Marcus I. Bursik

Computing granular avalanches and landslides

Geophysical mass flows—debris flows, volcanic avalanches, landslides—are often initiated by volcanic activity. These flows can contain O(106–107) m3 or more of material, typically soil and rock fragments that might range from centimeters to meters in size, are typically O(10 m) deep, and can run out over distances of tens of kilometers. This vast range of scales, the rheology of the geological material under consideration, and the presence of interstitial fluid in the moving mass, all make for a complicated modeling and computing problem. Although we lack a full understanding of how mass flows are initiated, there is a growing body of computational and modeling research whose goal is to understand the flow processes, once the motion of a geologic mass of material is initiated. This paper describes one effort to develop a tool set for simulations of geophysical mass flows. We present a computing environment that incorporates topographical data in order to generate a numerical grid on which a parallel, adap...

Sedimentation from turbulent jets and plumes

Theoretical models are developed for the sedimentation from the margins of a particle-laden, axisymmetric, turbulent, buoyant plume, in a still environment and for an axisymmetric turbulent momentum jet. The models assume that the mass of each individual size fraction of sediment carried in a parcel of fluid decreases exponentially with time. For relatively coarse particles, the fallout models predict that the sediment deposition beyond a distance r on the ground expressed in log units should decay linearly with distance away from the vent for the momentum jet and should decrease with r1/3 for the buoyant plume. The exponential decay constant J is proportional to the terminal fall velocity Vt of the particles in both cases and inversely proportional to the square root of the initial momentum flux M0 for the jet fallout (Jj ∝ VtMo−1/2) and to the third power of the initial buoyancy flux Fo for the plume fallout (Jp ∝ VtFo−1/3). Smaller particles are affected by reentrainment caused by the turbulent eddies sweeping ambient fluid back into the plume or jet and thus reincorporating some particles that were released from the flow at greater heights. This is taken into account by introducing a reentrainment coefficient, ϕ, into the theoretical models with the assumption that the coefficient has a constant value for a plume of given strength. In new experiments, fallout occurs from the margins of particle-laden, fresh water, buoyant jets, and plumes in a tank of salty water, and sedimentation is measured on the tank floor. Two experiments were weakly affected by reentrainment and show excellent agreement with the simple theory. For smaller particles and increasingly buoyant plumes and strong jets, particle reentrainment is important. The experimental data are fitted by the new reentrainment theory, confirming that values of the reentrainment coefficient are approximately constant for a given flow. A settling number, β, is defined as the ratio of the characteristic velocity of the jet or plume to the particle settling velocity. For β ≥ 1, reentrainment seems to reach an equilibrium state for which the reentrainment coefficient is a constant of value 0.1 for jets and 0.4 for plumes, irrespective of flow strength or particle size. The plume experiments indicate that the value of the reentrainment coefficient is strongly dependent on plume strength and particle size for β slightly less than 1. The general principles of sedimentation from turbulent plumes and jets are applied to the fallout of pumice from volcanic eruption columns and of metalliferous particles from black smokers on the ocean floor. For volcanic eruptions, the results provide an explanation for the near vent overthickening of tephra fall deposits and imply that lithic and pumice fragments from small lapilli up to at least 1 m diameter blocks are efficiently reentrained into eruption columns. The size of particles reentrained in hydrothermal plumes is predicted to vary from less than 100 μm in weakly buoyant plumes up to over 1000 μm in megaplumes.

