Search

Ongoing resurgence affects Campi Flegrei caldera (Italy) via bradyseism, i.e. a series of ground deformation episodes accompanied by increases in shallow seismicity. In this study, we perform a mathematical analysis of the GPS and seismic data in the instrumental catalogs from 2000 to 2020, and a comparison of them to the preceding data from 1983 to 1999. We clearly identify and characterize two overlying trends, i.e. a decennial-like acceleration and cyclic oscillations with various periods. In particular, we show that all the signals have been accelerating since 2005, and 90–97% of their increase has occurred since 2011, 40–80% since 2018. Nevertheless, the seismic and ground deformation signals evolved differently—the seismic count increased faster than the GPS data since 2011, and even more so since 2015, growing faster than an exponential function The ground deformation has a linearized rate slope, i.e. acceleration, of 0.6 cm/yr2 and 0.3 cm/yr2 from 2000 to 2020, respectively for the vertical (RITE GPS) and the horizontal (ACAE GPS) components. In addition, all annual rates show alternating speed-ups and slow-downs, consistent between the signals. We find seven major rate maxima since 2000, one every 2.8–3.5 years, with secondary maxima at fractions of the intervals. A cycle with longer period of 6.5–9 years is also identified. Finally, we apply the probabilistic failure forecast method, a nonlinear regression that calculates the theoretical time limit of the signals going to infinity (interpreted here as a critical state potentially reached by the volcano), conditional on the continuation of the observed nonlinear accelerations. Since 2000, we perform a retrospective analysis of the temporal evolution of these forecasts which highlight the periods of more intense acceleration. The failure forecast method applied on the seismic count from 2001 to 2020 produces upper time limits of [0, 3, 11] years (corresponding to the 5th, 50th and 95th percentiles, respectively), significantly shorter than those based on the GPS data, e.g. [0, 6, 21] years. Such estimates, only valid under the model assumption of continuation of the ongoing decennial-like acceleration, warn to keep the guard up on the future evolution of Campi Flegrei caldera.

Episodes of slow uplift and subsidence of the ground, called bradyseism, characterize the recent dynamics of the Campi Flegrei caldera (Italy). In the last decades two major bradyseismic crises occ...

Degassing-induced crystallization in volatile rich intermediate composition magmas results in material stiffening and strengthening that prior to solidification is reflected in non-Newtonian rheology. We explore the effects of a spectrum of such rheological regimes on eruptive style and morphologic evolution of lava domes, using a two-dimensional (2D) particle-dynamics model for a spreading viscoplastic (Bingham) fluid. We assume that the ductile magma core of a 2D synthetic lava dome develops finite yield strength, and that deformable frictional talus evolves from a carapace that caps the magma core. Our new model is calibrated against an existing analytical model for a spreading viscoplastic lava dome and is further compared against observational data of lava dome growth. Results indicate that a degassing-induced increase in strength of the injected magma causes a transition in the lava dome morphology from a dome with low surface relief evolving endogenously (with apparent bulk yield strength - 104 105 – 106 Pa) extruded through the dome carapace. The virtual lava dome with τ0a = 0.6 MPa shows good agreement with the observed dome heights observed at the Soufriere Hills Volcano, Montserrat during a period of endogenous growth. The calculated apparent flow viscosity (1.36 × 1011 Pa·s for τ0a = 0.6 MPa) is in the range of estimated viscosities (109 to 1012 Pa·s) for andesitic-dacitic crystal-rich lavas. Our model results indicate a strong correlation between apparent yield strength and dome morphology, with both controlled by degassing-induced crystallization and extrusion rate.

This study of burns patients from two eruptions of Merapi volcano, Java, in 1994 and 2010, is the first detailed analysis to be reported of thermal injuries in a large series of hospitalised victims of pyroclastic surges, one of the most devastating phenomena in explosive eruptions. Emergency planners in volcanic crises in populated areas have to integrate the health sector into disaster management and be aware of the nature of the surge impacts and the types of burns victims to be expected in a worst scenario, potentially in numbers and in severity that would overwhelm normal treatment facilities. In our series, 106 patients from the two eruptions were treated in the same major hospital in Yogyakarta and a third of these survived. Seventy-eight per cent were admitted with over 40% TBSA (total body surface area) burns and around 80% of patients were suspected of having at least some degree of inhalation injury as well. Thirty five patients suffered over 80% TBSA burns and only one of these survived. Crucially, 45% of patients were in the 40–79% TBSA range, with most suspected of suffering from inhalation injury, for whom survival was most dependent on the hospital treatment they received. After reviewing the evidence from recent major eruptions and outlining the thermal hazards of surges, we relate the type and severity of the injuries of these patients to the temperatures and dynamics of the pyroclastic surges, as derived from the environmental impacts and associated eruption processes evaluated in our field surveys and interviews conducted by our multi-disciplinary team. Effective warnings, adequate evacuation measures, and political will are all essential in volcanic crises in populated areas to prevent future catastrophes on this scale.

We provide structural and paleomagnetic evidence for major distributed rotational shear adjacent to and including the active Husavik-Flatey transform fault in the southern part of the Tjornes Fracture Zone of north Iceland. In the Flateyjarskagi peninsula immediately south of the fault, dike and bedding trends gradually change over about 11 km in approaching the transform fault main trace. In map view this pattern is seen as a progressive clockwise curvature of structural elements that equals or in some cases exceeds 90°. We present paleomagnetic analyses that verify large clockwise rotations of the same order as the curvature of structural elements observed. We compare our field and paleomagnetic observations with results from careful analog modeling that demonstrate how regional curvature of structures results from distributed strain that dominates the early stages of transform shear deformation, prior to the appearance of a transform fault system comprised of discrete shears. Our paleomagnetic results were obtained from one relatively undisturbed reference site, and three sites of deformed rocks in Flateyjarskagi and two in Tjornes peninsula. These results indicate large vertical-axis rotations, and the progressive deflection of dike trends throughout northern Flateyjarskagi also serves as a separate quantitative estimate of vertical-axis rotation. The clockwise vertical-axis rotation estimates of 71°–97° from dikes are less than those deduced from paleomagnetism, but the order of magnitude is correct and we attribute the differences as mainly reflecting imperfect bed attitude adjustments used in analyses. The problem is typical for deformed rocks from high-latitude sites with steeply inclined paleomagnetic vectors. We consider our field measurements of dikes as more accurate indicators of true tectonic rotation. This regional deformation of Miocene rocks occurred early in the evolution of the TFZ transform, probably beginning ca. 7 Ma, when an older rift zone present in western North Iceland jumped to its present NVZ location in northeast Iceland. The early deformations involved shear strain distributed over a region about 20-km broad and produced the curved structures. This phase was followed by the sequential development of more narrowly distributed discrete shears, generating the initial Husavik-Flatey transform fault.

One of the World's premier field geologists, Kristjan Saemundsson led immense geological mapping programs and authored or co-authored nearly all geological maps of Iceland during the past half century, including the first modern bedrock and tectonic maps of the whole country. These monumental achievements collectively yield the most inclusive view of an extensional plate boundary anywhere on Earth. When Kristjan began his work in 1961, the relation of Iceland to sea-floor spreading was not clear, and plate tectonics had not yet been invented. Kristjan resolved key obstacles by demonstrating that the active rifting zones in Iceland had shifted over time and were linked by complex transforms to the mid-ocean spreading ridge, thus making the concept of sea-floor spreading in Iceland acceptable to those previously skeptical. Further, his insights and vast geological and tectonic knowledge on both high- and low-temperature geothermal areas in Iceland yielded a major increase in knowledge of geothermal systems, and probably no one has contributed more than he to Icelandic energy development. Kristjan's legacy is comprised by his numerous superb maps on a variety of scales, the high quality papers he produced, the impactful ideas generated that were internationally diffused, and the generations of colleagues and younger people he inspired, mentored, or otherwise positively influenced with his knowledge and generous attitude.

