Why Did Ice In Antarctica Suddenly Appear 35 Million Years Ago? CO2, Says Study
    By News Staff | February 26th 2009 12:00 AM | 5 comments | Print | E-mail | Track Comments
    Ice in Antarctica suddenly appeared — suddenly in geologic terms being a little different than how we think of it — about 35 million years ago. For the previous 100 million years the continent had been essentially ice-free.  Even after Antarctica had drifted to near its present location, its climate remained subtropical but then, 35.5 million years ago, ice formed on Antarctica in only about 100,000 years, which is an "overnight" shift in geological terms. 

    What triggered the sudden shift?

    It wasn't global cooling, says Matthew Huber, assistant professor of earth and atmospheric sciences at Purdue University; or so some researchers believed.   "Previous evidence points paradoxically to a stable climate at the same time this event, one of the biggest climate events in Earth's history, was happening."

    Now a paper published this week in Science says they have evidence of widespread cooling and additional computer modeling of the cooling suggests it was caused by a reduction of greenhouse gases in the atmosphere.  So we need some  greenhouse gases - it's been 12,000 years since the last Ice Age and we'd like to keep it that way - just not too much.   

    "Our studies show that just over thirty-five million years ago, 'poof,' there was an ice sheet where there had been subtropical temperatures before," Huber says. "Until now we haven't had much scientific information about what happened."

    Before the cooling occurred at the end of the Eocene epoch, the Earth was warm and wet, and even the north and south poles experienced subtropical climates. The dinosaurs were long gone from the planet, but there were mammals and many reptiles and amphibians. Then, as the scientists say, poof, this warm wet world, which had existed for millions of years, dramatically changed. Temperatures fell dramatically, many species of mammals as well as most reptiles and amphibians became extinct, and Antarctica was covered in ice and sea levels fell. 

    History records this as the beginning of the Oligocene epoch, but the cause of the cooling has been the subject of scientific discussion and debate for many years. 

    The research team found before the event ocean surface temperatures near present-day Antarctica averaged 77 degrees Fahrenheit (25 degrees Celsius). 

    Mark Pagani, professor of geology and geophysics at Yale University, says the research found that air and ocean surface temperatures dropped as much as 18 degrees Fahrenheit during the transition.

    "Previous reconstructions gave no evidence of high-latitude cooling," Pagani says. "Our data demonstrate a clear temperature drop in both hemispheres during this time."

    The research team determined the temperatures of the Earth millions of years ago by using temperature "proxies," or clues. In this case, the geologic detectives looked for the presence of biochemical molecules, which were present in plankton that only lived at certain temperatures. The researchers looked for the temperature proxies in seabed cores collected by drilling in deep-ocean sediments and crusts from around the world.

    "Before this work we knew little about the climate during the time when this ice sheet was forming," Huber says.

    Once the team identified the global cooling, the next step was to find what caused it.

    To find the result, Huber used modern climate modeling tools to look at the prehistoric climate. The models were run on a cluster-type supercomputer on Purdue's campus.

    "That's what climate models are good for. They can give you plausible reasons for such an event," Huber says. "We found that the likely culprit was a major drop in greenhouse gases in the atmosphere, especially CO2. From the temperature data and existing proxy records indicating a sharp drop in CO2 near the Eocene-Oligocene boundary, we are establishing a link between the sea surface temperatures and the glaciation of Antarctica."

    Huber says the modeling required an unusually large computing effort. Staff at Information Technology at Purdue assisted in the computing runs.

    "My simulations produced 50 terabytes of data, which is about the amount of data you could store in 100 desktop computers. This represented 8,000 years of climate simulation," Huber says.

    The computation required nearly 2 million computing hours over two years on Pete, Purdue's 664-CPU Linux cluster.

    "This required running these simulations for a long time, which would not have been allowed at a national supercomputing center," Huber says. "Fortunately, we had the resources here on campus, and I was able to use Purdue's Pete to do the simulation."

    Additional members of the research team included David Zinniker at Yale; Robert DeConto and Mark Leckie at the University of Massachusetts, Amherst; Henk Brinkhuis at Utrecht University (Netherlands); and Sunita R. Shah and Ann Pearson at Harvard University. Zhonghui Liu, an assistant professor at the University of Hong Kong and a former postdoctoral fellow of Pagani's at Yale, was the study's lead author. 

    The research was supported in part by funding from the National Science Foundation.


    In numerical models iterated millions and millions of times, it is unlikely that any real result can be found. It becomes the philosophers stone that allows everyone to see what they want to see. It would also be nice to see some error bars on the estimates of global temperature.

    When we no longer have wind tunnels or test fixtures, i'll start believing computer models.

    There's a line between understanding the limitations of numerical models and how to calibrate results (especially with practically inconceivable volumes of data like climate results) and just being an old coot Luddite who thinks that 747s were ever actually built and tested in wind tunnels.

    The semiconductor industry is $250 billion and if you don't know the percentage of it that is actually tested by manufacturing or even put on a VNA before it tapes out, I won't spoil it for you.
    Ed Pardo
    While I find the article and the research interesting and well written  I have to disagree on modelling. This sounds like it is IPCC modelling which is known to be biased. The computer can only answer correctly for the data if the source code is correct.  The IPCC hypotheses are just that, not actually known to be correct and so far have not been able to recreate even recent climate accurately.  ie. GIGO  In order to prove this model you must demonstrate cause. ie. why did CO2 drop? At this point I know of no valid reason for a drop, CO2 or otherwise. ie. you cant put the horse before the cart.

    I would suggest running the same test on the data set for all known factors without GHGs included and see where else it might point. As far as I am aware, internal forcing is input as a constant, which it is not.  See the articles at this site and about Tectonic and Solar forcings.

    Aircraft are scale modelled first and tested in a wind tunnel, even the 747.
    I suggest there are a whole lot of "calibrations" in the IPCC models similar to the hurricane model predictions only a lot prettier to look at. Where i went to school, they might have thrown me out for like "calibrations". I personally doubt it is possible to model climate without an almost inconceivable scientific breakthrough. When i was in fluids and heat transfer classes, i was taught if my calculations were within 10% of the observed result for fluids and 30% in heat transfer, I had done my job. My prof. was a rhodes scholar and a graduate of the von Karman institute.

    I understand all planes are modeled in CFD and wind tunnels through multiple iterations before the first test article is built (and there are still surprises). In ship construction, they are CFD'd, model tested, and built (and there are always surprises). Both of these situations are infinitely simpler than a climate model and people loose lots of money when things aren't as predicted.

    I'd suggest they start looking more for anomalies in the earth's orbit, distance from the sun, solar activity, etc. I think there's plenty of evidence to show when that big "reactor" in the sky gets quiet, there's cold weather ahead.