Influence of Geometry and Rheology on Convergence Speed in Self-Sustained Andean-Type Subduction Systems

In a subduction system where an oceanic lithosphere dips beneath a continental lithosphere, the convergence speed (CS) is predominantly governed by ridge push and slab pull forces. However, numerical models have shown significant sensitivity to the geometry under the same physical parameters. This study aims to shed light on how subduction dynamics is affected by changes in both geometry and rheology, and explore an approach for simulating subduction that makes convergence speeds more consistent and stable by incorporating an effective partial melt region. A series of 2D simulations was conducted to investigate how the kinematics of subduction zones evolve in a self-sustained manner, where no external forces were applied to drive subduction. To achieve this, we used the geodynamic numerical code LaMEM to solve the set of constitutive equations of momentum, mass, and energy suited for geological processes. We also used the mineral assemblage code MAGEMin to compute density changes in relevant lithospheric and asthenospheric rocks. Furthermore, a pyrolytic composition was employed to parameterize the phase change from the asthenospheric mantle to the lower mantle, adopting a Clapeyron slope. In this study, an oceanic plate subducts over a low-viscosity region (LVR) representing a partial melt region. The goal was to demonstrate how the convergence speed varies as a function of both the LVR and the asthenosphere viscosities. To minimize friction between the lithospheric plates, the oceanic plate slides beneath a weak zone. The role of the oceanic plate geometry was studied by varying its horizontal length at the surface. We observed that the CS is inversely correlated with the length of the oceanic plate at the surface. Our study indicates that the LVR makes the convergence speeds more stable over time, simplifies adjustments, and reduces the drag force influence on the overall kinematics of the descending plate. In summary, such an approach minimizes the role of the plate length on the overall evolution of the system in numerical studies and facilitates more stable convergence speeds.