ABSTRACT The subduction thrust interface represents a zone of concentrated deformation coupled to fluid generation, flow, and escape. Here, we review the internal structure of the megathrust as exposed in exhumed accretionary complexes, and we identify a deformation sequence that develops as material entering the trench is subducted through the seismogenic zone. Initial ductile flow in soft sediment generates dismembered, folded, and boudinaged bedding that is crosscut by later brittle discontinuities. Veins formed along early faults, and filling hydrofractures with the same extension directions as boudins in bedding, attest to fluid-assisted mass transfer during the shallow transition from ductile flow to brittle deformation. In higher-metamorphic-grade rocks, veins crosscut foliations defined by mineral assemblages stable at temperatures beyond those at the base of the seismogenic zone. The veins are, however, themselves ductilely deformed by diffusion and/or dislocation creep, and thus they record fracture and fluid flow at a deeper brittle-to-ductile transition. The results of numerical models and mineral equilibria modeling show that compaction of pore spaces may occur over a wide zone, as underconsolidated sediments carry water under the accretionary prism to the region where the last smectite breaks down at a temperature of ≤150 °C. However, at temperatures above clay stability, no large fluid release occurs until temperatures reach the zone where lawsonite and, subsequently, chlorite break down, i.e., generally in excess of 300 °C. In thermal models and strength calculations along overpressured subduction interfaces, where phyllosilicates form an interconnected network that controls rheology, as is generally observed, the deep brittle-viscous transition—analogous to the base of the seismogenic zone—occurs at temperatures less than 300 °C. We therefore suggest that the seismogenic zone does not produce fluids in significant volumes; however, major fluid release occurs at or near the base of the seismogenic zone. These deep fluids are either trapped, thus enabling embrittlement and features such as episodic tremor and slow slip, or flow updip along a permeable interface. Overall, we highlight fluid production as spatially intermittent, but fluid distribution as controlled also by the permeability of a deforming zone, where secondary porosity is both generated and destroyed, commonly in sync with the generation and movement of fluids.

Abstract Numerical models are used to investigate the geometry of coseismic stress-field perturbation in the crust surrounding a reverse fault, a normal fault and a strike-slip fault. The results predict a coseismic stress drop in the upper crust and loading to high stress below the brittle–ductile transition (BDT) due to the taper-off in fault slip. Coseismic stress deflections occur for each fault type as a result of the coseismic stress redistribution and is at a maximum in the middle crust where fault slip tapers-off. The zone of high-stress deflection extends downwards to the base of the crust. During the post-seismic interval, the stress-field geometry recovers towards the pre-earthquake stress state, but simple stress-field geometries cannot be re-established. The numerical results indicate that: (1) stress deflection due to slip taper-off below the BDT is important for the stress perturbation throughout the crust; (2) predictions for coseismic stress deflection exclusively based on the fault-parallel shear-stress drop ratio systematically underestimate stress deflection in the entire crust; (3) stress rotation is persistent throughout the crust in seismically active regions; and (4) the geometry of secondary faults is expected to be affected by the perturbed stress field.

Abstract We report results of a study comparing numerical models of sandbox-type experiments. Two experimental designs were examined: (1) A brittle shortening experiment in which a thrust wedge is built in material of alternating frictional strength; and (2) an extension experiment in which a weak, basal viscous layer affects normal fault localization and propagation in overlying brittle materials. Eight different numerical codes, both commereiai and academic, were tested against each other. Our results show that: (1) The overall evolution of all numerical codes is broadly similar. (2) Shortening is accommodated by in-sequence forward propagation of thrusts. The surface slope of the thrust wedge is within the stable field predicted by critical taper theory. (3) Details of thrust spacing, dip angle and number of thrusts vary between different codes for the shortening experiment. (4) Shear zones initiate at the velocity discontinuity in the extension experiment. The asymmetric evolution of the models is similar for all numerical codes. (5) Resolution affects strain localization and the number of shear zones that develop in strain-softening brittle material. (6) The variability between numerical codes is greater for the shortening than the extension experiment. Comparison to equivalent analogue experiments shows that the overall dynamic evolution of the numerical and analogue models is similar, in spite of the difficulty of achieving an exact representation of the analogue conditions with a numerical model. We find that the degree of variability between individual numerical results is about the same as between individual analogue models. Differences among and between numerical and analogue results are found in predictions of location, spacing and dip angle of shear zones. Our results show that numerical models using different solution techniques can to first order successfully reproduce structures observed in analogue sandbox experiments. The comparisons serve to highlight robust features in tectonic modelling of thrust wedges and brittle-viscous extension.