WHEN LISA SLOAN began her modeling studies almost a decade ago, she found Scott Wing's data intriguing. She wanted to know what might have caused the warm winters Wing and others described. At the time, scientists assumed that warm ocean temperatures caused the warm inland winters of 55 million years ago, but no one had tested the assumption. Sloan decided to see if warm sea surface temperatures in a global climate model would cause the ancient conditions paleontologists described.
      The model she used was developed at the National Center for Atmospheric Research and runs on a supercomputer. It's a "terribly complex model developed by teams of dozens of scientists," she says. Climate researchers use the same type of model to predict today's greenhouse warming or El Nino effects. The model is based on the physics of the oceans and atmosphere. Researchers tuned it to simulate modern-day weather conditions and longer-term climate patterns.
      Like the fossil studies, then, paleoclimate modeling relies on the assumption that the present reflects the past -- in other words, that the physical laws linking the land, ocean, and atmosphere today also operated 55 million years ago.
      Sloan enters into the computer such information as the geographic location of land and sea, the elevation of the land, the presence or absence of ice sheets, the types of vegetation and soil, how much energy arrives from the sun, and the composition of the atmosphere. Then the model planet starts spinning and goes into orbit around the sun. To phrase it simply, Sloan says, "I put in all that stuff and let it compute."
      The computer then begins to crunch the numbers. It divides the Earth's surface into 16,200 squares and keeps track of approximately 75 variables for each square. It divides the atmosphere into 18 layers, giving itself almost 22 million points to calculate. It calculates and records these points every 30 minutes on the model Earth. (Time on the model Earth is defined by its rotation just as it is on the real planet -- one spin of the model Earth takes one day. The computer records the variables 48 times in one model day.)
      As the model planet spins, the computer tracks variables such as average annual temperature until the Earth has calmed down from the initial shock of unusual conditions Sloan supplied it with. It typically takes 12 model years for variables such as average annual temperatures to stop fluctuating in the computer model Sloan uses. Then Sloan lets it spin for a few more years. She analyzes the climate variables the computer produces for those years only. The entire process takes about 300 computer hours on a Cray supercomputer.
      When Sloan tested the idea that warm ocean temperatures could cause mild inland winters, the model failed to "predict" the climate conditions the fossils suggested. "I tried all kinds of outrageous temperatures in the ocean and still got cold winters," she says.
      Many paleontologists reacted strongly to her results, insisting the model must be at fault. Despite the backlash, Sloan thinks her results spurred on fossil gatherers who otherwise might have let their finds get dusty in desk drawers. They now felt invigorated to take a second look. Sloan says, "It brought more fossil data out and gave them a framework to put their data in and a windmill to tilt at."

WING AGREES that Sloan's 1990 paper did just that: it gave him a new way to think about the climate of the Eocene. Before the modelers started analyzing the reasons behind warm inland winters, Wing's main concern had been studying ancient plants and how the climate had affected them. Sloan's work made him realize that the implications of the paleoclimate extended beyond its effects on the local flora to understanding the global climate. "The models changed my thinking about why I do what I do," he says.
      He now focuses more on climate, in addition to his continuing work on the plants themselves. He hopes to find more fossil sites in continental interiors to either support or alter the current understanding of the Paleocene and Eocene climate. While at UC Santa Cruz, Wing is also working with Sloan on modeling studies.

IN THE YEARS following her 1990 paper, Sloan continued to seek the cause of the early Eocene's warm winters in continental interiors. Inspired by a controversial suggestion that high levels of greenhouse gases may have existed 55 million years ago, Sloan modeled three scenarios: the early Eocene world with preindustrial levels of carbon dioxide, double preindustrial levels, and six times preindustrial levels. None of these scenarios could predict both Wyoming's warm winters and the ocean temperatures, implying that greenhouse gases alone could not have caused that epoch's climate conditions.
      However, in combination with some other trigger, greenhouse gases may been responsible for the Eocene's warmth. To clarify what role -- if any -- greenhouse gases played in the climate warming and mild winters of 55 million years ago, Sloan needs more data about that epoch's greenhouse gas levels.
    Now she and Wing are changing the continent boundaries in the model, since ocean levels were different and the continental plates have moved in the past 55 million years. The two are also focusing on regional geography. Sloan recently added a large lake in her model of western North America and found warmer winters near the lake. Again she concludes that she needs more data -- in this case, about regional land elevation and where lakes were historically.
      Clouds are another factor that may have been important 55 million years ago. In recent work, Sloan has tried adding high clouds over the polar regions, and she finds that winters in mid- to high latitudes are much warmer when she does this. The model suggests they could be significant, but clouds' importance in the Eocene's climate will be hard to judge until scientists learn more about their location in ancient times. Researchers continually update the models to make them predict today's climate more precisely, in hopes of predicting future climes.
      But an inherent weakness in using models to study ancient or future climates remains: the models are tuned to today's climate. "It's like you've tuned a car to run really well at sea level," Sloan says, "but the minute you take it up to 9,000 feet, it's a dog." The added complication with climate models is that, unlike cars, they don't announce their struggles by sputtering and overheating. Researchers must seek out places where the models fail, and then improve them. "That doesn't give me and other people complete confidence that they're really accurate for other time periods," she says. "I've never considered that the models were completely right."
      A model's best use, then, is testing the potential merit of hypotheses.
      Today, few scientists disagree with Sloan that warm oceans could not have single-handedly warmed the climate interiors. The model results point to gaps in geologic and fossil data. No paleontologist or geologist has yet found evidence that clearly suggests a cause for the early Eocene climate. If Sloan manages to simulate the environment that paleontologists describe, she will still have no proof that those conditions actually existed. "Even if I did find a combination of model factors that produced the climate conditions dictated by the fossil evidence, it would take you right back to the geologic data," she says. Paleontologists' next step would then be to look for more crocodiles and palm trees at midlatitudes on continents not yet examined, such as Asia. Then fossil hunters could throw another challenge to modelers.
      "I tend to keep an open mind about how good the model is," she says. "The data interpretations aren't perfect either. You're looking at two imperfect worlds and trying to get them to match." Sloan doubts these two worlds ever will coincide. But by working together, scientists like Sloan and Wing seem likely to bring them closer together.

<< previous || next >>