Is Coupled Ice–Ocean Modeling Needed to Improve Projections of Thwaites Glacier?

Thwaites Glacier, the West Antarctic marine giant at the center of the fastest-retreating sector of the ice sheet, has been a top priority for projections of future sea level rise since observations of its ongoing mass loss were first recognized in the mid-2000’s. The observed retreat at Thwaites Glacier is thought to be strongly influenced by the intrusion of warm circumpolar deep water (CDW) onto the continental shelf, where CDW temperature, thermocline depth, and circulation control the delivery of heat to the ice shelf and drive basal melting. This hypothesized control on retreat has led to a push to improve the source of melt rate data used in forward simulations of Thwaites, ranging from advances in observations of melt to the development of fully coupled ice–ocean models which allow for realistically responsive ocean circulation. After over a decade into the effort to couple ice sheet models and general circulation models, we ask: Is Coupled Ice–Ocean Modeling Needed to Improve Projections of Thwaites Glacier? 

We present new experimental results from a coupling of the Ice-sheet and Sea-level System Model (ISSM) and the Massachusetts Institute of Technology general circulation model (MITgcm), examining the evolution of Thwaites over the 21st century. We test the response of ice loss and grounding line retreat to future climate scenarios, thermocline depth, and periodic variability in thermocline depth on inter-annual to decadal timescales, in one of the largest sensitivity testing experiments to date. Our results show that for periods of at least decadal signals, capturing realistic variability in melt rates does not have a significant impact on the current trajectory of Thwaites’ ice loss. While error in melt rates over longer timescales, due to error or uncaptured trends in thermocline depth and resulting heat flux, can impart substantive bias into the projections of mass loss, in our model these accumulate to a relative bias of less than 10% by 2100—an absolute bias on the order of 1 to 2 mm. Significantly, the overall trends of mass loss and patterns of grounding line retreat over this time period are broadly similar in our model to uncoupled schemes. Integrating these results with past work, we argue that coupled modeling, while a powerful tool with utility in other problems, should not be prioritized as an area of research necessary to improve current projections of mass loss from Thwaites Glacier.