"We have provided a plausible mechanism for how circulation changes can impact Arctic sea ice, but it is difficult to determine causality with observational evidence alone, because of the feedbacks between sea ice and the atmosphere," the study said.
Researchers were able to conduct several model experiments to show how high-latitude circulation impacted sea ice.
The researchers found 30 to 50 per cent of the overall decline in September Arctic sea ice since 1979 could be accounted for by natural variability.
They analysed three factors that affect atmospheric circulation and, therefore, sea ice - temperature, humidity and downward longwave radiation - by combining an atmospheric general circulation model with an ocean sea-ice model and reanalysis data. They found that circulation changes are contributing to up to 60 per cent of the sea-ice decline.
The authors then focused on whether the atmospheric circulation changes are natural or caused by human influence, and found that about 70 per cent of the atmospheric circulation changes are due to natural internal variability.
These findings highlight that high-latitude circulation affects sea ice and that gaining understanding of trends over the decades may increase the ability to predict sea-ice coverage over seasonal to decadal timescales.
In an editorial in Nature Climate Change, research scientist Neil Swart said the two main components driving recent changes in Arctic sea ice were an overall long-term loss of ice in response to external forcings such as greenhouse gases, and shorter term, random changes due to internal climate variability.
"The challenge, until now, is that there has been no clear understanding of the relative contributions of human-induced warming versus internal variability to the observed long-term decline in Arctic sea ice," he said.
"Qinghua Ding and colleagues illustrate that around half of the observed summer-time Arctic sea-ice loss has been driven by naturally induced changes in large-scale atmospheric circulation."
He said the results do not call into question whether human-induced warming has led to Arctic sea-ice decline.
"A wide range of evidence shows that it has," he said.
"Rather, the implication is that Arctic sea-ice is less sensitive to human-induced forcing than if one assumes that all loss observed to date is anthropogenically driven."
Commenting on the paper, Victoria University climate scientist Professor James Renwick said this meant the Arctic and Antarctic "aren't so very different after all".
"The standard picture is that Arctic ice is disappearing at speed because of feedbacks in the climate system and the strong effects of human-induced warming overwhelming relatively weak natural variability," Renwick said.
"At the same time the picture is reversed for the Antarctic, where natural variability is very large and appears to drown out the anthropogenic signal - hence the small increase in sea ice extent over the past 30 years."
Renwick noted how paper reported natural variability was important in both hemispheres, including linkages to the tropics.
"It happens to have acted to accelerate the loss of sea ice in the Arctic, while it has acted to slightly increase the amount of sea ice around the Antarctic.
"It is still the case that the direct thermal effect of greenhouse gas increase is a more important player in the Arctic than the Antarctic, but both poles experience a lot of natural variability."
Further, Renwick pointed out that just because climate models didn't get the short-term sequence of climate evolution exactly correct, this didn't mean they were wrong.
"The real world and the model both have natural variability and a human-induced trend," he said.
"The model sequence of natural variations is bound to be different to what's observed, since it is basically random noise.
"If the amount of noise is large enough, the trend may be affected."
