Abstract Magmatic systems are composed of melt accumulations and crystal mush that evolve with melt transport, contributing to igneous processes, volcano dynamics, and eruption triggering. Geophysical studies of active volcanoes have revealed details of shallow-level melt reservoirs, but little is known about fine-scale melt distribution at deeper levels dominated by crystal mush. Here, we present new seismic reflection images from Axial Seamount, northeastern Pacific Ocean, revealing a 3–5-km-wide conduit of vertically stacked melt lenses, with near-regular spacing of 300–450 m extending into the inferred mush zone of the mid-to-lower crust. This column of lenses underlies the shallowest melt-rich portion of the upper-crustal magma reservoir, where three dike intrusion and eruption events initiated. The pipe-like zone is similar in geometry and depth extent to the volcano inflation source modeled from geodetic records, and we infer that melt ascent by porous flow focused within the melt lens conduit led to the inflation-triggered eruptions. The multiple near-horizontal lenses are interpreted as melt-rich layers formed via mush compaction, an interpretation supported by one-dimensional numerical models of porous flow in a viscoelastic matrix.
Recent geochemical studies of MORB genesis suggest that at least some degree of chemical disequilibrium occues during the transport of magma to the surface. If disequilibrium transport does occur in the mantle, it would seem to preclude melt being distributed in a porous network on grain boundaries that could rapidly re-equilibrate with the solid. The questions remain however, as to how big a melt "channel" is required to produce disequilibrium and whether flow in such channels would violate assumptions inherent in the equations of magma migration. Using a series of simple physical scaling arguments, we quantify the requirements for chemical disequilibrium and lay out the conditions for which the melt migration equations are valid. These arguments show that a vein network with veins ∼ 10 cm apart is sufficient to cause significant disequilibrium. More precisely, these arguments show that to maintain equilibrium, the solid-state diffusion coefficient would need to increase by 2–4 orders of magnitude for every order of magnitude increase in channel spacing. Nevertheless, because the equations of magma migration are a macroscopic description of melt flow, they can readily describe even large scale networks of melt channels. By demonstrating the fundamental scalings governing the chemistry and motion of partial melts, these simple arguments show that, while the chemistry may be extremely sensitive to the microscopic distribution of melt, our physical understanding of magma migration is robust.
In the companion paper, equations for partially molten media were derived using two‐scale homogenization theory. This approach begins with a grain‐scale description and then coarsens it through multiple scale expansions into a macroscopic model. One advantage of homogenization is that effective material properties, such as permeability and the shear and bulk viscosity of the two‐phase medium, are characterized by cell problems, boundary value problems posed on a representative microstructural cell. The solutions of these problems can be averaged to obtain macroscopic parameters that are consistent with a given microstructure. This is particularly important for estimating the “compaction length” which depends on the product of permeability and bulk viscosity and is the intrinsic length scale for viscously deformable two‐phase flow. In this paper, we numerically solve ensembles of cell problems for several geometries. We begin with simple intersecting tubes, as this is a one parameter family of problems with well‐known results for permeability. Using the data, we estimate relationships between the porosity and all of the effective parameters by curve fitting. For the model of intersecting tubes, permeability scales as ϕ n , n ∼ 2, as expected, and the bulk viscosity scales as ϕ − m , m ∼ 1, which has been speculated but never shown directly for deformable porous media. The second set of cell problems adds spherical inclusions where the tubes intersect. For these geometries, the permeability is controlled the pore throats and not by the total porosity, as expected. However, the bulk viscosity remains inversely proportional to the porosity, and we conjecture that this quantity is insensitive to the specific microstructure. The computational machinery developed can be applied to more general geometries, such as texturally equilibrated pore shapes. However, we suspect that the qualitative behavior of our simplified models persists in these more realistic structures. In particular, our hybrid numerical‐analytical model predicts that for purely mechanical coupling at the microscale, all homogenized models will have a compaction length that vanishes as porosity goes to zero. This has implications for numerical simulations, and it suggests that these models might not resist complete compaction.
