Cambridge Publications |
|
The Conference Company |
April 8th - 12th, 2001 |
|
European Union of Geosciences |
Cambridge Publications |
|
The Conference Company |
April 8th - 12th, 2001 |
|
European Union of Geosciences |
(Wednesday April 11th 2001 at 12:15 in Room G0: Schweitzer Auditorium)
Ecole Normale Supérieure de Lyon
The paradigm of layered mantle convection was established nearly 20 years ago, mostly based on geochemical mass balance and heat budget arguments. It is now stumbling over the difficulty imposed by convection models to maintain a sharp interface in the mantle at mid-depth and by overwhelming tomographic evidence that at least some of the subducting lithospheric plates are currently reaching the core-mantle boundary. Discontinuities in the deep mantle, the D" apart, remain elusive. A further problem is the increasingly persuasive identification through geochemical means of various ingredients from ancient oceanic lithospheric plates, such as altered oceanic basalts, gabbros and pelagic sediments in the source of hot spot basalts. It is further being argued that ancient oceanic plateaus may also be a normal component of some ocean island basalts, notably those from the EM I clan.
The present situation, however, remains frustrating because the reasons why the layered convection model was defended in the first place are still there and do not find a proper answer with the model of homogeneous mantle convection. First, the imbalance between heat flow and heat production (Urey ratio) requires that the deep mantle is rich in U, Th, and K. Second, the imbalance of some refractory lithophile elements between the composition of the Earth estimated with a homogeneous mantle and the composition of chondrites leaves a number of 'paradoxes' unresolved. Third, convective mixing should take place with a characteristic time of less than 1 Gy and should essentially wipe out mantle isotopic heterogeneities.
It can be shown that, with the possible exception of lead, helium, and argon, the isochron relationships observed for many radioactive systems indicate time constants characteristic of the chemical differentiation of the mantle-crust system and have no chronometric implications. No more than Joly was finding the age of the oceans from the salt contained in seawater and rivers can we infer the age of the Earth from mantle-derived basalts. For all these systems, the mantle is therefore at steady state with 'no vestige of a beginning'. In addition, frustrating evidence that the lower mantle hides a geochemical 'black box', with a non-primitive composition and hardly accessible to observation, is mounting.
The question then arises of the transient between the primary differentiation of the planet 4.5 Gy ago and the modern state of mantle convection. Standard models of mantle convection do not properly address these issues. Evidence for the persistence of any leftover from the primary recipe rests entirely on the blurred signal of rare-gas isotopes and the remains of Early Archean mantle-derived volcanics that suffer preservation problems.
I will argue that, immediately after the early differentiation of the Earth 4.54 Gy ago which, because of a gravity stronger that that of the Moon and Mars, was probably not dominated by a huge magma ocean, mantle convection took off. Plate penetration and convection are strongly influenced by the abundance of plume heads carried by the lithospheric plates. Barren plates penetrate deeper than plates loaded with oceanic plateaus. Mantle heterogeneities are destroyed by convective stirring but permanently renewed by selective plate penetration. Hot spot instabilities have declined over the Earth's history and therefore the layering of the mantle has evolved substantially. Whole-mantle convection may be reconciled with geochemical evidence of modern and ancient basalts through the depth-dependent residence time permitted by selective subduction.
