A Brief Look at a Snowball Earth: The Sturtian Glaciation of the Neoproterozoic Era (750 Ma)

Abstract

Introduction

With the debate over global warming capturing the attention of many, it is not surprising that a great deal of current climate research is aimed at understanding the causes and effects of warmer climates. However, despite the likelihood that the 21st century will be an exceptionally warm century we are technically still in the midst of an ice age – the Pleistocene – that has persisted for nearly two million years. The Pleistocene ice age has been the focus of many climate studies that have helped us to better understand not only cold climates, but the Earth’s climate system in general. However, just as there have been periods in Earth history that were warmer even than what we expect from global warming, the Earth has experienced ice ages that were far colder than the Pleistocene.

Background

Between 542 million and 1 billion years ago, during the Neoproterozoic Era, the Earth twice dipped into deep freezes that most geologists consider to have been among the coldest climates in the history of our planet. A variety of evidence suggests that Earth experienced two broad intervals of widespread glaciation: the Sturtian glaciation, which occurred around 750 Ma (Ma=million years ago), and the Marinoan glaciation, which occurred around 635 Ma (see Figure 1). One of the more remarkable features of these glaciations is the determination, based on the characteristics and distribution of certain sedimentary rocks (see, e.g., Figure 2) and other geological data, that continent-scale ice sheets existed at sea level within 10 degrees of the equator – equivalent to the modern-day latitude of Costa Rica. Because the extreme cold conditions are thought to have produced snow and ice cover over much of the Earth's surface, the Sturtian and Marinoan glaciations have become known as "snowball Earth" intervals.

The possible occurrence of ice sheets in the tropics was a controversial topic until fairly recently, when data clearly supporting their existence came to light (Sohl et al., 1999). Still controversial is the question of whether the tropical oceans were also totally covered with sea ice during these extreme glacial intervals (e.g., Hoffman et al., 1998; Hoffman and Schrag, 2002). Proponents of total or near-total freeze-over of the tropical oceans – the "hard snowball" scenario – have argued that such occurrences had an enormous impact of the subsequent evolution of multicellular life on Earth, and in fact may have been the trigger for the "Cambrian explosion" of life forms that were the ancestors of much of modern multicellular life (Figure 3). Other researchers (e.g., Hyde et al., 2000; Chandler and Sohl, 2000) have argued for a very cold but less extreme climate, the "slushball" scenario, in which ice cover would have been extensive but broad areas of open ocean would have remained at mid to low latitudes. The broader ramifications of the glaciations' effects have thus provoked a great deal of interest in understanding just what climatic conditions might have been like during the Neoproterozoic glacial intervals, and what climate forcings may have been reponsible for producing such extremely cold conditions.

Snowball Earth: Evolution of a Hypothesis

The controversy over the possible occurrence of extreme glaciation in the Neoproterozoic has a long history. Brian Harland (1964) was the first scientist to propose a “worldwide glaciation” nearly 40 years ago, by combining observations about the widespread distribution of glacial deposits with paleomagnetic data that suggested that at least some of the glacial deposits were laid down in seas at low latitudes. Harland's proposal was the subject of much debate over the years, but basically remained unchanged until Joseph Kirschvink of Caltech developed a more elaborate "snowball Earth" hypothesis (Kirschvink, 1992). According to Kirschvink, the bizarre existence of continental-scale ice sheets at low latitudes could be accounted for if large areas of land were preferentially located in mid- to low latitudes; so from his perspective, the ice sheets simply grew where there was an available surface with sufficient snow accumulation. A concentration of land area in the tropics might also have enhanced global cooling by reflecting a greater amount of incoming solar radiation back into space, as land albedo values would have been higher than the ocean's. The mid- to high latitude oceans would be covered by pack ice, which Kirschvink suggested would reduce evaporation from the sea and cut off oceanic currents from wind patterns, inhibiting exchange of oxygen between the ocean and atmosphere and causing the ocean to stagnate. In this way, the "snowball Earth" hypothesis could also account for the renewed appearance of iron formations in the geologic record: with time, the stagnant ocean bottoms would become anoxic, accumulating reduced iron until ocean circulation was re-established and oxidized iron could be deposited in.

In 1998, Paul Hoffman of Harvard University and his colleagues extended the snowball Earth hypothesis to explain some unusual carbon isotope values in post-glacial cap carbonates associated with glacial deposits in Namibia (Hoffman et al., 1998). Taking the hypothesis considerably further than Kirschvink did, Hoffman et al. envisioned an ocean that was completely frozen over, cutting off any exchange of gases between the atmosphere and ocean. They suggested that recovery from this deep-frozen state occurred once atmospheric CO2, released into the atmosphere through volcanic outgassing, built up to high enough levels that permitted the greenhouse effect of the CO2 to overcome the albedo effect of the extensive snow and ice cover. The highly elevated levels of CO2 in the atmosphere would then briefly fuel a very hot climate in the aftermath of the snowball Earth, until the global carbon cycle could once again fall into equilibrium (Hoffman and Schrag, 2002).

