Sea ice surrounding Antarctica has reached a record high and is covering more of the southern ocean than it was in the late 1970s, but is only a third of the magnitude of sea ice loss in the rest of the world.

The phenomenon demonstrate the "diversity and complexity" of the Earth's environment, NASA/Goddard Space Flight Center reported.

"The planet as a whole is doing what was expected in terms of warming. Sea ice as a whole is decreasing as expected, but just like with global warming, not every location with sea ice will have a downward trend in ice extent," said Claire Parkinson, a senior scientist at NASA's Goddard Space Flight Center.

The Arctic has lost an average of 20,800 square miles of ice since the 1970s while the Antarctic has gained a total of 7,300 square miles. On Sept. 19, 2014, the Antarctic sea ice reached 7.72 million square miles for the first time since measurements began.

The warming climate is believed to be changing weather patterns and bringing cooler air to some regions such as the Antarctic.

"Part of it is just the geography and geometry. With no northern barrier around the whole perimeter of the ice, the ice can easily expand if conditions are favorable," Parkinson said.

The air around the Antarctic Peninsula is warming and sea ice is shrinking in the Bellingshausen Sea (the Ross Sea, where sea ice is shrinking, is to the west), suggesting  a low-pressure system centered in the Amundsen Sea could be becoming stronger and more frequent. The pattern would sweep cold air from the Antarctic continent over the Ross Sea. 

The researchers are also looking for other explanations behind the phenomenon. Ice is melting on the edges of the Antarctic continent, making the water colder. Snowfall could also be pushing the ice below the water, keeping it colder.

"There hasn't been one explanation yet that I'd say has become a consensus, where people say, 'We've nailed it, this is why it's happening,'" Parkinson said. "Our models are improving, but they're far from perfect. One by one, scientists are figuring out that particular variables are more important than we thought years ago, and one by one those variables are getting incorporated into the models."