Abstract: BDS-3 has launched global services, and the service area of the BDS Klobuchar (BDSklob) model has also expanded to a global scale. The global performance of the BDSklob model has also attracted great attention. In response to the low accuracy of the BDSklob model outside the service area and abnormal correction in polar regions during the BDS-2 period, this paper proposes a new BDSklob model refinement scheme-multi-source data refinement method-based on parameter refinement method, using empirical models IRI-Plas-2017, BDGIM model, and multi-source data from CODE’s GIM products. The results show that the refinement methods of various data sources have significantly improved the performance of the BDSklob model, especially in polar regions; BDSklob_ C (GIM product with CODE data source) has the highest processing accuracy; BDSklob_ B (data source is BDGIM) takes the second place in accuracy, but without the help of external data sources, refinement processing can be completed in the Beidou system; BDSklob_ the accuracy of I (data source is IRI model) is slightly poor, but based on the predictive ability of empirical models, it can meet the needs of real-time refinement processing.