LAUSR

Abstract We compare mantle dynamic topography predicted from mantle convection modelling with residual topography measurements for oceanic regions, where crustal basement thickness is 10.0 km or less. Measurements of residual topography, calculated by removing the isostatic effects of crustal thickness variation, bathymetry, sediments, ice and lithosphere thermal anomalies, from the observed topography, are inaccurate for continents and oceanic plateaus due to uncertainties in determining their crustal thickness and density. As a consequence, residual topography measurements for these regions are unsuitable for testing mantle dynamic topography predictions. Residual topography is more accurately determined for oceanic crust. We use global mapping of crustal basement thickness using gravity anomaly inversion to identify oceanic crust of 10.0 km thickness or less to select measured residual topography for comparison with predicted mantle dynamic topography. For these oceanic regions we compare mantle dynamic topography and residual topography and, using amplitude histogram matching and grid searches, compute the amplitude rescaling and shift which needs to be applied to predicted mantle dynamic topography to fit the observed residual topography. We examine three global compilations which use different approaches to determine mantle dynamic topography: (i) Steinberger (2007) , which uses seismic topography deeper than 220 km to determine mantle density; (ii) Flament et al. (2013) , which uses plate velocity and subduction history; and (iii) Steinberger et al. (2017) , which uses seismic tomography, including that above 220 km, to determine shallow upper mantle densities. Our analysis shows that for the Steinberger (2007) and Flament et al. (2013) compilations, the predicted mantle dynamic topography for oceanic regions requires a rescaling of approximately ×0.5 and a negative shift of approximately −500 m to match the observed residual topography. In contrast Steinberger et al. (2017) , which includes shallow upper mantle densities above 220 km, requires only a small shift (+50 m) but a greater scaling of ×0.375. Maps of renormalised (rescaled and shifted) mantle dynamic topography for Steinberger et al. (2017) show a close resemblance to measured residual topography.