 1 crystals is observed from those with thick rims (up to 100 µm), to those with thin rims, to those in which the rim is absent (Fig. 1f). In rimmed type 1 crystals the plagioclase in contact with the host glass is ~ An 40, whereas in rimless ones it is ~An 50 (i.e., the core composition). Rimmed type-1 crystals occur in pumices from all four eruptive phases, but rimless ones have only been observed in pumices from phase 4. Type 2 crystals are very rare; they are reversely zoned, with cores of An 36-30 mantled by rims identical to those of type 1 (Fig. 1e). The broad range of plagioclase compositions in Minoan pumice shows that the rhyodacite was the product of open-system magmatic processes involving multiple, compositionally diverse magmas.Trace elements were analysed in four rimmed type-1 crystals, one rimless type-1 crystal, some interstitial glasses and some inclusions of glass contained within the crystals ( Fig. 2; Tables S1 and S2). Mg, Sr and Ti are particularly useful elements for characterizing coexisting melt compositions and mixing time scales because they partition differently between melt and plagioclase 8 and diffuse at different rates. Mg diffuses faster than Sr 7 ; Ti probably diffuses slowly due to its high charge. Published An-dependent partition coefficients 8 were used to invert melt trace element contents to those of equivalent plagioclase (Fig. 2), and vice versa (Fig. 3).Rimmed type-1 crystals have core-to-rim gradients in all the three elements (Fig. 2). Rim compositions record equilibrium with the interstitial glass, but concentrations of Mg, Sr, and Ti in the cores and calcic inner cores are significantly higher than those calculated to be in equilibrium with the glass (green lines, Fig. 2). The rimless type-1 crystal has concentrations of Mg and Sr throughout that are too high to have been in equilibrium with the glass. None of these crystals resided in the host melt long enough for any of these the elements to reach total equilibrium with the host melt.This observation is reinforced by a comparison of calculated Mg, Ti, and Sr concentrations of melts in equilibrium with the different plagioclase zones (referred to as Mg melt , Sr melt , and Ti melt ), with those of Santorini lavas, pumices, and glasses. The latter represent an approximate liquid line of descent of the magmatic system 3 (Fig. 3a, b). The calculated c...

No abstract

Pyroclastic flows were formed at Soufrière Hills Volcano by lava-dome collapse and by fountain collapse associated with Vulcanian explosions. Major episodes of dome collapse, lasting tens of minutes to a few hours, followed escalating patterns of progressively larger flows with longer runouts. Block-and-ash flow deposit volumes range from <0.1 to 25 x 106 m3 with runouts of 1-7 km. The flows formed coarse-grained block-and-ash flow deposits, with associated fine-grained pyroclastic surge deposits and ashfall deposits. Small flows commonly formed lobate channelized deposits. Large block-and-ash flows in unconfined areas produced sheet-like deposits with tapering margins. the development of pyroclastic surges was variable depending on topography and dome pore pressure. Pyroclastic surge deposits commonly had a lower layer poor in fine ash that was formed at the current front and an upper layer rich in fine ash. Block-and-ash flows were erosive, producing striated and scoured bedrock surfaces and forming channels, many metres deep, in earlier deposits. Abundant accidental material was incorporated. Pyroclastic flow deposits formed by fountain collapse were pumiceous, with narrow sinuous, lobate morphologies and well developed levees and snouts. Two coastal fans formed where pyroclastic flows entered the sea. Their seaward extent was limited by a submarine slope break.

