Atmospheric Sciences & Global Change Division
Research Highlights
May 2011
World-renowned Expert Discusses Super-Parameterization
Representing clouds in global climate models
The impacts of climate change are already affecting human and environmental systems worldwide. Dr. David Randall, professor of Atmospheric Science at Colorado State University, has spent decades working on cloud processes and how they affect the climate. Dr. Randall showed that it is possible to produce a more accurate picture of the role of clouds on climate than typically portrayed in traditional climate models, using a novel technique called "super-parameterization." His talk on March 17th was part of the PNNL Frontiers in Global Change Seminar Series. The seminars, held at Pacific Northwest National Laboratory, allow experts to share results of studies and novel ideas in global climate change.
Not your typical model: Typical Global Climate Models (GCMs) have difficulty showing how clouds and aerosols-tiny natural and manmade bits of dirt, soot and other particles in the air—influence the Earth's climate. GCMs represent various fields in the model, like temperature, winds, clouds and humidity in the atmosphere, by laying out a grid over the earth, dividing it into a lot of boxes. Then GCMs use equations to predict the evolution of the fields in each grid box. It is easier to produce an accurate picture of the atmosphere when the boxes are small. But to make it feasible to perform a climate calculation over the whole globe for decades or centuries, it is ideal to minimize the number of boxes covering the earth.
Typical climate models divide the atmosphere into boxes about 200 by 200 kilometers (about 124 miles) horizontally, and by 100-2000 meters (300 feet to a mile or so) vertically. But important climate features such as clouds can be a few kilometers horizontally in size, making them small blips on the large boxes used in global models. Scientists resolve this dilemma by using special representations of clouds—called cloud parameterizations—that are designed to represent the effect of these small features in big GCM boxes. Cloud parameterizations work to describe the "ensemble average" effect of clouds by using simple rules to represent the effects of those clouds within each grid box. Over the last 40 years, cloud parameterizations have been an area of intense focus and on-going concern for climate modelers.
Adding the super in parameterization: Dr. Randall, along with his colleagues, has been exploring another path for representing clouds called a "super-parameterization." The super-parameterization tries to achieve a compromise between treating clouds as an ensemble, and resolving individual clouds. It achieves this by inserting a two-dimensional slice of a higher resolution model inside each bigger GCM box. The two-dimensional "cloud-resolving model" (CRM) consists of a set of columns spaced about 4 kilometers apart inside each GCM column. Though the CRM is very rough and does not really resolve clouds, it does use appropriate fundamental equations to describe basic cloud motions rather than the simpler formulations used in traditional GCMs. Super-parameterizations are used as the basis for increasing the accuracy of climate models that can be run more efficiently to simulate climate at a lower cost than a true global CRM, but more realistically than a traditional climate model.
"This method of representing clouds shows real promise. It addresses an area where we have real issues with traditional techniques," said Dr. Phil Rasch, Laboratory Fellow and Chief Scientist for Climate Science at PNNL.
What makes the Cloud-Resolving Model work: Including the super-parameterization component to the CRM provides researchers with a more accurate simulation and fewer approximations than used in traditional climate models. The model that includes the super-parameterization produces a realistic 30- to 60-day oscillation known as a Madden-Julian oscillation, an important feature in the tropical atmosphere. It also produces other climate features that are difficult for traditional climate models to simulate.
"We know that simulation is not the same as understanding, but it can help us to achieve understanding," said Dr. Randall.
Bio: Dr. David Randall is a Professor of Atmospheric Science at Colorado State University, and is the Director of the Center for Multiscale Modeling of Atmospheric Processes at National Science Foundation's Science and Technology Centers. He has published over 175 refereed journal articles. He has chaired or co-chaired the science teams of federally sponsored research projects, including the ARM Science Team, as well as numerous panels and boards. Among his awards, he has received NASA's Medal for Exceptional Scientific Achievement, the Meisinger Award of the American Meteorological Society, and NASA's Medal for Distinguished Public Service.