The properties of the ocean and atmosphere described by the snowball Earth hypothesis are not a part of the geologic record; they can only be inferred from a particular interpretation of the sedimentary deposits left behind. For difficult problems such as understanding the events associated with a snowball Earth state, climate simulations using a GCM offer the best way to evaluate whether a particular scenario is possible, given what we know of climate dynamics.

EdGCM Simulations of Snowball Earth Intervals

Paleoclimate modeling does present some unique challenges not posed by climate studies of the near future. Simulations of any time deep in Earth history need to take into account a sun that shone less brightly than at present, as well as sometimes radically different continental configurations. For the Archaean and Proterozoic eons in particular (before 542 Ma), there are also little or no data to constrain other major forcings we might want to examine, such as atmospheric CO2 levels. In such cases, the values we select for these forcings are arbitrary, but are reasonable estimates based upon other climatological or geological considerations.

For the purposes of the current Sturtian simulation, we altered key boundary conditions as follows:

* Solar Luminosity. As mentioned previously, models of stellar evolution have suggested that a G-type yellow star, such as the Sun, should have been less luminous earlier in Earth’s history (Figure 4). By the Neoproterozoic, the Sun would have increased its energy output but would still have been between 7% and 6% less luminous than today; we opted for a 6.19% reduction in luminosity for our Sturtian experiment. Solar radiation is reduced in the model by decreasing the total amount of shortwave radiation entering the top of the atmosphere, and is proportionally reduced at all wavelengths.

* Paleogeographic Distribution. We developed a paleocontinental reconstruction for the Sturtian glacial interval based upon the available paleomagetic data and other geologic constraints (Figure 5). Such reconstructions are necessarily tentative, as reliable paleomagnetic data are not abundant and age constraints on the relevant rocks are not well defined. However, the radically different continental confirguration for the Sturtian interval, as compared with modern geography, provides an opportunity to test the possible effects of varying land distribution on climate.

* Atmospheric CO2. The extremity of the Neoproterozoic ice ages suggests that these particular periods were times of CO2 drawdown. Previous simulations run for the younger Marinoan glaciation (Chandler and Sohl, 2000) suggested that reducing only CO2, even to a level of just 40 ppm, was still not quite sufficient to make the Earth cold enough to support continental-scale ice sheets in tropical latitudes. However, since we are using a slightly lower solar luminosity for this Sturtian experiment compared to the previous simulations, we opted to start modifying atmospheric CO2 with a "reasonable" value before proceeding to more extreme levels. The atmospheric CO2 level used in this simulation is 140 ppm, which is equivalent to half the accepted pre-industrial value.

* Ocean Heat Transports. The transport of heat by ocean circulation is critical to the distribution of temperatures on the planet. Today the oceans transport heat, on an annually averaged global basis, away from the tropics and subtropics and into the middle and high latitudes. For this Sturtian experiment. we have simulated the potential effects of a 50% decrease in poleward ocean heat transports, using ocean heat fluxes that yield zonally-averaged, meridional transports comparable to modern values. (The transports are necessarily modified since the Neoproterozoic ocean basin configurations are much different than modern.) Previous simulations (e.g., Chandler and Sohl, 2000) have shown that reducing ocean heat transports allows the polar regions to cool to a greater extent by sequestering heat in the tropics.

Analysis of Results

The likelihood of forming continental ice sheets at low latitudes depends largely on the existing temperature and moisture regime over low-latitude land areas; for an ice sheet to begin growing, there need to be at least some continental regions exist where annual snow accumulations withstand summertime melting. The extent to which sea ice develops is initially a function of temperature, but as it grows it forms a platform for the accumulation of more high-albedo snow, which can lead to further cooling. For a snowball Earth simulation, then, the climate variables of greatest significance include the surface air temperature, and snow and ice cover.

In order to determine just how much change we are introducing to a simulation by modifying the forcings, we also do "control runs" in which forcings such as solar luminosity, atmospheric greenhouse gases and ocean heat transports are set at their modern (1958) levels (i.e., both solar luminosity and ocean heat transports are at 100%, and the atmospheric CO2 level is 315 ppm). For paleoclimate experiments, we also do a control run with the appropriate paleogeography to make further comparisons simpler.

When we examine the annual average surface air temperatures of the Modern and Sturtian control runs (Figures 6 and 7), we find that the two runs are similar with respect to zonal (parallel to latitude) temperature patterns, with the Sturtian control run slightly warmer in both the tropics and at the South Pole. The South Pole is likely warmer for the Sturtian control run because it does not have a polar land mass like Antarctica that is surrounded by a climatically-isolating circumpolar current. The cause of the tropical warming may be directly related to the clustering of the land masses in low to mid-latitudes. This continental configuration appears to encourage zonal (Figure 8) rather than meridional (parallel to longitude; Figure 9) wind circulation, which in turn inhibits atmospheric heat transport away from the tropics.

