Prediction of Evapotranspiration Using Artificial Neural Network Model and Compared With Measured Values

A feature based Artificial Neural Network (ANN) model was developed for prediction of Evapotranspiration (ET) of eucalyptus. Six weather parameters namely maximum temperatures, minimum temperature, relative humidity first, relative humidity second, wind velocity and sunshine hour were used by the ANN model. The network was trained using the pattern matching capability of artificial neural network to recognize the pattern of daily metrological data. Results of ANN model training, testing and validation by back propagation technique were observed to be in good agreement with those of measured ET by lysimeter of eucalyptus plant. Correlation coefficient (r2) between measured and predicted ET during training phase were found 0.9810, during testing phase 0.8770 and during validating phase 0.9010.