Modeling the Interconnectivity of Non-stationary Polar Ice Sheets

Abstract

Understanding changes to the mass of the polar ice sheets is of crucial scientific and socioeconomic importance due to their effect on the wider Earth system and potential to contribute to future sea level rise. On monthly to multi-decadal timescales, there is much uncertainty around the extent to which the non-stationary, non-linear responses of the ice sheets interact through the atmosphere-ocean climate. We test and quantify the nature of non-stationarity and inter-dependence between ice mass balance time series for Greenland, West and East Antarctica using a multi-cointegration vector autoregression model, which has been used to show equivalence between simple climate models and emulations of complex physical processes. We focus on three alternative specifications by comparing I(2) models of cumulative ice mass balance with an I(1) model of the ice mass balance, exploring the model dynamics, and evaluating the out-of-sample forecasts against satellite observations. Our results support the I(2) model with a bipolar relationship between Greenland and West Antarctica and provide some of the first empirical evidence of tipping-points in the recent observed record. Long-term projections indicate that there is considerable risk of Greenland contributing more to sea level rise than under the IPCC’s extreme climate change scenario.