New paper shows climate models falsely predicted Antarctic sea ice would decline more than Arctic sea ice

A paper published today in Quaternary Science Reviews shows that climate models falsely predicted Antarctic sea ice would decline more than Arctic sea ice over the 20th century. In reality, Antarctic sea ice is currently near the highest levels recorded by satellites since 1979, and has almost completely offset the Arctic sea ice changes since 1979. The current Antarctic sea ice anomaly of +1 million square km is shown by the added red arrows below, and is about 2 million square kilometers above the decline falsely predicted by the mean of five climate models for the year 2000.

Note added red arrows show current sea ice anomalies in the Northern and Southern Hemispheres compared to 5 climate model simulations.

Fig. 7. Anomaly of annual mean sea ice area (in 10⁶ km²) simulated in five different models over the last millennium in the northern hemisphere and in the southern hemisphere. LOVECLIM1.2 results are in red, MPI-ESM-E1 in light blue, MPI-ESM-E2 in dark blue, MPI-ESM-P in violet, CCSM4 in green. The reference period is 1850–1980 AD and a 21-year running mean has been applied to the time series.

Fig. 6. Time series of the anomaly of ice extent (in 10⁶ km²) a) in the northern hemisphere in summer (September), b) in the northern hemisphere in winter (March), c) in the southern hemisphere in summer (March), d) in the southern hemisphere in winter (September). The results of ECHAM5/MPI-OM are in black (simulation covering the last 6000 years, Fischer and Jungclaus, 2011). Five simulations covering the last 8000 years with LOVECLIM1.1 using different model parameters are in green, yellow, red, magenta and violet (Goosse et al., 2007). The parameters that are varied are mainly related to the radiative scheme leading to climate sensitivities ranging from 1.6 to 3.8 K. An additional simulation with ECBILT-CLIO over the last 9000 years is in light green (Renssen et al., 2009). Compared to the other simulations, this longer simulation includes a forcing related to the presence of remains of the Laurentide during the early Holocene (effect on the surface albedo, elevation and freshwater forcing). Note that the plotted time series end in 1850 as some simulations does not include anthropogenic forcings. The reference period is 1000–1850 and a 51-year running mean has been applied to the time series. 

Modelling past sea ice changes



H. Goossea, , , D.M. Rocheb, A. Mairessea, M. Bergerd

• ^a Université catholique de Louvain, Earth and Life Institute, Georges Lemaître Centre for Earth and Climate Research, Place Pasteur 3, B-1348 Louvain-la-Neuve, Belgium


• ^b Laboratoire des Sciences du Climat et de l'Environnement (IPSL-CEA/INSU-CNRS/UVSQ), Gif-sur-Yvette, France


• ^c Cluster Earth & Climate, Faculty of Earth and Life Sciences, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands


• ^d Royal Institute of Technology, KTH Department of Mechanics, Stockholm, Sweden


• 
  

A dominant characteristic of the available simulations of past sea ice changes is the strong link between the model results for modern and past climates. Nearly all the models have similar extent for pre-industrial conditions and for the mid-Holocene. The models with the largest extent at Last Glacial Maximum (LGM) are also characterized by large pre-industrial values. As a consequence, the causes of model biases and of the spread of model responses identified for present-day conditions appear relevant when simulating the past sea ice changes. Nevertheless, the models that display a relatively realistic sea-ice cover for present-day conditions often display contrasted [opposite] response for some past periods. The difference appears particularly large for the LGM in the Southern Ocean and for the summer ice extent in the Arctic for the early Holocene (and to a smaller extent for the mid-Holocene). Those periods are thus key ones to evaluate model behaviour and model physics in conditions different from those of the last decades. Paleoclimate modelling is also an invaluable tool to test hypotheses that could explain the signal recorded by proxies and thus to improve our understanding of climate dynamics. Model analyses have been focused on specific processes, such as the role of atmospheric and ocean heat transport in sea ice changes or the relative magnitude of the model response to different forcings. The studies devoted to the early Holocene provide an interesting example in this framework as both radiative forcing and freshwater discharge from the ice sheets were very different compared to now. This is thus a good target to identify the dominant processes ruling the system behaviour and to evaluate the way models represent them.