Prediction in the Short Term

A severe thunderstorm, cumulonimbus cloud, moves across the plains east of Denver on June 10, 2004. Heavy rain and large hail is falling in the background, dark blue or blue-green area. A downdraft of cool, moist air produced by the rain and hail is pushing toward the camera. The strong winds produce fragmented cumulus clouds known as fractocumulus or scud, lighter, lower clouds in foreground. Some of the most threatening weather events—gusty thunderstorm winds, ice on highways and aircraft, turbulence aloft—are the toughest to predict. They may affect only a small area, and they can develop and dissipate within minutes.

Investigating the dynamics of weather systems with the aim of improving their prediction, estimating their limits of predictability, identifying the key physical processes that limit forecast skill, and developing improved methods of determining forecast skill at the mesoscale, is a major research effort at NCAR. For instance, a long-running line of research has clarified the factors that result in different types of thunderstorms. These types include weak, short-lived cells; narrow, fast-moving squall lines; isolated supercells that pack large hail or tornadoes; and mammoth storm clusters that dump swaths of heavy rain. Training tools are derived from this research, and now help forecasters decide how and when to warn the public as storms evolve.