The vertical extent of convection

The mantle transition zone (Section 2.8.5) may well influence the nature or even the vertical extent of convection in the mantle. If this zone represents a change in chemical composition, then it implies that convection currents do not cross it. In this case separate layers of convective circulation would occur above and below the transition zone, with heat transported by conduction across a thermal boundary layer within all or part of the transition zone.

The nature of the mantle transition zone is equivocal, but the majority view appears to be that it represents a region in which solid state phase changes take place, whereby the mineralogy of mantle material changes to higher pressure forms with depth, rather than representing a change in chemical composition (Section 2.8.5). For example, Watt & Shankland (1975) have shown, from an inversion of velocity-density data, that the mean atomic weight of the mantle shows no change across the transition zone. If this is the case, convection currents could cross the transition zone, as long as the phase changes take place very rapidly, and convection cells would then be mantle wide. The phase changes would have two important effects on convection, as they are temperature and pressure dependent and involve latent heat. In the case of olivine to spinel the change from low pressure to high pressure forms takes place at shallower than average depths in the cold descending currents and at greater than average depths in the hot ascending currents. Consequently, low-density minerals are created deeper on ascent and denser, high-pressure forms at shallower depths on descent. Their positive and negative buoyancies respectively then help to drive the convection cells. The phase change is also associated with a release or absorption of latent heat, the high- to low-pressure reaction being exothermic and the low to high-pressure reaction being endothermic. This causes steepening of the thermal gradient across the transition zone, so that the temperature in the lower mantle is 100-150°C higher than if the zone did not exist.

Tackley et al. (1993) have numerically modeled mantle convection in three dimensions with an endo-thermic phase change at the base of the transition zone. They suggest that cold downwelling material accumulates above 660 km and then periodically flushes into the lower mantle. This fits well with the results of seismic tomographic imaging of subduction zones, which suggests that some slabs flatten out within the transition zone and others penetrate the base of the zone and descend into the lower mantle (Section 9.4; (Plate 9.2 between pp. 244 and 245).

Thus, the transition zone may not be a barrier to mantle-wide convection, and a number of workers have presented evidence in accord with this premise. Kanasewich (1976) noted an organized distribution of plates, in which the Pacific and African plates are approximately circular with the smaller plates having an approximately elliptical form and arranged systematically between these two large plates. Kanasewich attributed this organization to convection that is mantle-wide. Davies (1977) conducted model experiments and concluded that only extreme viscosity contrasts would restrict convection to the upper mantle, and maintained that such contrasts do not exist. Elsasser et al. (1979) employed a scaling analysis in which the depth of convection is derived as a function of known parameters, and concluded that this depth is consistent with convection throughout the entire mantle. The topography on the base of the mantle transition zone has an amplitude of about 30 km (Shearer & Masters, 1992), which is an order of magnitude lower than predicted for a chemical, rather than a phase, change at this depth. Morgan & Shearer (1993) derived the buoyancy distribution in the mantle from seismic tomographic maps and concluded that there must be significant flow between the lower and upper mantle. However, other work, summarized by van Keken et al. (2002), suggests that the geochemical and isotopic pattern of trace elements found in oceanic volcanic rocks supports a model in which portions of the mantle have been chemically isolated for much of Earth history. This would suggest that the mixing implied by whole mantle convection has not occurred, and that layered convection is more likely. However, in the light of the geophysical evidence for mantle-wide convection many geochemists have derived models in which distinct chemical reservoirs can be preserved within this context (e.g. Tackley, 2000; Davies et al., 2002). It would seem, therefore, that convective circulation is most likely to be mantle-wide and not constrained by the transition zone.

0 0

Post a comment