No abstract

[1] In subduction zones, many observations indicate that the backarc thermal state is particularly hot and that the upper lithosphere is thin, even if no recent extension episode has occurred. This might result from free thermal convection favored by low viscosities in the hydrated mantle wedge. We perform 2-D numerical experiments of the convective mantle wedge interaction with both the downgoing slab and the overriding plate to test this hypothesis, explore its physical mechanism, and assess its dependencies on some relevant rock properties. Water transfers across the subducting plate and the mantle wedge are explicitly modeled by including in the calculation realistic hydration/dehydration reaction boundaries for a water-saturated mantle and oceanic crust. The rheology is non-Newtonian and temperature-, pressure-, and water content-dependent. For low strength reduction associated to water content, the upper plate is locally thinned by an enhanced corner flow. For larger strength reductions, small convection cells rapidly thin the upper plate (in less than 15 Myr) over the area in the overriding lithosphere hydrated by slab-derived water fluxes. As a result, the thinned region location depends on the subducting plate thermal state, and it increases with high convergence rates and low subduction dip angles. Other simulations are performed to test the sole effect of hydrous rock weakening on the upper plate/mantle convective interaction. They show that the thinning process is not influenced by the corner flow, but develops at the favor of a decoupling level induced by the formation of hydroxylated minerals inside the hydrated lithosphere. The erosion mechanism identified in these simulations allows us to explain the characteristic duration of erosion as a function of the hydrous strength reduction. We find that the presence of amphibole in the upper lithosphere in significant proportions is required down to a temperature of about 980°C, corresponding to an initial depth of $70 km, to strongly decrease the strength of the base of the lithosphere and trigger a rapid erosion (<15 Myr).

The role of seafloor roughness on the seismogenic behavior of subduction zones has been increasingly addressed over the past years, although their exact relationship remains unclear. Do subducting features like seamounts, fracture zones, or submarine ridges act as barriers, preventing ruptures from propagating, or do they initiate megathrust earthquakes instead? We address this question using a global approach, taking into account all oceanic subduction zones and a 117‐year time window of megathrust earthquake recording. We first compile a global database, SubQuake, that provides the location of a rupture epicenter, the overall rupture area, and the region where the largest displacement occurs (the seismic asperity) for MW ≥ 7.5 subduction interplate earthquakes. With these data, we made a quantitative comparison with the seafloor roughness seaward of the trench, which is assumed to be a reasonable proxy for the subduction interface roughness. We compare the spatial occurrence of megathrust ruptures, seismic asperities, and epicenters, with two roughness parameters: the short‐wavelength roughness RSW (12–20 km) and the long‐wavelength roughness RLW (80–100 km). We observe that ruptures with MW ≥ 7.5 tend to occur preferentially on smooth subducting seafloor at long wavelengths, which is especially clear for the MW > 8.5 events. At both short and long wavelengths, seismic asperities show a more amplified relation with smooth seafloor than rupture segments in general. For the epicenter correlation, we see a slight difference in roughness signal, which suggests that there might be a physical relationship between rupture nucleation and subduction interface roughness.

No abstract

[1] Recent statistical analysis by Lallemand et al. (2008) of subduction zone parameters revealed that the back-arc deformation mode depends on the combination between the subducting (v sub ) and upper (v up ) plate velocities. No significant strain is recorded in the arc area if plate kinematics verifies v up = 0.5 v sub À 2.3 (cm/a) in the HS3 reference frame. Arc spreading (shortening) occurs if v up is greater (lower) than the preceding relationship. We test this statistical law with numerical models of subduction, by applying constant plate velocities far away from the subduction zone. The subducting lithosphere is free to deform at all depths. We quantify the force applied on the two converging plates to sustain constant surface velocities. The simulated rheology combined viscous (non-Newtonian) and brittle behaviors, and depends on water content. The influence of subduction rate v s is first studied for a fixed upper plate. After 950 km of convergence (steady state slab pull), the transition from extensional to compressive stresses in the upper plate occurs for v s $ 1.4 cm/a. The effect of upper plate velocity is then tested at constant subduction rate. Upper plate retreat (advance) with respect to the trench increases extension (compression) in the arc lithosphere and increases (decreases) the subducting plate dip. Our modeling confirms the statistical kinematic relationship between v sub and v up that describes the transition from extensional to compressive stresses in the arc lithosphere, even if the modeled law is shifted toward higher rates of upper plate retreat, using our set of physical parameters (e.g., 100 km thick subducting oceanic plate) and short-term simulations. Our results make valid the choice of the HS3 reference frame for assessing plate velocity influence on arc tectonic regime. The subduction model suggests that friction along the interplate contact and the mantle Stokes reaction could be the two main forces competing against slab pull for upper mantle subductions. Besides, our simulations show that the arc deformation mode is strongly time dependent.

