Combined Application of Artificial Neural Network and Computational Methods to Estimate the Reference Evapotranspiration

Estimation of reference evapotranspiration (ETo) is essential for many issues i.e., irrigation and drainage, hydrology, environment, soil erosion and water resources. Using the artificial neural network (ANN) to estimate ETo is common in a lot of studies. But what has not been addressed in previous studies is using meteorological data as an input of neural network together with computational methods. In this study, calculated ETo by computational methods including Jensen-Haise, Turc, Hargreaves-Samani and pan evaporation methods accompanying with meteorological data were used as input data. Results showed that using the calculated ETo by Jensen-Haise method together with meteorological data as input data resulted in closer estimation to calculated ETo by Penman-Montieth-FAO among all of computational methods. Using the calculated ETo by other methods along with meteorological data improved the ETo estimation compared with using the meteorological data lonely, however, accuracy of ETo estimation by using these methods were still low.