Thermal anomalies of back-arc basins affected by double subductions

The dynamic evolution of back-arc basins (BABs) is widely studied under one subduction scenario. The anomaly high mantle potential temperature in some BABs (e.g., the Lau basin (LB) and the Manus basin (MB)) is attributed to the fast slab retreating. Recent studies, however, propose that the thermal anomalies in the North Fiji basin (NFB) and the West Philippine basin (WPB) may be induced by two face-to-face subductions, featured by, for instance, intersection of spreading ridges with trench and fertilized asthenospheric mantle. Nonetheless, this hypothesis is not well testified. In this study, we implement 2D thermo-mechanic models to investigate the thermal state of BABs under double subduction. We test four key parameters which control the double subduction modes, including the age of subducting oceanic plates, distance between two subductions, convergence rates and mantle viscosity. The model results show two typical modes: (1) symmetric mode, in which the two subductions behave concordantly, with slab retreat, back-arc spreading occurring at nearly the same time; (2) asymmetric mode, in which the two subductions behave discordantly. The mantle thermal anomalies under BABs are more prone to occur in asymmetric mode, in which the hot lower mantle return flow induced by earlier retreating slab upwells to the BABs triggered by the latter slab retreating. In cases with thermal anomalies, the mantle temperature at 100km depth and melting depth under the BABs could reach 1400 and 100 km respectively, which are larger than that (1300 and 70 km) in the one subduction models. Our model results render a new perspective for the interpretation of the thermal anomaly in the NFB and even the MB. The hot mantle return flow induced by earlier subduction may provide the thermal anomaly condition for the latter back-arc spreading. We argue that the presumed effect of hot mantle plume in both areas may be derived from deep mantle flow induced by multiple subductions. This interpretation is strongly based on the nonadiabaticity of the mantle in our model, which is plausible since multiple periods of mantle upwelling and downwelling occurred before newly generated BABs in regions of NFB or MB. This history may lead to the nonadiabaticity of the mantle as indicated by former numerical studies.