We introduce a doubly stochastic method for performing material failure theory based forecasts of volcanic eruptions. The method enhances the well known Failure Forecast Method equation, introducing a new formulation similar to the Hull-White model in financial mathematics. In particular, we incorporate a stochastic noise term in the original equation, and systematically characterize the uncertainty. The model is a stochastic differential equation with mean reverting paths, where the traditional ordinary differential equation defines the mean solution. Our implementation allows the model to make excursions from the classical solutions, by including uncertainty in the estimation. The doubly stochastic formulation is particularly powerful, in that it provides a complete posterior probability distribution, allowing users to determine a worst case scenario with a specified level of confidence. We apply the new method on historical datasets of precursory signals, across a wide range of possible values of convexity in the solutions and amounts of scattering in the observations. The results show the increased forecasting skill of the doubly stochastic formulation of the equations if compared to statistical regression., 27 pages, 13 figures

Lava domes are structures that grow by the extrusion of viscous silicic or intermediate composition magma from a central volcanic conduit. Repeated cycles of growth are punctuated by collapse, as the structure becomes oversized for the strength of the composite magma that rheologically stiffens and strengthens at its surface. Here we explore lava dome growth and failure mechanics using a two-dimensional particle-dynamics model. The model follows the evolution of fractured lava, with solidification driven by degassing induced crystallization of magma. The particle-dynamics model emulates the natural development of dome growth and rearrangement of the lava dome which is difficult in mesh-based analyses due to mesh entanglement effects. The deformable talus evolves naturally as a frictional carapace that caps a ductile magma core. Extrusion rate and magma rheology together with crystallization temperature and volatile content govern the distribution of strength in the composite structure. This new model is calibrated against existing observational models of lava dome growth. Results show that the shape and extent of the ductile core and the overall structure of the lava dome are strongly controlled by the infusion rate. The effects of extrusion rate on magma rheology are sensitive to material stiffness, which in turn is a function of volatile content and crystallinity. Material stiffness and material strength are key model parameters which govern magma rheology and subsequently the morphological character of the lava dome and in turn stability. Degassing induced crystallization causes material stiffening and enhances material strength reflected in non-Newtonian magma behavior. The increase in stiffness and strength of the injected magma causes a transition in the style of dome growth, from endogenous expansion of a ductile core, to stiffer and stronger intruding material capable of punching through the overlying material and resulting in the development of a spine or possibly inducing dome collapse. Simulation results mimic development of a megaspine upon the influx of fresh magma which leads to the re-direction of magma flow, creating a new shear zone and the switching of dome growth from one side to the other. Our model shows similar dome growth dynamics as observed at Soufriere Hills Volcano, Montserrat, indicating a strong correlation between extrusion rate and its subsequent effect on mechanical properties and variations in magma rheology.

We use histories of magma efflux and surface deformation from a continuously operating global positioning system (cGPS) to quantitatively constrain magma transfer within the deep crustal plumbing of the Soufriere Hills Volcano (SHV). Displacement records reach a surface aperture of approximately 11 km and are continuous over three successive cycles of eruption followed by a pause spanning 1995–2008, and we focus on data of this time period. The assumed geometry and flow topology is for twin vertically stacked spherical chambers pierced by a vertical conduit that transmits magma from the deep crust to the surface. For a compressible magma column within an elastic crust we use mean deformation rates measured at between 6 and 13 cGPS stations for periods of effusion then repose and the time-history of magma efflux to define optimal chamber depths and basal magma input. The best fit for a constrained constant basal input to the system is obtained for chambers at 5 and 19 km, and a constant magma input rate of approximately 1.2 m 3 s −1 . Eruptive then pause episodes are, respectively, characterized by synchronous deflation then inflation of both shallow and deep chambers. Throughout this period of three repeated episodes of effusion then repose, the total effusive volume ( c. 0.95 km 3 dense rock equivalent, DRE) has been sourced half from the lower chamber ( c. 0.5 km 3 ) and half from below this chamber ( c. 0.45 km 3 ). A consistent observation, repeated through three episodes, is that the eruption restarts as the shallow chamber regains its original volume following the pause and that eruption rearrests when the shallow chamber has deflated by a small but constant volume change ( c. 16–22 Mm 3 ). This magmatic metering is consistent with a control on eruption periodicity that involves overpressured breaching of the shallow chamber followed by underpressured sealing. We contrast these observations with other contemporary models that have consistently placed an upper chamber at a depth of approximately 5–6 km, and deeper chambers at 12 km and deeper.

Since 1995 the eruption of the andesitic Soufriere Hills Volcano (SHV), Montserrat, has been studied in substantial detail. As an important contribution to this effort, the Seismic Experiment with Airgunsource-Caribbean Andesitic Lava Island Precision Seismo-geodetic Observatory (SEA-CALIPSO) experiment was devised to image the arc crust underlying Montserrat, and, if possible, the magma system at SHV using tomography and reflection seismology. Field operations were carried out in October–December 2007, with deployment of 238 seismometers on land supplementing seven volcano observatory stations, and with an array of 10 ocean-bottom seismometers deployed offshore. The RRS James Cook on NERC cruise JC19 towed a tuned airgun array plus a digital 48-channel streamer on encircling and radial tracks for 77 h about Montserrat during December 2007, firing 4414 airgun shots and yielding about 47 Gb of data. The main objecctives of the experiment were achieved. Preliminary analyses of these data published in 2010 generated images of heterogeneous high-velocity bodies representing the cores of volcanoes and subjacent intrusions, and shallow areas of low velocity on the flanks of the island that reflect volcaniclastic deposits and hydrothermal alteration. The resolution of this preliminary work did not extend beyond 5 km depth. An improved three-dimensional (3D) seismic velocity model was then obtained by inversion of 181 665 first-arrival travel times from a more-complete sampling of the dataset, yielding clear images to 7.5 km depth of a low-velocity volume that was interpreted as the magma chamber which feeds the current eruption, with an estimated volume 13 km3. Coupled thermal and seismic modelling revealed properties of the partly crystallized magma. Seismic reflection analyses aimed at imaging structures under southern Montserrat had limited success, and suggest subhorizontal layering interpreted as sills at a depth of between 6 and 19 km. Seismic reflection profiles collected offshore reveal deep fans of volcaniclastic debris and fault offsets, leading to new tectonic interpretations. This chapter presents the project goals and planning concepts, describes in detail the campaigns at sea and on land, summarizes the major results, and identifies the key lessons learned.

We analyze the textures and mineralogy of Merapi tephra generated during explosive VEI 3–4 eruptions over the past 2000 years, and compare these data with those observed for Merapi dome and flow lavas. We find that the Merapi pumiceous tephra and lava textures differ significantly with respect to small-size crystal populations, but that phenocryst textures are generally similar. A similar initial phase of crystallization is indicated for tephras and lavas in mid-crustal (> 10 km depth) reservoirs. Subsequent textural differences are mainly affected by ascent rate and degassing during ascent, and, for dome lavas, with temporary storage in shallower reservoirs. These differences also correspond to different eruptive styles. Our analyses include study of pumices, lava, and breadcrust-bomb samples, including samples from some of the most recent explosive episodes prior to the 2010 eruption. Textural analyses of youthful breadcrust bomb samples yield insights on one type of transition between effusive and explosive eruptive styles, involving pressure build-up under a degassed crystalline shallow-conduit plug. In general, comparison of the crystal size distributions and calculated residence times among the effusive and explosive eruptive styles suggests that the two main magma-product types resided for similar lengths of time in a mid-crustal reservoir, before ascending toward the surface and either erupting explosively (tephra), or stagnating in a shallow magma chamber prior to extrusion (lava). The interpretation is supported by the occurrences of amphibole in pristine condition in tephra and in altered state in lava. Finally the 1872 and 2010 explosive eruptions are examined and compared with others over the past three millennia. The 2010 bulk-rock compositions overlap with products of other major explosive Merapi eruptions, such as the Ngrangkah, Tegalsruni, Temusari, and Kepuharjo tephras. The 2010 products show a gradational late-stage mafic enrichment, dissimilar to 1872 which showed little (

