February 2003
Understanding
cloud systems:
Are researchers closing in on a general theory of convective cloud systems?
We’ve all had the experience of getting rained on from one small cloud even though blue sky is plainly visible all around. The reason is that rain can come from an individual cumulus cloud just as easily as from a massive cloud system that hangs over an entire region.
Mitch Moncrieff.
Researchers, however, have yet to figure out the precise mechanisms that drive the behavior of either individual clouds or the larger-scale systems that comprise them. Cloud-scale models can yield some insight on what makes a single cloud tick, but on a global scale the collective behavior of clouds is still a challenge. That’s partly because climate models, which are fundamental for research into regional or global climate patterns, lack the resolution to accurately capture events that occur on scales of less than about 250 kilometers (155 miles). As a result, they have to estimate, or parameterize, the behavior of a group of clouds that each may be as small as 10 kilometers (6.2 miles) or less.
“The small scale is very important,” explains Mitch Moncrieff of the Mesoscale and Microscale Meteorology Division. “At small scales, there’s an awful lot of instability and huge exchanges of energy. Convective rainfall is intense.”
Mitch and his colleagues at MMM, including Wojciech Grabowski, Changhai Liu, and Piotr Smolarkiewicz, are working to fill that gap by studying the evolution of cloud systems ranging in size from about one kilometer (0.6 mile) to thousands of kilometers. They are looking at natural processes that span widely differing timescales: raindrops can develop in seconds, for example, whereas intraseasonal oscillations in the tropics may last for months.
A major focus of the research is understanding the difference between events that occur seemingly at random (localized showers and thunderstorms caused by convection, as opposed to steady rain over large areas) and events that proceed in an organized way (the formation of clusters of cumulus cloud systems that can span many hundreds of miles). A computer model may parameterize cumulus clouds by specifying some number of them in a given area without trying to identify each specific cloud. But the techniques used to parameterize each cloud’s random behavior may be missing some aspect that connects the cloud features.
Mitch and his colleagues are finding that convective systems on both large and small scales appear to be governed by similar natural laws—an understanding that could shed light on the physics of entire cloud systems. If the finding holds up, it would indicate that clouds, which often seem to be acting randomly, actually are behaving in a coherent—and partially predictable—fashion. For example, large organized systems are often born from localized showers and thunderstorms.
The finding could also mark an important step toward creating a general theory of precipitating systems. As Mitch puts it, constructing such a theory of cloud systems and how to parameterize them “is the Holy Grail of precipitation prediction.”
The implications go beyond rainfall. Clouds have profound effects on long-term climate trends because they reflect and absorb sunlight and transport water vapor, which is the most important of the greenhouse gases that warm the atmosphere. Quantifying the evolution of cloud systems is critical for understanding the cycling of water and energy in the atmosphere, as well as for making accurate numerical weather forecasts. (Part of the reason that weather prediction models cannot forecast precipitation accurately is because they fail to model the formation of cloud systems.)
So far, Mitch has focused largely on clouds in the tropics, where convection is the dominant atmospheric process. Understanding tropical convection will help weather prediction in midlatitudes because the effects of tropical convection ripple across to midlatitudes within a few days. “If we are ever to predict midlatitude weather at its deterministic limit of about two weeks, we must understand tropical convection,” Mitch says.
What is learned about the tropics, in other words, can be applied to the entire earth.
The long shadow of a few showers
Scientists have long understood the basics of convection: warm, moist air bubbles up through colder air aloft. It’s the complex interplay of convection with the atmosphere around it, sometimes at a great distance, that remains a challenge. For example, tropical storms over the Indian Ocean and western Pacific Ocean may be related to the timing of larger-scale pulses of wind and storminess that cross this region from west to east every 30 to 60 days. This cycle, known as the Madden-Julian Oscillation (discovered by NCAR’s Rol Madden and Paul Julian, formerly of NCAR), in turn affects the evolution of clouds and weather across large parts of the Americas and the Atlantic.
When working with regional and high-resolution global weather prediction models, researchers face another challenge: the models do not correctly handle the growth of large storm systems. As a storm system grows big enough to exceed the size of a single grid cell in the model, it may jump over an artificial threshold: instead of parameterizing the growing system, the model begins to resolve it. “We’ve shown that surrogate and distorted cloud systems may develop, generating a new kind of uncertainty in modern numerical weather prediction models,” Mitch says.
To better understand precipitating convection and clouds, Mitch and his colleagues are drawing on a variety of disciplines, including mathematics, cloud physics, fluid dynamics, and numerical models. Their advanced cloud-system-resolving models calculate cloud systems directly in a process called “superparameterization.” This new approach was developed by comparing the models and parameterizations with data collected during major field experiments, such as the Tropical Ocean and Global Atmosphere Program’s Coupled Ocean-Atmosphere Response Experiment (TOGA COARE) and NASA’s Tropical Rainfall Measuring Mission (TRMM).
While it may not be possible to anticipate the location of a single rain cloud in a partly sunny sky, MMM’s findings should lead to increasingly detailed estimates of the timing and general location of storms in the next few years.
“One of the ultimate goals of our research,” Mitch says, “is to provide a foundation for more accurate and longer-range weather forecasting.” •David Hosansky
Also in this issue...
Coming soon:
New Mesa Lab attractionsMentoring talk offers tips for nonscientists
