The development of slabs in the upper mantle: Insights from numerical and laboratory experiments

We have performed numerical and laboratory experiments to model subduction of oceanic lithosphere in the upper mantle from its beginnings as a gravitational instability to the fully developed slab. A two‐dimensional finite element code is applied to model Newtonian creep in the numerical experiments. Scaled analog media are used in the laboratory, a sand mixture models the brittle crust, silicone putty simulates creep in the lower crust and mantle lithosphere, and glucose syrup is the asthenosphere analog. Both model approaches show similar results and reproduce first‐order observations of the subduction process in nature based on density and viscosity heterogeneities in a Stokes flow model. Subduction nucleates slowly and a pronounced slab forms only when the viscosity contrast between oceanic plate and mantle is below a threshold. We find that the subduction velocity and angle are time‐dependent and increase roughly exponentially over tens of millions of years before the slab reaches the 670‐km discontinuity. The style of subduction is controlled by the prescribed velocity of convergence, the density contrast between the plates, and the viscosity contrast between the oceanic plate and the mantle. These factors can be combined in the buoyancy number F which expresses the ratio between driving slab pull and resisting viscous dissipation in the oceanic plate. Variations in F control the stress in the plates, the speed and the dip of subduction, and the rate of trench retreat, reproducing the contrasting styles of subduction observed in nature. The subduction rate is strongly influenced by the work of bending the lithosphere as it subducts.