A method is introduced to directly estimate the model coefficients of global navigation satellite system (GNSS) satellite clock errors using global or local GNSS carrier phase and pseudorange observations. The primary objective is to ensure the timely delivery of GNSS satellite clock error services. The model coefficients are estimated simultaneously with tropospheric delay, receiver clock error, and phase ambiguity parameters. These estimated model coefficients are then utilized to provide a new final product and construct the broadcast ephemeris and high-precision real-time services (RTS) of the satellite clock error. The results indicate that using the estimated model coefficients simplifies providing final products and significantly reduces the memory space required for user storage. By estimating the parameters over 1 h and utilizing a combination of quadratic polynomial and eighth-order harmonic-based functions, the GNSS static and kinematic precise point positioning (PPP) performance is found to be comparable to using the international GNSS service (IGS) final products. Evaluating the constructed broadcast ephemeris, which utilizes linear and quadratic polynomials, demonstrates varying degrees of improvement across four service schemes of 1, 2, 3, and 4 h ages. These improvements enhance the accuracy and timeliness of the broadcast ephemeris satellite clock error service. The estimation is conducted over 1 h for high-precision RTS, employing a combination of quadratic polynomial and eighth-order harmonic-based functions. The experiment shows that at a 1-min update interval, the average standard deviation for GPS, BDS-3, and Galileo satellite clocks achieves 0.103, 0.098, and 0.08 ns, respectively. When comparing the real-time (RT) broadcast of the estimated satellite clock correction coefficients with update intervals of 1, 2, 5, and 10 min to the IGS RT archived products, there is a slight degradation in GPS PPP positioning errors. However, for BDS-3 and Galileo, the four schemes provide comparable or even better performance than the archived RT products. Overall, the proposed method saves time when fitting model coefficients to satellite clock errors and ensures reliable real-time PPP positioning, especially in situations with poor communication.
A method is introduced to directly estimate the model coefficients of global navigation satellite system (GNSS) satellite clock errors using global or local GNSS carrier phase and pseudorange observations. The primary objective is to ensure the timely delivery of GNSS satellite clock error services. The model coefficients are estimated simultaneously with tropospheric delay, receiver clock error, and phase ambiguity parameters. These estimated model coefficients are then utilized to provide a new final product and construct the broadcast ephemeris and high-precision real-time services (RTS) of the satellite clock error. The results indicate that using the estimated model coefficients simplifies providing final products and significantly reduces the memory space required for user storage. By estimating the parameters over 1 h and utilizing a combination of quadratic polynomial and eighth-order harmonic-based functions, the GNSS static and kinematic precise point positioning (PPP) performance is found to be comparable to using the international GNSS service (IGS) final products. Evaluating the constructed broadcast ephemeris, which utilizes linear and quadratic polynomials, demonstrates varying degrees of improvement across four service schemes of 1, 2, 3, and 4 h ages. These improvements enhance the accuracy and timeliness of the broadcast ephemeris satellite clock error service. The estimation is conducted over 1 h for high-precision RTS, employing a combination of quadratic polynomial and eighth-order harmonic-based functions. The experiment shows that at a 1-min update interval, the average standard deviation for GPS, BDS-3, and Galileo satellite clocks achieves 0.103, 0.098, and 0.08 ns, respectively. When comparing the real-time (RT) broadcast of the estimated satellite clock correction coefficients with update intervals of 1, 2, 5, and 10 min to the IGS RT archived products, there is a slight degradation in GPS PPP positioning errors. However, for BDS-3 and Galileo, the four schemes provide comparable or even better performance than the archived RT products. Overall, the proposed method saves time when fitting model coefficients to satellite clock errors and ensures reliable real-time PPP positioning, especially in situations with poor communication.