We analyze and compare the textures of Merapi lavas (basalts and basaltic andesites) ranging in age from Proto-Merapi through modern activity, with the goal of gaining insights on the temporal evolution of Merapi's magmatic system. Analysis of textural parameters, such as phenocryst and microphenocryst crystallinity, coupled with crystal size distribution theory, provides information about the storage and transport of magmas. We combine textural analyses with geochemical investigations for a comprehensive comparison of erupted lavas over time. The chemical analyses identify crystal growth processes in magma chambers and underline differences between sample groups. Our work suggests the occurrence of two distinct histories, presumably associated with (at least) two generally distinct types of rheological behaviors and storage/transport systems. These behaviors are associated with different plagioclase growth patterns, with both groups influenced by late-stage shallow decompression degassing-induced microlite crystallization. Both groups contain amphibole crystals that indicate an early period of mid-crustal to deep-crustal storage of water-rich magmas. Dome lavas from the 20th century eruptive activity indicate quasi-steady-state nucleation-and-growth evolution interspersed with episodes of reheating and textural coarsening, suggesting residence in magma storage at multiple depths, both > 10 km, and

We identify and quantify controls on the timescales and magnitudes of cyclic (periodic) volcanic eruptions using the numerical model DOMEFLOW ( de’ Michieli Vitturi et al., 2010 ) which was developed by the authors for magma systems of intermediate composition. DOMEFLOW treats the magma mixture as a liquid continuum with dispersed gas bubbles and crystals in thermodynamic equilibrium with the melt and assumes a modified Poiseuille form of the viscous term for fully developed laminar flow in a conduit of cylindrical cross-section. During ascent, magma pressure decreases and water vapor exsolves and partially degasses from the melt as the melt simultaneously crystallizes, causing changes in mixture density and viscosity. Two mechanisms previously proposed to cause periodic eruption behavior have been implemented in the model and their corresponding timescales explored. The first applies a stick–slip model in which motion of a shallow solid plug is resisted by static/dynamic friction, as described in Iverson et al. (2006) . For a constant magma supply rate at depth, this mechanism yields cyclic extrusion with timescales of seconds to tens of seconds with values generally depending on assumed friction coefficients. The second mechanism does not consider friction but treats the plug as a high-viscosity Newtonian fluid. During viscous resistance, pressure beneath the degassed plug can increase sufficiently to overcome dome overburden, plug weight, and viscous forces, and ultimately drive the plug from the conduit. In this second model cycle periods are on the order of hours, and decrease with increasing magma supply rate until a threshold is reached, at which point periodicity disappears and extrusion rate becomes steady (vanishingly short periods). Magma volatile content for fixed chamber pressure has little effect on cycle timescales, but increasing volatile content increases mass flow rate and cycle magnitude as defined by the difference between maximum and minimum extrusion rates. Increasing mass flow rate also increases the magnitude of predicted deformation cycles. Both models have been applied to dome-building eruptions using conditions appropriate for the Soufriere Hills volcano, Montserrat. Results suggest that the modeled stick–slip mechanism cannot explain the cycles of extrusion and explosion well-documented at Montserrat (timescales of hours). However, a subset of the simulations for the viscous plug model are consistent with the Montserrat data and therefore demonstrate the feasibility of this second formulation.

To better understand the volcanic phenomena acting on Montserrat, the SEA-CALIPSO seismic experiment (Seismic Experiment with Airgun-source - Caribbean Andesitic Lava Island Precision Seismo-geodetic Observatory) was conducted in 2007 December with the aim of imaging the upper crust and the magmatic system feeding the active Soufriére Hills Volcano. The 3-D survey covered an area of about 50 × 40 km and involved the deployment of 247 land stations and ocean-bottom seismometers (OBSs). A subset of the data, recorded by four OBSs and four land stations on a southeast to northwest line, has been analysed, and traveltimes have been inverted to obtain a 2-D seismic velocity model through the island. Inverted phases include crustal and sediment P waves and wide-angle reflections. The resulting velocity model reveals the presence of a high velocity body (3.5-5.5 km s-1) beneath the island, with highest velocities beneath the Soufriére and Centre Hills, corresponding primarily to the cores of these volcanic edifices, built of a pile of andesite lava domes and subsequent intrusions. In the offshore region, velocities in the surficial sediment layer vary from 1.5 to 3.0 km s-1, consistent with a mainly calcareous and volcaniclastic composition. A wide-angle reflector is observed at a depth of ∼1200 m below the seabed, and appears to deepen beneath the island. The upper crust beneath this reflector has velocities of 4.0-6.0 km s-1 and is inferred to correspond to plutonic and hypabyssal rocks and sedimentary material of the old arc. The high velocity region beneath the island, extends into the crust to a depth of at least 5 km, and is believed to be caused by an intrusive complex, possibly of intermediate composition. A low velocity zone, as would be expected in the presence of an active magma chamber, was not observed perhaps due to the limited resolution beneath ∼5 km depth. Our results so far provide the first wide-angle seismic constraints on the upper crustal structure of the island to a depth of 10 km, and will help understanding the processes that drive volcanism at Montserrat and other island arc volcanoes. © 2009 The Authors Journal compilation © 2009 RAS.

Strikingly similar examples of edifice collapse and directed blast are the 18 May 1980 eruption of Mount St. Helens (MSH), Washington, USA, and the 30 March 1956 eruption of Bezymianny Volcano (BZ), Kamchatka, Russia. In these cases, flank failures led to near-instantaneous decompression and fragmentation of intra-edifice cryptodome magma, which produced catastrophic, laterally directed blasts. In both instances, the blast products consisted of juvenile material with bimodal density/vesicularity distributions: low- and high-density modes at 1900 and 2400 kg m− 3 for BZ, 1600 and 2300 kg m− 3 for MSH, although the proportion of high-density material is greater at BZ. Blast materials also exhibit striking variety in groundmass crystallinity ( 90 vol.%) despite having fairly uniform pheno-crystallinities, suggesting that degassing-driven groundmass crystallization occurred to varying extents within cryptodome magma at both volcanoes. New bulk-rock H2O and ∂D measurements confirm that progressive open-system outgassing occurred prior to both blasts. The correlations between crystallinity, clast density, and bulk H2O contents suggest that syn-blast magma expansion was modulated both by non-uniform volatile distribution within the cryptodome and rheological controls associated with non-uniform crystal content. Spatial heterogeneities in volatiles and crystallinity within a given cryptodome are attributed to distance from the wallrock margin, which probably correlates with timing of magma injection. The greater proportion of high-density material at BZ is speculatively related to lower blast energy compared with MSH.

The transient dynamics of magma ascent during dome-forming eruptions were investigated and the effects of magma chamber pressure perturbations on eruption rate are illustrated. The numerical model DOMEFLOW, developed by the authors for this work, is applied to the problem. DOMEFLOW is a transient 1.5D isothermal two-phase flow model of magma ascent through an axisymmetric conduit of variable radius, which accounts for gas exsolution, bubble growth, crystallization induced by degassing, permeable gas loss through overlying magma and through conduit walls, as well as viscosity changes due to crystallization and degassing. For runs in which chamber pressure increases, the time required to reach the new steady state (transition time) is a complex function of the pressure perturbation, while for decreasing chamber pressure, transition time is a monotonic function of the magnitude of the pressure perturbation. The transition to the new steady state is mainly controlled by magma compressibility, travel time (time required for one parcel of magma to travel from chamber to surface), and the time over which the pressure perturbation occurs. Results of many runs (> 300) were analyzed using dimensional analysis to reveal a general relationship which predicts the temporal evolution of magma effusion rate for a given sudden increase in chamber pressure; the product of the change in steady-state extrusion rate and the time required to reach the new steady state is linearly proportional to the normalized change in chamber pressure, the volume of the conduit, and the ratio of top and bottom conduit radii, and inversely proportional to the cubic root of volatile fraction. This relationship is used to interpret observed variations in two ongoing dome-building eruptions, the Soufriere Hills volcano, Montserrat, and Merapi volcano, Indonesia.

