Convective motion in the earth's mantle is considered
to be induced by cooling at its upper surface,
radioactive heating within, and heat flow coming from the core.
However, the characteristics of mantle flow are likely
to differ from those of simple Benard-Rayleigh-type thermal convection
due to material and geometric heterogeneities,
particularly those relating to viscosity variations,
the phase boundary near 670 km depth and the distribution of internal heating.
One geological factor likely to exert particular influence on the mantle convection patterns is the existence of oceanic and continental plates. Such plates impose horizontally heterogeneous conditions at the top boundary of the mantle and result in different patterns of convection than occur for a homogeneous boundary condition. Continental plates consist of lower-density materials than oceanic plates, and in contrast rest stably on the top of the mantle without sinking. It is also the case that continental plates are thicker than oceanic plates. Analysis of seismic data reveals that a positive seismic velocity anomaly region commonly extends to 400 km depth beneath continents (Jordan 1975). Thus, the continental plates also behave as thermal conduction layers and affect the convective motion beneath.
The Wilson cycle of periodic continental collision and breakup is believed to be one tectonic phenomenon caused by the existence of the continental plates themselves (Wilson, 1966). Since the continental plate behaves as a thermal conduction layer, it impedes the transport of heat to the surface, and is therefore presumed to result in the fluid beneath it becoming warmer. This suggests that upwellings are easily produced beneath continental plates. These features are considered to cause continental breakup and subsequent re-collision (Anderson, 1982). Observations of seismic data and gravity anomalies reveal a negative density anomaly beneath the Atlantic Ocean and Africa, which is inferred to correspond to the broken remnants of the latest super continent, "Pangea" (Harger et al., 1985).
Many studies have been performed to investigate the effects of continental plates on mantle convection. Laboratory experiments of thermal convection by Guillou and Jaupart (1995) employed a thermal conduction layer partly covering the top of a fluid layer to represent the effects of a continental plate. Those authors showed that an upwelling emerges beneath the thermal layer and that the horizontal dimension of the convection cells is equal to the horizontal extent of the thermal conduction layer above. Laboratory experiments by Yanagisawa (1995) utilized a floating sponge plate to investigate the temporal development of upwellings beneath a continental plate and their response to changes in the position of the plate. Numerical calculations (e.g. Gurnis 1988, Lowman and Jarvis 1993) have also succeeded in treating the collisions and breakups of plates; the upwellings induced under the plates break up and distribute the plates, and then reassemble them again.