Populus euphratica is pivotal to ecological restoration in arid regions, while seed cleaning constitutes a major bottleneck for intensive seedling cultivation. To address the lack of simulation parameters for Populus euphratica seeds in cleaning device design, this study adopted a combined approach of physical experiments and simulations for parameter calibration. First, physical tests were conducted to measure the intrinsic parameters, contact parameters, and actual repose angle of the seeds. Subsequently, a discrete element model was established using Discrete Element Method software (EDEM). The model accuracy was verified by a comparative analysis of the experimental, theoretical, and simulated values of seed suspension velocity. Taking the actual repose angle of 18.69 degrees as the optimization target, the Plackett-Burman test, Steepest climbing test, and Box-Behnken test were performed for parameter calibration. Results showed that the simulated repose angle with optimal parameters reached 18.82 degrees, with a relative error of 0.696%. A two-sample t-test confirmed no significant difference (p > 0.05). This study provides valid parameters for the DEM simulation of Populus euphratica seeds. Furthermore, the proposed calibration method offers a reference for the fine calibration of seed parameters.