Porosity‐driven convection and asymmetry beneath mid‐ocean ridges

Abstract

[1] Seismic tomography of the asthenosphere beneath mid-ocean ridges has produced images of wave speed and anisotropy that are asymmetric across the ridge axis. These features have been interpreted as resulting from an asymmetric distribution of upwelling and melting. Using computational models of coupled magma/mantle dynamics beneath mid-ocean ridges, I show that such asymmetry should be expected if buoyancy forces contribute to mantle upwelling beneath ridges. The sole source of buoyancy considered here is the dynamic retention of less dense magma within the pores of the mantle matrix. Through a scaling analysis and comparison with a suite of simulations, I derive a quantitative prediction of the contribution of such buoyancy to upwelling; this prediction of convective vigor is based on parameters that, for the Earth, can be constrained through natural observations and experiments. I show how the width of the melting region and the crustal thickness, as well as the susceptibility to asymmetric upwelling, are related to convective vigor. I consider three causes of symmetry breaking: gradients in mantle potential temperature and composition and ridge migration. I also report that in numerical experiments performed for this study, the fluid dynamical instability associated with porosity/shear band formation is not observed to occur.

Geodynamic Models of Melt Generation and Extraction at Mid-Ocean Ridges

“…The large sensitivity of crustal thickness to potential temperature arises because melt focusing pools the difference in melt production over a large area. crustal thickness has a maximum at ~ 2 cm yr -1 because the relative contribution of buoyancy-driven, active upwelling (compared to platedriven, passive upwelling) is largest here (see Katz, 2010, for details). The fit to data was achieved by increasing the clapeyron slope of the solidus, dP/dT, to give a depth of initial melting of ~ 65 km, and reducing the potential temperature (τ 0 = 1,633 K) to achieve the observed scale of crustal thickness.…”

“…The channel may be generated self-consistently (e.g., Katz, 2008) or may follow an approximate parameterization (Hebert and Montési, 2010). Variations in crustal thickness and lava chemistry are now used to test the importance of segmentation (Gregg et al, 2009;Weatherley and Katz, 2010;Hebert and Montési, 2011;Montési et al, 2011), buoyancy (Katz, 2010), and ridge migration (Toomey et al, 2002;Conder et al, 2002;Katz et al, 2004). New under- …”

“…This concept represents a small perturbation to corner flow. The dynamics of this combination were explored in subsequent studies that investigated the effect of ridge migration, pressure gradients, and thermal anomalies on the shape of the melt region Toomey et al, 2002;Conder, 2007;Katz, 2010). and the ability of these models to cross boundaries and integrate both geophysical (Gregg et al, 2007) and geochemical/petrological observations (Shaw et al, 2010) provide the opportunity to examine fundamental questions in mantle dynamics and the formation of new oceanic crust at mid-ocean ridges.…”

“…models (squares + lines) after Katz (2010) compute active mantle flow, melting, and melt migration for a homogeneous mantle. The three curves correspond to three mantle potential temperatures.…”

Multilevel and local time-stepping discontinuous Galerkin methods for magma dynamics

“…While the spatial discretization of these problems was very efficient and simple, these methods cannot be easily extended to high-order methods on general grids due to extended stencils required for reconstruction [21]. Finite volume methods have been used to model melt migration beneath the mid-ocean ridge [15,23], and these methods provide the aforementioned geometric flexibility for lower order. However, to increase the order of accuracy in more than one dimension requires costly reconstruction.…”