A dyke conduit has been shown to be a realistic model for the shallow magmatic feeder system for the dome forming eruption at Soufriere Hills Volcano, Montserrat. Here we use a three-dimensional Finite Element model to examine the ground deformation that can be expected due to the pressurization of a dyke conduit. We find that the generated deformation has a bilateral symmetry with nearly no displacement in the direction of dyke strike, and a maximal ground deformation about 1 km away from the vent in the perpendicular direction. Resultant surface deformation is mainly triggered by the upper part of the feeder system, where the dyke opens into a cylindrical conduit. We apply our deformation model to investigate tilt data collected in 1997 in order to infer the orientation of the dyke. We obtain a best-fit for a NNW–SSE trending dyke, which matches observations of the ground displacement field obtained by differential GPS and the alignment of main structural geologic features of southern Montserrat.

Assessments of pyroclastic flow (PF) hazards are commonly based on mapping of PF and surge deposits and estimations of inundation limits, and/or computer models of varying degrees of sophistication. In volcanic crises a PF hazard map may be sorely needed, but limited time, exposures, or safety aspects may preclude fieldwork, and insufficient time or baseline data may be available for reliable dynamic simulations. We have developed a statistically constrained simulation model for block-and-ash type PFs to estimate potential areas of inundation by adapting methodology from Iverson et al. (Geol Soc America Bull 110:972–984, 1998) for lahars. The predictive equations for block-and-ash PFs are calibrated with data from several volcanoes and given by A = (0.05 to 0.1)V2/3, B = (35 to 40)V2/3, where A is cross-sectional area of inundation, B is planimetric area and V is deposit volume. The proportionality coefficients were obtained from regression analyses and comparison of simulations to mapped deposits. The method embeds the predictive equations in a GIS program coupled with DEM topography, using the LAHARZ program of Schilling (1998). Although the method is objective and reproducible, any PF hazard zone so computed should be considered as an approximate guide only, due to uncertainties on the coefficients applicable to individual PFs, the authenticity of DEM details, and the volume of future collapses. The statistical uncertainty of the predictive equations, which imply a factor of two or more in predicting A or B for a specified V, is superposed on the uncertainty of forecasting V for the next PF to descend a particular valley. Multiple inundation zones, produced by simulations using a selected range of volumes, partly accommodate these uncertainties. The resulting maps show graphically that PF inundation potentials are highest nearest volcano sources and along valley thalwegs, and diminish with distance from source and lateral distance from thalweg. The model does not explicitly consider dynamic behavior, which can be important. Ash-cloud surge impact limits must be extended beyond PF hazard zones and we provide several approaches to do this. The method has been used to supply PF and surge hazard maps in two crises: Merapi 2006; and Montserrat 2006–2007.

We report on detailed laboratory experiments designed to illustrate the frictional behavior of granular volcanic debris. The materials include pyroclastic flow debris from Soufriere Hills Volcano, Montserrat, and from Stromboli Volcano, Italy, and lahar deposits from Mount St. Helens. Experiments were conducted in a servo-controlled, double-direct shear apparatus under conditions of displacement-control and with monitored temperature and humidity. The effects of loading velocity, normal stress, grain-size range, and saturation state were examined for normal stresses from 0.75 to 8 MPa. The Soufriere Hills debris was sampled in the field using a 2 mm sieve. Samples from Stromboli and MSH represent the original bulk material. The influence of grain-size and size distribution were examined in detail for the Soufriere Hills material for 1) two narrow size ranges (3–4 /, 0.063–0.125 mm; and 0–1 /, 0.5–1 mm), 2) a wide size range (3–0 /, 0.125–1 mm), and 3) the natural size distribution in the range 0–1 mm. These four data sets show remarkably uniform properties: coefficients of residual internal friction varied from 0.62 to 0.64 and coefficients of peak internal friction varied from 0.66 to 0.69, with zero cohesion in each case. For the natural grain-size distribution, we find a small but clear increase in residual sliding friction with increasing slip velocity in the range 10 to 900 μm/s, i.e. velocity strengthening frictional behavior. For Soufriere Hills pyroclastic material the coefficient of internal friction changes very little when water saturated. The frictional characteristics were remarkably similar for the three volcanoes. For the natural grain-size distributions, the coefficient of residual internal friction ranged from 0.61 to 0.63 and peak internal friction values ranged from 0.65 to 0.66. Pre-loaded and over-compacted materials sheared at reduced normal stress gave peak coefficients as high as 0.80. Our results imply that granular volcanic debris should often fail via stable creep, but may exhibit stick-slip instability under some conditions. © 2008 Elsevier B.V. All rights reserved.

Styles of dome collapse at Soufriere Hills volcano (SHV; November 1995–March 1998) are classified by relations between extrusion rate prior to collapse and collapse volume. Four separate modes of collapse behavior are apparent. Notably, moderate rates of extrusion are shown to result in two disparate modes of collapse: small-to-large collapses on steeply inclined failure planes that switch to collapse volumes an order of magnitude larger that cut deeply into the dome core. For constant effusion rates, this bifurcation in behavior is explained by the monotonic growth of a soft core that ultimately promotes the development of a deep-seated failure over previously favored shallow failure modes. Models are developed to test this hypothesis that first constrain magnitudes of cohesive and frictional strength with observed dome collapse morphologies and volumes. Evaluations of dome strengths confirm the important role of a soft core in promoting deep failure. A nested model representing a cohesive dome core, surrounded by a frictional rind, with constant rate of magma input, confirms the observed bifurcation in behavior, and for invariant effusive activity. Importantly, failure volumes are shown to increase by close to an order of magnitude for a few percent change in the proportion of dome core comprising cohesive material. This model is capable of replicating, a posteriori, the approximate timing of failure for both small (250 m) and large (325 m) domes. The timing and style of the 17 September 1996 and June to November 1997 collapses are honored.

On many volcanoes, there is evidence of a relationship between dome collapse and periods of high precipitation. We propose a mechanism for this relationship and investigate the conditions that optimize failure by this process. Observations of elongate lobes that evolve through exogenous growth of lava domes reveal that they commonly develop tensile fractures perpendicular to the direction of motion. These cracks can increase in depth by localized cooling and volumetric contraction. During periods of high rainfall, water can fill these cracks, and the increase in fluid pressure on the base of the lobes and within the crack can trigger the collapse of the hot exogenous lava domes. Using limit-equilibrium analysis, it is possible to calculate the water and vapor forces acting on the rear and base of the potentially unstable part of the lobe. The model presented is rectangular in cross-section, with material properties representative of andesitic dome rocks. Vapor pressures at the base of cracks are sealed by the penetrating rainfall, which forms a saturated cap within the lobe. This leads to an increase in fluid pressurization both through the underlying gas pressure and the downslope component of the liquid water cap. Fluid pressurization increases as the penetration depth increases. This rainfall penetration depth is dependent on the thermal properties of the rocks, antecedent temperature, lobe geometry, and the intensity and duration of precipitation. Dominant parameters influencing the stability of the lobe are principally lobe thickness, duration and intensity of rainfall, and antecedent lobe temperature. Our modeling reveals that thicker lobes are intrinsically more unstable due to the amplification of downslope forces in comparison to cohesive strength. The increase in the duration and intensity of rainfall events also increases the potential for collapse, as it leads to deeper liquid penetration. Deeper penetration depths are also achieved through lower antecedent temperatures since less fluid is lost through vaporization. Thus, the potential for rain-triggered collapse increases with time from emplacement.

The way in which strands of uncertain volcanological evidence can be used for decision-making, and the weight that should be given them, is a problem requiring formulation in terms of the logical principles of Evidence Science. The basic ideas are outlined using the explosion at Galeras volcano in Colombia in January 1993 as an example. Our retrospective analysis suggests that if a robust precautionary appraisal had been made of the circumstances in which distinctive tornillo signals were detected at Galeras, those events might have been construed as stronger precursory evidence for imminent explosive activity than were the indications for quiescence, given by the absence of other warning traits. However, whilst visits to the crater might have been recognised as involving elevated risk if this form of analysis had been applied to the situation in January 1993, a traditional scientific consideration of the available information was likely to have provided a neutral assessment of short-term risk levels. We use these inferences not to criticise interpretations or decisions made at the time, but to illustrate how a structured, evidence-based analysis procedure might have provided a different perspective to that derived from the conventional scientific standpoint. We advocate a formalism that may aid such decision-making in future: graphical Bayesian Belief Networks are introduced as a tool for performing the necessary numerical procedures. With this approach, Evidence Science concepts can be incorporated rationally, efficiently and reliably into decision support during volcanic crises.

