Comparison of the reference evapotranspiration estimations by data mining methods and Crop Water Requirement System project in Alborz province

The reference evapotranspiration (ET0) is an important factor for determining the plant water requirements and irrigation scheduling, which is usually estimated by widely accepted equation of Penman Monteith FAO-56 method. The aim of this study was to evaluate the performance of artificial neural networks (ANNs), random forest (RF) and support vector machine (SVM) methods for estimating the daily ET0 in Alborz province. A ten- year-data (2010-2020) of five meteorological synoptic stations of namely MeshkinDasht, Hashtgerd, Eshtehard, Taleghan and Karaj were used for estimation of ETo. The obtained values were compared with the provided data of a national project entitled Crop Water Requirement System. According to the results, the best agreement was found in Meshkin Dasht and Karaj stations. Besides, among the applied approaches, the ANNs method had the highest accuracy comparing to other methods. The values of EF and NRMSE in the ANNs method were determined 0.96 and 0.11, respectively for both training and testing steps in Meshkin Dasht station. While, these values for the RF method were determined 0.96 and 0.11, for training stage and 0.95 and 0.12 for testing stage, respectively. The obtained results in the Karaj station showed that EF and NRMSE in the ANNs method were 0.96 and 0.11, respectively for the training and 0.95 and 0.12 for the testing stage. These values for the RF method were 0.96 and 0.12 for the training stage respectively and 0.95 and 0.13 for the testing. Considering the higher accuracy of the ANNs and RF methods, these approaches can be recommended for estimating the daily ET0 across Alborz province.