Climate scientists testing a new mathematical and statistical method that converts projections of future climate outcomes in a warming world into reliable probabilities have found there is a significant chance the Arctic could be ice-free even if world leaders meet the Paris targets of 1.5°C and 2°C.

The research using 31 climate models published today in Nature Communications calculates there is a 6% (1 in 17) chance of an ice-free Arctic at global temperatures 1.5°C warmer than the preindustrial era rising to a 28% (1 in 4) chance at 2°C. At 2.4°C the odds go up to 50%.

“Currently, if every country meets its agreed commitments, it is estimated global temperatures are on track to be 3.1°C – 3.5°C above pre-industrial temperatures, which means the probability of an ice-free Arctic is much greater than 50:50,” said one of the paper’s authors, Prof Jason Evans from the ARC Centre of Excellence for Climate Extremes. 

“These results emphasise the importance of staying within the Paris Agreement targets as without that the chances of an ice-free Arctic increase quickly. This is a fundamental shift in the climate system that will have global consequences.” 

“This is just one example of how this new method can present climate change data in a form that is understandable and useful to decision makers. Probabilities can be calculated for sea level rise, or many other variables produced by climate models.”

Climate models are not developed independently. Often sharing computer code, techniques and assumptions. Until now there was no solid statistical theory that dealt with dependant models and converted their results to probabilities. To solve this problem, the research team has developed an entirely novel statistical method.

In the new method climate model projections are considered hypothesis of the future climate state. The method then calculates the probability of each hypothesis and their combination. Importantly, the method does not assume that the hypothesis excludes one another when performing this calculation. As well as considering the degree of dependence between models (or hypothesis), it also considers the skill at capturing real-world observations with better performing models getting higher probabilities.

“Up to now, there was no established mathematical framework to assign probabilities on non-exclusive theories,” said Dr Roman Olson, the lead author and researcher at the Institute for Basic Science, Center for Climate Physics (ICCP) in South Korea. 

“While we only tested the new approach on climate models, we are eager to see if the technique can be applied to other fields, such as stock market predictions, plane accident investigations, or in medical research.”