Prediction of daily reference evapotranspiration using hybrid artificial intelligence method based on empirical mode decomposition

Evapotranspiration (ET) is one of the essential components of the hydrological cycle, which plays a crucial role in the study of a watershed water balance. In calculating the water requirement of plants, it is essential to calculate the reference evapotranspiration, and then, the crop evapotranspiration is estimated using the calculated value. In the present research, for accurate determining of daily reference evapotranspiration of Lake Urmia watershed, three stations of the watershed, including Urmia, Mahabad, and Khoy, were selected and daily reference evapotranspiration values were calculated based on the standard FAO-Penman-Monteith method. The best input parameters for modeling reference evapotranspiration were selected based on Malo’s coefficient. The MT model, which is used in the current study, is one of the inference-classification algorithms. To deal with the complexity and instability of time series data, the empirical mode decomposition (EMD) preprocessing algorithm was used. The results of the methods were compared with the empirical relationships of Romanenko and Schendel. The results of this study show that although the tree modeling method performs relatively equal and sometimes weaker than the ANN method, the combination with EMD technique increases the accuracy of the model and reduces the error in daily ET0 prediction. According to the results, the EMD-MT method in correlation coefficient index for Urmia, Mahabad, and Khoy stations increased 1.02%, 4.39%, and 2.04%, respectively. Also, among the empirical relations, the Romanenko relation is more accurate than the Schendel equation, and it is a reliable empirical model.