Understanding Lithosphere and Mantle Dynamics with Numerical Models Constrained by Observations

Abstract

Numerical studies play an important role in understanding lithospheric and mantle dynamics. In this thesis, we first develop and use multiphysics geodynamic models to study the evolution of subduction. Our geodynamic models are constrained by different geological and geophysical observations, including topography. We then use 3D numerical simulations of dynamic rupture with off-fault inelastic deformation to study the scaling between damage zone thickness and fault width. Finally, we study the mechanical strength and anisotropy in the continental collision region with flexural models and gravity and topography data. Topography is valuable data for investigating lithosphere and mantle dynamics and constraining numerical studies. Topography prediction with forward models is well established at plate interiors, while it is still difficult to predict realistic topography at subduction zones. We use multiphysics geodynamic models to tackle this problem. Our models incorporate a true free surface, phase changes, and elasto-visco-plastic rheology. We also include surface processes, water migration and water weakening. We study the influences of different geophysical, petrological, and geochemical processes on topography and subduction zone evolution and show that surface geometry, surface processes, elasticity, and oceanic crust all strongly influence the stress state and deformation within plates, water weakening decouples the overriding plate and the subducting slab at the mantle wedge region and contributes to the initiation of overriding plate failure, and oceanic crust has a similar effect with sediments lubricating the subduction interface. Free slip surface topography and free surface topography have substantial differences, and free surface topography is influenced by different processes by adjusting the force balance. Application to the New Hebrides subduction zone suggests that deformation within a detached slab segment caused by the impact of the slab segment on the strong lower mantle explains the origin of the isolated deep earthquakes in the transition zone beneath the North Fiji Basin, and the difference in the seismic intensities between northern and southern deep earthquake clusters is caused by transition from strong deformation to weak deformation after the impact. We apply our multiphysics approach to investigate the influence of inherited lithospheric heterogeneity on subduction initiation at the Puysegur Incipient Subduction Zone (PISZ) south of New Zealand. Our predictions fit the morphology of the Puysegur Trench and Ridge and the deformation history on the overriding plate. We show how a new thrust fault forms and evolves into a smooth subduction interface, and how a preexisting weak zone can become a vertical fault inboard of the thrust fault during subduction initiation, consistent with two-fault system at PISZ. The model suggests that the PISZ may not yet be self-sustaining. We propose that the Snares Zone (or Snares Trough) is caused by plate coupling differences between shallower and deeper parts, that the tectonic sliver between two faults experiences strong rotation, and that low density material accumulates beneath the Snares Zone. We then turn to the scaling between damage zone thickness and fault width. Field observations indicate that damage zone thickness scales with accumulated fault displacement at short displacements but saturates at a few hundred meters for displacements larger than a few kilometers. To explain this transition of scaling behavior, we conduct 3D numerical simulations of dynamic rupture with off-fault inelastic deformation on long strike-slip faults. We find that the distribution of coseismic inelastic strain is controlled by the transition from crack-like to pulse-like rupture propagation associated with saturation of the seismogenic depth. The yielding zone reaches its maximum thickness when the rupture becomes a stable pulse-like rupture. Considering fracture mechanics theory, we show that seismogenic depth controls the upper bound of damage zone thickness on mature faults by limiting the efficiency of stress concentration near earthquake rupture fronts. We obtain a quantitative relation between limiting damage zone thickness, background stress, dynamic fault strength, off-fault yield strength, and seismogenic depth, which agrees with first-order field observations. Our results help link dynamic rupture processes with field observations and contribute to a fundamental understanding of damage zone properties. Finally, we investigate the interactions between mechanical strength and lithospheric deformations. Variation of lithospheric strength controls the distribution of stress and strain within plates and at plate boundaries. Simultaneously, deformation caused by localized stress and strain reduces the lithospheric strength. We calculate the effective elastic thickness, Te, which is a proxy of lithospheric strength, and its anisotropy at the Zagros-Himalaya belt and surrounding regions. Te varies from