Filtering Errors in LEO Trajectories Obtained by Kinematic GPS with Floated Ambiguities

We have developed and tested a way to precisely locate a low earth orbiter (LEO) carrying an on-board GPS receiver. The idea is to first use long-range kinematic GPS (lacking dynamic constraints), and then make an orbit fit (constrained by orbit dynamics) to the resulting trajectory, in order to filter out the kinematic errors. These errors tend to be rather large in satellite trajectories, because both the fast-changing subset of GPS spacecraft in view from the LEO, and the very long baselines between this and the fixed ground stations, together make ambiguity floating imprecise, and ambiguity fixing unreliable. The procedure outlined here requires a relatively small number of ground sites distributed around the world, selected from the much larger set of IGS stations on the basis of their consistently good performance. This method could be useful when processing altimetry and other satellite data requiring good geolocation. We have implemented this method at Goddard SFC using almost entirely pre-existing software. As an example, we have calculated two 24-hour orbit estimates of the oceanographic satellite TOPEX/ Poseidon, and another two for the satellite JASON. The resulting orbits agree to better than 5 cm RMS in height and 17 cm in three-dimensional RMS with the NASA Goddard Space Flight Center Precise Orbit Estimates (POE). Those POE, distributed in the case of TOPEX as part of the Geophysical Data Records, have been derived exclusively from DORIS Doppler and laser tracking data, so they provide an entirely independent way to verify our GPS-based results.