Abstract/Summary

Ocean heat transport (OHT) has been proposed as a major driver of sea ice extent on multidecadal timescales. This thesis brings new insight into the mechanisms behind this relationship, with implications for uncertainties arising in simulations of present and future climate with coupled general circulation models (GCMs). Using a novel implementation of a zonally-averaged energy-balance model (EBM), it is shown that sea ice is intrinsically more sensitive to ocean that atmospheric heat transport (AHT). The ratio of sensitivities to the two heat transports mainly depends on large-scale atmospheric radiation parameters. A simple equation is derived relating changes in sea ice, OHT, and surface temperature, revealing how the sea ice sensitivity to OHT arises from emergent constraints on the top-of-atmosphere radiation balance. Simulations by GCMs exhibit strong anticorrelation between poleward OHT and sea ice extent in both hemispheres, applying to both internal and forced (future) multidecadal variability. These relationships are captured and explained by the aforementioned EBM equation. Different qualitative processes are exhibited in each hemisphere, robust across 20 GCMs and with analogues in the EBM. In the Arctic, OHT converges along the Atlantic sea ice edge, efficiently eroding the ice edge and enhancing AHT to higher latitudes. Poleward OHT into the Southern Ocean is released relatively uniformly under the Antarctic sea ice pack. Under rising greenhouse-gas emissions, GCMs simulate a wide range of projected sea ice losses in both hemispheres. This is strongly related to biases in the mean-state and future change of poleward OHT. These results motivate the need for improved ocean representation in GCMs to better constrain future sea ice projections.