slowly. For the first time, observers can determine how far the slabs penetrate into the mantle, providing first-order information on the deeper aspects of the convection patterns associated with plate tectonics.
In addition, the shape of the slab is a direct reflection of the tectonic forces associated with convection. Observations of significant distortion of slabs in certain regions give evidence for variations in rock properties with depth and for background flows in the mantle convection pattern. This conclusion is supported by evidence of changes in the distribution and focal mechanisms of earthquakes with depth. All indications are that tectonic forces do vary along subduction zones, but the patterns are still indistinct.
Although many aspects of mantle convection are reasonably well understood, major scientific questions remain unresolved. These include the vertical structure and multiple scales of mantle convection and the efficiency of mantle mixing.
Plate tectonics is the surface manifestation of mantle convection. The rigid plates are the uppermost thermal boundary layers of mantle convection. These cool layers are rigid on geological time scales and behave as plates. But the plates become denser because of thermal contraction. Eventually, they become gravitationally unstable and founder into the mantle at subduction zones, defined by the ocean trenches. The weight of descending plates is a major force driving plate tectonics.
There is direct seismic evidence that the slabs of material subducted into the mantle at ocean trenches descend to depths of 670 km. An unresolved question is whether the downward limbs can penetrate this depth. The evidence is contradictory, and the views of experts are divided. If thermal convection penetrates this density barrier, whole-mantle convection occurs. If convection does not penetrate, then separate convection cells develop in the upper and lower mantle and mantle convection is layered. If mantle layers do not mix, significant variations in chemical composition and temperature could characterize the interior. There is also the possibility that both styles of convection can coexist. The amount of material transported across the entire mantle is currently the single largest uncertainty in understanding the Earth's thermal and chemical evolution.
Just as the plates are thermal boundary layers at the top of the convecting mantle, there should be boundary layers at the base of the convecting system. For layered mantle convection this boundary layer would be the result of heat transfer from the lower mantle; for whole mantle convection, it would be the result of heat transfer from the core. The gravitational instabilities in these lower boundary layers may generate ascending mantle plumes that are responsible for intraplate volcanism, such as that in Hawaii.
The fate of plates that founder into the mantle at ocean trenches has also been a subject of controversy. These plates are layered. The basaltic ocean crust extends to a mean thickness of about 6 km. Beneath that crust is a zone that has been depleted of basalt and is primarily composed of the refractory mineral olivine. This layer, which has a thickness of approximately 50 to 100 km, is gravitationally buoyant. Simple mass balance calculations show that ocean crust must be recycled through the mantle on a time scale of about 1.7-billion-years or less. Therefore, present-day basaltic ocean crust has been processed through the plate tectonic cycle several times.
One hypothesis for the fate of the subducted ocean crust is that convection stirs it into the bulk of the mantle until it is nearly homogenous. Another hypothesis suggests that significant density differences between the basaltic ocean crust, which transforms to a dense phase called eclogite at depth, and the olivine-rich mantle result in gravitational segregation, with the depleted mantle rock overlying the crustal rock. The essential question that must be answered is whether convective mixing can homogenize the mantle before the buoyancy differences can cause layering.
The effect of an increase in temperature is to decrease the density of rock by a fractional amount—a few percent per 1000°C. Density variations caused by lateral temperature variations drive mantle convection. By the same token, the convective flows induce temperature differences. The flow field and temperature field are coupled.
Assuming that lateral variations of seismic velocities in the deep mantle can be ascribed to temperature variations, laboratory measurements of the changes in acoustic velocity with temperature can be used to infer the density variations in the mantle. In this way the buoyancy forces associated with mantle convection are obtained directly from seismic tomography.
Variations of the external gravity field can be calculated from the inferred density variations, taking into account not only the density distribution within the convecting mantle but dynamic topog-