Continuous monitoring of surface deformation at Long Valley Caldera, California, with GPS

Continuous Global Positioning System (GPS) measurements at Long Valley Caldera, an active volcanic region in east central California, have been made on the south side of the resurgent dome since early 1993. A site on the north side of the dome was added in late 1994. Special adaptations for autonomous operation in remote regions and enhanced vertical precision were made. The data record ongoing volcanic deformation consistent with uplift and expansion of the surface above a shallow magma chamber. Measurement precisions (1 standard error) for “absolute” position coordinates, i.e., relative to a global reference frame, are 3–4 mm (north), 5–6 mm (east), and 10–12 mm (vertical) using 24 hour solutions. Corresponding velocity uncertainties for a 12 month period are about 2 mm/yr in the horizontal components and 3–4 mm/yr in the vertical component. High precision can also be achieved for relative position coordinates on short (<10 km) baselines using broadcast ephemerides and observing times as short as 3 hours, even when data are processed rapidly on site. Comparison of baseline length changes across the resurgent dome between the two GPS sites and corresponding two-color electronic distance measurements indicates similar extension rates within error (∼2 mm/yr) once we account for a random walk noise component in both systems that may reflect spurious monument motion. Both data sets suggest a pause in deformation for a 3.5 month period in mid-1995, when the extension rate across the dome decreased essentially to zero. Three dimensional positioning data from the two GPS stations suggest a depth (5.8±1.6 km) and location (west side of the resurgent dome) of a major inflation center, in agreement with other geodetic techniques, near the top of a magma chamber inferred from seismic data. GPS systems similar to those installed at Long Valley can provide a practical method for near real-time monitoring and hazard assessment on many active volcanoes.

Emplacement of pyroclastic flows during the 1998–1999 eruption of Volcán de Colima, México

After three years of quiescence, Volcan de Colima reawakened with increasing seismic and rock fall activity that reached its peak on November 20, 1998, when a new lava dome forced its way to the volcano’s summit. The new lava rapidly reached the S–SW edge of the summit area, beginning the generation of Merapi-type pyroclastic flows that traveled down La Lumbre, and the El Cordoban Western and Eastern ravines, reaching distances of 3, 4.5, and 3 km, respectively. On December 1, 1998, the lava flow split into three fronts that in early 1999 had reached 2.8, 3.1, and 2.5 km in length, advancing down the El Cordoban ravines. The lava flow fronts disaggregated into blocks forming pyroclastic flows. One of the best examples occurred on December 10, 1998. As the lava flow ceased moving in early 1999, activity became more explosive. Strong blasts were recorded on February 10, May 10, and July 17, 1999. The last event developed a 10-km-high eruptive column from which a pyroclastic flow developed from the base, traveling 3.3 km SW from the summit into the San Antonio–Montegrande ravines. Regardless of the mechanism of pyroclastic-flow generation, each flow immediately segregated into a basal avalanche that moved as a granular flow and an upper ash cloud in which particles were sustained in turbulent suspension. When the basal avalanche lost velocity and eventually stopped, the upper ash cloud continued to move independently as a dilute pyroclastic flow that produced a massive pyroclastic-flow deposit and an upper dune-bedded surge deposit. The dilute pyroclastic flow scorched and toppled maguey plants and trees, and sandblasted vegetation in the direction of the flow. At the end of the dilute pyroclastic-flow path, the suspended particles lifted off in a cloud from which a terminal ash fall was deposited. The basal avalanche emplaced block-and-ash flow deposits (up to 8 m thick) that filled the main ravines and consisted of several flow units. Each flow unit was massive, monolithologic, matrix-supported, and had a clast-supported steep front (ca. 1.5 to 2 m thick) composed of boulders up to 1.7 m in diameter. The juvenile lithic clasts had an average density of 1800 kg/m3. The dilute pyroclastic flow emplaced overbank deposits, found on valley margins or beyond the tip of block-and-ash flow deposits. They consist from bottom to top of a massive medium to coarse sand-size flow layer (2–4 cm thick), a dune-bedded surge layer (2–10 cm thick), and a massive silt-size layer (0.5 cm thick). The total estimated volume of the pyroclastic-flow deposits produced during the 1998–1999 eruption is 24×105 m3.

