Berkeley Lab Science Beat Magazine banner Berkeley Lab logo
October 4, 2004
How Good are Computer Climate Models at Prediction?
A Berkeley Lab Researcher Tests Their Predictions

Michael Wehner, a climate researcher in the Scientific Computing Group of Berkeley Lab's Computational Research Division, has received a National Science Foundation grant to analyze the results of three new climate models as a means of determining their predictive quality. Each of the three models will be run to predict both past and future climate change patterns. The results will also be compared with observational climate data to see how the predictions and observations correlate.

One aspect of Michael Wehner's climate research involves the study of extreme weather events. (Click here for an animation relating hurricane wind speed and atmospheric pressure.)

"Some features will be correct, some will be wrong, and some will be difficult to determine," said Wehner. "Our goal is to learn about the ability of the models to predict reality, and hopefully we will learn more about the reality of climate change in the process."

When finished, his work will be incorporated into an international report on climate change. The report is designed to advise governments on the state of scientific research on climate change so that policies can be set.

The first part of Wehner's research will build on his fruitful collaboration with Ben Santer of the Program for Climate Model Diagnosis and Intercomparison at Lawrence Livermore National Laboratory. In a series of studies, the team led by Santer has used the Parallel Climate Model (PCM) and observational data to show that statistically significant changes to the tropospheric temperature, lower stratospheric temperature, and tropopause height have occurred in the recent past — and that these changes were the result of human activities. This work was reported in the July 25, 2003, edition of Science magazine.

Wehner will extend a multivariate climate-change detection analysis, which he and colleagues have already applied to the older PCM, to early simulations from the three most recent U.S. coupled climate models: the Goddard Institute for Space Studies model, the Geophysical Fluid Dynamics Laboratory model, and the Community Climate System Model from the National Center for Atmospheric Research, or CCSM3.

The second part of Wehner's research is to characterize the ability of coupled climate models to simulate extreme weather events. While the mathematical formalisms necessary for rigorous statistical description of extreme events are quite well developed, application to climate models is still in its infancy.

The ability of these models to accurately simulate historical climate changes in the twentieth century will determine the credibility of their predictions of climate change during the twenty-first century and beyond.

"Five years ago, the results from the climate change models were at odds with the observational data," Wehner said. "But now, with our ability to simulate sea, ice, and land systems, we are finding real similarities between the observational data, particularly from satellites, and the models."

Wehner's proposal, titled "Multivariate Climate Change Detection," was one of several funded as part of the U.S. Climate Variability and Predictability (CLIVAR) program's Climate Model Evaluation Project (CMEP). The objective of CMEP is to increase diagnostic research into the quality of model simulations, leading to more robust evaluations of model predictions and a better quantification of uncertainty in projections of future climate. U.S. CLIVAR is a joint project of the National Science Foundation, the National Oceanic and Atmospheric Administration, NASA, and the Department of Energy, and is part of an international, interdisciplinary research effort within the World Climate Research Programme.

The first step is getting the necessary climate data in place. Wehner is currently transporting data files from Oak Ridge National Laboratory and the National Center for Atmospheric Research in Boulder, Colo. Wehner will then run the models using this data.

The results of the CMEP analyses will be presented at a CLIVAR workshop in Hawaii from March 1 to 4, 2005. A workshop report summarizing the presentations will be furnished to the lead authors of the relevant chapters for the Intergovernmental Panel on Climate Change (IPCC) Assessment Report Four, a comprehensive review of current knowledge about climate change, which will be published in 2007.

"Until now, most of the science for the IPCC has been done by Europeans, and the time is right for Americans to begin making a major impact in this field by becoming a major source of the science, not just analyzing the European data," Wehner said. "By showing that our models, such as the CCSM3, are credible, we are poised to make a major contribution. This is really a high point in my career."

Additional information