SCIENTISTS concerned about global warming are looking to the past to clean their crystal ball.
They are forced to do this because their forecasts have been fuzzy. The computer programs they use to model climate give confusing results. They need a reality check.
To get it, some of the modelers are joining with a global community of experts in Earth's ecological history to see if their computer climate models can simulate what has already happened.
They call their effort the Paleoclimate Modeling Intercomparison Project. That's a big name for an unprecedented kind of climate research.
``The getting together of people who do climate modeling with people who study past climates is a first in our field,'' says meteorologist John Kutzbach at the University of Wisconsin in Madison. He adds that it's ``exciting and fun to see a real mix of [such] scientists working together for the first time.''
They need to help each other in order to understand what's happening to our climate today. Dr. Kutzbach explained in a recent briefing for science writers that climate forecasts suffer from lack of accountability. Their moment of truth is decades in the future. But when those same computer programs are used to hindcast the past, scientists know what the correct answer to the test should be. ``That's very comforting when you're working with climate models - believe me!'' Kutzbach says.
His Madison meteorological colleague Reid Bryson agrees. Dr. Bryson says he has a ``firm conviction that the past is the key to the future.'' He explains that ``if one cannot understand changes in past climates and their consequences, then one has no hope of predicting the future climates and environments.''
A major problem with present computer climate models is their inconsistency in handling global warming. The models used in different research centers around the world generally agree that a doubling of the atmosphere's carbon- dioxide concentration should raise Earth's average planetary temperature by about 2 to 4 degrees C over the next several decades. But they start squabbling among themselves when they try to predict what this would mean in terms that really matter, namely regional weather.
The Intergovernmental Panel on Climate Change - the world body specifically concerned with possible global warming - reviewed this situation last year. It reported that confidence in these regional estimates is low. ``This is a bit disconcerting as we're trying to make policy plans based upon these kinds of [climate] models,'' Kutzbach observes.
The new paleoclimate project will deal with this weakness directly. Project members are preparing for a meeting next fall to lay out a series of experiments to explore the differences among the climate-simulation models.
Every model will reconstruct the past using identical conditions of solar heating, glaciation, and other so-called climate-forcing influences. This ensures that any differences in what the models predict will be due to the way different modelers have written their computer programs and to the assumptions they have put into those programs.
As Bryson points out, each of these computer models is ``a formal statement of how the scientist thinks the real world works ... [and] is nothing more than the modeler's opinion.'' The comparison tests should highlight the differences in those opinions.
Project researchers also hope to discover how they should change the common opinions that shape all their models. They want to learn more about key elements, such as vegetation changes that they have been neglecting, and to discover unsuspected weaknesses.
Ample historical data
There's a wealth of data about the past 21,000 years - the time period for this research - to help them. Archaeologists can link habitation patterns to environmental conditions. Lake sediments and tree rings reflect changing precipitation patterns. Ice cores such as those drilled from the Greenland ice sheet contain samples of ancient air and clues to atmospheric chemistry. Pollen remains can reflect local and regional vegetation. Scientists can use these and other types of ``proxy'' data to determine past climatic conditions even though no one directly measured the winds and rain or took the temperature.
In fact, Kutzbach notes that pulling together these diverse data into a coherent resource will be one of the major fruits of the project.
``This is the first time that the paleoecological data of the world is really being settled into one place,'' he says. He adds that this reflects the work of ``hundreds of people who studied their individual favorite lake or their individual favorite bog.''
In trying to make their computer simulations match what this data record shows, the modelers start with the assumption that the main factor driving climate change has been change in solar heating due to cyclical change in Earth's relation to the sun. The idea harks back to the ancient Greeks. But it is now identified with the Yugoslav astronomer Milutin Milankovitch.
There are three of these Milankovitch cycles:
* The tilt of Earth's spin axis away from vertical affects how high the sun appears in the sky. Now at an angle of 23.5 degrees, this tilt varies between 21.5 and 24.5 degrees over 41,000 years.
* Earth's orbit stretches and contracts every 100,000 years, varying between a more elliptical and a more nearly circular shape. This changes the difference between Earth's closest and farthest distance from the sun.
* Earth's axis wobbles like a top. It points to different places in the sky over about a 23,000-year period. This determines whether the summer or winter season of a hemisphere occurs near perihelion - Earth's closest approach to the sun.
Right now, we're closest to the sun during Northern Hemisphere winter and farthest in summer. It was the other way around 6,000 years ago. There were hotter summers and colder winters. That's the kind of situation Kutzbach and his fellow modelers will be feeding into their climate programs to see if these can reproduce the regional climatic conditions that actually prevailed.``That's going to make us either more sure or less sure about the simulations of future climate that will be coming out over the next decade or so,'' Kutzbach says.