Supercomputer To Advance Studies Of Abrupt Climate Change

The U.S. Department of Energy presented an INCITE Award to the Center for Climatic Research - part of the university's Nelson Institute for Environmental Studies - this week to conduct climate simulation experiments on a Cray supercomputer at the Oak Ridge National Laboratory in Tennessee.

Project leader Zhengyu Liu, a UW-Madison professor of atmospheric and oceanic sciences and environmental studies, estimates that the award is worth hundreds of thousands of dollars in prized supercomputer time.

The award provides 420,000 processor hours over the next year and is expected to be renewed for the same number of hours in 2008. Running a 420,000-hour project on a single-processor desktop computer would take almost 48 years.

The Energy Department launched the INCITE (Innovative and Novel Computational Impact on Theory and Experiment) program in 2003 to support "computationally intensive" research projects in the national interest. UW-Madison was one of 14 universities nationally to receive INCITE Awards this year.

Scientists at the National Center for Climatic Research (NCAR) and at the Oak Ridge and Argonne national laboratories will collaborate in the effort, using a sophisticated computer model at NCAR designed to simulate interactions of the atmosphere, oceans, and vegetation.

Computer models are key tools in understanding and predicting climate change, but complex climate models require massive amounts of computer capacity to process all of the relevant data.

The centerpiece of the UW-Madison-led experiments will be the first continuous computer simulation of the past 21,000 years of global climate change. This period in the earth's history included the last major glacial retreat and a significant increase in atmospheric carbon dioxide (CO2), the "greenhouse gas" most associated with global warming.

Simulating climates of the past is crucial to improving the accuracy of climate models. Scientists compare the results of simulation experiments with physical evidence of past climate change, such as fossilized vegetation, to validate and refine the models.

The better such models become at mimicking past climates, the more credible they are as predictors of future climates under various scenarios, such as steady increases in atmospheric CO2. Liu and his collaborators will extend their supercomputer simulation 2,000 years into the future with an eye toward revealing potentially abrupt climate changes in the decades and centuries ahead.

Most predictions of future climates emphasize changes in average temperatures and precipitation over relatively long periods of time. Some scientists now say that focusing on averages overshadows the likelihood of rapid, dramatic changes that could have more significant impacts on human societies and plant and animal life.

Abrupt climate changes have been fairly common in the earth's recent history. For example, Liu points to scientific evidence that the parched Sahara region of northern Africa transformed from a wetter - and much greener - region thousands of years ago into the world's largest hot desert in less than a century.

The supercomputer experiments "may lead us to a new era of understanding abrupt climate change in the next several decades," says Liu.