No abstract

AbstrctSantorini volcanic field has had 12 major (1–10 km3 or more of magma), and numerous minor, explosive eruptions over the last ~ 200 ka. Deposits from these eruptions (Thera Pyroclastic Formation) are well exposed in caldera-wall successions up to 200 m thick. Each of the major eruptions began with a pumice-fall phase, and most culminated with emplacement of pyroclastic flows. Pyroclastic flows of at least six eruptions deposited proximal lag deposits exposed widely in the caldera wall. The lag deposits include coarse-grained lithic breccias (andesitic to rhyodacitic eruptions) and spatter agglomerates (andesitic eruptions only). Facies associations between lithic breccia, spatter agglomerate, and ignimbrite from the same eruption can be very complex. For some eruptions, lag deposits provide the only evidence for pyroclastic flows, because most of the ignimbrite is buried on the lower flanks of Santorini or under the sea. At least eight eruptions tapped compositionally heterogeneous magma chambers, producing deposits with a range of zoning patterns and compositional gaps. Three eruptions display a silicic–silicic + mafic–silicic zoning not previously reported. Four eruptions vented large volumes of dacitic or rhyodacitic pumice, and may account for 90% or more of all silicic magma discharged from Santorini. The Thera Pyroclastic Formation and coeval lavas record two major mafic-to-silicic cycles of Santorini volcanism. Each cycle commenced with explosive eruptions of andesite or dacite, accompanied by construction of composite shields and stratocones, and culminated in a pair of major dacitic or rhyodacitic eruptions. Sequences of scoria and ash deposits occur between most of the twelve major members and record repeated stratocone or shield construction following a large explosive eruption.Volcanism at Santorini has focussed on a deep NE–SW basement fracture, which has acted as a pathway for magma ascent. At least four major explosive eruptions began at a vent complex on this fracture. Composite volcanoes constructed north of the fracture were dissected by at least three caldera-collapse events associated with the pyroclastic eruptions. Southern Santorini consists of pryoclastic ejecta draped over a pre-volcanic island and a ridge of early- to mid-Pleistocene volcanics. The southern half of the present-day caldera basin is a long-lived, essentially non-volcanic, depression, defined by topographic highs to the south and east, but deepened by subsidence associated with the main northern caldera complex, and is probably not a separate caldera.

International audienceThe ability of a dense pyroclastic flow to maintain high gas pore pressure, and hence low friction, during runout is determined by (1) the strengths and longevities of gas sources, and (2) the ability of the material to retain residual gas once those sources become ineffective. The latter is termed the gas retention capacity. Gas retention capacity in a defluidizing granular material is governed by three timescales: one for the evacuation of bubbles (t be ; brief and not considered in this paper), one for hindered settling from the expanded state (t sett), and one for diffusive release of residual pore pressure from the non-expanded state (t diff). The relative magnitides of t sett and t diff depend on bed thickness, t sett dominating in thin systems and t diff in thick ones. Three pyroclastic flow materials, two ignimbrites and a block-and-ash flow sample, were studied experimentally to investigate expansion behaviour under gas flow and to determine gas retention times. Effects of particle size were evaluated by using two size cuts (<4 mm and <250 μm) from each sample. Careful drying of the materials was necessary to avoid effects of humidity-related cohesion. Two sets of experiments were carried out: (1) expansion in the non-bubbling regime at 50–200°C, (2) bed collapse tests from the initially bubbling state at 50–550°C. Provided that gas channelling was avoided by gentle stirring, all the samples exhibited a regime of uniform expansion prior to the onset of bubbling. Fine particle size (in particular high fines content), low particle density and high temperature all favoured smoother fluidization by increasing the maximum expansion possible in the non-bubbling state. An empirical equation describing the uniform expansion of the materials was determined. High temperature also favoured greater gas partitioning into the dense phase of the bubbling bed, as well (in finer-grained samples) as higher voidage in the settled bed. Large values of t sett and t diff were favoured by fine particle size. Temperature had less influence, suggesting that experimental results at low temperatures (50–200°C) can be extrapolated to higher temperatures. Gas retention times provide insight into the ability of pyroclastic flows in expanded (t sett) or non-expanded (t diff) flow states to retain gas once air ingestion or gas production have become ineffective. Finer-grained pyroclastic flows are expected to retain gas longer, and hence to have higher apparent ‘mobilities', than coarser-grained ones of comparable volume, as has been observed on Montserrat