Future Projections over Antarctica from UKESM1.0-ice simulations

Antony Siahaan1, Robin S. Smith2, and Paul R. Holland1

1British Antarctic Survey,  2NCAS/Department of Meteorology, University of Reading

Many advances have been achieved since the last newsletter article discussing the UKESM1 model configuration with interactively coupled Greenland and Antarctic ice sheets (https://ukesm.ac.uk/portfolio-item/preliminary-results-ukesm1-antarctic-coupling/). In this article, we give a brief overview of some new results for Antarctica based on some future projection simulations made with this model, namely UKESM1.0-ice. This configuration, UKESM1.0-ice, has been frozen since the development and preliminary tests were completed (Smith et al., 2021) where the N96-eORCA1 configuration was chosen due to its stable computation and more acceptable simulation  of the Southern Ocean circulation than the N96-eORCA025 configuration.

We carried out some 21st century runs with this model under the CMIP6 anthropogenic forcing scenarios (Shared Socioeconomic Pathway) SSP1-1.9 and SSP5-8.5, making them the first future projection simulations using a coupled Earth system model with two-way coupling between both atmosphere and ocean components to dynamic models of the Greenland and Antarctic ice sheets. A small initial condition ensemble of four members was used for each scenario, where each set of initial conditions were shared by an SSP1-1.9 and an SSP5-8.5 ensemble member. Creating initial conditions for such a coupled model is generally complicated and it will continuously need improvements in future work. The process of setting up our four initial conditions is described in Siahaan et al. (in review) where we made use of various data from historical UKESM1 simulations, modern observations, and present-day outputs from different sources.

All simulations in the ensemble remained stable throughout the 21st century. For each emission scenario, the simulations show similar ice sheet and basal melting responses across the ensemble members regardless of their initial climate states. The SSP1-1.9 scenario runs do not show major changes of melt rate pattern (Figs. 1a,b) or area-integral melt flux (Siahaan et al., in review) during the 21st century. The only exception is under the Amery Ice Shelf, as indicated in Fig. 1b, where the melt rates ensemble mean are high close to the grounding line in the last 10 years of the run (2090 to 2100).

Figure 1: Antarctic ice shelf melt rates (m/yr) (a) average 2020-2030 all ensemble mean, (B) average 2090-2099 SSP1-1.9 ensemble mean, and (C) average 2095-2100 SSP5-8.5 ensemble mean.

Clear differences in basal melt responses between the SSP1-19 and SSP5-85 scenarios are found under the Ross and Filchner ice shelves (Fig. 1a) where very strong responses to the SSP5-8.5 forcing are brought about by sustained warm water intrusions into the shelf regions which begin around 2070 and 2100 respectively, driven by freshening on the continental shelf and slope (Siahaan et al., in review). Since shelf salinity is an important factor in the intrusion, freshwater biases we have in the initial conditions in these two regions will need to be improved in future developments. On the other hand, only limited changes were simulated under ice shelves in the Amundsen Sea in the SSP5-8.5 runs, where melting under the Pine Island Glacier and Thwaites ice shelves is no higher than in the SSP1-1.9 simulations. This is not unexpected given the low resolution of the ocean domain in the Amundsen sector. 

Figure 2: Antarctica Surface Mass Balance (SMB) (m/yr). (a) All ensemble mean for period 2020-2030. Locations with statistically significant difference (95% Student’s t- confidence interval) between SSP1-1.9 and SSP5-8.5 ensemble means are hatched. (b) SSP1-1.9 ensemble mean difference between 2090-2100 period and 2020-2030 period. Locations where the difference is statistically significant (95% Student’s t- confidence interval) are hatched. (c) SSP5-8.5 ensemble mean difference between 2090-2100 period and 2020-2030 period, almost everywhere is statistically significant. Black and grey lines indicate the ice sheet grounding lines and ice shelf fronts, respectively.   

Similar to the ice shelf melt responses, the ice sheet surface mass balance (SMB) responses in the SSP1-1.9 simulations do not show significant change over the century as indicated in Figs. 2a-b.In contrast, for the SSP5-8.5 simulations, the SMB on the grounded ice sheet and floating ice shelves show opposing trends (Fig. 2c). A strong increase in SMB on the grounded ice sheet follows the increase in atmospheric greenhouse gas concentrations, with a rapid increase in snowfall and negligible surface melting (Siahaan et al., in review). On the ice shelves, a large increase in surface melting and runoff dominates the SMB trend, leading to net loss of ice mass from the surface (Siahaan et al., in review).

