Contact Us

Predicting Antarctic Climate Using Climate Models

Thomas J. Bracegirdle (1)*, Nicholas E. Barrand (2), Kazuya Kusahara (3), Ilana Wainer (4)

(1) British Antarctic Survey, Cambridge, UK. tjbra[at]
(2) School of Geography, Earth and Environmental Sciences, University of Birmingham, UK.
(3) Antarctic Climate and Ecosystems Cooperative Research Centre, University of Tasmania, Hobart Tasmania, Australia.
(4) Instituto Oceanografico, Universidade de Sao Paulo, Sao Paulo, Brazil

Climate models are the main tool for making quantitative estimates of how Antarctic climate may change over the 21st century. There is high agreement on some aspects of the predictions provided by models, but improvements in understanding are needed in key components of the Antarctic climate system, such as sea ice and coastal ocean-ice shelf processes. In the near term (on timescales of a few years) the climate change signal is small compared to natural cycles (associated with phenomena such as El Niño), the remote impacts of which on the Antarctic atmosphere are difficult to predict. In the longer term (on multi-decadal timescales) the reliability of climate model predictions is limited by uncertainty over human emissions pathways, the realism of climate models, and feedbacks between other elements of the Earth System (e.g. ice sheets).

Climate models and their use in simulating Antarctic climate

The climate of Antarctica is defined here to include the conditions of the atmosphere, ocean, snow and ice across the Antarctic continent and the surrounding Southern Ocean. At a given location a warming climate will be characterised by a long-term background shift (i.e. the signal) combined with shorter-term variability from year to year (i.e. noise). The World Meteorological Organization standard practice is to define the climate of a region based on a 30-year average of parameters of interest (such as temperature). On shorter time scales there are large variations associated with daily weather and major multi-year cycles such as El Niño/La Niña. Changes in climate should therefore be viewed from a multi-decadal perspective, which is the approach here. It is important to note however that there are also components of natural variability that operate over multiple decades and can in some cases affect the effectiveness of even 30-year means for tracking the baseline climate change signal [1].

The use of climate models in predicting future climate change is the main focus here. However, it is acknowledged that alternative approaches, such as the use of past analogues [2], can provide alternative views of future Antarctic climate. All major state-of-the-art climate models, which are produced at approximately 30 different modelling centres around the world, are founded on well-established physical laws of geophysical fluid dynamics, such as Newton’s laws of motion. However, limits on computer power mean that model calculations are generally still done at a large scale for global simulations, which involves representing the atmosphere as a series of boxes, which are typically at present 100 km across. This thus presents a major challenge in modelling smaller-scale phenomena (e.g. clouds) and physical characteristics (e.g. complex mountainous terrain) realistically.

Assuming that greenhouse gases will continue to increase in concentration to the year 2100, there is high confidence in a number of changes predicted. Under a medium-intensity assumption of human influence (i.e. an approximate doubling of carbon dioxide concentrations by that point) there is high agreement between the different climate models for the following:

  • Antarctic-wide terrestrial annual mean surface warming will occur (two thirds of climate models in the range 1.8°C to 3.3°C) [3].
  • Antarctic-wide terrestrial annual mean snow accumulation rate will increase (by 8% to 18%) [3].
  • Total Southern Hemisphere annual mean sea ice coverage will retreat (by 24% to 42%) [3].
  • The coastal sea-ice production will decrease, along with increased melting of land ice – both have been shown to cause a weakening of the primary global ocean circulation, the thermohaline circulation [4].
  • Southern Ocean water masses, such as the Antarctic Intermediate Water (AAIW) will warm and freshen as the densities at which water masses form become significantly lower. ([5], [6]). The AAIW is important for climate change because it is within this water mass that the highest concentration of anthropogenic CO2 is found [7].
  • Increases in snowfall will be accompanied by increased rates of ice discharge [e.g. [8],9]. Any negative contribution to sea level from increased snowfall may therefore be countered by faster rates of ice flow and increased discharge of inland ice to the ocean.

Challenges in estimating future climate

To account for the difficulties inherent in predicting human behaviour, the general approach taken in the climate science community is to consider a range of plausible ‘what if’ scenarios for anthropogenic greenhouse gas emissions, with no explicit judgement over which might be more likely [10]. The climate change estimates based on these scenarios are therefore referred to as ‘projections’ rather than predictions [e.g. see Figure 1].

