Abstract: In medium-long baseline GNSS kinematic relative positioning, the correlation of errors between the reference station and the rover station decreases with the increase of the baseline length. Consequently, the ambiguity parameters cannot be fixed instantaneously, which may decrease the precision of positioning. In multiple reference station scenarios, we can adaptively switch the reference station to a closer one to form a more rational baseline to guarantee the precision of RTK positioning. Aiming at solving the re-convergence of positioning results caused by reinitiating ambiguity after switching reference stations, a method of adaptively switching GNSS RTK reference stations in real time is proposed in this paper. By imposing the double-difference ambiguity between the two reference stations before and after switching, the precise prior information of double-difference ambiguity between the new reference station and rover station is derived, thus circumventing the ambiguity re-initialization after switching reference stations and maintaining the continuity of high-precision positioning results. The method described in this paper can be used in real-time positioning and can meet the demand for large-scale RTK high-precision continuous positioning. Numerical examples using data of Hong Kong CORS stations show that the proposed method can solve the re-convergence issue of positioning solutions after switching stations and continuous high-precision positioning results can be obtained after switching reference stations.

Key words: medium-long baseline, kinematic relative positioning, double-difference ambiguity, RTK, reference station switching