Geol 456/656 - The Mechanism of Plate Tectonics

Thermal convection in the mantle drives plate tectonics. Two older theories had to be discounted first.

heat loss by conduction
200-600 km loss of circumference
orogeny by thrusting
discovery of radioactivity
- greater age of earth than Kelvin's cooling
- internal heat source
doesn't explain extensional regions

continental lithosphere originally continuous
universal gravitation G decreasing with time?
Pangaea breakup during 3x area expansion last 300 Ma
earth moment of inertia constraints
- expansion would slow rotation to conserve angular momentum
- rotation also slows by lunar tides
- tides also affect lunar orbit, month
- Devonian coral growth rings constrain lunar month
- day 24 s longer every million years
- only allows 0.5% expansion
ancient earth radius
- sample sites of same age on stable continent
- compute paleolatitudes from paleomagnetics
- known latitudes and distance gives paleo-radius
- 400 Ma radius within 3% of present
how to explain observed subduction? pre-Mesozoic drift?

temperature gradient at surface about 25 C/km
extrapolates to 2500 C at 100 km, but not molten since S-waves pass
temperature (T) gradient must decrease at depth
- shallow heat sources
- convection can provide same heat flux w/ lower T gradient
less heat flow from older regions (200 -> 40 mW/sq m)
radioisotopes in continental granites provide 70%
ocean crust has few radioisotopes, 96% of heat from below
solid convection in the mantle transports this heat

finite strength argument overcome by melting without convection
convection parameters for Newtonian fluid (power law of one, too low)
- Rayleigh number R - convects above 2380, 6000 in upper mantle
- Reynolds number Re - low for laminar flow, high for turbulent, laminar in mantle
- Taylor number T - less than one, convection pattern does not depend on rotation
- Nusselt number Nu - ratio of total heat transport over conductive, 10
numerical or lab modeling extremely difficult
non-Newtonian rheology increases critical Rayleigh number

if 670 discontinuity is change in chemical composition, convection will not cross it
with 2-layer convection the 670 would be a thermal boundary layer
subducted slabs would pile up at 670, seismic tests
convection cells may change shape or distribution above and below
circular shape of Pacific and African plates suggests whole-mantle cells
return flow not balanced per plate, only globally
this new GeoSat geoid radar-alteimetry image from NOAA and UCSD shows spreading and subduction rate differences in the morphology of the ridges and trenches (click above for a larger image, or here to see the original work).

Morgan's (1971, 1972) drive by radial spreading from hotspot tops can't put a consistent direction on the plates.
Largest hotspot, Hawaii, should be elevated.
But Kellog (Oct. 1995 Geology Colloquium) showed episodes of giant plume volumes have occurred, and may work together with ridge upwelling to augment plate motions.

Direct viscous drag on lithosphere base by mantle convection cell.
Thermal boundary layer is the asthenosphere.
Convection generally follows ridge and trench location.
No cells beneath continents.
Tension at ridges, compression at trenches.
Zone of drag is partially-molten less-viscous asthenosphere, so convection would have to move at 5 times as fast as lithosphere. (Harder for hotspots to punch through?)
Large convection cells, >2500 km, implies simpler plate boundaries than appear.
How can large cells move relative to each other, as in migrating triple junctions?
How can there be small plates?
Asthenospheric conduction only accounts for 10% of mantle heat output.

Top of convection cell is the oceanic lithosphere.
Thermal boundary layer is the lithosphere.
Ridge push from asthenospheric upwelling.
Slab pull at trenches, with denser phases deeper.
Better heat transfer from mantle.
Compression at ridges, tension at trenches.
But (unlike Kearey and Vine) earthquakes show only normal mechanisms at ridges, and in slabs any normal mecanisms are balanced by thrusts on the other side of the slab, from bending (no overall extension).
Agrees with present plate motions (from Forsyth and Uyeda, 1975):
- Plate velocity is independent of plate area. Mantle drag suggests larger plates should be faster.
- Plates with more downgoing slabs move faster.
- Plates with more continental area move more slowly.
Convection very closely follows ridge and trench location.
Easy to have small plates, complex boundaries.

Continental crust generates enough heat to allow only occasional hotspots below.
Convection geometry the same as oceanic lithosphere geometry.
Bathymetry, gravity, and geoid are all high over plumes and low over slabs.
Kearey and Vine assert no correspondence between ridges and geoid highs, but new GeoSat data (above) dispute that.
Any spreading centers not at geoid highs could be passive and not driven by any mantle upwelling.
Elongation of geoid anomalies explained by Kearey and Vine as small-scale convection now appears with GeoSat data to be related only to fracture zones, which may record ancient geometries of continental breakup.
Seismic velocity tomography gives geometry of hot and cold plumes in the mantle.
Seismic anisotropy tomography shows flow directions from preferred orientations of the long axes of olivine crystals in shear.
Little velocity correlation with surface tectonics below 200-400 km.
Harvard plume images (as above) show that two major plumes in the lower mantle diverge laterally in the upper mantle, forming the more complex systems underlying plate tectonics.
Difficult to resolve nature of convection at the 670 discontinuity.
Anisotropy results verify plate motion directions in upper mantle and lithosphere; essentially no anisotropy in the lower mantle.

Below are ridge-root and continent-root investigations from global surface-wave tomography by Yu-Shen Zhang of U.C. Santa Cruz