Climate simulation capabilities advanced with $3.6 million grant

University scientists have received a $3.6 million grant from the National Science Foundation to develop software during the next five years that will enable them to rapidly conduct advanced climate simulations.

“All the big questions are fair game to us, from why Mars was warm and wet early in its history, to how the Earth could have completely frozen over 600 million years ago, to the role of water vapor in clouds in global warming,” said Raymond Pierrehumbert, Professor in Geophysical Sciences and the College.

The project aims to speed up the pace of climate simulation research by making modern software design features more widely and freely available to the research community. Climate researchers currently depend upon complex and unwieldy computer programs that require months of work if they want to adjust their models. But that could change with the adoption of an open source (free on the Internet) programming language called Python, said David Beazley, Assistant Professor in Computer Science and the Physical Sciences Collegiate Division.

Working with Pierrehumbert and Beazley on the project are eight colleagues at Chicago, Argonne National Laboratory, the Computation Institute and the University of Southern California. The Computation Institute, a joint project of the University and Argonne, brings the most advanced tools of computer science to bear on problems in the sciences, the arts and the humanities.

Among Python’s virtues is its flexibility for scientific applications, said Beazley, the author of Python Essential Reference, which has been translated into four languages. “You can hook it into specialized programs, like climate modeling or molecular dynamics or anything you want to do,” he said.

Beazley said using Python was a boon to the molecular dynamics simulations he performed as a scientist working at Los Alamos National Laboratory. “We were doing more simulations in a month than we had done in the previous three years of the project combined,” he said.

Python also permits scientists to easily test small parts of a simulation in development.

“If you’re trying to build a new system, you find out immediately if something is going wrong, and you can try out pieces interactively without having to debug the whole system all at once,” Pierrehumbert said. He and his colleagues plan to adapt Python to existing simulations that combine the effects of the ocean and the atmosphere on climate. Coupling the oceanic and atmospheric effects is vital to advanced climate modeling, Pierrehumbert said. “There are various phenomena that occur when you hook the atmosphere and the ocean together that don’t occur with each one individually,” he said.

Pierrehumbert and his colleagues have used a coupled simulation to test the “snowball Earth” hypothesis. The hypothesis stems from geologic evidence indicating that the Earth once was completely covered by ice.

“Those of us who do theory and simulation are having a hard time making the snowball idea work,” Pierrehumbert said. “When you couple the atmosphere to the ocean and let the ocean transport a little bit of warm ocean water––and warm can mean just one or two degrees above freezing––from the tropics to the ice margin, that is extremely good at keeping the ice from advancing into the equatorial regions.”

Another question he will try to answer with Python software is how the climate of Mars could have been warm and wet early in its history. Today Mars is cold and dry, but NASA space probes have produced strong evidence of surface features that were produced long ago by flowing water. Exactly what could account for the difference has become a controversial matter.

Many climatologists, including Pierrehumbert, favor the theory that a dense carbon dioxide atmosphere could have made Mars warmer in the distant past. But in order to test the theory, they need simulations that include weather patterns, which current models lack.

“Almost all the modeling so far has been very simple, with no weather in it,” Pierrehumbert said. “But almost all the interesting problems for early Mars now require simulation that has weather in it.”

Such research has implications for understanding Earth’s own climate. “If you study the climate of one planet, you can get into a rut and think that because the climate behaves a certain way now, that it always has to behave that way,” Pierrehumbert said. Also, he said, “Other planets give you a test of your basic physics in a wildly different set of circumstances.”

Mars climatological research has helped to stimulate research on the role of aerosols, suspended particles and dust, in global warming on Earth, Pierrehumbert said. Dust is as important to the climate of Mars as water is to Earth’s climate. But the distribution of dust particles and of tiny droplets of sulfuric acid, which result from burning dirty coal, also is a key but poorly understood factor in global warming on Earth, Pierrehumbert said.