Pyroclastic density currents (PDCs) represent a formidable challenge for volcanology. This is true in terms of scientific understanding of their dynamics, and also in terms of assessment of their hazard and risk. These phenomena occurred in the famous, lethal 79 AD Pompeii eruption of Mt. Vesuvius, although recognition in “modern” times dates from observations made in 1902 at Montagne Pelee. Major progress in understanding has developed in the last few decades, due both to direct observations and improved modeling capability. Currently, PDC generation and propagation mechanisms are deeply investigated by field reconstructions, and also via detailed experimental and modeling studies. In this chapter we review the current status of knowledge on the dynamics of PDCs, and highlight the key processes and impacts as revealed by recent eruptive events. Some current approaches and modeling studies aimed at PDC hazard assessment are then illustrated, with specific reference to short-to-medium term hazard assessments carried out at the Soufriere Hills volcano at Montserrat (West Indies, UK), and to long-term assessments at the high-risk Vesuvius volcano and Campi Flegrei caldera (Italy). We conclude this review with remarks on the current and evolving state of knowledge of this dangerous volcanic phenomenon, and implications on the challenges for hazards mitigation. Significant limitations exist in our current ability to accurately forecast zones at risk from some PDCs, and in practice, hazard evaluations require cautious good judgment and not over-reliance on computations based on oversimplified algorithms and inputs.

Gravitational collapses of the lava dome at Soufriere Hills Volcano on 25 June and 26 December 1997 generated pyroclastic surges that spread out over broad sectors of the landscape and laid down thin, bipartite deposits. In each case, part of the settling material continued to move upon reaching the ground and drained into valleys as high-concentration granular flows of hot (120-410°C) ash and lapilli. These surge-derived pyroclastic flows travelled at no more than 10 m s -1 but extended significantly beyond the limits of the parent surge clouds (by 3 km on 25 June and by 1 km on 26 December). The front of the 25 June flow terminated in a valley about 50 m below a small town that was occupied at the time. Despite their small deposit volumes (5-9 x 10 4 m 3 ), the surge-derived pyroclastic flows travelled as far as many of the Soufriere Hills block-and-ash flows on slopes as low as a few degrees, reflecting a high degree of mobility. An analysis of the deposits from 26 December suggests that sediment accumulation rates of at least several millimetres per second were sufficient to generate pyroclastic flows by suspended-load fallout from pyroclastic surges on Montserrat. Surge-derived pyroclastic flows are an important, and hitherto underestimated, hazard around active lava domes. At Montserrat they formed by sedimentation over large catchment areas and drained into valleys different from those affected by the primary block-and-ash flows and pyroclastic surges, thereby impacting areas not anticipated to be vulnerable in prior hazards analyses. The deposits are finer-grained than those of other types of pyroclastic flow at Soufriere Hills Volcano; this may aid their recognition in ancient volcanic successions but, along with valley-bottom confinement, reduces the preservation potential.

During the 1995 to 1998 phase of dome growth at Soufriere Hills Volcano on Montserrat, we documented instability of the steep southern rim of English9s Crater, known as Galway9s Wall. The horseshoe-shaped English9s Crater provided good evidence for previous sector collapses, and assessments undertaken in late 1996 anticipated the possibility of a partial sector collapse and a SW-directed explosion, hazards previously unrecognized on Montserrat. A change from predominantly endogenous to exogenous growth of the lava dome at the end of 1996 eased the stress on the southern sector. However, rapid dome growth in November and December 1997 led to severe reloading and eventual sector failure at the base of the buried Galway9s Wall and in the adjacent hot-spring area. This failure resulted in the debris avalanche and lateral blast of 26 December 1997. Similar sector collapses at a number of small volcanoes in the Caribbean, as well as worldwide, are evidence that edifice instability develops commonly in dome-forming eruptions. The hazards from a sector collapse and a consequent lateral blast are extreme, and monitoring operations and disaster planning at such volcanoes should focus on these, as well as on the more common hazards of conventional pyroclastic flows associated with dome growth.

Dome growth at Soufriere Hills Volcano halted in early March 1998. After dome growth ceased, seismicity reduced significantly, but activity related to dome disintegration and degassing of magma at depth continued. A sustained episode of pyroclastic flows on 3 July 1998 marked the single largest collapse from March 1998 to November 1999. This led to a remarkable episode of dome collapses, low-energy explosions and ash-venting that resulted in the regular production of ash plumes, commonly reaching 1.5-6 km above sea level (a.s.l), but sometimes up to 11 km a.s.l., and the development of a small block-and-ash cone around the explosion crater. During the period of this residual activity, higher levels of activity occurred approximately every five to six weeks. This periodicity was similar to the cycles observed during active dome growth during 1995 to 1998, and probably had a similar cause. The relatively high level of observed activity caused continued concern regarding volcanic hazards and their potential to impact upon the resident population. Vigorous magma extrusion resumed in November 1999. The activity of the intervening period is attributed to the continued cooling and degassing of the dome, conduit and deep magma body, the impact of rising volcanic gases in the volcanic edifice, and limited magma flow in the conduit.

In 1997 Soufriere Hills Volcano on Montserrat produced 88 Vulcanian explosions: 13 between 4 and 12 August and 75 between 22 September and 21 October. Each episode was preceded by a large dome collapse that decompressed the conduit and led to the conditions for explosive fragmentation. The explosions, which occurred at intervals of 2.5 to 63 hours, with a mean of 10 hours, were transient events, with an initial high-intensity phase lasting a few tens of seconds and a lower-intensity, waning phase lasting 1 to 3 hours. In all but one explosion, fountain collapse during the first 10-20 seconds generated pyroclastic surges that swept out to 1-2 km before lofting, as well as high-concentration pumiceous pyroclastic flows that travelled up to 6 km down all major drainages around the dome. Buoyant plumes ascended 3-15 km into the atmosphere, where they spread out as umbrella clouds. Most umbrella clouds were blown to the north or NW by high-level (8-18 km) winds, whereas the lower, waning plumes were dispersed to the west or NW by low-level ( 1 and ballistic blocks were thrown as far as 1.7 km from the dome. Each explosion discharged on average 3 x 10 5 m 3 of magma, about one-third forming fallout and two-thirds forming pyroclastic flows and surges, and emptied the conduit to a depth of 0.5-2 km or more. Two overlapping components were distinguished in the explosion seismic signals: a low-frequency (c. 1 Hz) one due to the explosion itself, and a high-frequency (>2 Hz) one due to fountain collapse, ballistic impact and pyroclastic flow. In many explosions a delay between the explosion onset and start of the pyroclastic flow signal (typically 10-20 seconds) recorded the time necessary for ballistics and the collapsing fountain to hit the ground. The explosions in August were accompanied by cyclic patterns of seismicity and edifice deformation due to repeated pressurization of the upper conduit. The angular, tabular forms of many fallout pumices show that they preserve vesicularities and shapes acquired upon fragmentation, and suggest that the explosions were driven by brittle fragmentation of overpressured magmatic foam with at least 55 vol% bubbles present in the upper conduit prior to each event.

