Improved precise kinematic LEO orbits based on the raw observation approach

Observing Earth's gravity field variations by means of precise orbits is nowadays a common practice next to dedicated gravity field satellite missions. The reason for using kinematic orbits of low earth orbiting (LEO) satellites for gravity field estimates instead of dynamic or reduced-dynamic orbits is based on the fact that kinematic orbits do not include a priori information about the gravity field. To achieve highly accurate gravity field estimates through the use of LEO satellite positions, it is of paramount importance to determine these positions as precisely as possible. We investigated whether it is possible to improve the accuracy of the kinematic orbits that had been processed by our working group in the past. For this purpose we modified our processing approach, which is based on an iterative least-squares adjustment using raw observations, and re-processed the kinematic orbits for several satellite missions such as CHAMP, GRACE, GRACE Follow-On, Jason, MetOp, Swarm, Sentinel, TanDEM-X and TerraSAR-X. The modifications include that we are now processing global navigation satellite system (GNSS) constellations in-house and that the phase ambiguities are fixed to integer values. In this paper we discuss these modifications. We show that the consistent in-house processing of the GNSS products already significantly improves the orbit determination, since the use of the these GNSS products mitigates the performance degradation. Furthermore, we present that an additional enhancement through integer ambiguities can be achieved.