Snowball Earth

In the early 1960s, Brian Harland, a geologist at Cambridge University, observed that rocks on several continents, dating from the Neoproterozoic era (approximately 800–680 million years ago), contain glacial debris. Some of the glacial debris included carbonate rocks, which are known to form in the tropics (e.g., in the presentday Bahama Banks). This conclusion later gained additional support from paleomagnetic data. One potential explanation is that the entire Earth was covered by ice and snow during the Neoproterozoic. This is known as the “snowball Earth” hypothesis.

One early problem was understanding how a global ice age could have commenced. During the 1960s, the Russian climate scientist Mikhail Budyko used a computer simulation to establish that a runaway ice-albedo feedback effect could lead to global glaciation. The term albedo refers to the amount of the sun’s energy that is reflected by the Earth’s surface. As glaciers grow in extent, they reflect more of the sun’s energy, which causes the atmosphere to cool. This in turn causes the glaciers to grow. Budyko showed that if the glaciers extended beyond a certain critical point, this ice-albedo feedback could lead to a global ice age.

A second obstacle was understanding how a global ice age could ever end once it began. In the early 1990s, Joseph Kirschvink of the California Institute of Technology observed that during a global ice age, the carbon cycle would shut down. Volcanoes sticking up through the ice cover would continue to add carbon dioxide to the atmosphere. Having nowhere else to go, the carbon dioxide would then accumulate over millions of years until a runaway greenhouse effect caused the ice to melt. One important rival to the snowball Earth hypothesis is the high obliquity hypothesis. If the tilt of the Earth’s axis had been much different during the Neoproterozoic, the poles could have received more solar energy than the tropics. If so, it would be possible to explain the evidence for glaciers in the tropics, without supposing that the entire planet had frozen over.

In his widely cited 1992 paper, Kirschvink also proposed an explanation for banded iron deposits observed in Neoproterozoic glacial debris. Iron is not soluble in seawater in the presence of oxygen. During a true snowball Earth episode, the oceans would have become deoxygenated over time. Iron from thermal vents would build up in the seawater. Then, when the ice finally melted and oxygen was once again exchanged between the oceans and atmosphere, oxidized iron would have been left along with the debris from the retreating glaciers.

During the 1990s, two Harvard scientists, Paul Hoffman and Daniel Shrag, gathered additional, highly suggestive evidence that seemed to favor the snowball Earth theory. They found that in many places, the Neoproterozoic glacial debris occurs right below thick layers of carbonate rock (which are known as “cap carbonates”), and they showed how Kirschvink’s proposal could account for this. During a snowball Earth episode, very large amounts of carbon dioxide would have built up in the atmosphere. As the ice receded and the carbon cycle resumed, large amounts of carbon would have been washed out of the atmosphere during storms and ended up in the form of carbonate rock on the ocean floor. More controversially, Hoffman and Shrag also studied the ratio of carbon-12 to carbon-13 isotopes in the cap carbonates. They argued that an unusual dip in the carbon isotope ratio signified a temporary shutdown of photosynthetic activity in the Earth’s oceans.

Challenges to the Snowball Earth Theory

One potentially serious challenge to the snowball Earth theory comes from paleontology. Today, most geologists agree that there were at least two major ice ages during the Neoproterozoic: the Sturtian, around 750 million years ago, and the Varanger, around 590 million years ago. The second of these episodes occurred shortly before the Cambrian explosion of metazoan life. However, a true snowball Earth episode would have killed off nearly all eukaryotic life, and it is not clear that there was enough evolutionary time for life to recover from a global ice age. Some scientists have used computer models to show that softer versions of the snowball Earth episode might have been possible—for example, a mostly ice-covered planet with massive continental ice sheets in the tropics, but largely ice-free tropical oceans.

Although scientists generally agree that there was low-latitude glaciation during the Neoproterozoic, they continue to use a combination of fieldwork and numerical modeling techniques to work out the details. The snowball Earth scenario remains an intriguing live hypothesis.