Between 3 January 1996 and 1 December 1998 at Soufriere Hills Volcano, Montserrat, there were 27150 seismic signals attributable to rockfalls and pyroclastic flows. Large (1-4 x 10 6 m 3 ) and major (>4 x 10 6 m 3 ) dome collapses began to occur after the extrusion rate reached 2 m 3 s -1 and the lava dome exceeded 30 x 10 6 m 3 in volume. Large to major collapses occurred on 26 occasions and were usually associated with periods of elevated extrusion rate (6 13 m 3 s -1 ), intense hybrid earthquake swarms and/or inflation-deflation cycles of crater-rim ground deformation. During these cycles, gas-rich pulses of magma, 140 000 320 000 m 3 in volume, were intruded into the dome and, soon thereafter, dome collapses from the headwalls of shear lobes were generated. Large dome collapses also occurred after 10 March 1998, when magma extrusion ceased, but these had no seismic precursors. These events represent structural failures from over steepened canyon-like walls and were followed by intense degassing, suggesting that gas pressure build-up within the relict dome may have played some role. Rockfall counts and durations established from seismic data show variations that correlate with extrusion rate. Using pyroclastic flow runout and rockfall duration data as proxies for event magnitude, power law relationships between frequency (total number of events) and magnitude have been found. Seismic signals associated with rockfalls and pyroclastic flows commonly comprise both high-frequency and long-period components. Intense degassing from the dome is interpreted as the source for the long-period component. These results indicate that rockfalls and pyroclastic flows generated by dome collapse at Montserrat are not simply the result of passive dome failure, but are intimately related to discharge of pressurized gas and pulses of magma extrusion. Pyroclastic flows were usually sourced from lobe headwalls, where lava was hot and gas-rich and where fragmentation of the micro vesicular andesite lava occurred more readily.

This paper presents numerical models of the Vulcanian explosions that occurred in 1997 at Soufriere Hills Volcano. Plume evolution and velocities were calculated for the well-documented and typical explosions of 6 and 7 August 1997, and these data and other observations were compared to transient, axisymmetric, multiphase flow simulations of coupled conduit evacuation and pyroclastic dispersal. Pre-explosion conduit conditions were estimated from Montserrat data, using a simple gas solubility law and assuming that conduit magma flow had stagnated with a constant overpressure prior to the explosions. Reference simulation input parameters include conduit diameter of 30 m, crater diameter of 300 m, meltwater content of 4.3±0.5°, grain sizes of 30, 2000 and 5000 µm, and conduit overpressure of l0MPa. The numerical simulations of the explosions resolved highly unsteady vent exit conditions such as velocity, pressure and mass flux, and the spatial and temporal dispersal of pyroclasts during the initial few minutes was investigated using one gas phase and two or three solid phases representing pyroclasts of different size. Our simulations produced transitional eruptive regime behaviour, dividing the erupted mass into a portion that generated a radial pyroclastic current fed by a collapsing column, and a convective portion that generated a buoyant plume. This behaviour generally mimicked the observed explosions. The movement of different particle sizes was tracked, with fine particles dominantly influencing the convective behaviour of the central plume and ash plume thermals generated above the pyroclastic currents. Simulated initial vent velocities ranged from 85 to 120ms -1 , collapse heights ranged from 450 to 1370 m above the vent, initial pyroclastic current velocities ranged from 40 to 60 ms -1 with surge runouts to 1.8 km, drawdown depths in the conduit were a few hundred metres, and simulated pyroclastic current deposit temperatures ranged between 135 and 430°C. Subsets of these results are in reasonable agreement with observed and measured parameters of the 1997 explosions. The best match was intermediate between our reference simulation, which assumed no loss of volatiles from the conduit during rise from the magma reservoir and which appeared too energetic, and another simulation in which much volatile leakage was assumed. The results suggest that volatile depletion in the conduit was an important factor in influencing the dynamic behaviour of the Vulcanian explosions on Montserrat.

Emplacement of a debris avalanche in the White River valley of southern Montserrat (Lesser Antilles), on 26 December 1997, was caused by sector failure of the south flank of the active Soufriere Hills volcano. Pre- and post-emplacement surveys of the region indicate a debris avalanche deposit volume of about 40–50 × 106 m³. This avalanche is modeled as the gravitational flow of a homogeneous continuum governed by a basal friction law. Mass and momentum equations are depth-averaged over the slide thickness. Numerical results show that the observed distribution of debris and duration of emplacement is simulated well for a Coulomb-type friction law with a dynamic friction coefficient dependent upon the thickness and the velocity of the flowing mass.

During the sequential dome collapse of Merapi Volcano on 22 November 1994, a broadband seismic station on the western slope was the only operational seismic equipment that provided continuous on-scale recording of the event. According to visual and seismic observations, the collapse activity lasted about 10 h. We divide the activity into two phases: (I) a period with fluctuating but generally increasing seismic activity associated with 24 moderate to large dome-collapse nuees ardentes, lasting about 40 min and culminating with the largest event at 10:54 (all times reported as local time); and (II) activity from 11:42 until 20:00 described by scattered clusters of individual rockfalls and 20 nuees ardentes. The broadband data were evaluated using the assumption that avalanches with the same source areas and descent paths exhibit a linear relation between source volume and recorded seismic-amplitude envelope area. A result of the analysis is the determination of the volume of selected individual events. From the field surveys, the total volume of the collapsed dome lava is 2.6 Mm3. Discounting the volumetric influence of rockfalls, the average size of the 44 nuees ardentes is therefore about 60,000 m3. The largest collapse event at 10:54 is estimated to involve 260,000 m3, based on an analysis of the seismicity. The remaining 23 phase I events averaged 60,000 m3, with the total volume of all phase I events accounting for 63% of the unstable dome. The 20 phase II events comprised 37% of the total volume and averaged 47,000 m3. The methods described here can be put to practical use in real-time monitoring situations. Broadband data were essential in this study primarily because of the wide dynamic range.

A study of emplacement temperatures was carried out for the largest of the 22 November 1994 nuee ardente deposits at Merapi Volcano, based mainly on the response of plastic and woody materials subjected to the hot pyroclastic current and the deposits, and to some extent on eyewitness observations. The study emphasizes the Turgo–Kaliurang area in the distal part of the area affected by the nuee ardente, where nearly 100 casualties occurred. The term nuee ardente as used here includes channeled block-and-ash flows, and associated ash-clouds of surge and fallout origins. The emplacement temperature of the 8 m thick channeled block-and-ash deposit was relatively high, ∼550°C, based mainly on eyewitness reports of visual thermal radiance. Emplacement temperatures for ash-cloud deposits a few cm thick were deduced from polymer objects collected at Turgo and Kaliurang. Most polymers do not display a sharp melting range, but polyethylene terephthalate used in water bottles melts between 245 and 265°C, and parts of the bottles that had been deformed during fabrication molding turn a milky color at 200°C. The experimental evidence suggests that deposits in the Turgo area briefly achieved a maximum temperature near 300°C, whereas those near Kaliurang were

The common occurrence of tree and pole blow-down from pyroclastic currents provides an opportunity to estimate properties of the currents. Blow-down may occur by uprooting (root zone rupture), or flexure or shear at some point on the object. If trees are delimbed before blow-down, each tree or pole can be simulated by a cylinder perpendicular to the current. The force acting on a cylinder is a function of flow dynamic pressure, cylinder geometry, and drag coefficient. Treated as a cantilever of circular cross-section, the strength for the appropriate failure mode (rupture, uprooting or flexure) can then be used to estimate the minimum necessary current dynamic pressure. In some cases, larger or stronger standing objects can provide upper bounds on the dynamic pressure. This analysis was treated in two ways: (1) assuming that the current properties are vertically constant; and (2) allowing current velocity and density to vary vertically according to established models for turbulent boundary layers and stratified flow. The two methods produced similar results for dynamic pressure. The second, along with a method to approximate average whole-current density, offers a means to estimate average velocity and density over the height of the failed objects. The method is applied to several example cases, including Unzen, Mount St. Helens, Lamington, and Merapi volcanoes. Our results compare reasonably well with independent estimates. For several cases, we found that it is possible to use the dynamic pressure equations developed for vertically uniform flow, along with the average cloud density multiplied by a factor of 2–5, to determine average velocity over the height of the failed object.

