Forecasting Rainfall Using Adaptive Neuro-Fuzzy Inference System (ANFIS)

Rainfall is one of the most complex and difficult elements of the hydrologic cycle. The trend and forecasting of rainfall is very difficult to understand and to model due to the complexity of the atmospheric processes that generate rainfall. This paper investigates the development of an efficient model to forecast monthly monsoon rainfall for Gandhinagar station using Adaptive Neuro Fuzzy Inference System (ANFIS). Eight models are developed using various membership functions and climatic parameters as inputs. In this study, the generalized bell-shaped built-in membership function (gbell) has been used as a membership function in both Hybrid and Back propagation method for ANFIS. The four evaluation parameters Root mean square error (RMSE), Correlation Coefficient (r), Coefficient of Determination (R 2) and Discrepancy ratio (D) are used to evaluate the developed model. The study reveals that hybrid Model with seven membership functions and using three inputs, temperature, relative humidity and wind speed gives best result to forecast rainfall for study area.