Traditional weather forecasting relies on a number of atmospheric prediction methods which could involve the use of certain model assumptions, statistics, and complex approximation schemes. This is for simulating the meteorological behaviour, and the system dynamic developments to give prediction to various types of synoptic flow, pressure systems and some weather events. It often needs to process and assimilate very large amounts of data from several sources plus intuitive perception to make a routing forecast. This paper presents a preliminary study of using artificial neural networks to help process the meteorological data so as to learn the relevant characteristics for forecasting the rainfall in Hong Kong. The simulation illustrates the capabilities of the networks for the analysis and represnetaiton of data. It shows that the approach has produced reasonable accurate weather forecast, paving the way to enhance and improve the qualitative analysis of our meteorological systems in the region.
展开▼
