Direct observation of mantle plumes has so far proven elusive and highly debateable at best. Even the msot advanced imaging techniques for the mantle suffer from abidguity, lack of vertical resolution and difficult interpretation to mention some factors
Seismic tomography is currently widely used in a variety of publications, though the validity of some of the iamges produced and their interpretation is still under debate.
Seismic tomography produces 3 dimensional seismic velocity datasets based on P and S wave travel times. Following interpretation 2 dimensional 'slices' can be taken and velocity anomalies can be displayed. Usually negative veolcity anomalies are thought to suggest high temperature material reducing the seismic velocity and positive anomalies are cooler regions.
Seismic tomography example - (Ritsema 1999) imaging under Africa and Iceland
This is perhaps one of the more successful applications of seismic tomography which led to sensible and understandeable results. Shear (s) wave velocities were studied under Africa (relating to the Cenozoic flood basalts in the Afar region and the currect east African rift) and Iceland (thought to be the coincidence of a mantle plume and a mid ocean ridge)
The studies found that the two regions have vastly different low-velocity anomalies - the anomaly beneath Africa appears to be more prominent and originates at the core-mantle boundary beneath the south-eastern Atlantic ocean then rising at an angle to reach Africa almost 45° of longitude away. The Iceland anomaly on the other hand appears to be confined to the upper mantle and is more intense.
Seismic tomography images with section orientations from Iceland (Ritsema, 1999)
Limitations of global seismic tomography - Some observations by Don Anderson (Anderson, 2007)
Don Anderson presents a number of issues associated with the limitations of seismic tomopgraphy and comments on the inclination of many to rely too much on images produced by seismic tomography. Some of the following are the main concerns with interpreting and compiling global seismic tomography images :
* Limited ray cover - some areas of the mantle recieved limited cover of seismic rays due to both the distribution of earthquakes and seismic recording stations. D. Anderson argues that this lower density of data cannot be compensated by inverse modelling and therefore results are much less reliable in some areas despite being shown in the same plots.
* Resolution limitations as a result of ray geometry - this is an intrinsic problem to seismic tomography which means seismic tomography images suffer a much lower vertical resolution due to the near-vertical incidence of earthquake-derived waves and plumes are expected to be constrained vertical columns of low density material so imaging these is particularly difficult and results often mislead by not explicitly stating that vertical resolution is much lower than horizontal.
* Images shown are subject to user-bias base don the colour schemes used - It is well known that humans respond more to changes from certain colours than others and the colour scales and representations in seismic tomography images can be altered to highlight certain features which may otherwise be obscured but other peculiarities may remain hidden
* Strong dependance on the line of section - certain areas of the mantle show interesting features only when a certain vertical section is used, due to the difficulty in representing a 3D dataset in 2D slices seismic tomography is always a limited and often biased representation of reality.
Modelling of convection currents in the mantle has proven very difficult and demanding, both in understanding the effects of rheology, plates, planet geometry and many other factors and in terms of computing power. Numerical models were not possible until the 1980's with the rapid advance in computing power, but even today models have to be efficient and relatively simple.
Following seismic and rheological studies it is generally agreed that the mantle is very likely to convect, in a non-chaotic manner suggested by a very high Rayleigh number (a number governing the relative strength of convection and diffusion transfers) and a virtually infinite Prandlt number (a similar constant comparing momentum diffusivity and thermal diffusivity)
There is an excellent collection of model video captures made by U. Hansen & J. Schmalzl at Westfälische Wilhelms Unversity of Münster, but they range in size up to over 20MB, therefore I will only show some screen captures of the models. Users should visit the web page to have a look at some of the other models
Convection at high Rayleigh and infinite Prandtl number Red is hotter uwellings
Strongly temperature dependent viscosity – pulsed Plumes