Kan Wang, National Time Service Center, Chinese Academy of Sciences, Xi’an, China; School of Earth and Planetary Sciences, Curtin University, Perth, Australia; University of Chinese Academy of Sciences, Beijing, China; Ahmed El-Mowafy, School of Earth and Planetary Sciences, Curtin University, Perth, Australia; Chris Rizos, School of Civil and Environmental Engineering, UNSW, Sydney, Australia

View Abstract Sign in for premium content


In recent years, mega-constellation Low Earth Orbit (LEO) satellites have been proposed as an augmentation to the Global Navigation Satellite System (GNSS) for positioning on the ground, especially for those in measurement environments with limited satellite visibility. The fast geometry change of these LEO satellites also reduces the convergence time of Precise Point Positioning (PPP) techniques. To realize the benefits brought by these LEO satellites, their precise orbits and clocks need to be delivered to users, which would typically be based on processing the GNSS signals collected onboard LEO satellites. Assuming that this will be possible in the future, during data reception, storage and transmission, however, data gaps could exist in the collected GNSS measurements, which would result in gaps in the LEO clock estimates. The transmission of the LEO satellite clock corrections to users could also experience outages. In this study, taking the Ultra-Stable Oscillator (USO) onboard GRACE FO-1 as an example of LEO satellites that has similar operational conditions to the expected LEO mega-constellations, three different models are proposed for bridging clock gaps varying from 1 to 60 minutes. Model A considers its mid- to long-term systematic effects, Model B bridges the gaps using low-order polynomials employing the data near the gap, and Model C exploits the benefits of both Models A and B. Results show that Model A results in larger errors than the other two models for short clock gaps, while Model B could lead to a dramatic increase in the bridging errors for long gaps, e.g., 1h. Applying Model C for the USO on GRACE FO-1, the mean absolute bridging errors (in range) are within 1cm for gaps shorter than 10min, and within 0.2m for gaps not exceeding 1h. Increasing the polynomial degree of Model C from quadratic to cubic can lead to a reduction in the mean absolute bridging errors to mm- to cm-level.