Lecture 9: Snowball Earth

Abiol 574

Low Latitude Glaciations

• Paleomagnetic data indicate low-latitude glaciation at 2.3 Ga, 0.75 Ga, and 0.6 Ga

• Paleoproterozoic glaciations (~2.3 Ga) may be triggered by the rise of O2 and the corresponding loss of CH4

• Late Precambrian glaciations studied by Hoffman et al., Science 281, 1342 (1998)

The Great Infra-Cambrian Ice Age

W. Brian Harland & M.J.S. Rudwick, Scientific American 211 (2), 28-36, 1964

Courtesy of Joe Kirschvink

Using MagneticData to

DeterminePaleolatitudes

Courtesy of Adam Maloof

Polar

Equatorial

Periglacial Outwash Varves From the Elatina Formation, South

Australia

Courtesy of Joe Kirschvink

Late Precambrian Geography*

(according to Scotese)

Hyde et al., Nature, 2000* glacial deposits

Possible Explanations for Low-Latitude Glaciation

1. High obliquity hypothesis (George Williams, papers since1970’s)

– Equatorial glaciation predicted for obliquities exceeding 54o

– Explains how the photosynthetic algae and other life forms survived (polar regions remain ice-free)

– Difficult to explain dynamically– Inconsistent with evidence (just shown) for

high-latitude glaciation

Possible Explanations for Low-Latitude Glaciation

(cont.)2. “Slushball Earth” model (Hyde et al.,

Nature, 2000)– Tropics remain ice-free (+25o to 25o

latitude) photosynthetic algae survive– Climate state is metastable and, hence,

improbable– Requires large mountain ranges on paleo-

Australia that do not appear to have existed

– Has difficulty explaining cap carbonates…

Ghaub Glaciatio

n(Namibia)

Glacial Tillite

Courtesy of Joe Kirshvink

Maieberg“cap”

Hoffman et al.,Science, 1998

Maieberg Formation(400 m thickness)

• The cap is more commonly defined as the bottommost 1-5 m of fine- grained carbonate

Possible Explanations for Low-Latitude Glaciation

(cont.)3. “Snowball Earth” model (Joe

Kirshvink, 1990)– Easy to explain from a climatic

standpoint– Accounts for cap carbonates (indeed, it

predicts them!)– Accounts for reappearance of BIFs– “Hard Snowball” model (km-thick ice

everywhere) poses significant problems for the photosynthetic biota

Triggering a Snowball Earth

• Need to get CO2 levels below ~2.5 PAL at 600 Ma, for S/S0 = 0.95 (Hyde et al.)

• Possible ways to do this1. Continental rifting created new shelf area,

thereby promoting burial of organic carbon (Hoffman et al., Science, 1998)

2. Clustering of continents at low latitudes allows silicate weathering to proceed even as the global climate gets cold (Marshall et al., JGR, 1988)

Snowball Triggers (cont.)

3. Release of CH4 from methane clathrates increased Ts CO2 decreased. Then, when the clathrates were exhausted, the Snowball was launched (Schrag et al, G3, 2002)

4. Mid-Proterozoic CH4 levels were high because of a Canfield ocean (sulfidic, no O2), then they dropped because of an increase in either O2 or sulfate (Pavlov et al., Geology, 2003)

In any case, ice albedo feedback takes overwhen the polar caps reach some criticallatitude (near 30o)

Caldeira and Kasting, Nature (1992)

Modern Earth

Recovering from a Snowball Earth episode

• Volcanic CO2 builds up to ~0.1 bar over a time span of ~30 m.y. (old result—Caldeira and Kasting)– This estimate may be too low

(Pierrehumbert, Nature, 2004)– Our “thin-ice” model recovers 10 times

more quickly, i.e., in about 3-5 m.y.

Problem: CO2 may condense at the winter pole

R. Pierrehumbert, Nature (2004)

CO2 saturationtemperature forpCO2 = 0.2 bars(172 K)

Recovering from a Snowball Earth episode

• Once the meltback begins, the ice melts catastrophically (within a few thousand years)

• Surface temperatures climb briefly to 50-60oC– Post-Snowball temperatures would be 10-15o lower

in the thin-ice model

• CO2 is rapidly removed by silicate (and carbonate) weathering, forming cap carbonates

How did photosynthetic life survive the Snowball

Earth?• Refugia such as Iceland?

– Tidal cracks, meltwater ponds, tropical polynyas? (Hoffman and Schrag, Terra Nova, 2002)

• “Jormungand” model (Abbot et al., JGR, 2011)

• “Thin ice” model (C. McKay, GRL, 2000) – Tropical ice remains thin due to penetration

of sunlight

Jormungand state

• According to Abbot et al., liquid water remained present in a thin, wavy strip near the equator

• This model looks like a mythical Norwegian sea serpent that was big enough to circle the Earth and grasp its tail in its mouth

• Was it open water, though, or might this region have been covered with thin ice?

