Daily reference evapotranspiration prediction based on climatic conditions applying different data mining techniques and empirical equations

Considering evapotranspiration takes a basic role in the hydrologic cycle, water resources management, and irrigation water requirements. Evapotranspiration estimation is not an easy case because of the number of direct and indirect effects. The ability of the M5 model tree (M5T); adaptive neuro-fuzzy inference system (ANFIS); support vector machines (SVM); Hargreaves-Samani, Ritchie, Turc, and Penman FAO 56 empirical equations; and multilinear regression (MLR) for modeling daily reference evapotranspiration is investigated. Daily climatic data, air temperature (T), relative humidity (RH), wind speed (U), and solar radiation (SR) from De Soto County, Florida, USA, station are used as inputs for the training of the models and calculation of equations. Mean square error (MSE), mean absolute error (MAE), and correlation coefficient statistics are computed to evaluate the performances of the created models. A total comparison is done between all results to underline how effective is soft computing techniques. Also, the impact of each meteorological parameter on evapotranspiration is investigated by using ANFIS, MLR, and SVM methods as a part of the parameter effect study. According to the error calculations and correlation coefficient, Turc empirical formula found better than other empirical equations. All data-driven techniques gave better results than empirical equations. The highest correlation coefficient is calculated for ANFIS, and the minimum errors are calculated for radial basis function SVM.