Now, Penn State geoscientists believe that these glaciers could only have formed after the Earth's oceans were entirely covered by thick sea ice.

"There is strong geologic evidence of tropical glaciation at sea level during those times," Dr. David Pollard, research associate, Penn State College of Earth and Mineral Sciences' Environmental Institute, told attendees at the spring meeting of the American Geophysical Union Tuesday (May 29) in Boston. "We wanted to determine how low-level tropical glaciers could have formed."

Ice can accumulate in the tropics only if temperatures are below freezing or around freezing with large amounts of snowfall. Tropical glaciers exist today only on high mountain peaks such as the Andes and Mt. Kilimanjaro, and do not reach anywhere near sea level.

Pollard and James K. Kasting, professor of geosciences, first looked at the possibility that tropical ice sheets formed before the oceans completely froze into a snowball Earth, when equatorial oceans were still ice-free and could supply enough moisture for substantial snowfall.

During the lead-up to a snowball Earth episode, the Earth gradually cools because the amount of carbon dioxide in the Earth's atmosphere decreases. Relatively fast weathering of silicate rocks on large tropical landmasses causes this decrease that locks up carbon.

As the earth cools, the oceans begin freezing. The high reflectivity of the snow and ice that covers the northern and southern oceans reflects, rather than absorbs, the sun's heat and further cools the planet.

This cooling takes place slowly until the oceans are frozen to about 30 degrees latitude, or from the North Pole down to New Orleans, La. and from the South Pole up to the tip of South Africa.

"This is the coldest that the Earth can get before all the entire ocean surface freezes," says Pollard. "Beyond this, there is no stable point at say 20 or 10-degrees latitude: instead, the ice-reflectivity feedback becomes unstable and the system collapses rapidly to a snowball Earth with all oceans ice covered."

The researchers adjusted a global climate model, GENESIS, to the coldest point just before the collapse and used climate outputs of temperature and precipitation to drive a dynamic ice-sheet model.

They used paleomagnetic reconstructions of land mass distributions for 750 and 540 million years ago, but, because the locations of major mountain chains are unknown that long ago, they put mountains analogous to the Andes all around the edges of tropical land masses in their ice-sheet model.

"Ice sheets did form on the tops of these mountains," says Pollard. "However, the ice sheets never flowed down to sea level, where we find glacial deposits. Tropical temperatures were still too warm and melted the ice before it could flow down from the mountains."

The researchers conclude that it is unlikely that tropical sea level glacial deposits formed before the collapse into snowball Earth. However, having them form after the oceans freeze also seemed problematic because once the oceans are frozen, the rates of precipitation decrease drastically, to only a few millimeters per year.

"However, in further simulations with the global climate model for full snowball conditions, snowfall did exceed evaporation of snow and ice in some land areas, allowing a slow build up of tropical ice sheets that would eventually flow to the sea," says Pollard. "It would have taken several thousand years to form big ice sheets this way, but since it takes several million years to reverse snowball Earth, there would have been plenty of time for the ice to form."

Also, snowfall rates would have been gradually increasing during that time as carbon dioxide built up. Researchers have estimated that it required a buildup of carbon dioxide by volcanic outgassing to 300 times today's levels to bring Earth out of snowball Earth, which accounts for the millions of years necessary to reverse the process.

Some scientists question whether life could have survived a full snowball-Earth episode, and therefore suggest that the Earth never passed beyond the critical point with sea ice down to about 30 degrees latitude.

However, the Penn State results imply that full snowball Earth must have occurred in order to produce the observed tropical glacial deposits at sea level. Others have suggested that oceanic life could have survived full snowball episodes below gaps in the ice around volcanic island, or in tropical oceans where sunlight may have limited sea ice thickness to only a few meters. - By A'ndrea Elyse Messer

[Contact: David Pollard, James K. Kasting, A'ndrea Elyse Messer]

30-May-2001