RAINFALL-RUNOFF MODELING USING ARTIFICIAL NEURAL NETWORKS AND ADAPTIVE NEURO-FUZZY INFERENCE SYSTEM MODELS

摘要

In this study, rainfall-runoff modeling was carried out in Hajighoshan watershed using artificial neural network (ANN) and adaptive neuro-fuzzy interface system (ANFIS) with different inputs (current day rainfall; current rainfall and pervious day rainfall; current rainfall, pervious day rainfall and two previous day) methods. Root mean squared error (RMSE), mean absolute error (MAE) and correlation coefficient (R) statistics are employed to evaluate the performance of the ANNs and ANFIS models for forecasting runoff. Comparison of the obtained results reveals that the ANFIS model outperforms the ANN models. Based on the results of test stage, ANFIS with RMSE=7.11 m~3 s~(-1), MAE=2.18m~3 s~(-1) and R=0.60 is superior to rainfall-runoff modeling than the ANN with RMSE=6.03m~3 s~(-1), MAE=1.97 m~3 s~(-1) and R=0.39.