The Effects of Topography on Sedimentation from Particle-Laden Turbulent Density Currents

ABSTRACT We present results from a series of laboratory experiments that illustrate the influence of changes in channel topography (depth or width) on the sedimentation patterns produced by steady, particle-laden currents. On a planar surface in a channel with constant width, such currents created deposits that thinned exponentially with distance. As long as a current was not blocked to produce a bore, topographic features consisting of constrictions, ridges, and sudden openings caused no significant deviation from exponential deposit thinning or discontinuity in thickness, even when there was a transition in flow regime caused by the topographic feature. Changes in the channel width did affect the distance over which the deposit thinned, with the deposits associated with wider parts of the channel thinning more rapidly with distance. In contrast, if the topographic change was large enough to partially reflect the flow, producing an upstream-propagating bore, then the deposit did not thin exponentially. The results are consistent with a model in which the current is assumed to be turbulent and well mixed. In this case, sedimentation occurs at a rate proportional to the channel width, the settling speed of the particles, and their concentration within the current, and inversely proportional to the current discharge. Such a model predicts that as a flow passes through a topographic control that does not produce a bore, the deposit continues to thin exponentially, even if the flow undergoes a transition from the subcritical to the supercritical regime. The results suggest that in natural systems changes in current speed and flow regime do not in themselves produce changes in deposit thickness or gradient in thickness unless the flow is partially blocked and a fraction of the flow is reflected upstream.

Subplinian eruption mechanisms inferred from volatile and clast dispersal data

Abstract Subplinian tephras erupted during the North Mono eruption of 1350 A.D. contain abundant obsidian clasts. The weight percent of water in the clasts from each tephra bed can be used to infer the depths from which the clasts originated within the magmatic system. A comparison of the depths of origin with inferred eruption dynamics and local bedrock stratigraphy suggests that the clasts were eroded from conduit walls during repeated withdrawal of magma from a replenished shallow magmatic system. Furthermore, the obsidian was preferentially formed at and eroded from sections in the conduit where increased cooling and magma fragmentation took place due to proximity to groundwater or to the surface. Finally, the data are compatible with magma withdrawal during separate explosive phases resulting from the movement of a disruption surface downward into magma columns that became larger in cross-sectional area with each explosive phase.

A laboratory study of ash flows

Hot ash flows propagate from volcanic vents during some explosive volcanic eruptions. As a flow entrains and heats ambient air and simultaneously deposits pyroclasts, part of the flow may become less dense than the air and ascend from the flow. If entrainment of air is more important than sedimentation in controlling the flowbehaviour, then a large, buoyant ash cloud may form above the flow. Conversely, if sedimentation of particles dominates, then a much smaller fraction of the particles are able to ascend in the buoyant cloud. We have carried out a series of analog laboratory experiments to simulate the motion and buoyancy generation in dilute ash flows propagating along the ground. The first series of experiments investigated buoyancy generation through entrainment, using mixtures of methanol, ethylene glycol and water. The second series examined the role of sedimentation upon the generation of buoyancy, using particle-laden, fresh-water currents propagating in a saline ambient fluid. The experiments showed that the rate of entrainment increases while the sedimentation rate decreases as the downward slope of the terrain increases, suggesting that massive ash-laden clouds are more likely to rise from flows traveling on steeper slopes. Changes in slope angle also control the motion of ash flows. The experiments suggest that flow separation and hydraulic jumps resulting from sudden changes in slope angle can enhance entrainment or sedimentation and increase buoyancy. Observations of the the interactions of ash flows with topography at Redoubt Volcano in April 1990 and at Mount St. Helens on May 18, 1980, are consistent with our laboratory observations.

