Abstract

High-resolution uppermantle tomographic models are interpreted in terms of plate tectonics, hotspots and plume theories. Ridges correlate with very low velocity areas to a depth of 100 km, probably a result of passively induced upwelling and partial melting. Past positions of ridges also exhibit very low seismic velocities in the uppermantle. At depths greater than 100 km, some low velocity anomalies (LVA) may record past positions of migrating ridges. Buoyant upwellings induced by spreading do not track the migration of surface ridges; they lag behind.

At depths greater than about 150 km many LVA (Atlantic and Indian oceans) are more closely related to hotspots, and past positions of ridges than to current ridge locations. In the upper 200 km of the mantle, back-arc and continental extension areas are generally slower than hotspot mantle, possibly reflecting partially molten and/or hydrous mantle. The Pacific ocean ridges tend to be LVA, and probably hot, to about 400 km depth.

The surface locations of hotspots, ridges and continental basaltic magmatism seem to require a combination of hot uppermantle and suitable lithospheric conditions, presumably the existence of tensile stresses. The high-velocity regions of the upper 200 km of the mantle correlate with Archaean cratons. Below 300 km the regions of generally fast seismic velocity, and therefore cold mantle, correlate with regions probably underlain by ancient slabs, where the uppermantle may be cooled from below. A moving plate, overriding a hot region, and being put into tension, will behave as if it were being impacted from below by a giant plume head.

At sublithospheric depths there are very large LVA (VLVA) in the Pacific and Indian oceans and in the North and South Atlantics. The large continental and oceanic flood basalt provinces seem to have formed over these large, presumably hot, regions. These VLVA do not appear to be plume heads nor is there any obvious damage to the lithosphere under the present locations of flood and plateau basalt provinces. The uppermantle does not appear to be isothermal; the LVA are not restricted to hotspot locations.

We suggest that LVA are hotcells in the uppermantle which reflect, in part, the absence of subduction cooling. Plate tectonic induced rifting causes massive magmatism if the break occurs over hotcells, i.e. low-seismic velocity regions. Flood basalts (CFB) may result from the upwellings of already hot, even partially molten, mantle. In contrast to plume heads and plume tails, hotcells are robust features which are fixed relative to one another. They are most pronounced in parts of the mantle that have not been cooled by subduction. There is a close relationship between CFB initiation sites, LVA and ridges and, we believe, hotcells.