Abbot et al. (2011)

Courtesy of Dale Andersen

Antarctic Dry Valleys

McKay’s “thin ice”model was inspiredby his visits to theAntarctic lakes

Image from: Land ProcessesDistributed Active ArchiveCenter, USGS

http://LPDAAC.usgs.gov 

Lake Bonney (Taylor Valley)

Courtesy of Dale Andersen

• Photosynthetic life thrives beneath ~5 m of ice

McMurdo Sound dive hole

Courtesy of Dale Andersen

Ice thickness2.5-3 m

One of the intrepid explorers

Courtesy of Dale Andersen

Life Magazine, Dec. (2004)

Jellyfish photographedbeneath 2 m of Antarcticsea ice

Photograph byNorbert Wu

Windows through the ice (McMurdo)

Courtesy of Dale Andersen

“Hard” Snowball EarthIce Thickness

Fg

Ts -27o C (Hyde et al., 2000)

Toc 0oC

Let = thermal conductivity of ice z = ice thickness T = Toc – Ts

Fg = geothermal heat flux

z

Ice Transmissivity (400-700 nm)

C. McKay, GRL (2000)

Photosynthetic limit

Possiblesolution at

equator

Problems with McKay’s Model

• Unrealistic treatment of radiative transfer within the ice– Treated ice as being a pure absorber,

whereas it mostly scatters radiation in the visible

• Did not account for equatorward flow of sea glaciers (Goodman and Pierrehumbert, 2003)

“New” numerical model(Dave Pollard—Penn State)

• 1.5-D energy-balance climate model (EBM), similar to models developed by Budyko, Sellers, North, Caldeira and Kasting

• Fully coupled to dynamic sea glacier model– In contrast, Goodman and Pierrehumbert’s sea

glacier model was only weakly coupled to a climate model

• Realistic treatment of ice radiative transfer

Absorption Spectrum of Ice

Warren et al., JGR 107, 3167 (2002)

Visible IR

Two waveband treatment of ice radiative transfer

2-stream scattering model in the visible (0.3-0.7 m)– 60% of Sun’s energy in this wavelength region– Albedo dominated by scattering by bubbles and

brine inclusions within the ice– Single scattering albedo: 0 = kscat/(kabs+kscat)– Ice that forms slowly (as it would here) is very

clear• Fixed albedo of 0.5 in the near-IR* (changed)

– 40% of solar energy– Albedo dominated by specular reflection from

the ice surface

Sea Glacier Schematic Diagram

Ocean

Ice

Atmosphere

90oN or S

LatitudeEquator

To

Ta

qa

vh

hs Snow

Ts

Ts

• Ice is snow-covered at high latitudes where P-E > 0• Sea glacier flow follows Weertman (1957), Morland (1987), and MacAyeal (1996), as modified for global 2-D symmetry by Goodman and Pierrehumbert (2003)

Transition from Slushball to Snowball and back

No sea ice flow

Ice flow included

4 W/m2 = 1 CO2 doubling20 W/m2 = 5 CO2 doublings

= 32 PAL (~0.02 bar at 600 Ma)

Pollard and Kasting, JGR,2005

Collapse

Recovery

Solutionsfor no seaice flow

Bubbly ice(0 = 0.999)

Clear ice(0 = 0.994)

Sea ice flowincluded

Bubbly ice(0 = 0.999)

Clear ice(0 = 0.994)

Principal Modeling Results

• Ice remains thin (2-3 m) between +10o and −10o latitude for 0 < 0.995

• 4-6% of the incident sunlight is transmitted photosynthesis continues over a wide region

• The ice thins and the albedo drops as the surface warms system recovers in ~5 m.y. (as compared to 30 m.y. in the Caldeira and Kasting model and even longer in Pierrehumbert’s model)

Problems with the thin-ice model

• The solution is not robust because changing the input parameters slightly can result in Hard Snowball solutions (because the sea glaciers flow all the way to the equator)

• Tropical seas that are protected from sea-glacier flow could still have thin ice, however

The Mediterranean—a possible Snowball refuge

Conclusions• A thin-ice Snowball Earth solution is possible

under some circumstances. A more detailed physical model is required to say whether or not it is expected.

• Such a model does a good job of explaining cap carbonates (unlike the Slushball model)

• Photosynthetic life survives this catastrophe much more easily than in the hard Snowball model the paleontologists should like it