GFDL - Geophysical Fluid Dynamics Laboratory

Change in future climate due to Antarctic ice melt

November 19th, 2018

Benjamin Bronselaer, Michael Winton, Stephen M. Griffies, Ronald J. Stouffer, William J. Hurlin, Keith Rodgers, Joellen L. Russell. Nature. DOI: 10.1038/s41586-018-0712-z

Ice sheet melt is a known neglected forcing in climate model simulations, contributing to uncertainties in climate projections. This is the first study to directly implement estimates of Antarctic ice sheet melt in a climate simulation, showing the actual change in climate projections due to the freshwater input. The authors used a large ensemble to confidently separate the freshwater signal from natural variability and show when we can expect these freshwater-induced effects to become significant.

The study shows that adding Antarctic ice sheet melt estimates to GFDL’s Earth System Model (ESM2M) significantly alters future climate projections. The addition of ice sheet melt to a 21st century climate projection reduces global warming. However, this comes at the more significant cost of potential increased future sea level rise through increased ocean warming. We also show that the ice sheet melt causes more precipitation just north of the equator, and less precipitation south of the equator, due to a Northwards shift in the Inter-tropical convergence zone. This means reduced future drying in areas like Central America, but increased drying in Australia. The authors propose that the freshwater input associated with the ice melt should be included in all future climate simulations to improve 21st century projections.

Observations over the satellite era show a positive trend in Southern Ocean sea ice cover. This observed trend can be simulated without ice sheet melt through natural variability, but it is far more likely to occur in climate model experiments, in a scenario with ice melt. The ice sheet melt also causes sea ice expansion until the middle of the century, suggesting that the observed trend is also likely the beginning of a longer trend.

To determine the climate impact of Antarctic ice sheet melt, the authors took a previously published estimate of 1950-2100 Antarctic Ice sheet melt and applied it directly to 10 members of a 30-member simulation using ESM2M (RCP8.5). The freshwater is put in at the ocean’s surface, in the three closest grid boxes around Antarctic coast. It is has no seasonal cycle.

Time series of the anomalies in a, global-mean surface air temperature anomaly, b, difference between Northern and Southern hemisphere precipitation, c, Southern Hemisphere sea-ice extent, and d, ocean temperature around the Antarctic coast at 400m depth, relative to the 1950–1970 mean. Orange shows the standard ensemble with out meltwater and blue shows the meltwater ensemble. Solid lines show ensemble means, the dark shading shows the 95% uncertainty in the mean and the light shading shows the full ensemble spread of 20-year means. The solid black line shows the difference between the orange and blue lines, and the applied meltwater flux is shown in grey (scaled to the mean of the final five years of the meltwater-induced SAT anomaly). The green bar indicates the period when the standard and meltwater ensembles diverge. The inset in panel c shows the period 1980–2020, with the double black line showing the observed sea ice extent anomaly from the National Snow and Ice Data Center (NSIDC)46, relative to the 1980–2000 mean, and the thin grey line showing the unsmoothed observations. The inset in panel d shows the 2080–2100 meltwater-induced anomaly in the ocean temperature around the Antarctic coast at a depth of 400 m, relative to the standard ensemble without meltwater. Hatching indicates where the anomalies are not significant at the 95% level.