Changes up to the mid 21st century will not necessarily follow the long-term warming trend [11]. A key consequence for policy makers is that regions that have been warming very rapidly in recent decades could potentially switch to a period of cooling on time scales of a few years, before background warming takes over. Currently there is significant research effort being directed at seasonal and decadal prediction in an effort to bridge the gap to longer-term climate-change timescales [12].

Figure 1. Climate model simulations of change in surface air temperature (temperature 2m above the surface) by the end of the 21st century (2069-2098) following a range of low (RCP2.6) medium (RCP4.5) and high (RCP8.5) scenarios of known important climate drivers such as greenhouse gas increases and stratospheric ozone recovery [10]. Changes are all relative to the period 1970-1999 in ‘historical’ climate model simulations with observed levels of greenhouse gases and other known natural and anthropogenic factors. Information from 41 climate models was combined based on a methodology detailed in [19]. The climate model dataset used was the Coupled Model Intercomparison Project Phase 5 (CMIP5) dataset, which fed into the most recent IPCC report. The 41 climate models are: ACCESS1.0, ACCESS1.3, BCC-CSM1.1, BCC-CSM1.1(m), BNU-ESM, CanESM2, CCSM4, CESM1(BGC), CESM1(CAM5), CESM1(WACCM), CMCC-CESM, CMCC-CM, CMCC-CMS, CNRM-CM5, CSIRO-MK3.6.0, EC-EARTH, FGOALS-g2, FIO-ESM, GFDL-CM3, GFDL-ESM2G, GFDL-ESM2M, GISS-E2-H, GISS-E2-H-CC, GISS-E2-R, GISS-E2-R-CC, HadGEM2-AO, HadGEM2-CC, HadGEM2-ES, INM-CM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR, MIROC-ESM, MIROC-ESM-CHEM, MIROC5, MPI-ESM-LR, MPI-ESM-MR, MRI-CGCM3, MRI-ESM1, NorESM-M, NorESM-ME.

Key challenges in representing the Antarctic climate system in climate models

  • The Southern Ocean. Like the cyclonic storms in the atmosphere, small cyclones/anticyclones (or eddies) occur within the ocean. A major issue for modelling the ocean is that these eddies are much smaller in size than their atmospheric counterparts and therefore small boxes are required to properly represent them mathematically [13]. At present there is not, in general, enough computational power available to do this realistically and the effects of these oceanic cyclones must be approximated.
  • Representation of atmosphere and ice over complex terrain. The Antarctic Peninsula is a region of particular importance due to the sensitivity of land ice to surface melting [14] and resulting impacts both on global sea level and regional ecosystems. However, the complex terrain of the high mountains cannot be accurately resolved in most climate models. To address this issue additional climate model studies are being conducted that focus computational effort on a specific location such as the Antarctic Peninsula and therefore enable the use of smaller grid boxes and improved representation of complex terrain [15].
  • Sea ice. Sea ice is one of the most difficult components to get right in climate models [16]. Since sea ice is in direct contact with both the atmosphere and ocean, almost all of the above issues with modelling Antarctic climate have an impact on reproducing realistic patterns of sea ice extent and thickness. Accuracy in reproducing observed sea ice extent is possibly the major concern when making judgements about the reliability of future climate projections from a given climate model.
  • Antarctic clouds. Little is known about cloud formation in the relatively pristine Antarctic atmosphere [17]. Clouds are a major control of surface temperature and therefore feed into the above-mentioned challenges of representing sea ice and the Southern Ocean in climate models. Field campaigns to measure cloud properties over the Southern Ocean and Antarctica are a key component of ongoing research with a particular focus on addressing climate model biases in representing Southern Ocean sea-surface temperatures.
  • Ice-ocean interaction at glacier fronts and ice-shelf cavities. Evidence increasingly shows ice fronts, floating ice tongue bases, and sub-ice-shelf cavities as key environments driving the dynamics of glacier and ice sheet mass loss (and therefore sea level rise). These environments are poorly observed and, presently, have limited or simplified representation in ice sheet models [4].

A SCAR Advisory Group on Antarctic Climate Change and the Environment (ACCE) compiles annual updates on the science of Antarctic climate, which are presented to the ATCM. These updates build on the main 2009 ACCE report ( [18]).


Antarctic Climate Change and the Environment (ACCE) report published by SCAR


Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report Working Group 1


SCAR report on 21st century climate projections to be produced by the Scientific Research Programme Antarctic Climate Change in the 21stCentury (AntClim21)


Approval of the IPCC Special Report on the Ocean and Cryosphere in a Changing Climate (SROCC)


SCAR report on 21st century climate projections to be produced by the Scientific Research Programme Antarctic Climate Change in the 21st Century (AntClim21)


Approval of the IPCC Sixth Assessment Report Working Group 1.