Figure 3: Time-series of ice mass budget. Top row: cumulative anomaly of ice mass above flotation relative to the initial condition; middle: rate of discharge across grounding lines; bottom row: SMB rate on the grounded ice. The grey, black, pink and red lines represent the SSP1-1.9 ensemble members, SSP1-1.9 ensemble mean, SSP5-8.5 ensemble members and SSP5-8.5 ensemble mean respectively. The Antarctic Peninsula region is considered part of West Antarctica. In each row (budget), the axis range is set the same for all columns (regions). 

Steady responses to the SSP1-1.9 forcing lead to limited changes in the ice sheet behaviour where the mass of ice above flotation on Antarctica (Figs. 3a-c) decreases more slowly than is currently observed (The IMBIE Team, 2018). UKESM1-ice therefore simulates a smaller increase in Global Mean Sea Level (GMSL) for this scenario than many other projections in the AR6 report (Fox-Kemper et al., 2021). In the Amundsen sector, although the basal melting under the Pine Island Glacier and Thwaites ice shelves are within observational uncertainty limits  (Siahaan et al., in review), our SSP1-1.9 simulations are unable to simulate the observed increasing rate of ice discharge across the grounding line in the West Antarctica (Fig. 3f). This needs further investigation as we develop UKESM1-ice further. 

Despite significant surface warming and strong melting beneath the large ice shelves, our SSP5-8.5 simulations do not produce a significant sea level rise contribution from Antarctica by 2100 (Figs. 3a-c). This is due to the large increase in SMB over the grounded ice sheet (Fig. 2c & Figs. 3g-i), which outweighs changes in the discharge of ice across the grounding line (Figs. 3d-f). The former dominates because of the rapid response of precipitation to climate change, while the impact of basal melt in the latter takes longer to be significant. Since ice sheets respond to changes in climate on centennial timescales, our simulations of the 21st century do not capture the full potential impact of the changes triggered in the Antarctic Ice Sheet. 

Despite some biases and shortcomings, these results have demonstrated promising capabilities that UKESM1.0-ice has for further research into Earth system projection simulations. Those results also hint at some important areas of future work needed to reduce the sea level rise projection uncertainty. Among them are salinity bias reduction in the Ross and Weddell shelves, improvements of ocean model simulation in the Amundsen Sea and enhancement of key ice sheet model components.


  1. Fox-Kemper, B., H.T. Hewitt, C. Xiao, G. Aðalgeirsdóttir, S.S. Drijfhout, T.L. Edwards, N.R. Golledge, M. Hemer, R.E. Kopp, G. Krinner, A. Mix, D. Notz, S. Nowicki, I.S. Nurhati, L. Ruiz, J.-B. Sallée, A.B.A. Slangen, and Y. Yu, 2021: Ocean, Cryosphere and Sea Level Change. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 1211–1362, doi:10.1017/9781009157896.011.
  2. Siahaan, A., Smith, R., Holland, P., Jenkins, A., Gregory, J. M., Lee, V., Mathiot, P., Payne, T., Ridley, J., and Jones, C.: The Antarctic contribution to 21st century sea-level rise predicted by the UK Earth System Model with an interactive ice sheet, The Cryosphere Discuss. [preprint], https://doi.org/10.5194/tc-2021-371, in review, 2021.
  3. Smith, R. S., Mathiot, P., Siahaan, A., Lee, V., Cornford, S. L., Gregory, J. M., Payne, A. J., Jenkins, A., Holland, P. R., Ridley, J. K., and Jones, C. G.: Coupling the U.K. Earth System Model to dynamic models of the Greenland and Antarctic ice sheets, J Adv Model Earth Syst, https://doi.org/10.1029/2021MS002520, 2021.
  4. The IMBIE Team: Mass balance of the Antarctic Ice Sheet from 1992 to 2017, Nature, 558, 219–222, https://doi.org/10.1038/s41586-018-0179-y, 2018.