Evolving ice and ocean loads drive viscoelastic deformation of the solid Earth and perturbations of its gravitational potential and rotation, collectively referred to as glacial isostatic adjustment (GIA). Accurate sea-level models, both paleo reconstructions and future forecasts, must account for GIA physics, but software capable of capturing the full range of processes involved at the necessary scales remains in its infancy. This comparative lack of capability adds a significant source of uncertainty into sea-level models that propagates all the way through to coastal hazard assessments.

Here, we report on recent developments in Australian GIA modelling through the G-ADOPT codebase. This finite element framework solves the governing global GIA equations at lengthscales down to 1 km and timescales ranging from months to millions of years, including in the presence of large lateral viscosity variations and arbitrarily complex rheologies. Crucially, it enables automatic derivation of adjoint sensitivity kernels, permitting optimisation of ice histories and Earth structure using standard GIA datasets including geological sea-level indicators, tide gauges, and satellite geodesy. As the G-ADOPT project continues to mature, this data assimilation approach will permit characterisation of model uncertainties and their propagation into increasingly robust global forecasts of future sea-level change.