Constraints of Relative Sea Level Change on the Mantle Viscosity and the Late Pleistocene Deglaciation History

Abstract Two commonly used ice models that are constructed using glacial isostatic adjustment (GIA) modeling are the ICE‐6G and ANU ice models. In this study, we examined the relationships among mantle viscosity, ice models and relative sea level (RSL) data through an analytic GIA model. In general, almost all the pairs of RSL data sets and ice models we considered appear to be consistent with a mantle viscosity structure with a factor of 10–20 viscosity increase from the upper to lower mantles. By using the GIA model with the viscosity structure that produces a minimum model‐data misfit, we constructed the temporal and spatial distributions of misfit (i.e., misfit maps) to different RSL data sets, for both the ICE‐6G and ANU ice models. While the misfit maps at different times clearly show that more observations are desired to constrain ice models, we propose that the spatial and temporal misfit maps should be used to revise the existing ice models to further improve the fit to RSL data. In our initial proof of concept attempts to modify ICE‐6G by adding more ice to it, the three modified ICE‐6G ice models we considered all significantly improve the fit to the far‐field RSL data, although additional effort is needed to reduce misfit to near field RSL data. Finally, we emphasize that RSL at different far‐field sites may differ by up to ∼25 m at the Last Glacial Maximum (LGM) (∼26 ka), suggesting the need for a sufficiently large number of far‐field RSL data in determining the total melt ice volume since the LGM.