Other information:

1. T.Fan,  C. Deser, D.P. Schneider, Recent Antarctic sea ice trends in the context of Southern Ocean surface climate variations since 1950. Geophysical Research Letters 41(7), 2419-2426 (2014). doi: 10.1002/2014GL059239

2. P.A.Mayewski,  et al., Potential for Southern Hemisphere climate surprises. Journal of Quaternary Science 30(5), 391-395 (2015). doi:org/10.1002/jqs.2794

3. T.J.Bracegirdle,  W.M. Connolley, J. Turner, Antarctic climate change over the twenty first century. Journal of Geophysical Research-Atmospheres 113, (2008) doi: 10.1029/2007jd008933.

4. K.Kusahara, H. Hasumi, Modeling Antarctic ice shelf responses to future climate changes and impacts on the ocean. Journal of Geophysical Research-Oceans 118(5), 2454-2475 (2013). doi: 10.1002/jgrc.20166

5. M.Goes, M., et al., Changes in subduction in the South Atlantic Ocean during the 21st century in the CCSM3. Geophysical Research Letters 35(6)(2008).  doi: 10.1029/2007GL032762

6. S.M.Downes,  N.L. Bindoff,  S.R. Rintoul, Impacts of climate change on the subduction of Mode and Intermediate Water Masses in the Southern Ocean. Journal of Climate 22(12), 3289-3302 (2009). doi: 10.1175/2008JCLI2653.1

7. B.I.McNeil,  B. Tilbrook,  R.J. Matear, Accumulation and uptake of anthropogenic CO2 in the Southern Ocean, south of Australia between 1968 and 1996. Journal of Geophysical Research-Oceans 106(C12), 31431-31445 (2001). doi: 10.1029/2000JC000331

8. R.Winkelmann, et al., Increased future ice discharge from Antarctica owing to higher snowfall. Nature 492(7428), 239-+ (2012).  doi: 10.1038/nature11616

9. N.E.Barrand, et al., Computing the volume response of the Antarctic Peninsula ice sheet to warming scenarios to 2200. Journal of Glaciology 59(215), 397-409 (2013). doi: 10.3189/2013JoG12J139

10. M.Meinshausen,  et al., The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change 109(1-2), 213-241 (2011). doi: 10.1007/s10584-011-0156-z

11. E.Hawkins, R. Sutton, Time of emergence of climate signals. Geophysical Research Letters 39, L01702 (2012). doi: 10.1029/2011GL050087

12. M.Osman,  C.S. Vera,  F.J. Doblas-Reyes, Predictability of the tropospheric circulation in the Southern Hemisphere from CHFP models. Climate Dynamics 46(7), 2423-2434 (2015). doi: 10.1007/s00382-015-2710-2

13. R.Hallberg,  Using a resolution function to regulate parameterizations of oceanic mesoscale eddy effects. Ocean Modelling 72, 92-103 (2013). doi: 10.1016/j.ocemod.2013.08.007

14. N.E.Barrand,  et al., Trends in Antarctic Peninsula surface melting conditions from observations and regional climate modeling. Journal of Geophysical Research-Earth Surface 118(1), 315-330 (2013). doi: 10.1029/2012JF002559

15. J.M.van Wessem, , et al., Improved representation of East Antarctic surface mass balance in a regional atmospheric climate model. Journal of Glaciology 60(222), 761-770 (2014). doi: 10.3189/2014JoG14J051

16. J.Turner, et al., An initial assessment of Antarctic sea ice eExtent in the CMIP5 models. Journal of Climate 26(5), 1473-1484 (2013). doi: 10.1175/JCLI-D-12-00068.1

17. Bromwich, D.H., et al., Tropospheric clouds in Antarctica. Reviews of Geophysics 50(2012). doi: 10.1029/2011RG000363

18. J.Turner, , et al., Antarctic climate change and the environment: an update. Polar Record 50(3), 237-259 (2014). doi: 10.1017/S0032247413000296

19. T.J.Bracegirdle,  D.B. Stephenson, Higher precision estimates of regional polar warming by ensemble regression of climate model projections. Climate Dynamics 39(12), 2805-2821 (2012). doi: 10.1007/s00382-012-1330-3