Numerical weather prediction for GATE

Abstract The results of a few numerical weather prediction experiments that utilize GATE data are presented. Aside from some simple forecasts based on single‐level models, experiments with a multilevel model examine the roles of some of the physical processes. They include the influences of diurnally varying radiative effects, effects of cloud feedback on shortwave and longwave radiation, the heat balance of land surfaces, and the influence of West African orography in the 3‐ to 4‐day range of numerical weather prediction. Data sources for these experiments include surface and upper air observations from the World Weather Watch, GATE ships, upper winds from commercial aircraft, low‐ and high‐level cloud winds from geostationary satellites, winds from GATE research aircraft, and special surface observations from a ship data collection. The data analysis is based on a successive correction method where a subjective interface is also incorporated. The initial data are further subjected to a static as well as a dynamic initialization process. The major results are that single‐level forecasts are very promising with this data set if they are carried out at 700 mb. The multilevel adiabatic experiment produces very poor predictions due to lack of adequate vertical coupling between the lower and the upper troposphere and due to its inability to describe the broad‐scale monsoons. The diurnally varying heat balance of the earth's surface and cloudiness in radiative calculations appear to be important in the 3‐ to 4‐day range of prediction of African disturbances. In this paper we show some interesting vertical structure diagrams of African waves based on the results of numerical weather prediction, which exhibit a close similarity to the structures of composite GATE disturbances constructed by Professor R. J. Reed and his associates.