Accuracy/Precision of USNO Predicted Clock Estimates for GPS Satellites
Abstract
The United States Naval Observatory (USNO) Earth Orientation Department GPS Analysis Division (GPSAD) processes data from hundreds of GNSS receivers daily, computing multiple GNSS carrierphase- based product sets 365 days/year. Quantities estimated typically include high-precision (1-3 cm) GNSS satellite orbits/clock corrections, receiver clock corrections, station coordinates, earth-orientation parameters (EOPs), and troposphere-delay estimates. The goal of the research presented herein is to improve the accuracy of USNO ultra-rapid satellite clock predictions. Reason: Precise point positioning (PPP [1]) experiments conducted in Summer/Fall 2012 using USNO ultra-rapid predicted satellite clocks, satellite orbits and EOPs [2][3] indicated that improved position accuracy (output) would require improved satellite clock prediction accuracy (input). The current USNO clock prediction algorithm uses linear modeling. A quadratic prediction algorithm tested degraded the accuracy of satellite clock predictions rather than improving it. Error analysis of results obtained using the extant linear prediction algorithm suggested that prediction errors for Block IIA satellites equipped with Cs clocks were larger than prediction errors associated with other satellites, consistent with results obtained in [2]. We were unable to identify an error source uniquely impacting those satellites. However, we demonstrated that PPP positioning might be improved by deleting measurements from Block IIA satellites. This suggests that PPP position errors caused by poorly estimated satellite clocks could be mitigated by de-weighting or deleting the measurements associated with those satellites.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2012
- Accession Number
- ADA580019
Entities
People
- Christine Hackman
Organizations
- United States Naval Observatory