Implications of mantle plume structure for the evolution of flood basalts

Abstract

Morgan [1,2] suggested that continental flood basalts appear as the first volcanic expression of new mantle plumes. Experimental studies in viscous liquids have shown that new, or "starting", plumes should consist of a large bulbous head followed by a narrow feeder conduit. Analysis of the plume flow [3] indicates that, if the plume ascent is driven by thermal buoyancy, the head will entrain the surrounding mantle as it rises. The head cools and enlarges, and develops a compositional zonation of source and entrained mantle. Uncontaminated, relatively hot material from the plume source continues to flow up the trailing conduit making the temperature of the plume axis greater than that of the remainder of the head. We explore the implications of this plume structure by comparing the physical and chemical characteristics of two flood basalt provinces (the Deccan and Karoo) with predictions of the dynamical model. The chronology, tectonics and geochemistry of the two provinces all fit well with the starting plume hypothesis. We attribute the sudden onset and short duration of continental flood volcanism, over an equant area 2000-2500 km across, to melting the plume head and its subsequent decline to a narrow chain of volcanic activity, ∼ 200 km wide, to melting in the plume tail. A surface uplift of 500-1000 m is predicted but this gives way to subsidence due to lateral spreading of the plume head before the onset of the main period of volcanism. A period of enhanced subsidence is then predicted to occur as magma escapes from the mantle and loads the earth's surface, followed by slow subsidence over 109 years as the plume's thermal anomaly gradually decays. The timing and duration of volcanism has not been predicted with certainty, but activity is expected to begin as a burst and to die away rapidly over a total time of order 20 Ma, in agreement with dating which indicates that the bulk of the magmas in each province were ejected within 2-3 Ma and followed by smaller volumes over a further 5-10 Ma. The model predicts that the high-temperature picritic melts associated with continental flood basalts are derived from hot, relatively uncontaminated plume-source mantle at the plume axis and that the more voluminous tholeiitic basalts are produced by melting of cooler hybrid mantle in the plume head. This explains for why the picrites of the Karoo and Deccan are strongly enriched in highly incompatible elements, consistent with melting of an OIB-type source, whereas the associated basalts are weakly enriched in incompatible elements, consistent with derivation from a mixed OIB-lower-mantle source. Show more