Abstract: Sixteen daily scale ET₀ calculation models were developed using the particle swarm optimization algorithm and random forest (PSO-RF) based on daily meteorological data collected from 24 meteorological stations in the Hanjiang dasin and it’s vicinity from 1960 to 2017, and were compared with Hargreaves-Samani, Irmark-Allen and Makkink, which applicability and portability were evaluated. The results showed that: （1）the non-linear relation between daily ET₀ and each input parameter can be well displayed by PSO-RF model. The PSO-RF2 established by T_max, T_min and R_a had sufficient accuracy (MAE=0.402 mm·d⁻¹, RMSE=0.575mm·d⁻¹, R²=0.863). With the increase of meteorological elements, the computational accuracy of PSO-RF model was improved（MAE decreased by 37.9%, RMSE decreased by 36.3%, R² increased by 6.78%）.（2）When the input factors and number of models are the same, the PSO-RF model had higher simulation accuracy than Hargreaves-Samani, Irmark-Allen, and Makkink models(MAE decreased by 43.6%, RMSE decreased by 34.5%, R² increased by 4.12%).（3）The PSO-RF model had strong generalization ability and portability in the Hanjiang basin. High accuracy was shown in mutual portability of PSO-RF models established in different zone (MAE= 0.159mm·d⁻¹, RMSE=0.245mm·d⁻¹, R²=0.974). Therefore, the model based on PSO-RF can be used as a recommended model for the calculation of daily ET₀ in Hanjiang basin in the absence of meteorological data.

Key words: