In this section we look at what drives the plates. The fundamental energy source within the Earth is heat, primarily provided by radioactive decay in the mantle. Converting this to kinetic energy is through convection. There are several ways on investigating how this works.
One approach is to focus on the plates themselves. The key two options are "ridge push" (forcing plates apart at mid-ocean ridges) and "slab pull" (dragging plates down into the asthenosphere). We can investigate these and establish that empirically the fastest moving plates are those that have more trenches around them. The amount of constructive plate boundary seems to have little relationship to velocity.
We can continue to explore processes at ridges. If there was dynamic pushing from ridges by upwelling mantle then the sea floor should be pushed up. Check out the bathymetry of some "typical" ridge in the mid Atlantic.
If the more important plate-derived drive is from slab pull then we need to examine how this works on the scale of the mantle. There are two main models for mantle convection: double-layered and single layered. The models place different emphasis on the role of the "mantle transition zone" - a discontinuity at about 670km down in the Earth. There are different types of evidence that can be used. First off - the chemistry of basalts erupted at different settings is a clue to how mixed the mantle is. By clicking here you can see that basalts erupted at the mid ocean ridges (MORB) has a distinct source to those arising at hot spot volcanoes (ocean islands - OIB). As MORB comes from shallow decompression melting of the mantle (click to recap mantle melting) - its source lies in the upper mantle. And MORB has surprisingly consistent chemistry around the world implying that the upper mantle (at least below the oceans) is rather homogeneous. OIB is distinct and presumably comes from deeper levels. So there are at least two distinct chemical reservoirs in the mantle - implying that the mantle is not mixed up as might be the case if the whole mantle was convecting. So at first site the chemistry argues for two (or more) layered convection.
The other arguments are seismological. Earthquakes in subduction zones continue down into the mantle but stop at the 670km transition zone. Furthermore, the earthquakes in the slabs just above the transition zone are commonly compressional implying that the slabs are meeting with resistance to further subduction. Again this points to double layer convection. But modern tomographic images give a different picture which strongly argues for whole mantle convection.
The other key phenomenon operating in the mantle apart from plate-type convection are mantle plumes. These can be modelled as jets of hot upwells that originate at the core-mantle boundary and which rise up to the lithosphere (click for animation of plume. The key is that these must originate below plates because plates move with respect to them. We record this as hot-spot traces. You can explore some of this plate and plume interaction here.
Putting this together, we can explore the ideas put forward by Geoff Davies. He considers the Earth as a single layer convection system with the plates forming the upper thermal boundary layer and the core-mantle boundary forming the lower one. The whole behaviour can be schematically represented in a famous cartoon generated by Davies. In the modern whole the plate mode of convection is responsible of 90% of the heat loss from the mantle - while the plume mode covers the remaining 10%. However, these proportions could have changed through Earth history.
Models of convection