Tomographic images of earth's interior are often used to show subduction. However, "low velocity volumes of the mantle detected by tomography can be due to lateral variations in composition rather than in temperature, i.e., they can be even higher density areas rather than hotter, lighter buoyant material as so far interpreted." "As extreme examples, gold or lead have high density but low seismic velocity. Therefore the interpretation of tomographic images of the mantle where the red (lower velocity) areas are assumed as lighter and hotter rocks can simply be wrong, i.e., they may even be cooler and denser. With the same reasoning, blue (higher velocity) areas, which are assumed as denser and cooler rocks may even be warmer and lighter." Sometimes pieces are assumed to have detached from slabs and fallen away, and appear as blobs on tomographic images. "Tomographic images are based on velocity models that often overestimate the velocity of the asthenosphere where usually the detachment is modeled. Therefore the detachment disappears when using slower velocity for the asthenosphere in the reference velocity model, or when generating regional tomographic images with better accuracy."
"If oceanic lithosphere is heavier than the underlying mantle, why are there no blobs of lithospheric mantle falling in the upper mantle below the western, older side of the Pacific plate?" Slab dip has been thought to be related to its age, as in "the western Pacific subduction zones because the subducting western Pacific oceanic lithosphere is older, cooler and therefore denser. However, the real dip of the slabs worldwide down to depths of 250 km shows no relation with the age of the downgoing lithosphere."
We are supposed to believe that "the 700 km-long West Pacific slab... should pull and carry the 10,000 km-wide Pacific plate, 33 times bigger, overcoming the shear resistance at the plate base and the opposing basal drag induced by the relative eastward mantle flow."
"The negative buoyancy [(dense enough to sink)] of slabs should determine the pull of plates, but it has been shown that the dip of the subduction zones is not correlated with the age and the thermal state of the downgoing plates." "In fact there are slabs where, moving along strike, the age of the downgoing lithosphere varies, but the dip remains the same, or vice versa, the age remains constant while the dip varies (Philippines). There are cases where the age decreases and the dip increases (Western Indonesia), and other subduction zones where the age increases and the dip decreases (Sandwich). This shows that there is not a first order relationship between slab dip and lithospheric age."
Finally, no matter how it is computed, "results do not support a correlation between slab length percentage (length of the trench compared to the length of the whole boundary around the plate) and plate velocity."
"This long list casts doubts on the possibility that slab pull can actually trigger subduction... and drive plate motions."
Why should the lithosphere start to subduct at all? "Hydrated and serpentinized oceanic lithosphere that has not yet been metamorphosed by the subduction process... is still less dense." Slab pull has been calculated to become potentially efficient only at a certain depth, around 180 km. Shallower than that, how is subduction initiated?
This raises a basic problem for plate tectonics theory
"Why convection in Earth's mantle gives rise to plate tectonics is not obvious. The top thermal boundary layer is supposed to be very stiff because the viscosity of silicate rocks is strongly temperature-dependent. This temperature dependency is so strong that so-called stagnant lid convection should be the most likely mode of mantle convection; the entire surface should be covered by just one single plate, not by a number of rigid plates." "What is needed for the self-consistent generation of plate tectonics is... a mechanism to initiate subduction." Consider moderate-age oceanic lithosphere, say 100 million years old. Even at a "depth range of 10-45 km, oceanic lithosphere of this age is too stiff to be deformed by any reasonable tectonic stress, resulting in a fatal bottleneck for the operation of plate tectonics." "The currently available estimate suggests that yield strength for this 'semi-brittle' regime is still on the order of 600-800 MPa [6000-8000 bars]." "It is known that the self-consistent numerical modeling of plate tectonics (i.e., plate tectonics naturally arising from buoyancy distribution and given rheology, not imposed by boundary conditions) is impossible with such high yield stress. In all previous attempts to simulate plate tectonics in a self-consistent fashion, therefore mantle rheology is modified in one way or another" to get the desired result.26 (Emphasis added)