Multiphase (MP) and low frequency (LF) earthquakes with spectral peak amplitudes at 3–4 and 1 Hz, respectively, are two common types of shallow volcanic earthquakes previously recognized at Merapi Volcano. Their mechanisms are poorly understood but MPs have been temporally associated with lava dome growth. We conducted a seismic experiment in January–February 1998, using four broadband seismographs to investigate the nature of seismic activity associated with dome growth. During our experiment, Merapi experienced mild dome growth with low-level seismic activity. We compare our data to that recorded on a local short-period (SP) network, with the following preliminary results. MP and LF events as recorded and classified on the short-period network instruments were recognized on the broadband network. Frequency spectrograms revealed similar patterns in the near summit region at widely separated broadband stations. Higher frequency spectra than previously recognized were identified for both MP and LF events, and were strongly attenuated as a function of radial distance from the source. Thus the spectral characteristics of these events as recorded on far-field stations are not fully indicative of the source processes. In particular, many events classified as LF-type appear to have much high frequency energy near the source. This aspect of these so-called LF earthquakes, and their association with very-long-period (VLP) pulses, suggests that many events identified in the far-field as LF events are in actuality a variety of the MP event and involve similar source processes. Broadband records indicated that simple large-amplitude VLP pulses were embedded in MP and LF wavetrains. From event to event these pulses were similar in their waveforms and had periods of 4 s. VLP events embedded in LF and MP earthquakes were located using particle motions. The epicenters were clustered in a central region of the dome complex, and preliminary source depths were within about 100 m of the dome surface, suggesting a source region deep within the dome or the uppermost conduit. A similar source location was established by study of MP high-frequency onsets. Our broadband data suggests that we have recorded both elastic seismic waves and a simple embedded pulse that is interpreted to represent a surface tilt at the seismometer site. The inferred tilt indicates an inflation and subsequent deflation, possibly caused by a gas pressure pulse or episodic shallow magma transport with stick-slip movement of the conduit wall.

Information on Merapi eruptive activity is scattered and much is remotely located. A concise and well-documented summary of this activity has been long needed to assist researchers and hazard-mitigation efforts, and the aim of this paper is to synthesize information from the mid-1700s to the present. A descriptive chronology is given, with an abbreviated chronology in a table that summarizes events by year, assigns preliminary Volcanic Explosivity Index (VEI) ratings and Hartmann classifications, and provides key references. The history of volcano monitoring is also outlined. The study reveals that a major difference in eruption style exists between the twentieth and nineteenth centuries, although the periodicity between larger events seems about the same. During the twentieth century, activity has comprised mainly the effusive growth of viscous lava domes and lava tongues, with occasional gravitational collapses of parts of oversteepened domes to produce the nuees ardentes—commonly defined as “Merapi-type”. In the 1800s, however, explosive eruptions of relatively large size occurred (to VEI 4), and some associated “fountain-collapse” nuees ardentes were larger and farther reaching than any produced in the twentieth century. These events may also be regarded as typical eruptions for Merapi. The nineteenth century activity is consistent with the long-term pattern of one relatively large event every one or two centuries, based on the long-term eruptive record deduced by others from volcanic stratigraphy. It is uncertain whether or not a “recurrence-time” model continues to apply to Merapi, but if so, Merapi could soon be due for another large event and its occurrence with only modest (or inadequately appreciated) precursors could lead to a disaster unprecedented in Merapi's history because the area around the volcano is now much more densely populated.

Syn-eruptive degassing of volcanoes may lead to syn-eruptive crystallization of groundmass phases. We have investigated this process using textural and compositional analysis of dome material from Merapi volcano, Central Java, Indonesia. Samples included dome lavas from the 1986‐88, 1992‐93, 1994 and 1995 effusive periods as well as pyroclastic material deposited by the November 1994 dome collapse. With total crystallinities commonly in excess of 70% (phenocrysts 1 microlites), the liquids present in Merapi andesites are highly evolved (rhyolitic) at the time of eruption. Feldspar microlites in dome rocks consist of plagioclase cores (Ab63An29Or8) surrounded by alkali feldspar rims (Ab53An5Or42), compositional pairs which are not in equilibrium. A change in the phase relations of the ternary feldspar system caused by degassing best explains the observed transition in feldspar composition. A small proportion of highly vesicular airfall tephra grains from the 1994 collapse have less evolved glass compositions than typical dome material and contain rimless plagioclase microlites, suggesting that the 1994 collapse event incorporated less-degassed, partially liquid magma in addition to fully solidified dome rock. As decompression drives volatile exsolution, rates of degassing and resultant microlite crystallization may be governed by magma ascent rate. Microlite crystallinity is nearly identical among the 1995 dome samples, an indication that similar microlite growth conditions (PH2 O and temperature) were achieved throughout this extrusive period. However, microlite number density varied by more than a factor of four in these samples, and generally increased with distance from the vent. Low vent-ward microlite number densities and greater microlite concentrations down-flow probably reflect progressively decreasing rates of undercooling at the time of crystal nucleation during extrusion of the 1995 dome. Comparison between dome extrusion episodes indicates a correlation between lava effusion rate and microlite number density, suggesting that extrusion slowed during 1995. Crystal textures and compositions in the 1992‐93 and 1994 domes share the range exhibited by the 1995 dome, suggesting that transitions in crystallization conditions (i.e., rates of undercooling determined by effusion rate) are cyclic. q 2000 Elsevier Science B.V. All rights reserved.

Following the eruption of January 1992, episodes of lava dome growth accompanied by generation of dome-collapse nuees ardentes occurred in 1994–1998. In addition, nuees ardentes were generated by fountain-collapse in January 1997, and the 1998 events also suggest an explosive component. Significant tilt and seismic precursors on varying time scales preceded these events. Deformation about the summit has been detected by electronic tiltmeters since November 1992, with inflation corresponding generally to lava dome growth, and deflation (or decreased inflation) corresponding to loss of dome mass. Strong short-term (days to weeks) accelerations in tilt rate and seismicity occurred prior to the major nuees ardentes episodes, apart from those of 22 November 1994 which were preceded by steadily increasing tilt for over 200 days but lacked short-term precursors. Because of the combination of populated hazardous areas and the lack of an issued warning, about 100 casualties occurred in 1994. In contrast, the strong precursors in 1997 and 1998 provided advance warning to observatory scientists, enabled the stepped raising of alert levels, and aided hazard management. As a result of these factors, but also the fortunate fact that the large nuees ardentes did not quite descend into populated areas, no casualties occurred. The nuee ardente episode of 1994 is interpreted as purely due to gravitational collapse, whereas those of 1997 and 1998 were influenced by gas-pressurization of the lava dome.

Edifice deformations are reported here for the period 1988–1995 at Merapi volcano, one of the most active and dangerous volcanoes in Indonesia. The study period includes a major resumption in lava effusion in January 1992 and a major dome collapse in November 1994. The data comprise electronic distance measurements (EDM) on a summit trilateration network, slope distance changes measured to the upper flanks, and other data collected from 1988 to 1995. A major consequence of this study is the documentation of a significant 4-year period of deformation precursory to the 1992 eruption. Cross-crater strain rates accelerated from less than 3×10 −6 /day between 1988 and 1990 to more than 11×10 −6 /day just prior to the January 1992 activity, representing a general, asymmetric extension of the summit during high-level conduit pressurization. After the vent opened and effusion of lava resumed, strain occurred at a much-reduced rate of less than 2×10 −6 /day. EDM measurements between lower flank benchmarks and the upper edifice indicate displacements as great as 1 m per year over the four years before the 1992 eruption. The Gendol breach, a pronounced depression formed by the juxtaposition of old lava coulees on the southeast flank, functioned as a major displacement discontinuity. Since 1993, movements have generally not exceeded the 95% confidence limits of the summit network. Exceptions to this include 12 cm outward movement for the northwest crater rim in 1992–1993, probably from loading by newly erupted dome lava, and movements as much as 7 cm on the south flank between November 1994 and September 1995. No short-term precursors were noted before the November 1994 lava dome collapse, but long-term adjustments of crater geometry accompanied lava dome growth in 1994. Short-term 2-cm deflation of the edifice occurred following the November 1994 dome collapse.

