Alternative heuristics equations to the Priestley–Taylor approach: assessing reference evapotranspiration estimation

• Published in: Theoretical and applied climatology 2019-10, Vol.138 (1-2), p.831-848
• Main Authors: Shiri, Jalal, Nazemi, Amir Hossein, Sadraddini, Ali Ashraf, Marti, Pau, Fakheri Fard, Ahmad, Kisi, Ozgur, Landeras, Gorka
• Format: Article
• Language: English
• Subjects: Accuracy; Air temperature; Analysis; Aquatic Pollution; Aridity; Atmospheric Protection/Air Quality Control/Air Pollution; Atmospheric Sciences; Calibration; Climate science; Climatology; Earth and Environmental Science; Earth Sciences; Evapotranspiration; Evapotranspiration estimates; Gene expression; Mathematical models; Meteorological data; Model accuracy; Modelling; Net radiation; Original Paper; Performance assessment; Problem solving; Procedures; Radiation; Radiation balance; Radiation data; Records; Solar radiation; Solar radiation data; Vapor pressure; Vapour pressure; Waste Water Technology; Water Management; Water Pollution Control; Weather stations
• ISSN: 0177-798X; 1434-4483
• DOI: 10.1007/s00704-019-02852-6

The radiation-based Priestley–Taylor (PT) model is a common way of modeling evapotranspiration (ET 0 ) when only air temperature and solar radiation records are obtainable. However, this method presents the crucial drawback of requiring an initial local calibration. Most of the existing calibration procedures usually utilize similar patterns (data) for calibrating and testing the ET 0 models. In the current research, the PT model and its calibrated versions were evaluated by utilizing meteorological data from 29 weather stations in Iran. Thoroughgoing local and external data scrutinizing procedures (using k-fold validation) were adopted to evaluate the models. Additionally, the externally calibrated models were reevaluated considering vapor pressure deficit (VPD) and net radiation ( R n ) records. A similar plan was replicated for assessing the performance of gene expression programming (GEP)–based equations fed with the air temperature and solar radiation data. Both local and external calibrations of the PT model generally improved the accuracy of original PT for estimating ET 0 , especially in humid and arid stations. External calibration using VPD and R n data was found to be comparable to locally calibrated and non-calibrated PT models. Local GEP models generally performed more accurately than locally calibrated and non-calibrated PT models. Higher accuracy was also obtained using external GEP models in comparison with non-calibrated PT models.