Pliocene Antarctic ice sheet model ensembles with joint constraints from reconstructed sea level and margin retreat

The warm Pliocene was a period of comparable atmospheric carbon dioxide concentrations to modern, but with sea levels up to ~20 m higher. High Pliocene sea level implies collapse of the West Antarctic ice sheet, and mass loss from East Antarctica. Modelling studies have sought to reproduce Pliocene deglaciation, and use sea level reconstructions as a constraint on future projections, despite their large uncertainties. We simulated the Pliocene Antarctic ice sheet under warm Pliocene climate with the BISICLES ice sheet model, capturing grounding line and ice stream dynamics down to 4 km resolution. Our perturbed parameter ensemble approach explores uncertainties in basal sliding, surface mass balance processes, bedrock-ice sheet interactions, ice shelf basal melt sensitivity to ocean forcing and choice of climate model. We simulated a mean Antarctic sea level contribution of 1.85 m and a range of -15.90 to 28.27 m, largely driven by uncertainty in the perturbed basal sliding parameter. We applied a joint calibration, combining a Pliocene Antarctic sea level contribution range and a comparison of regional grounding line with reconstructed Pliocene retreat. This reduced the mean to 1.46 m. The calibration reduced the simulated range by a factor of ~4, and was more effective in reducing uncertainty than comparing to sea level reconstructions alone. Further ensembles explored initial condition uncertainty, and the impact of perturbing the control. The Pliocene initial condition was tested for a subset of main ensemble simulations (mean = -2.35 m), increasing the mean contribution by 11.45 m with all simulations passing the joint calibration. We perturbed the control simulation for the same subset of ensemble members. This increased the mean Antarctic contribution by 7.78 m, and by 8.45 m in combination with the two Pliocene data constraints. We demonstrate a modelling framework that captures important interactions between the ice sheet and other components of the Earth system, whilst being efficient for ensemble studies. Moreover, we used two Pliocene data constraints to rule out ensemble members. This Pliocene-calibrated modelling framework can be run under future climate scenarios, to reduce uncertainty in projections of Antarctica’s long-term contribution to sea level under anthropogenic climate change.