<p>Accurate estimation of reference crop evapotranspiration (<i>ET</i>_<i>0</i>) is fundamental for developing crop irrigation strategies and regional water resource management plans. This study improves <i>ET</i>_<i>0</i> estimation accuracy in the Shiyang River Basin by using the FAO-56-recommended Penman-Monteith (PM) model as the reference to evaluate and recalibrate ten temperature- and radiation-based empirical models. Before calibration, all models exhibited similar seasonal patterns to the PM model, with values increasing from early spring to a peak in July and then declining. However, considerable differences in estimation accuracy were observed among the models. The Irmak-Allen (IA), De Bruin-Keijman (DK), and Priestley-Taylor (PT) models showed the highest consistency with the PM model. After calibration of empirical parameters, all models demonstrated varying degrees of improvement in root mean square error (<i>RMSE</i>), relative root mean square error (<i>RRMSE</i>), and mean absolute error (<i>MAE</i>), along with increases in the index of agreement (<i>IOA</i>), Nash-Sutcliffe efficiency (<i>NSE</i>), and coefficient of determination (<i>R</i>^<i>2</i>). A comprehensive evaluation of model performance indicated that the Makkink (Mak), Irmak (IK), and PT models ranked among the top three in terms of overall accuracy and consistency. For practical applications, the radiation-based Mak and IK models are recommended when sufficient meteorological data are available, while the temperature-based Hargreaves (Har) model is more suitable under data-scarce conditions. The calibration of empirical parameters improved the applicability and reliability of <i>ET</i>_<i>0</i> estimation in the Shiyang River Basin.</p>