—Rainfall is often defined by stochastic process due to its random characteristics, i.e. space and time dependent and it is therefore, not easy to predict. In general, rainfall is a highly non-linear and complicated phenomenon. In order to acquire an accurate prediction, advanced computer modeling and simulation is required. Artificial Neural Network (ANN) has been successfully used to predict the behavior of such non-linear system. Among the different types of ANN models used, Backpropagation Network (BPN) and Radial Basis Function Networks (RBFN) are the two common ANN models that had produced valuable results. However, there was no study conducted to research on which, among these two methods, is the better model for rainfall forecast. Therefore, this study will fill this gap by comparing the capabilities of these two ANN models in rainfall forecast using metrological data from year 2009 to 2013 obtained from Malaysian Meteorological Department for Kuching, Sarawak, Malaysia. From the research, it is concluded that, BPN (MSE≈0.16, R≈0.86) performs better as compared to RBFN (MSE≈0.22, R≈0.82). The strengths and weaknesses of these models are also presented in this paper.
—Artificial neural network, backpropagation, radial basis function, rainfall, classification.
Soo See Chai is with the Department of Computing and Software Engineering, Faculty of Computer Science and Information Technology, University of Malaysia Sarawak (UNIMAS), Malaysia (e-mail: email@example.com).
Wei Keat Wong was with Faculty of Computer Science and Information Technology, University of Malaysia Sarawak (UNIMAS), Malaysia.
Kok Luong Goh is with the International College of Advaned Technology Sarawak (ICATS), Sarawak, Malaysia (e-mail: firstname.lastname@example.org).
Cite: Soo See Chai, Wei Keat Wong, and Kok Luong Goh, "Rainfall Classification for Flood Prediction Using Meteorology Data of Kuching, Sarawak, Malaysia: Backpropagation vs Radial Basis Function Neural Network," International Journal of Environmental Science and Development vol. 8, no. 5, pp. 385-388, 2017.