Improving Rainfall Forecasting Efficiency Using Modified Adaptive Neuro-Fuzzy Inference System (MANFIS)

Rainfall is one of the most complicated effective hydrologic processes in runoff prediction and water management. The adaptive neuro-fuzzy inference system (ANFIS) has been widely used for modeling different kinds of nonlinear systems including rainfall forecasting. Adaptive Neuro-Fuzzy Inference Systems (ANFIS) combines the capabilities of Artificial Neural Networks (ANN) and Fuzzy Inference Systems (FIS) to solve different kinds of problems, especially efficient in rainfall prediction. This paper after reconsidering conventional ANFIS architecture brings up a modified ANFlS (MANFlS) structure developed with attention to making ANFIS technique more efficient regarding to Root Mean Square Error (RMSE), Correlation Coefficient (R ²), Root Mean Absolute Error (RMAE), Signal to Noise Ratio (SNR) and computing epoch. The modified ANFIS (MANFIS) architecture is simpler than conventional ANFIS with nearly the same performance for modeling nonlinear systems. In this study, two scenarios were introduced; in the first scenario, monthly rainfall was used solely as an input in different time delays from the time (t) to the time (t-4) to conventional ANFIS, second scenario used the modified ANFIS to improve the rainfall forecasting efficiency. The result showed that the model based Modified ANFIS performed higher rainfall forecasting accuracy; low errors and lower computational complexity (total number of fitting parameters and convergence epochs) compared with the conventional ANFIS model.