New Approach to Climate Modeling

Center factors in variability of clouds.

+ More

By Marlene Cimons, National Science Foundation

Scientists have developed a revolutionary new approach to climate prediction based on a vastly improved—and more realistic—understanding of how clouds behave in the atmosphere.

Researchers at the Center for Multiscale Modeling of Atmospheric Processes at Colorado State University have created what director David A. Randall describes as “a dramatically new way of representing clouds in climate models.”

Clouds are complex, with wide-ranging effects on the weather and long term climate. They influence temperature, precipitation, and the way pollutants are distributed. They reflect sunlight, warm the Earth by blocking infrared radiation to space, move air, and produce rain and snow.

Accurate climate predictions are critically important, given the potential harmful effects of a warming Earth on the environment, global health, international security, agriculture and the economy, as well as on extreme weather events, such as droughts, floods, tornadoes and hurricanes.

“There is absolutely no question in my mind that climate change is quite real,” says Randall, also a professor of atmospheric science. “If we continue for another few decades without taking any action, there will be major changes that people and all the other life on this planet will have to adapt to.”

The new climate modeling approach involves a more accurate way of factoring in the variability of clouds and their significant impact on weather and climate, feedback that has been missing until now.

“It’s a really big change which will allow us to make predictions about climate change between now and the end of the century that we think will be better than has been possible before,” Randall says. “We are working on them now, and expect to do the first such simulation this year.”

The center, which is beginning its sixth year, is one of the National Science Foundation’s seventeen current Science and Technology Centers. NSF supports the center with $37 million in funding spread over ten years.

Colorado State University is the lead institution for the center, which also has research partners at Colorado College, the National Center for Atmospheric Research, the State University of New York at Stony Brook, the University of California at Berkeley, the University of California at San Diego, the University of California at Los Angeles, the Scripps Institution of Oceanography, the University of Colorado, the University of Utah and the University of Washington.

The center also collaborates with, among others, NASA and the National Oceanic and Atmospheric Administration (NOAA).

When scientists study and predict weather and climate, they use computer models, or simulations. In these, the world is divided into grid boxes, or “cells,” each one about the size of the state of Delaware. Every cell contains values for such properties as temperature, humidity and wind speed, and computers use mathematical equations to calculate the value of these variables over time. Researchers develop weather forecasts by updating the numbers in the cells for a few days, and climate predictions by doing the same thing over many years.

Clouds are important to the results of these computations, but they are much smaller in size than the grid cells, which makes it very difficult to calculate their effects. Although scientists have developed models that make grid cells small enough to simulate clouds—about the size of a city block—the models are quite slow and only can be used for small areas and over short time periods.

The solution: center scientists have developed a new “middle ground” approach that bridges the gap between the Delaware-sized grids and the city block-sized grids. It involves adding small grid cells to a tiny fraction of each Delaware-sized grid, essentially simulating realistic cloud processes in a representative sample of the grid, rather than everywhere.

Center researchers have compared this approach to conducting a public opinion poll, where researchers extrapolate data using responses from representative random samples, rather than from querying the entire population.