One of the principal goals of basic research in atmospheric science is to improve our ability to predict weather and climate. This requires not only a thorough understanding of atmospheric dynamical and physical processes, but also finding the proper means of representing these processes in sophisticated numerical models. Clear progress has been made in this field in the last decade and the principal investigator has been responsible for many important advances. Nevertheless, we have not begun to approach the theoretical limit of useful atmospheric prediction. This award provides support for basic research aimed at improving numerical weather prediction, with particular emphasis on the tropics. The research involves a number of related studies, most of which employ a state-of-the-art, multilevel global spectral model developed by Dr. Krishnamurti and his colleagues. Among the many important topics that will be investigated are model initialization with emphasis on humidity distribution; parameterization of cumulus convection; the impact of increased model resolution; representation of cloud processes, particularly radiation; systematic errors related to representation of momentum flux in the models; and maintenance of vertical consistency in the various physical parameterizations. The emphasis in this research is on the tropics, but the results have direct and important implications for improving medium range numerical weather prediction for the midlatitudes because, on this timescale, weather in one part of the globe influences weather in other regions. One of the most serious shortcomings in global models today is the poor representation of the tropical regions. This research should improve both our fundamental understanding of the tropical atmosphere and its representation in numerical models.
