Neuro-Genetic Hybrid Approach for Rainfall Forecasting

Weather is certainly the most important factor over which man has no control, and hence it has created dominance on the success or the failure of agricultural enterprises. Most important efforts since long time have been on weather and rain forecasting. However, the unpredictable nature of rainfall has not changed. Meteorologists can neither solve nor evaluate the problem of effective rainfall merely from tables of frequency, amount and intensity of rainfall or from physical phenomena in the atmosphere. It is a task in which several disciplines and sub-disciplines overlap. The present study investigates the ability of hybrid approach of artificial neural network (ANN) and genetic algorithm (GA) in forecasting rainfall. A standard feed forward network (FFN) is utilized for performing the prediction task. Moreover GA is used to determine the optimal structure of ANN.