Moving on to the Sturtian snowball Earth simulation, it is immediately obvious that this run (Figure 10) is significantly colder than the Sturtian control run. A difference (or anomaly) map comparing the two Sturtian runs (Figure 11) shows clearly that the snowball Earth run ranges between 3 to 40 degrees C colder than the control, although the maximum cooling is not at the poles as one might expect. Instead, the maximum cooling occurs in the upper mid-latitudes and lower polar latitudes, in a zonal pattern bracketing the continents. In this case, the reduced ocean heat transports appear to have had the anticipated effect of the cooling the higher latitudes.

However, we note that the annual average temperatures on the equatorial continents, such as Laurentia, are above the freezing point over broad areas (Figure 10). Such a temperature pattern is not conducive to the formation and persistence of continental ice sheets, or to the presence of extensive sea ice cover in the tropics.

Figures 12 and 13 show the annual average snow and ice cover for the Sturtian control and snowball Earth runs, respectively. Again, it is clear that the snowball Earth run has a significantly greater snow and ice cover, with the greatest increase over the control run in the upper mid-latitudes and lower polar latitudes (Figure 14). As with the surface air temperature, the reduced ocean heat transports appear to have enhanced cooling in the higher latitudes; here it is reflected as greater sea ice coverage. However, it is also again clear that the annual snow and ice cover does not even come close to what would be needed to establish continental ice sheets in the tropics, or to substantially freeze over the ocean.

At this stage of our investigation, we can surmise that, just like our previous simulations for the Marinoan glaciation (Chandler and Sohl, 1999), we have not yet identified the all the most likely forcings - or their appropriate values - to adequately simulate continental glaciation during a snowball Earth interval. It is also not possible then, purely on the basis of the model results, to determine what percentage of sea ice cover would be a realistic estimate for this time period. As far as the model results are concerned, the "hard snowball" vs. "slushball" debate is not resolved.

Additional experiments employing still more extreme forcings may yet be able to bring us closer to a better simulation of the Sturtian glacial interval. Given the effect that greenhouse gases have on the planet's ability to stay warm, a further reduction in atmospheric CO2, in conjunction with a reduction in atmospheric methane (CH4), may provide the key.

Summary of Results

We find that by reducing solar luminosity to 6.19% less than modern, bringing atmospheric CO2 levels down to 140 ppm, and cutting ocean heat transports by 50 %, the GCM is able to produce significant overall cooling of the Earth. The altered paleogeography also contributes slightly to the cooling effect at higher latitudes by enhancing zonal rather than meridional circulation. However, the modified forcings were still not sufficient to reproduce the first-order characteristics of the Sturtian snowball Earth glaciation, i.e., the annual average temperatures and degree of snow/ice cover needed on tropical continents to initiate or maintain large-scale ice sheets. Since we could not reproduce the continental aspect of the glaciation in this simulation, we cannot make any definitive statement about whether the oceans did freeze over, based solely on the model results. Additional simulations that explore more extreme forcings, such as additional decreases in greenhouse gas levels, may be able to shed more light on the controversy.

References

Chandler, M.A., and Sohl, L.E., 2000. Climate forcings and the initiation of low-latitude ice sheets during the Neoproterozoic Varanger glacial interval. Journal of Geophysical Research–Atmospheres, 105(D16): 20,737-20,756.

Dalrymple, G.B., 1991. The Age of the Earth. Stanford, California: Stanford University Press, 474 pp.

Gough, D.O., 1981. Solar interior structure and luminosity variations. Solar Physics, 74: 21-34.

Harland, W.B., 1964. Critical evidence for a grat Infracambrian glaciation. Geologische Rundschau, 54: 45-61.

Hoffman, P.F., Kaufman, A.J., Halverson, G.P., and Schrag, D.P., 1998. Snowball Earth. Science, 281: 1342-1346.

Hoffman, P.F., and Schrag, D.P., 2002. The snowball Earth hypothesis: Testing the limits of global change. Terra Nova, 14: 129-155. Available online at http://www.eps.harvard.edu/people/faculty/hoffman/TerraNova.PDF

Hyde, W.T., Crowley, T.J., Baum, S.K., and Peltier, W.R., 2000. Neoproterozoic "snowball Earth" simulations with a coupled climate/ice-sheet model. Nature, 405: 425-429.

Kirschvink, J.L., 1992. Late Proterozoic low-latitude global glaciation: The snowball earth. In: The Proterozoic Biosphere, A Multidisciplinary Study, edited by J.W. Schopf and C. Klein. New York: Cambridge University Press, pp. 51-52.

Sohl, L.E., Christie-Blick, N., and Kent, D.V., 1999. Paleomagnetic polarity reversals in Marinoan (ca. 600 Ma) glacial deposits of Australia: Implications for the duration of low-latitude glaciation in Neoproterozoic time. Geological Society of America Bulletin, 111: 1120-1139.