Transitions in the style of mantle convection at high Rayleigh numbers

The pattern and style of mantle convection govern the thermal evolution, internal dynamics, and large-scale surface deformation of the terrestrial planets. In order to characterize the nature of heat transport and convective behaviour at Rayleigh numbers, Ra, appropriate for planetary mantles (between 10⁴ and 10⁸), we perform a set of laboratory experiments. Convection is driven by a temperature gradient imposed between two rigid surfaces, and there is no internal heating. As the Rayleigh number is increased, two transitions in convective behaviour occur. First we observe a change from steady to time-dependent convection at Ra≈10⁵. A second transition occurs at higher Rayleigh numbers, Ra≈5×10⁶, with large-scale time-dependent flow being replaced by isolated rising and sinking plumes. Corresponding to the latter transition, the exponent β in the power law relating the Nusselt number Nu to the Rayleigh number (NuRa^β) is reduced. Both rising and sinking plumes always consist of plume heads followed by tails. There is no characteristic frequency for the formation of plumes.     

Thermal and rheological controls on the formation of mafic enclaves or banded pumice

Magma mixing can occur in a fluid manner to produce banded pumice or in a brittle manner to form enclaves. We propose that the critical control on mixing style is a competition between developing networks of crystals in the intruding magma that impart a strength to the magma and melting and disrupting those networks in the host. X-ray computed tomography analysis demonstrates that banded pumice from the 1915 Mt. Lassen eruption lacks crystal networks. In contrast, rhyodacite hosts with mafic enclaves from Chaos Crags contain well-developed networks of large crystals. We present a one-dimensional conductive cooling model that predicts mixing style, either ductile or brittle, as a function of magma compositions, temperatures, and the size of the intruding dike. Our model relies on three assumptions: (1) Mixing is initiated by the injection of a hot dike into a cooler magma body with a yield strength; (2) when magma crystallinity exceeds a critical value, 13 vol% plagioclase, the magma develops a yield strength; and (3) when total crystallinity exceeds 40 vol%, the magma has a penetrative crystal network and is effectively solid. Importantly, because the two magmas are of different compositions, their crystallinities and viscosities do not have the same variations with temperature. As the intruding magma cools, it crystallizes from the outside in, while simultaneously, host magma temperature near the intruder rises. Mixing of the two magmas begins when the host magma is heated sufficiently to (1) disrupt the crystal network and (2) initiate convection. If the shear stress exerted by the convecting host magma on the dike is greater than the yield strength of the dike margin (and dike crystallinity does not exceed 40 %), then fluid mixing occurs, otherwise enclaves form by brittle deformation of the dike. Application of the model to magma compositions representative of Lassen and Chaos Crags shows that emplacement of dikes <1 m thick should produce enclaves, whereas thicker dikes should generate fluid mixing and form banded pumice within days to weeks of emplacement. Similar relationships apply to other modeled magmatic systems, including Pinatubo, Unzen, and Ksudach/Shtuybel’ volcanoes. For all studied systems, the absolute size of the intruding dike, not just its proportion relative to the host, influences mixing style.     

Heat transfer coefficients of natural volcanic clasts

Heat transfer coefficients used in numerical simulations of volcanic eruptions are typically borrowed from industrial settings where the coefficients are well determined for non-permeable, machined (spherical) materials. Volcanic clasts, in contrast, are permeable and have irregular shapes. We performed a series of laboratory experiments to determine heat transfer coefficients for natural volcanic particles. We measured the surface and interior temperatures during cooling at wind speeds ranging from 0 to 10 m/s. We also measured the permeability and density of the particles. We find that the permeability of the particles has little effect on clast cooling. In the absence of any wind, heat loss occurs by free convection, and we find no relationship between the heat transfer coefficient and particle density. However, for non-zero Reynolds numbers (finite wind speed), the heat transfer coefficient decreases with increasing porosity. We obtain a correlation for the dimensionless heat loss, or Nusselt number, of the form Nu = 2 + aRe^1/2Pr^1/3 where a is a density dependent coefficient given by a = 0.00022ρ + 0.31, with ρ in kg/m³, and Re and Pr are the Reynolds number and Prandtl number, respectively. Compared with non-porous particles, heat transfer coefficients for natural pumice clasts are reduced by a factor of 2–3 for particles with similar Re. Numerical simulations show that this leads to an increase in depositional temperature by 50–90 °C.     

An experimental facility for investigating hydromagmatic eruptions at high-pressure and high-temperature with application to the importance of magma porosity for magma-water interaction

An experimental facility has been developed to investigate magma-water interaction (MWI). The facility operates in a high-pressure and high-temperature environment, with temperatures up to 1,200°C and pressures up to 200 MPa. Cylindrical sample-holders (20 by 180 mm in size) are heated conductively to yield a three phase (melt, crystals and gas) system, and then water (or other fluid) is injected into the sample through a capillary tube (diameter 0.5 mm, length ca. 1,000 mm) under controlled conditions. Pressure, volume and temperature changes are continuously recorded during every phase of the experiments. To test this facility, MWI is studied at subliquidus temperatures (800 and 900°C) and pressure (8 MPa), using a leucite tephrite sample with two different initial grain sizes. Because of the grain-size dependence of sintering, the two starting materials produce magmas with different textures at the same temperature: porous magma for large initial grain sizes and dense magma for small initial grain sizes. In these experiments 1.5 g of water at room temperature is injected into 6.0 g of partially molten sample at velocities ranging from 1 to 3 m/s. We find that the extent of fragmentation and transport caused by MWI are mainly controlled by the texture of the interacting sample with explosive interaction occurring only for porous magmas.     

Rounding of pumice clasts during transport: field measurements and laboratory studies

Volcanic clasts in many pyroclastic density current deposits are notably more round than their counterparts in corresponding fall deposits. This increase in roundness and sphericity reflects different degrees of comminution, abrasion and breakup during transport. We performed experimental measurements to determine an empirical relationship between particle shape and mass loss caused by particle–particle interactions. We consider, as examples, pumice from four volcanoes: Medicine Lake, California; Lassen, California; Taupo, New Zealand; Mount St Helens, Washington. We find that average sample roundness reaches a maximum value once particles lose between 15% and 60% of their mass. The most texturally homogeneous clasts (Taupo) become the most round. Crystal-rich pumice abrades more slowly than crystal-free pumice of similar density. Abrasion rates also decrease with time as particles become less angular. We compare our experimental measurements with the shapes of clasts in one of the May 18, 1980 pyroclastic density current units at Mount St Helens, deposited 4–8 km from the vent. The measured roundness of these clasts is close to the experimentally determined maximum value. For a much smaller deposit from the 1915 Lassen eruption, clast roundness is closer to the value for pumice in fall deposits and suggests that only a few volume percent of material was removed from large clasts. In neither field deposit do we see a significant change in roundness with increasing distance from the vent. We suggest that this trend is recorded because much of the rounding and ash production occur in proximal regions where the density currents are the most energetic. As a result, all clasts that are deposited have experienced similar amounts of comminution in the proximal region, and similar amounts of abrasion as they settle through the dense, near-bed region prior to final deposition.