Numerical Modelling of Subduction Plate Interface, Technical Advances for Outstanding Questions

Abstract:

The subduction zone interface is the place of the largest earthquakes on earth. Compared to the size of a subduction zone itself, it constitutes a very thin zone (few kilometers) with effective rheological behaviour that varies as a function of pressure, temperature, loading, nature of the material locally embedded within the interface as well as the amount of water, melts and CO₂. Capturing the behaviour of this interface and its evolution in time is crucial, yet modelling it is not an easy task. In the last decade, thermo-mechanical models of subduction zone have flourished in the literature. They mostly focused on the long-term dynamics of the subduction; e.g. flat subduction, slab detachment or exhumation. The models were validated models against PTt path of exhumed material as well as topography. The models that could reproduce the data all included a mechanically weak subduction channel made of extremely weak and non cohesive material. While this subduction channel model is very convenient at large scale and might apply to some real subduction zones, it does not capture the many geological field evidences that point out the exhumation of very large slice of almost pristine oceanic crust along localised shear zone. Moreover, modelling of sismological and geodetic data using short term tectonic modelling approach also point out that large localised patches rupture within the subduction interface, which is in accordance with geological data but not with large-scale long-term tectonic models.

I will present how high resolution models permit to produce slicing at the subduction interface and give clues on how the plate coupling and effective location of the plate interface vary over a few millions of year time scale. I will then discuss the implication of these new high-resolution long-term models of subduction zone on earthquake generation, report progress in the development of self-consistent thermomechanical codes which can handle large strain, high resolution and fluids pathways together with adaptative time stepping schemes permitting to switch from static long term-tectonic approach to seismic cycle modelling within the same frame work and discuss how the results of these forward models can be used to better assess coupling at the subduction zone interface from geodetic, seismologic and geologic record.