Item

Abstract

The deviations of the microwave signal transmitting point of Global Navigation Satellite Systems (GNSS) satellites with respect to the corresponding center-of-mass were and still are significant bias sources for GNSS-based terrestrial reference frames (TRF). Satellite phase center offsets (PCOs) and variations (PVs) are usually derived from GNSS observations from the tracking network of the International GNSS Service (IGS). Because of the strong correlation between the scale of the TRF and the satellite PCOs in the z-direction (z-PCOs), a no-net-scale (NNT) condition relative to, for instance, the International Terrestrial Reference Frame (ITRF) is commonly applied. One approach for estimating transmitter phase centers without constraining the scale is to consider space-based GNSS observations. Within this study, we estimate the satellite PCOs of the Global Positioning System (GPS) based on zero-difference ionosphere-free observations from six low Earth orbiters (LEOs) and ground networks with different numbers of stations. The six LEOs are different in their orbits, GPS receiver equipment, and tracking characteristics. We jointly estimate orbits, station coordinates, and GPS satellite PCOs in an integrated processing. The horizontal offsets are shown to significantly benefit from the LEOs onboard observations by reducing the correlation between these offsets and the GPS satellite orbit and attitude. Without applying a no-net-scale condition to the ground network, the GPS z-PCOs are estimated. By adding six LEOs, the correlation coefficients between the GPS z-PCOs and the scale is reduced significantly (about from 0.85 to 0.3), consequently, the precision of estimation is improved. When including the six LEOs, the GPS z-PCOs estimated without applying NNS are very stable. The estimated GPS z-PCOs have a −231 mm difference in average to the values in igs14_2134.atx and the corresponding scale to IGS14 reference frame is +1.89 part per billion. The estimated GPS z-PCOs agree with previous studies based on Swarm satellites and Galileo. The improvement due to different numbers of LEOs and the impact of LEO z-PCO errors on the estimation are discussed.