The influence of lateral Earth structure on inferences of global ice volume during the Last Glacial Maximum

The mapping between far-field relative sea level (RSL) records and changes in ice volume or global mean sea level (GMSL) involves a correction for glacial isostatic adjustment (GIA). This mapping is thus sensitive to uncertainties inherent to GIA modeling, including the spatio-temporal history of ice mass changes and viscoelastic Earth structure. Here, we investigate the effect of incorporating lateral variations in Earth structure on predicting far-field sea level in order to determine if this source of model uncertainty significantly impacts estimates of global ice volume at the Last Glacial Maximum (LGM). We consider a set of forty 3-D simulations that sample different Earth model parameters: the adopted lithospheric thickness, the seismic velocity model used to infer lateral temperature variations, the scaling factor used in the conversion from temperature to viscosity, and the spherically averaged “background” viscosity profile. In addition, we consider results based on two ice histories. We present global maps of the differences between these simulations and a set of 1-D simulations at the LGM, as well as RSL histories at 5 locations that have been previously considered in estimates of ice volume at LGM: Barbados, two sites at the Great Barrier Reef, Bonaparte Gulf and Sunda Shelf. We find that the difference between inferences of global mean sea level (GMSL) at LGM based on 3-D and 1-D Earth models peaks in Barbados with differences ranging from ∼2.5 to 11 m, with a mean of ∼6–7 m. At the other sites, the difference ranges from ∼2 to −8 m, with mean differences between ∼0 and −3 m. After comparing different pairs of simulations, we conclude that, in general, the impact of varying the seismic model, lithospheric thickness model, background 1-D model, and scaling factor from temperature to viscosity is significant at far-field sites. Finally, while we do not find a consistent signal at the above far-field sites that would help to reconcile the LGM ice volumes estimated from GIA studies and those estimated from summing regional ice sheet reconstructions, the impact is nonetheless large enough that GIA analyses of RSL records in the far field of ice sheets should include 3-D viscoelastic Earth models.