The influence of diffusive convection on sedimentation from buoyant plumes

Abstract Diffusive convection driven by the differential diffusion of density altering fluid properties may enhance the scavenging of particles from natural buoyant plumes. For single-phase (fluid–solute–heat) systems this phenomenon has been extensively studied because salt fingering generated at the oceanic thermocline is a major mechanism of salt transport in the oceans. However, the influence of this process on particle laden plumes, for example, fluvial plumes in lakes and estuaries volcanic clouds and seafloor hydrothermal plumes is largely unknown. In this paper, we present direct experimental measurements of the interfacial particle flux at the plume base which can be applied to predict particle scavenging from natural buoyant plumes. Particle flux is measured using a light attenuation technique employing a chain of photodiodes which average concentration over a large number of fingers. The results are in good general agreement with earlier studies based on finger velocity. Flux measurements cover a wide range of conditions from those where diffusive convection dominates to those where settling and diffusive convection are of a similar magnitude. For very small particles double diffusive (salt finger) theory is applicable to two component particulate systems as suggested by earlier studies [Green, T., 1987. The importance of double diffusion to the settling of suspended material. Sedimentology 34, 319–331]. Two component diffusive convection theory is extended to three components in order to predict particle scavenging from marine fluvial plumes which involve the diffusion of sediment, salt and heat. For larger particles which settle significantly the flux can be approximated by adding the double-diffusive and settling fluxes. A theory to predict particle transport through the lower layer and sedimentation at the bed is developed based on the observation of strong convection below the plume. Application of our theory and experimental results indicates that while double diffusion may significantly influence the longitudinal distribution and vertical sorting of deposits from lacustrine plumes, the diffusive convection process is generally insignificant in marine plumes. Observational evidence for lacustrine double diffusion based on water column measurements is presented.

Development of lithic-breccias in the 1982 pyroclastic flow deposits of El Chichon Volcano, Mexico

Abstract Pyroclastic flow deposit F1 (volume 0.02 km 3 ) produced during Phase III of the 1982 eruption of El Chichon Volcano, Mexico, contains basal lithic breccias. The breccia layers are well exposed in El Platanar gully between 2 and 4 km east of the volcano crater. The lithic breccias are inversely graded as defined by lithics, dense juvenile blocks, and pumice supported in a coarse sandy matrix composed of the same constituents. The contact between the main body of the pyroclastic flow deposit and the lithic breccias is generally sharp and planar but not erosive. In some outcrops it is gradational, and is only shown by an alignment of lithic clasts. The origin of these beds is interpreted to be due to a hydraulic jump in the moving pyroclastic flow formed after a pronounced slope break (from 11° to 3°), at the site where the flow began to be funnelled into the El Platanar gully. We have investigated the possible modes of formation of the lithic breccias with analog laboratory experiments. The experiments show that coarse localized segregations could form through a number of mechanisms. The field observations interpreted with the assistance of the laboratory results suggest that pyroclastic flow 1 (which produced deposit F1) moved as a kind of density stratified flow, with a basal lithic-rich zone transporting larger particles and an upper, less-dense zone transporting smaller particles in suspension. At the slope break, flow 1 lost competence and dumped the largest particles, forming a piled-up breccia. Downstream, somewhat smaller lithic particles may have been deposited as ballistics from a low-angle, jetlike structure comprising a hydraulic jump. This deposit thins with distance downstream. Once the deposit was sufficiently thick on its upstream end, particles may have been re-entrained into the jet, to be deposited further along the flow in a low-angle, downstream prograding, bouldery dune. Reverse grading in the dune beds may result from kinetic sieving. We suggest that planar contacts between the lithic breccias and the associated pyroclastic flow deposit may be the result of strong density and velocity gradients between the two parts of the flow.

Advances in studies of dense volcanic granular flows

The collapse and decrepitation of a lava dome at the summit of a volcano generally results in the generation of dense granular flows, often referred to as block and ash flows. As the dome particles propagate from the source, they break apart by internal pressure as well as collision. The propagation of block and ash flows can be simulated to some accuracy with a depth averaged numerical model of the equations of continuity and momentum for a material with a frictional resistance. However, important features of such flows, such as the influence of remote stress through force chains, erosion of the volcano substrate, and shock formation and pressurization upon particle break up are poorly understood. In the near future, the influence of these factors will be incorporated into depth averaged models. Various numerical techniques based on particles will some day yield results that can be compared not only with bulk flow properties, but to the internal layering of block and ash flow deposits.