Abstract. The Greenland ice sheet is one of the largest contributors to global mean sea-level rise today and is expected to continue to lose mass as the Arctic continues to warm. The two predominant mass loss mechanisms are increased surface meltwater run-off and mass loss associated with the retreat of marine-terminating outlet glaciers. In this paper we use a large ensemble of Greenland ice sheet models forced by output from a representative subset of the Coupled Model Intercomparison Project (CMIP5) global climate models to project ice sheet changes and sea-level rise contributions over the 21st century. The simulations are part of the Ice Sheet Model Intercomparison Project for CMIP6 (ISMIP6). We estimate the sea-level contribution together with uncertainties due to future climate forcing, ice sheet model formulations and ocean forcing for the two greenhouse gas concentration scenarios RCP8.5 and RCP2.6. The results indicate that the Greenland ice sheet will continue to lose mass in both scenarios until 2100, with contributions of 90±50 and 32±17 mm to sea-level rise for RCP8.5 and RCP2.6, respectively. The largest mass loss is expected from the south-west of Greenland, which is governed by surface mass balance changes, continuing what is already observed today. Because the contributions are calculated against an unforced control experiment, these numbers do not include any committed mass loss, i.e. mass loss that would occur over the coming century if the climate forcing remained constant. Under RCP8.5 forcing, ice sheet model uncertainty explains an ensemble spread of 40 mm, while climate model uncertainty and ocean forcing uncertainty account for a spread of 36 and 19 mm, respectively. Apart from those formally derived uncertainty ranges, the largest gap in our knowledge is about the physical understanding and implementation of the calving process, i.e. the interaction of the ice sheet with the ocean.

[1] Dynamic and thermodynamic regimes of ice shelves experiencing weak (≲1 m year À1 ) to strong (~10 m year À1 ) basal melting in cold (bottom temperature close to the in situ freezing point) and warm oceans (bottom temperature more than half of a degree warmer than the in situ freezing point) are investigated using a 1-D coupled ice/ocean model complemented with a newly derived analytic expression for the steady state temperature distribution in ice shelves. This expression suggests the existence of a basal thermal boundary layer with thickness inversely proportional to the basal melt rate. Model simulations show that ice shelves afloat in warm ocean waters have significantly colder internal ice temperatures than those that float in cold waters. Our results indicate that in steady states, the mass balance of ice shelves experiencing strong and weak melting is controlled by different processes: in ice shelves with strong melting, it is a balance between ice advection and basal melting, and in ice shelves with weak melting, it is a balance between ice advection and deformation. Sensitivity simulations show that ice shelves in cold and warm oceans respond differently to increase of the ocean heat content. Ice shelves in cold waters are more sensitive to warming of the ocean bottom waters, while ice shelves in warm waters are more sensitive to shallowing of the depth of the thermocline.

The effect of the North Atlantic Ocean on the Greenland Ice Sheet through submarine melting of Greenland's tidewater glacier calving fronts is thought to be a key driver of widespread glacier retreat, dynamic mass loss and sea level contribution from the ice sheet. Despite its critical importance, problems of process complexity and scale hinder efforts to represent the influence of submarine melting in ice sheet-scale models. Here we propose parameterizing tidewater glacier terminus position as a simple linear function of submarine melting, with submarine melting in turn estimated as a function 5 of subglacial runoff and ocean temperature. The relationship is tested, calibrated and validated using datasets of terminus position, runoff and ocean temperature covering the full ice sheet and surrounding ocean from the period 1960-present. We demonstrate a statistically significant link between multi-decadal tidewater glacier terminus position and submarine melting and show that the proposed parameterisation has predictive power when considering a population of glaciers. An illustrative 21st century projection is considered suggesting that tidewater glaciers in Greenland will undergo little further retreat in a low 10 emissions RCP2.6 scenario. In contrast, a high emissions RCP8.5 scenario results in a median retreat of ∼6 km, with 35% of glaciers experiencing retreat exceeding 10 km. Our study provides a long-term and ice sheet-wide assessment of the sensitivity of tidewater glaciers to submarine melting and proposes a practical and empirically validated means of incorporating ocean forcing into models of the Greenland ice sheet.