Influence of past and present-day plate motions on spherical models of mantle convection: implications for mantle plumes and hotspots

We explore the influence of tectonic plate motions, given by the no-net-rotation (NNR)-NUVEL-1 model, on a 3-D spherical model of mantle convection in which plates can be coupled to the underlying mantle flow in a dynamically consistent manner. We first derived a reference convection model in which only the NUVEL-1 geometry of the tectonic plates is prescribed. The plate rotations are then predicted on the basis of the buoyancy forces in the mantle, ensuring a dynamical balance of torques acting on the plates. This dynamically consistent reference convection model, which is based on a simple two-layer viscosity profile, yields the main features of plate tectonics: linear subduction zones, passive diverging zones and four mantle plumes. We next developed a time-dependent convection model, which is initiated with the average radial temperature field extracted from the reference convection model, and in which we imposed the NNR-NUVEL-1 plate velocities. This convection simulation yields six focused upwelling plumes, whose location and number is very similar to the primary terrestrial hotspots which have been recently identified (Courtillot et al. 2003). In all convection models incorporating the NNR-NUVEL-1 plate motions, we find that the surface heat flow and mantle potential temperature stay essentially constant, demonstrating the compatibility between the observed NNR-NUVEL-1 velocities and the internal buoyancy forces in the mantle. To determine the robustness of these results we carried out complementary convection simulations incorporating the past 120 Ma history of tectonic plate evolution. These simulations yielded shifting 'hotlines' at the core-mantle boundary, but the locations of the overlying hotspot plumes remained relatively stable. The configuration of the hotlines in the convection experiments with and without evolving surface plate geometries are very similar to each other, showing that convection models with present-day plate configurations are sufficient for capturing the essential characteristics of the present-day thermal structure in the mantle. In a final experiment, the prescribed NNR-NUVEL-1 plate velocity constraint is released, allowing the plates to rotate freely in response to the underlying mantle flow. We find that the initial evolution of this model is characterized by a strong stability of the mantle thermal structures, in particular the upwelling plumes. An important and novel feature of the convection model with free plate motions is the predicted opening of the African plate along the East African Rift boundary, which occurs in response to the large-scale mantle flow and does not appear to require the presence of upwelling plumes directly beneath the rift.