Integration of Fuzzy C-Means and Artificial Neural Network for Short-Term Localized Rainfall Forecast in Tropical Climate

Abstract

This paper evaluates the performance of a rainfall forecasting model. In this paper Artificial Neural Network (ANN) and Fuzzy C-Means (FCM) clustering algorithm are combined and used to forecast short-term localized rainfall in tropical climate. State forecast (raining or not raining) and value forecast (rain intensity) are tested using a number of trained networks. Different types of ANN structured were trained with a combination of multilayer perceptron with back propagation network. Levenberg-Marquardt, Bayesian Regularization and Scaled Conjugate Gradient training algorithm are used in the network training. Each neurons uses linear, logistic sigmoid and hyperbolic tangent sigmoid as transfer function. Input parameter preliminary analysis, data cleaning and FCM clustering were used to prepare input data for the ANN forecast model. Meteorological data such as atmospheric pressure, temperature, dew point, humidity and wind speed have been used as input parameters. The predicted rainfall forecast for 1 to 6 h ahead are compared and analyzed. 1 h ahead for state and value forecast yield high accuracy. Result shows that, the combined of FCM-ANN forecast model produces better accuracy compared to a basic ANN forecast model.