Slope failures resulting from structural instability of andesitic volcanic edifices can generate mobile debris avalanches that travel long distances down or beyond the flanks of volcanoes. More than 20 major slope failures have occurred worldwide over the past 500 years, a rate exceeding that of caldera collapse. Hazards derive from the debris avalanches themselves, from associated explosive activity that ranges from vertical eruptions (often accompanied by pyroclastic currents) to devastating directed blasts, from associated lahars, and from tsunamis. Collapses of growing lava domes are more frequent, are similar, in many ways, to edifice collapse, and can directly generate devastating pyroclastic currents. This paper examines some aspects of current understanding of edifice and lavadome instability. The primary focus of the presentation is on mechanisms and factors associated with collapse, the geometric factors, augmented loading by magma, localized strength reduction by physical and chemical changes (the latter commonly associated with hydrothermal processes), strain weakening, pore-fluid (water or gas) pressure enhancement, retrogressive failure, time-dependent failure, and seismic shaking. Some aspects of material property evaluation, analysis procedures, and implications on monitoring are also discussed. Case examples discussed include edifice instability at Mt St Helens, USA, and Soufriere Hills volcano, Montserrat, the stability of lava spines at Mont Pelee, Martinique, and Lamington, Papua New Guinea, and lavadome stability at Soufriere Hills. The topics bear on understanding hazardous edifice and dome failures, and the measures to anticipate such failures.

An important type of gravitational lava-dome failure involves significant internal gas overpressure that influences the failure process. For the first time, this paper presents a rigorous analysis of the mechanics of gas-pressurized dome failure. We develop diffusion models to calculate gas overpressures in a lava dome, and embed these pressure data into stability analyses to demonstrate how gas pressurization can initiate instability. The results bear on the mechanistic understanding of hazardous dome failures. Gas pressurization can promote deep-seated failure, and the timing of conduit pressurization can control the timing of dome collapse. Theory explains why dome collapse can be delayed for cases of oscillating effusion. Dome collapses and explosive events linked to collapse of lava domes can be hazardous. Monitoring criteria are considered to anticipate whether or not such collapses are likely to be explosive.

Investigation of well-exposed volcaniclastic deposits of Shiveluch volcano indicates that large-scale failures have occurred at least eight times in its history: approximately 10,000, 5700, 3700, 2600, 1600, 1000, 600 14 C BP and 1964 AD. The volcano was stable during the Late Pleistocene, when a large cone was formed (Old Shiveluch), and became unstable in the Holocene when repetitive collapses of a portion of the edifice (Young Shiveluch) generated debris avalanches. The transition in stability was connected with a change in composition of the erupting magma (increased SiO2 from ca. 55-56% to 60-62%) that resulted in an abrupt increase of viscosity and the production of lava domes. Each failure was triggered by a disturbance of the vol- canic edifice related to the ascent of a new batch of vis- cous magma. The failures occurred before magma in- truded into the upper part of the edifice, suggesting that the trigger mechanism was indirectly associated with magma and involved shaking by a moderate to large volcanic earthquake and/or enhancement of edif- ice pore pressure due to pressurised juvenile gas. The failures typically included: (a) a retrogressive landslide involving backward rotation of slide blocks; (b) frag- mentation of the leading blocks and their transforma- tion into a debris avalanche, while the trailing slide blocks decelerate and soon come to rest; and (c) long- distance runout of the avalanche as a transient wave of debris with yield strength that glides on a thin weak layer of mixed facies developed at the avalanche base. All the failures of Young Shiveluch were immediately followed by explosive eruptions that developed along a similar pattern. The slope failure was the first event, followed by a plinian eruption accompanied by partial fountain collapse and the emplacement of pumice flows. In several cases the slope failure depressurised the hydrothermal system to cause phreatic explosions that preceded the magmatic eruption. The collapse-in- duced plinian eruptions were moderate-sized and ordi- nary events in the history of the volcano. No evidence for directed blasts was found associated with any of the slope failures.

Volcanic eruptions are episodic despite being supplied by melt at a nearly constant rate. We used histories of magma efflux and surface deformation to geodetically image magma transfer within the deep crustal plumbing of the Soufrière Hills volcano on Montserrat, West Indies. For three cycles of effusion followed by discrete pauses, supply of the system from the deep crust and mantle was continuous. During periods of reinitiated high surface efflux, magma rose quickly and synchronously from a deflating mid-crustal reservoir (at about 12 kilometers) augmented from depth. During repose, the lower reservoir refilled from the deep supply, with only minor discharge transiting the upper chamber to surface. These observations are consistent with a model involving the continuous supply of magma from the deep crust and mantle into a voluminous and compliant mid-crustal reservoir, episodically valved below a shallow reservoir (at about 6 kilometers).

Breakout of magmatic activity at Soufriere Hills volcano, Montserrat, was preceded by a tenfold increase in rate of earthquake occurrence. A new model of subcritical rock failure shows that this increase is consistent with the growth, possibly episodic, of the magma conduit at a rate controlled by progressive weakening of the host country rock. The preferred weakening mechanism is stress corrosion, by which circulating juvenile and hydrothermal fluids chemically attack the country rock and promote failure at stresses smaller than the rock's theoretical strength. The results illuminate the potential for slow-cracking models to enhance eruption forecasts using the inverse-rate technique combined with traditional monitoring methods.

Telemetered high-resolution tiltmeters were installed in Montserrat in summer of 1995, in December 1996, and in May 1997. The 1995 installations, several km from the Soufriere Hills vent, were too distant to yield useful data. However, the 1996 and 1997 installations on the crater rim revealed 6–14 h inflation cycles caused by magma pressurization at shallow depths (< 0.6 km below the base of dome). The tilt data correlated with seismicity, explosions, and pyroclastic flow activity, and were used to forecast times of increased volcanic hazard to protect scientific field workers and the general public.

Varied seismicity has accompanied growth and collapse of the lava dome of the Soufriere Hills Volcano, Montserrat. Earthquakes have been classified as either volcano-tectonic, long-period or hybrid, and daily variations in the numbers of events have mapped changes in the style of eruption. Repetitive hybrid earthquakes were common during the first months of dome growth. In July 1996 the style of seismicity changed and regular, short-lived hybrid earthquake swarms became common. This change was probably caused by an increase in the magma flux. Earthquake swarms have preceded almost all major dome collapses, and have accompanied cyclical deformation, thought to be due to a build-up of pressure in the upper conduit which is later released by magma moving into the dome.

IODP Expedition 340 successfully drilled a series of sites offshore Montserrat, Martinique and Dominica in the Lesser Antilles from March to April 2012. These are among the few drill sites gathered around volcanic islands, and the first scientific drilling of large and likely tsunamigenic volcanic island-arc landslide deposits. These cores provide evidence and tests of previous hypotheses for the composition and origin of those deposits. Sites U1394, U1399, and U1400 that penetrated landslide deposits recovered exclusively seafloor-sediment, comprising mainly turbidites and hemipelagic deposits, and lacked debris avalanche deposits. This supports the concepts that i/ volcanic debris avalanches tend to stop at the slope break, and ii/ widespread and voluminous failures of pre-existing low-gradient seafloor sediment can be triggered by initial emplacement of material from the volcano. Offshore Martinique (U1399 and 1400), the landslide deposits comprised blocks of parallel strata that were tilted or micro-faulted, sometimes separated by intervals of homogenized sediment (intense shearing), while Site U1394 offshore Montserrat penetrated a flat-lying block of intact strata. The most likely mechanism for generating these large-scale seafloor-sediment failures appears to be propagation of a decollement from proximal areas loaded and incised by a volcanic debris avalanche. These results have implications for the magnitude of tsunami generation. Under some conditions, volcanic island landslide deposits comprised of mainly seafloor sediment will tend to form smaller magnitude tsunamis than equivalent volumes of subaerial block-rich mass flows rapidly entering water. Expedition 340 also successfully drilled sites to access the undisturbed record of eruption fallout layers intercalated with marine sediment which provide an outstanding high-resolution dataset to analyze eruption and landslides cycles, improve understanding of magmatic evolution as well as offshore sedimentation processes. This article is protected by copyright. All rights reserved.