CUMULONIMBUS SHORT TIME FORECASTING USING ARTIFICIAL NEURAL NETWORK WITH SELECTED RADIOSONDE INDICES IN JAKARTA, INDONESIA

Abstract

This research proposed an accurate forecasting of Cumulonimbus (Cb) Cloud development events. Cb plays a vital role for mostly weather hazard in Indonesia, from thunderstorm, lighting, small tornadoes and torrential ice. Neural network has been used in numerous meteorological applications including weather forecasting. By using the data from radiosonde observation indices and data cloud classification from Multifunctional Transport Satellites (MTSAT), neural network was used as a classifier, also reduced atmospheric indices to select best variables by using Principle component analysis (PCA). The location interest of this research is Jakarta Area in Indonesia with data observation from Cengkareng Meteorological Station. An application to classify the Cb occurrences was shown and evaluated.