Almost since the proposal of the mantle plume model there has been a debate as to how plumes arise, where they originate, how they propagate, the stability, logevity and recycle time etc etc. Initially it was proposed that plumes were thermal anomalies arising at the core mantle boundary driven by radiogenic heating from the core and mantle, but in many examples this not thought to be the case currently (for example surface flow over Hawaii and the Ontong Java Plateau show no significant rise as may be expected from a large thermal anomaly)

Depth of Origin

There evidence for a variety of origins for mantle plumes according to seismic tomography, for example Hawaii and Iceland appear to be associated with very shallow anomalies whereas the east Africa/Afar plume appears to have a very deep origin at the core mantle boundary (see case study) The difficulty in establishing the depth of mantle plume origins is determining whether enough instability exists at the different known discontinuities in the mantle (e.g. the 410km, 660km and the D'' discontinuities) and how if once initiated those plumes rise and change their properties (such as viscosity, thermal diffusivity etc) as well as that of the surroundings(Condie et al. 2001). Various models, both analogue and numerical have shown high sensitivity of plume propagation to various factors such as the contrast in viscosity between plume and surroundings as well as the individual viscosities, the excess temperature and Rayleigh number calculations (see modelling page). These factors and competing explanations are still far from a definite answer and much material has been produced on the issue which is beyond the scope of this website. I suggest Condie's book for an excellent summary in fairly recent findings viewed from a geochemist's point of view.

Hot, Wet, Chemical or Decompressional?

It well known that convection is driven by density changes in a viscous fluid which can arise due to heating, change of phase or mineralogical changes in the mantle. Studies of island arcs all tend to agree that volcanism is caused by addition of volatiles to the continental plate, especially water from the subducting plate. It therefore makes sense that mantple plumes could be driven by addition of water at some depth, a similar argument may be used with concentrations of different minerals feeding plumes (for example in the reheated slabs model)

Thermally driven plume theory has met with a lot of difficulty lately as generally the thermal anomalies required to produce plumes consistent with current observations require a temperature rise higher than what is observed currently. Interestingly thermally driven convection models have successfully shown a periodicity of volcanism as anomalies rising in pulses(see the Hansen et al. web page). On the other hand Large Igneous provinces show little evidence of thermal uplift and subsequent subsidence similar to what we see at spreading riges, suggesting some other process of formation (see the LIPs page)