Geology Reference
In-Depth Information
depends on a chain of no less than eight elemental ratios. U and K require a further
ratio each. Workman and Hart [186] begin with a reconstruction of whole-rock
composition, which requires partition coefficients and modal abundances, then a
melting model to estimate pre-melt composition. Both studies focus on establishing
the composition of what they call the depleted mantle, to the exclusion of more
enriched components.
Thus it may be that the 'depleted MORB mantle' is more enriched than either of
these estimates. This would imply that both enriched heterogeneities and the mean
composition of the MORB source would then be proportionately more enriched.
10.6.7 Basaltic gradient through the upper mantle?
A final reason why previous estimates of the trace element content of the mantle may
be too low comes not from geochemistry but from the numerical models of mantle
stirring discussed in Chapter 9. Figure 9.8 reveals a gradient in the concentration
of basaltic tracers through the upper mantle. The tracers, representing subducted
oceanic crust, which is denser than average in the upper mantle, tend to settle
through the upper mantle, leaving the top of the upper mantle depleted by about
a factor of 2 relative to the lower mantle. The effect is strongest in the early, hot
mantle, but it persists into the present in a number of models like those in Figures 9.5
and 9.10. This mechanical depletion reduces the amount of melting and results in
the oceanic crust being much thinner than if the upper mantle had the average
basaltic content.
The persistence of this effect into the present was a little surprising, because
an earlier model run at present conditions (in other words, with steady heating
rather than heating that declines with radioactive decay) showed no gradient in
the upper mantle [123]. Therefore the behaviour of several evolving models was
checked, with results shown in Figure 10.18. The evolving models all show a clear
gradient through the upper mantle, whereas the steady model shows very little
gradient. The steady model also did not develop an accumulation at the base of the
mantle.
The final degree of depletion is not easy to discern from Figure 10.18. It is better
represented by the comparisons of actual crustal thickness and 'fertile' crustal
thickness shown earlier (Figures 9.8 and 9.11). These show that the final crustal
thickness is 70-90% of the fertile thickness, suggesting a depletion of 10-30%.
The difference in basal accumulation among the models is easier to understand.
Evidently the lower mantle is viscous enough under present conditions to prevent
significant settling to the base of the mantle. On the other hand, settling can occur in
the early, hot mantle when the lower mantle is less viscous. Once an accumulation
