Abstract:Accurate estimation of reference crop evapotranspiration(ET0)is greatly significance for regional water resources management and irrigation decision-making. The Hargreaves-Samani model(HS)is currently recognized as the ET0 estimation model with the simplest structure and highaccuracy. In order to further improve the prediction accuracy of the HS model, the bee colony theory and the data of 20 weather stations in Guangxi Basin(1961—2019)are used to calibrate the HS model globally in this manuscript. The data of 1961—2000 are used to calibrate the HS model, and the data of 2001—2019 are verified on the daily, monthly, and annual scales. The results showed that among the globally optimized empirical parameters C, m and a, the parameter a varies greatly with the topography, while the difference between the parameters C and m is small. The HS model after calibration(average MAE and R2 were 1.06 mm/d and 0.86)was better than the original HS model(MAE and R2 were 2.20 mm/d and 0.68). On the daily, monthly and annual time scales, both the calibrated HS model and the original HS model overestimate ET0, but the calibrated HS model is closer to the ET0 calculated by the PM model. Therefore, in areas only with temperature data, it is recommended to use thecalibrated HS model to estimate ET0.