The Arctic's rapidly melting sea ice continues to outrun even our most dire predictions for the future, and that's not out of line with the past.
A new and improved model, based on the last warm period in Earth's history, now suggests shallow pools of rain and melt water could bring about the end of summer sea ice considerably sooner than we thought.
If what's happening to the Arctic right now is anything like the last interglacial period, scientists say there's a chance it could be virtually free of sea ice in only 15 years.
"The prospect of loss of sea ice by 2035 should really be focussing all our minds on achieving a low-carbon world as soon as humanly feasible," says Louise Sime, a palaeoclimate modeller at the British Antarctic Survey (BAS).
Past projections from the United Nations' Intergovernmental Panel on Climate Change (IPCC) show summer sea ice persisting over a million square kilometres until at least 2050 or even beyond 2100, but recently, that timeline has begun to look far too optimistic.
Just this year, an analysis of numerous climate models found that even in the best case scenario, Arctic sea ice would slip below this mark, making the region virtually "ice-free" before the mid-point of the century.
Of course, this doesn't end the debate; there are still so many subtle factors to consider. But Sime says we can gain a better understanding of the future by looking back at Earth's last warm period, which began roughly 130,000 years ago and was much warmer than today.
While previous models of this time do not show ice-free summers in the Arctic, Sime and her colleagues found the opposite using improved model physics and incorporating subtle feedback systems in the climate.
Their model suggests the Arctic was very likely to have been ice-free during the summers of the last interglacial period, and this was enhanced by the presence of melt ponds - even more so than clouds or ocean currents, which have historically been given more weight in the warming Arctic.
Melt ponds occur in the Arctic's late spring and summer, when rain and melting ice and snow gather into shallow pools of blue. Slightly darker than the ice surrounding them, these scattered bodies of water reduce surface reflectance and absorb significantly more solar radiation than the frozen ground.
Some studies have shown melt ponds actually enhance surrounding ice melt and increase the potential of phytoplankton blooms in the ocean underneath. It can also make sea ice unsteadier and lead to fractures, revealing the ocean underneath and contributing to further heat absorption.
If what happened to melt ponds back then happens in the future, the authors predict summer sea ice might disappear from the Arctic in the next few decades - anywhere between 2035 and 2086.
And it's probably on the sooner side.
Half the models they looked at predicted sea-ice-free conditions between 2030 and 2040, and even in the worst case scenario, where we do nothing to curb emissions and populations and economies continue to grow unfettered, the authors found the latest disappearance of sea ice would come in 2066.
This study, of course, is not a direct measurement of today's sea ice melt, nor does it examine winter temperatures or seasonal changes in sea ice. It's a prediction, based on what happened in the hottest days of years past and what will happen in the future using our current understanding of the atmosphere, land, ocean and ice.
It's an imperfect estimate, but the findings do support more recent models that imply sea ice is on its way out far sooner than we had hoped for, in large part because of overlooked feedback systems such as melt ponds.
"The ability of the [new] model to realistically simulate the very warm LIG Arctic climate provides independent support for predictions of ice-free conditions by summer 2035," the authors conclude.
"This should be of huge concern to Arctic communities and climate scientists."
The study was published in Nature Climate Change.