We have developed a new approach to characterize the seafloor roughness seaward of the trenches, as a proxy for estimating the roughness of the subduction interface. We consider that abrupt elevation changes over given wavelengths play a larger role in the seismogenic behavior of the subduction interface than the amplitude of bathymetric variations alone. The new database, SubRough, provides roughness parameters at selected spatial wavelengths. Here we mainly discuss the spatial distribution of short‐ (12–20 km) and long‐wavelength (80–100 km) roughness, RSW and RLW, respectively, along 250‐km‐wide strips of seafloor seaward of the trenches. Compared with global trend, seamounts show distinct roughness signature of much larger amplitudes at both wavelengths, whereas aseismic ridges only differ from the global trend at long wavelengths. Fracture zones cannot be distinguished from the global trend, which suggests that their potential effect on rupture dynamics is not the consequence of their roughness, at least not at these wavelengths. Based on RLW amplitude, segments along subduction zones can be defined from rough to smooth. Subduction zones like the Solomons or the Ryukyus appear dominantly rough, whereas others like the Andes or Cascadia are dominantly smooth. The relative contribution of smooth versus rough areas in terms of respective lateral extents probably plays a role in multipatch rupture and thus in the final earthquake magnitude. We observe a clear correlation between high seismic coupling and relatively low roughness and conversely between low seismic coupling and relatively high seafloor roughness.

S U M M A R YNumerical simulations of 2-D Rayleigh-Bénard convection are designed to study the development of convection at the base of the cooling lithosphere. A zero temperature is suddenly imposed at the mantle surface, which has initially a homogeneous temperature. The strongly temperature-dependent viscosity fluid is heated from within, in order to balance the internal temperature drift resulting from global fluid cooling. For a while, the lithosphere cools approximatively as a conductive half-space and the lithospheric isotherms depth grows as the square root of age. As instabilities progressively develop at the base of the lithosphere, lithospheric cooling departs from the half-space model. We propose two different parametrizations of the age of the first dripping instability, using boundary layer marginal stability or quantifying the characteristic timescale of the instability exponential growth as a function of the Rayleigh number and of the viscous temperature scale. Both parametrizations account very well for our numerical estimates of onset times, but with a slightly better adjustment of the viscous temperature scale dependence in the second case. The absolute value of the onset time depends on the amplitude and location of initial temperature perturbations within the box and on the initial temperature structure of the thermal boundary layer (TBL). Furthermore, thermal perturbations of finite amplitude located within the lithosphere (such as the ones induced by transform faults, for example) strongly reduce the age of the first dripping instability. However, the onset time parametrization derived from transient cooling experiments well adjusts the lithospheric age of the first drip instability below a lithosphere cooling perpendicularly to the ridge axis. This study emphasizes the role of the initial topography of the lithospheric isotherms on the development of instabilities.

Two-dimensional simulations using a thermomechanical model based on a finite-difference method on a staggered grid and a marker in cell method are performed to study the plumelithosphere interaction beneath moving plates. The plate and the convective mantle are modelled as a homogeneous peridotite with a Newtonian temperature-and pressure-dependent viscosity. A constant velocity, ranging from 5 to 12.5 cm yr −1 , is imposed at the top of the plate. Plumes are generated by imposing a thermal anomaly of 150 to 350 K on a 50 km wide domain at the base of the model (700 km depth); the plate atop this thermal anomaly is 40 Myr old. We analyse (1) the kinematics of the plume as it impacts the moving plate, (2) the dynamics of time-dependent small-scale convection (SSC) instabilities developing in the low-viscosity layer formed by spreading of hot plume material at the base of the lithosphere and (3) the resulting thermal rejuvenation of the lithosphere. The spreading of the plume material at the base of the lithosphere, characterized by the ratio between the maximum down-and upstream horizontal (dimensionless) velocities in the plume-fed sublithospheric layer, Pe up /Pe down depends on the ratio between the maximum plume upwelling velocity and the plate velocity, Pe plume /Pe plate . For fast plate velocities and sluggish plumes (low Pe plume /Pe plate ), plate motion drags most plume material and downstream flow is dominant. As Pe plume /Pe plate increases, an increasing part of the plume material flows upstream. SSC systematically develops in the plume-fed sublithospheric layer, downstream from the plume. Onset time of SSC decreases with the Rayleigh number. For vigorous plumes, it does not depend on plate velocity. For more sluggish plumes, however, variations in the plume spreading behaviour at the base of the lithosphere result in a decrease in the onset time of SSCs with increasing plate velocity. In any case, SSC results in uplift of the isotherm 1573 K by up to 20 km relative to its initial equilibrium depth at the impact point.