We review physical and chemical constraints on the mechanisms of melt extraction from the mantle beneath mid–ocean ridges. Compositional constraints from MORB and abyssal peridotite are summarized, followed by observations of melt extraction features in the mantle, and constraints from the physical properties of partially molten peridotite. We address two main issues. (1) To what extent is melting 'near–fractional', with low porosities in the source and chemical isolation of ascending melt? To what extent are other processes, loosely termed reactive flow, important in MORB genesis? (2) Where chemically isolated melt extraction is required, does this occur mainly in melt–filled fractures or in conduits of focused porous flow?Reactive flow plays an important role, but somewhere in the upwelling mantle melting must be 'near fractional', with intergranular porosities less than 1%, and most melt extraction must be in isolated conduits. Two porosity models provide the best paradigm for this type of process. Field relationships and geochemical data show that replacive dunites mark conduits for focused, chemically isolated, porous flow of mid–ocean ridge basalt (MORB) in the upwelling mantle. By contrast, pyroxenite and gabbro dikes are lithospheric features; they do not represent conduits for melt extraction from the upwelling mantle. Thus, preserved melt extraction features do not require hydrofracture in the melting region. However, field evidence does not rule out hydrofracture.Predicted porous flow velocities satisfy 230Th excess constraints (ca. 1 m yr-1, provided melt extraction occurs in porous conduits rather than by diffuse flow, and melt-free, solid viscosity is less than ca. 1020 Pa s. Melt velocities of ca. 50 m yr-1 are inferred from patterns of post–glacial volcanism in Iceland and from 226Ra excess. If these inferences are correct, minimum conditions for hydrofracture may be reached in the shallowest part of melting region beneath ridges. However, necessary high porosities can only be attained within pre–existing conduits for focused porous flow. Alternatively, the requirement for high melt velocity could be satisfied in melt–filled tubes formed by dissolution or mechanical instabilities.Melt–filled fractures or tubes, if they form, are probably closed at the top and bottom, limited in size by the supply of melt. Therefore, to satisfy the requirements for geochemical isolation from surrounding peridotite, melt–filled conduits may be surrounded by a dunite zone. Furthermore, individual melt–filled voids probably contain too little melt to form sufficient dunite by reaction, suggesting that dunite zones must be present before melt extraction in fractures or tubes.
Dynamic models are presented to investigate the consequences of melting and melt transport for stable trace element geochemistry in open systems. These models show that including explicit melt transport in 2-D adds non-trivial behaviour because melts and residues can travel and mix along very different trajectories. Calculations are presented for both equilibrium and disequilibrium transport, and passive and active mid-ocean ridge flows. These calculations demonstrate that trace elements are sensitive to mantle dynamics and can readily distinguish between different end-member flow fields. Passive, plate-driven flow with strong melt focusing produces enrichments of incompatible elements. Active small-scale solid convection within the partially molten region, however, can lead to extreme dilution of incompatible elements, suggesting that this form of convection may not be significant beneath normal ridges. These calculations provide additional predictions about across-axis trends of geochemical variability and estimate the variation in concentrations that can occur even for a constant source. Many of these results are not seen in geochemical models that neglect melt transport and we discuss how this new behaviour affects the inferences drawn from simpler models.
Geochemical and field evidence suggest that melt transport in some regions of the mantle is localized into mesoscale “channels” that have widths of 0.1–100 m or larger. Nevertheless, the mechanisms for formation of such channels from a grain‐scale distribution of melt are poorly understood. The purpose of this paper is to investigate one possible mechanism for channel formation: the reaction infiltration instability (RII). We present numerical solutions of the full equations for reactive fluid flow in a viscously deformable, permeable medium. We show that dissolution in a compactible solid with a vertical solubility gradient can lead to significant flow localization such that >90% of the melt flux is channelized in <20% of the available area. In particular, the ability of the solid to compact enhances the instability by forming impermeable regions between channels. The combination of reaction, diffusion, and solid compaction leads to strong selection of preferred length scales with channel spacing smaller than the compaction length (δ ∼ 10 2 − 10 4 m). We explore the evolution of dissolution channels over parameter space and show that the behavior of the full nonlinear solutions is consistent with predictions from linear stability analysis. We also briefly consider the behavior of the instability in the presence of melting due to adiabatic decompression and demonstrate that significant localization can occur even in the presence of uniform melting and compaction. Using the linear analysis to extend these results for parameters expected in the Earth's mantle suggests that robust channel systems could form through the RII from a homogeneous system in ∼100,000 years with dominant channel spacing of 1–200 m.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)
