Winds of change in Arctic sea-ice sensitivity

New results suggest that climate models underestimate Arctic ice loss caused by atmospheric circulation changes compared with real life observations. Instead, models sort-of compensate for the lack of atmospheric wave motions by simulating too much ice loss in response to anthropogenic forcing.
Published in Earth & Environment
Winds of change in Arctic sea-ice sensitivity

What is Arctic sea-ice sensitivity?

Increases in atmospheric concentrations of  greenhouse gases released into the atmosphere by industry, agriculture and other human activities are clearly warming the planet. Perhaps the most tangible sign of this is the rapid melting of Arctic sea ice and the Greenland ice sheet. Reducing uncertainties around the future of the Arctic is of paramount importance because the Arctic is warming faster than the global average. Arctic sea ice loss per unit of CO2 emissions is known as Arctic sea ice sensitivity, the accurate quantification of which is still uncertain.

How to improve its estimates?

A common, but debatable, practice to evaluate the 'accuracy' of climate models' sea ice sensitivity is to directly compare the polar ice melt simulated by the models with the ice melt observed from satellite measurements. However, this approach is missing a step that has been overlooked in research to date. The main problem is that computer models of the climate system give much less weight to ice melting caused by planetary-scale atmospheric circulation compared to real observations. Therefore, we need to consider and account for that models may miss some mechanisms operating in nature when assessing the climate sensitivity of the Arctic.

How to quantify the winds' influence?

 Using a modelling approach, we have quantified atmospheric circulation-induced melting in a complex coupled ice-ocean-atmosphere-land climate model (Community Earth System Model v.2), which explicitly incorporates the physics of the melting of the Greenland ice sheet and Arctic sea ice. In this experiment, the modelled wind field was replaced by a wind field based on real observations from a so-called climate reanalysis, while the atmospheric concentration of carbon dioxide in the model was kept constant. This allowed to accurately separate wind-induced ice melting from the direct melting caused by carbon dioxide emissions. Arctic winds are responsible for up to 50 percent of the melting seen over the last four decades. On the contrary, the same figure is just over two percent in the climate models used in the latest report of the Intergovernmental Panel on Climate Change (IPCC). The extent to which carbon dioxide forcing is responsible for wind changes in the models does not influence our methodology, making this model evaluation technique quite general and informative.

Where do changes in the Arctic winds come from?

Our paper suggests that ocean-atmosphere interactions in the tropical Pacific generate planetary-scale wave motions that propagate into the Arctic (so-called teleconnections) and enhance the atmospheric circulation's rotational component. This creates something similar to a heat-dome over the Arctic, while enhancing ice melt. Models on the other hand simulate much weaker  teleconnections, instead, they compensate for the missing wind effect by 'over-emphasizing' the greenhouse warming. It is important to note that this in no way implies that the role of humans in causing climate change is in question, but rather that our article draws attention to the real dangers of overtuning models to replicate past climate change, possibly for the wrong reasons. We still are unsure of whether what portion of the observed winds might be caused by anthropogenic forcing, which remains a challenge to properly quantify and thus should motivate future research.

Future implications

 After accounting for the mismatch between observations and models, the rate of future ice melt predicted by the models slows down. This slow-down yields that the timing of the first sea-ice free Arctic Ocean is also delayed; by at least a decade compared to the previous estimates. These are direct consequences of the fact that the models miss the wind effect.

Fig. 1.  For three different emissions scenarios (Community Earth System Model 2, red; Max Plank Institute Grand Ensemble, blue; and 21 models from the Coupled Model Intercomparison Project CMIP6, black), cumulative probability density functions for the Arctic to become sea-ice-free are shown (solid lines). If atmospheric-circulation observations are accounted for, the functions shift (dashed lines; uncertainties indicated by thin dashed lines): all three ensembles signal a delay of roughly ten years for the Arctic to become sea-ice-free (dates below the horizontal axis).

Please sign in or register for FREE

If you are a registered user on Research Communities by Springer Nature, please sign in