Styles of seamount subduction and overriding plate deformation

The subduction of seamounts and its accompanying crustal deformation of the overriding plate is thought to have a large effect on the occurrence of megathrust earthquakes. Subducted seamounts can generally only be observed using seismic-reflection studies, which have shown that seamounts can subduct intact down to 30-40 km depth. On the other hand, there is evidence for accreted seamounts in e.g. the Costa Rica and Makran subduction zones. Because such observations only provide snapshots in space and time, little is still known about the exact evolution of seamount subduction and its effect on overriding-plate deformation and subduction zone seismicity through time. We investigate the different styles of seamount subduction and how these influence seismicity and overriding plate deformation. We use seismo-thermo-mechanical (STM) models with a visco-elasto-plastic rheology simulating seamount subduction over millions of years in a 2D realistic subduction setting. The momentum, mass and energy equations are solved and a strongly slip rate dependent friction allows for the spontaneous development of faults. The use of a realistic rheology allows us to evaluate faulting patterns and the state of stress in the overriding plate caused by seamount subduction. We find three scenarios for seamount subduction by varying the rock properties cohesion (C) and pore fluid pressure ratio (λ): (1) cutting off of the seamount at the trench leading to frontal accretion; (2) intact subduction through the trench, followed by flattening and stretching of the seamount; and (3) intact subduction of the seamount until seismogenic depths. Scenario’s 1 and 2 are most common, while scenario 3 only occurs under a limited range of material parameters. Particularly, a cohesion of the seamount and upper oceanic crust larger than 20 MPa is required for intact seamount subduction. Decreasing λ on locations with ample amounts of fluids increases the strength of the sediments, upper oceanic crust and seamount, but does not lead to intact seamount subduction. Subduction scenario’s 2 and 3 show more crustal deformation and seismicity within the fore-arc than subduction of a smooth interface (scenario 1 and models without a seamount). Seismicity patterns are also affected by λ and C. A low λ results in shorter and shallower megathrust ruptures and higher cohesions decrease the recurrence interval. Furthermore, the seamount itself introduces more frequent nucleation of smaller events at its edge.