Evidence for dehydration-modulated small-scale convection in the oceanic upper mantle from seafloor bathymetry and Rayleigh wave phase velocity

Abstract It is well established that the evolution of oceanic lithosphere departs from the predictions of the half-space cooling model (HSC). Understanding the magnitude, spatial distribution and physical origin of this departure is fundamental to our understanding of plate tectonics and the Earth's thermal evolution. Here we analyze Rayleigh wave phase velocities and seafloor topography separately along spreading-history trajectories, which track each piece of seafloor since its formation at the mid-ocean ridge, in the Pacific and the Atlantic basins. We identify the age along each trajectory at which HSC fails to satisfy the observations. Bathymetry and seismic velocity yield consistent results on whether or not HSC gives a statistically better fit to the observations than the plate cooling model at 80% of the locations we examine. Locations of HSC failure identified by velocity but not topography overlap with SS precursor detections in the Pacific, suggesting partial melting at these places. For the locations where the two data sets agree, in the Atlantic the probability of HSC failure is correlated with the axial depth of the present-day ridge. Along trajectories for which the present-day ridge is shallow (deep), HSC failure occurs later (earlier). A similar dependence in the Pacific is difficult to observe since both ridge depth and HSC failure age vary little. We show that the patterns of HSC failure are best explained by small-scale convection in the oceanic upper mantle for which the onset time is governed by rheological stratification created by partial melting at the ridge.