Hyperwarming

Studies suggest that long-term climate models up to the year 2300 are missing key positive feedbacks that could send global temperatures towards levels high enough to melt the ice, if not over the entire Gamburtsevs, then at least large parts of even Antarctica for the first time in over 30 million years:

In particular, the release of methane from melting Arctic permafrost has not yet been factored in. Methane is a potent greenhouse gas, but remains in the atmosphere for only 10 years on average before it reacts with hydroxyl radicals in the air to form CO2. However, a large release of methane from melting permafrost could swamp the hydroxyl supply, allowing the methane to linger in the atmosphere for 15 years or more, further amplifying the warming. 

Some feedbacks never before considered might also come into play... In the future oceans may store less carbon. Normally some atmospheric carbon is lost at sea, buried in the carcasses of tiny marine animals. But sediment from the Eocene contains little carbon, suggesting that this process failed during the last hothouse...

Reversing this?