Hybrid Genetic Algorithm−Based BP Neural Network Models Optimize Estimation Performance of Reference Crop Evapotranspiration in China

Precise estimation of reference evapotranspiration (ET0) is significant importance in hydrologic processes. In this study, a genetic algorithm (GA) optimized back propagation (BP) neural network model was developed to estimate ET0 using different combinations meteorological data across various climatic zones and seasons China. Fourteen locations were selected represent five major climates. Meteorological datasets 2018–2020, including maximum, minimum mean air temperature (Tmax, Tmin, Tmean, °C) diurnal range (∆T, °C), solar radiation (Ra, MJ m−2 d−1), sunshine duration (S, h), relative humidity (RH, %) wind speed (U2, m s−1), first subjected correlation analysis determine which variables suitable as input parameters. Datasets 2018 2019 utilized for training the models, while 2020 testing. Coefficients determination (r2) 0.50 0.70 adopted threshold values selection correlated run models. Results showed that U2 had least r2 with ET0, followed by ∆T. Tmax greatest Ra Tmin. GA significantly improved performance BP models zones, accuracy GABP higher than GABP0.5 (input based on > 0.50) best reduced errors, especially autumn winter whose errors larger other radiation/temperature highly recommended promising tool modelling predicting locations.