Prediction of daily reference evapotranspiration using adaptive neuro-fuzzy inference system

The estimation of evapotranspiration from vegetated surfaces is a basic tool to compute water balances and to estimate water availability and requirements. Reference evapotranspiration (ET₀) just reflects weather conditions. Adaptive Neuro-Fuzzy Inference System(ANFIS), has the capacity of non-linear mapping between input layer and output layer by fuzzy inference, and has storing and learning ability with the information of the neural network at the same time. In this paper, the computation of daily ET₀ by ANFIS is presented, comparing the results with the ET₀ calculated through FAO Penman-Monteith method in the same period. Sunlight hour and maximum air temperature are as input variables in ANFIS according to regression analysis between every weather factor. The ANFIS for ET₀ estimator is built from training data, whose array list includes 1827 data of five years. The result of testing data of 213 datum groups, to estimate ET₀ using the ANFIS, is acceptable comparing with the result of Penman-Monteith method.