Physical controls on dynamics of spontaneously retreating subduction zones

Abstract

Retreating subduction zones are geological features present all over the world, such as Scotia region, between South America and Antarctica or, according to some studies, in the Caribbean. Retreating subduction zones consist in a plate having a downward motion towards the Earth’s mantle, which retreats due to slab pull. They are expressed by curved trenches at the surface. Subduction zones are at the interface between our planet’s surface and its mantle and deep structures. What happens there is influenced by deep mechanisms and the properties of the mantle, but also by what is happening at the surface, such as loading and unloading of the trench, the subducting and overriding plate, or the physical properties of the subducted material. In this work, I am interested in understanding how these systems move around, what influences their course both in depth and at the surface. To address this research question, I need to take into account many characteristics, from the composition and mechanical evolution of the Earth’s layers, to the thermal structure of our planet. I am interested in the retreat dynamics as well as the topography evolution. This is a 3D problem in essence. I performed 3D thermo-mechanical simulations of subduction initiation followed by self-sustained slab retreat. I first focus on the thermo-mechanical aspect of the problem. As the slab is allowed to travel thanks to fracturing occurring at its edges (at the so called STEP-faults [Subduction- Transform Edge Propagator]), I pay special attention to fracturing processes in these regions. Using the I3ELVIS thermo-mechanical code, I vary the physical properties of the Earth’s layers and resulting deformation mechanisms. I show that localized deformation at the edges of the slab leads to converging STEP-faults and a narrowing of the slab. This is eventually followed by its detachment when the plate is narrow enough so that the slab pull fully detaches it. A contrario, I show that less localized deformation leads to diverging fracturing paths and a widening of the slab which eventually stops when the resistance to slab pull is high enough for the fracture to stop propagating. When a slab retreats in between continents, I observe that it does not follow the passive margins, for both localized and less localized deformation mechanisms. In the second part of this work, I focus more precisely on the interactions between deep and surface processes in retreating subduction zones. I use a new mass conserving code for material diffusion and advection at the surface of the model, FDSPM (Finite Differences Surface Processes Model), coupled to the thermo-mechanical code I3ELVIS. I perform experiments aiming at understanding how the thickness of the sediment cover and surface processes affect the retreat dynamics for slabs retreating in an oceanic domain or along a continental margin. I find that the thicker the sediment cover, the slower the slab retreats as the sediments accumulate in large accretionary prisms which act as a lock on the system. I also note that the sediment cover affects the retreat trajectory when the slab is retreating along a passive margin. There, the thicker the sediment cover, the more the slab deviates from the continent. The effect of surface processes on the slab dynamics also strongly depends on the amount of sediments available in the system. If there are very little sediments available, I find that the more efficient the surface processes, the slower the retreat rate. However, when a lot of sediments are available, such as along a passive margin, the more efficient the surface processes, the faster the retreat. This is likely due to erosion of the accretionary wedge which decreases the locking of the subduction interface. In the third part of this thesis, I focus on other elements which could affect slab retreat. I define a synthetic case inspired by the Caribbean to assess if the slab dynamics I observe in my models are consistent with observations in this region. There, slab retreat would be towards a mid-oceanic ridge, which implies that the lithosphere is getting younger in the direction of slab retreat. As this is likely to influence the slab’s behavior, I model retreating slabs with an age transition in the lithosphere, in the direction of slab retreat. In addition, as the slab would be retreating along the South American margin, a large amount of sediments are brought in the retreating system from the continent. I test how this influences the subduction evolution, and this, varying the surface processes efficiency. I find that an age transition in the subducting plate is changing the slab dynamics drastically. Instead of having a retreating slab which eventually breaks off via necking somewhere in the ocean, I systematically observe an early slab break off when the slab leaves the continent margin. There, instead of necking, I see a fast propagation of the STEP-fault located along the continent margin. The fault rotates toward the STEP located on the ocean side, triggering the formation of a horizontal tear in the slab. The whole slab detaches about 1 Ma after the horizontal tearing started. I observe a trench uplift in the topography, following the tear propagation. The surface processes have a less striking effect than the age transition. They again affect the slab trajectory and retreat rate. The more efficient the surface processes, the longer the slab retreats along the continental margin, the shallower the slab dip, the faster the retreat after subduction initiation, but the slower the retreat once an accretionary wedge forms and is large enough (about 3 Ma after subduction initiation in my experiments). When comparing these models with the topography in the Caribbean, I also observe the formation of an asymmetric accretionary wedge with a higher elevation close by the continent and a deep trench on the ocean side. However, in terms on slab structure my observations do not fit what was suggested by recent studies, such as for instance, the presence of a horizontal tear at shallow depth starting along the continent. If there has been some slab retreat in the Caribbean, other mechanisms are necessary to explain the evolution of that region. As an outlook to this work, I started working on understanding the effects of localization of the deformation on slab retreat trajectory using an analytical approach. After building a first work balance for slab retreat I looked for an expression of the slab retreat rate as a function of quantities that I defined as constants in a first approximation, such as the slab width, its thickness and the mantle rheology. The aim would be to further develop this work balance to find an expression the slab width as a function of the area affected by deformation at the tip of the STEP-faults.