A Weather Nowcasting Model for Rainfall in Sri Lanka

Abstract

Weather forecasting is the application of science and technology to predict the state of the atmosphere for a future time for a given location. Forecasting the weather for next six hours is often referred as nowcasting. During this time range, it is possible to forecast smaller features such as showers and thunderstorms with a reasonable accuracy. Several numerical and computational models have been developed to forecast the rainfall for the Sri Lanka. However, country has faced to some difficulties in nowcasting the rainfall. Some of the major reasons are the lack of information that is needed and the use of traditional numerical models. In recent years researches conducted by the University of Colombo had shown that forecasting rainfall for Sri Lanka can be improved by using neural networks. In the research we focus on building a nowcasting model for rainfall using Artificial Neural Networks. The model is developed to Colombo, the capital of Sri Lanka. The objective of the study is to build 3 hour ahead and 6 hour ahead rain nowcasting models. For the development of the models, both ground level observations and satellite images are used. The results are impressive when compare with the statistical nowcasting. Both 3 hour ahead and 6 hour ahead nowcasting models have successfully reached to their destinations. However 3 hour nowcasting models show greater improvement than 6 hour ahead nowcasting models, in nowcasting for Sri Lanka.