Uncertainties in Ensemble Predictions of Future Antarctic Mass Loss with the fETISh Model

Abstract:

Marine ice sheet models should be capable of handling complex feedbacks between ice and ocean, such as marine ice sheet instability, and the atmosphere, such as the elevation-mass balance feedback, operating at different time scales. Recent model intercomparisons (e.g., SeaRISE, MISMIP) have shown that the complexity of many ice sheet models is focused on processes that are either not well captured numerically (spatial resolution issue) or are of secondary importance compared to the essential features of marine ice sheet dynamics. Here, we propose a new and fast computing ice sheet model, devoid of most complexity, but capturing the essential feedbacks when coupled to ocean or atmospheric models. Its computational efficiency guarantees to easily tests its advantages as well as limits through ensemble modelling.

The fETISh (fast Elementary Thermomechanical (marine) Ice Sheet) model is a vertically integrated hybrid (SSA/SIA) ice sheet model. Although vertically integrated, thermomechanical coupling is ensured through a simplified representation of ice sheet thermodynamics based on an analytical solution of the vertical temperature profile, including strain heating and horizontal advection. The marine boundary is represented by a flux condition similar to Pollard & Deconto (2012), based on Schoof (2007). Buttressing of ice shelves is taken into account via the Shallow-Shelf Approximation (SSA). The ice sheet model is solved on four staggered finite difference grids for numerical efficiency/stability. Numerical tests following EISMINT, ISMIP and MISMIP are performed as a prerequisite.

The fETISh model is forced with different ice-shelf melt rates and basal sliding perturbations to allow comparison with recent model intercomparisons of the Antarctic ice sheet (e.g., SeaRISE, Favier et al. (2013)). These forcings are further completed with a set of scenarios involving ice-shelf buttressing and unbuttressing. All experiments are carried out on different spatial resolutions of the BEDMAP2 datasets, leading to an ensemble of dynamic Antarctic mass change over time scales ranging from 100-500 years, according to the scenarios presented. Analysis of the ensemble compared to previous ensemble experiments (e.g. SeaRISE) shows that uncertainties due to model setup/specifications can be greatly reduced.