THE ROLE OF MATNLE-DEPLETION AND MELT-RETENTION BUOYANCY IN SPREADING-CENTER SEGMENTATION

Numerical experiments are used to examine the structure of mantle flow beneath the axis of a spreading center. Buoyancy results from the depletion of residual mantle in iron relative to magnesium as melt is extracted (mantle-depletion buoyancy) and from the presence of low-density melt (melt-retention buoyancy). 3-D buoyant mantle flow arises spontaneously from an initially 2-D solution for low mantle viscosity and low spreading rate. At high viscosity and high spreading rate, initially 2-D solutions remain 2-D. This may explain the fundamentally different structure of fast and slow spreading centers. In a uniform viscosity halfspace, the along-axis wavelength of 3-D buoyant upwelling scales with the maximum depth of melting, the only length scale in this system. For reasonable maximum depths of melting (60-90 km), along-axis wavelengths of 200-300 km are preferred, longer than the 50-100 km segmentation length of slow spreading centers. In a viscosity-layered halfspace, the thickness of the asthenosphere introduces another length scale. A wavelength of segmentation comparable to the asthenosphere thickness also develops in our numerical experiments. This suggests the possibility that a wavelength corresponding to the spacing between gravity lows may be controlled by the asthenosphere thickness, while the spacing of major fracture zones corresponds to the longer wavelength (almost-equal-to 3 times the maximum depth of melting) intrinsic to the melting region. The along-axis structure of 3-D flow varies from narrow, focused upwellings, at low spreading rates and mantle viscosities, to broad regions of upwelling at high spreading rates and mantle viscosity. To allow an along-axis crustal thickness variation no larger than that which is observed (almost-equal-to 3-4 km), the highly focused upwelling and crustal production predicted at slow spreading rates requires appreciable along-axis transport of melt.