Determination of GPS Orbits to Submeter Accuracy

Orbits for satellites o f the Global Positioning System (GPS) have been determined with submeter accuracy. Tests used to assess orbit accuracy include orbit comparisons from independent data sets, orbit prediction, ground baseline determination, and formal errors. One satellite tracked for 8 hours each day shows rms errors below I m even when predicted more than 3 days outside of a 1-week data arc. Differential tracking of the GPS satellites in high Earth orbit provides a powerful relative positioning capability, even when a relatively small continental U. S. fiducial tracking network is used with less than one-third of the full GPS constellation. To demonstrate this capability, baselines of up to 2000 km in North America were also determined with the GPS orbits. The 2000-km baselines show rms daily repeatability of 0.3 to 2 parts in lo8 and agree with very-long-baseline interferometry (VLBI) solutions at the level o f 1.5 parts in lo8. This GPS demonstration provides an opportunity to test different techniques for high-accuracy orbit determination for high Earth orbiters. The best GPS orbit strategies included data arcs of at least I week, process noise models for tropospheric fluctuations, estimation of GPS solar pressure coefficients, and combined processing of GPS cam'er phase and pseudorange data. For data arcs of 2 weeks, constrained process noise models for GPS dynamic parameters significantly improved the solutions.