Adaptive neuro-fuzzy inference system based model for rainfall forecasting in Klang River, Malaysia

Runoff prediction still represents an extremely important issue in applied hydrology. On the other hand, rainfall is one of the most complicated effective hydrologic processes in runoff prediction. For a developing country such as Malaysia which is prone to flood disaster having such an expert model for runoff forecasting is a very vital matter. In this article, an adaptive neuro-fuzzy inference system (ANFIS) model is proposed to forecast the rainfall for Klang River in Malaysia on monthly basis. To be able to train and test the ANFIS and ANN models, the statistical data from 1997 to 2008, was obtained from Klang gates dam data. The optimum structure and optimum input pattern of model was determined through trial and error. Different combinations of rainfall were produced as inputs and five different criteria were used in order to evaluate the effectiveness of each network and its ability to make precise prediction. The performance of the ANFIS model is compared to artificial neural network (ANN) model. The five criteria are root mean square error (RMSE), Correlation Coefficient (), and Nash Sutcliffe coefficient (NE), gamma coefficient (GC) Spearman correlation coefficient (SCC). The result indicate that the ANFIS model showed higher rainfall forecasting accuracy and low error compared to the ANN model. Furthermore, the rainfall estimated by this technique was closer to actual data than the other one.