Mantle Flow and Melt Transport in the Mantle beneath Oceanic Spreading
Centers
Bill Hammond and Rob Dunn
 
INTRODUCTION
 
Rob and I are splitting this discussion on the transport of melt beneath
the ridge axis.  In keeping with this class's bottom up approach to the mid ocean
ridges system, I will start with the deeper process of melt percolation
through grain scale interstices, and relate the presence of melt and melt migration
to broader scale geodynamic processes.

Hopefully by the end you will have an appreciation for the complexity of the
problem, and if at the end, you think it all makes perfect sense, its probably because
I left something out!
 

I. SOME MOTIVATING QUESTIONS: [Figure]

•  How does melt get focused towards the ridge axis?

• What is the shape of the melting region?

• To what extent is the melt interconnected through this region?

• What is the scale of mantle convection beneath the ridge?


 i.e. how deep does the ridge reach?

II. Microstructural look at melt pockets.  - We want to understand the physics of melt migration so here is the place to start.

• Melt forms at grain boundaries: therefore when the melt first mobilizes, it can be characterized as:
• Flow through a porous medium  - [Figure of square grains model - Bear, 1972]
• basic model with square grains
• Darcy's Law
• Bulk velocity of fluid proportional to pressure gradient
• dp/dx from difference in melt density and matrix density
• Connected vs Unconnected matrix - need connectivity in order for porous flow to occur
• Discuss models for melt pockets as function of dihedral angle - [Figure- Schmeling]
• Experimental work of Waff, Faul & Kholsedt has gone a long way toward showing us how the melt really looks. [Figure - Faul, JGR 1997 & Figure Riley & Kholstedt]
• Important because:  [Figure]
1. Want you to understand how we think about the migration of melt in the region where it is generated.
2. Has a direct effect on the local melt % and hence the spacial distribution of physical properties dependent on melt fraction, like density, viscosity, elastic moduls
3. Only interconnected melt escapes to form oceanic crust.  thus the things we learn from the crust, petrology, etc may only reflect part of the melt that has been generated.

 III. How does the mantle melt?

    Dana & others will say more but...

• See Figure of P/T path of parcel of mantle as it convects through MOR system.
• Mantle parcel intersects solidus and begins to melt
• parcel continue along trajectory, possibly losing melt as it goes
• eventually turns and flows horizontally and loses more heat conductively to the surface, so temperature goes down but depth remains approximately constant.
• path intersects solidus again and melt dissappears.
• note multiple possible paths for different streamlines withen the flow
• note that as mantle rises its temperature drops according to adiabatic decompression, and thus its temperature may not be much more that the surrounding mantle, even though it came from the hotter depths. - thus there may not be additional bouyancy due to thermal anomaly from ascending mantle.
• Now in position to look at some ideas for how melt gets focussed. [Figure]
• Melt solidus lid - sloping P/T boundary could cause pressure gradients and barrier to flow that guide melt towards the ridge.
• Anisotropic permeability - because of shear strain, elongation and alignment of periodotite mineral grains can occur, aka Lattice Preferred Orientation (LPO).  Implies that melt pockets will also be aligned and hence could be a source for preferred direction of flow.
• Mantle flow could be strongly focussed (see below) and hence melting could preferentially occur directly beneath the axis.
• Decompacting boundary layers - e.g. Sparks & Parmentier
• Other types of fracturing --> see Rob's talk.

IV. Why and How Does the Mantle Flow?

• Ridge is essentially a passive feature that responds to large scale plate motions.
• Fluid flow problem [Figure - show some solutions] with lithospheric plate velocity and thickening with sqrt(t) as boundary conditions.
• Geodynamic models of flow: explore the parameters...
• Fast vs. Slow spreading ridges [Figure - Parmentier] determined from 3 dimensional modelling where the viscosity and plate velocities were varied and diapiric flow was observed or not observed.
• Passive vs. Bouyant models [Figure - Blackman, et al.] - melt retention (non-connectivity) vs. melt extraction (connectivity).
• In melt retentive case, additional bouyancy is supplied by the presence of melt
• Note the differences here in the distribution of the melt, the velocity of mantle and the concentration of the upwelling in the central region beneath the ridge axis.
• The question of passive vs. bouyant upwelling was addressed in the MELT seismic experiment.
• [Figure, Su and Buck, 1993]
• temperature dependent viscosity
• thermal bouyancy
• depletion affecting density/viscosity
• melt retention
• melt fraction dependent viscosity
• migrating ridge [Figure - Schouten et al.]
• strain induced petrofabric providing anisotropic viscosity/thermal properties