Lithosphere-mantle interactions

We study the interaction between lithospheric dynamics and mantle convection to establish a comprehensive model of large-scale deformation that can be tested using a variety of surface observables. In chapter 1 we explore subduction and upper-mantle convection using fluid-dynamic models. Comparing results from analog and numerical experiments, we show that key observations in subduction zones can be explained using fluid flow. We analyze contrasting styles of back-arc deformation based on the balance between viscous dissipation and gravitational energy release in the subducting plate. By combining a wide range of geological and geophysical data in chapter 2 we are able to reconstruct the subduction history of the Central Mediterranean. Subduction initiated slowly and lead to episodic back-arc spreading, as predicted from our models for predominantly slab-pull driven subduction. Based on these regional studies, we estimate global deformation in the lithosphere from spherical-circulation calculations. Chapter 3 describes a comprehensive and quantitative comparison between mantle models from tomography and geodynamics. Establishing such a benchmark for three-dimensional structure is important if we want to move from the mapping phase of tomography to hypothesis testing. To examine the dynamic self-consistency of our flow models, we study the torques on the plates in chapter 4 and reevaluate arguments of the plate-driving force discussion. We substantiate the finding that slab pull contributes most to the driving forces but find that a range of models can explain plate motions. Velocity inversions are shown to have limited sensitivity to parameterized plate-boundary forces, here included in a velocity model for the first time. Chapter 5 deals with a first application of our deformation model: the prediction of finite-strain development in the uppermost mantle and its comparison with seismic anisotropy. We can fit a range of anisotropy data for oceanic and young continental plate-regions and compare our global models with surface-wave derived azimuthal anisotropy. Alignment of fast propagation axes with finite strain leads to low misfits and better results than the absolute plate motion hypothesis. Three-dimensional flow modeling with realistic plate geometries may provide the needed tool to interpret seismic anisotropy quantitatively as an indicator of mantle flow.