Mitigating the Impact of Predicted-Satellite-Clock Errors on GNSS PPP Positioning

摘要

The United States Naval Observatory (USNO) is testing the use of predicted values of (a) GNSS satellite orbits,(b) GNSS satellite clock corrections and (c) earth orientation parameters (EOP) as input to GNSS precise point positioning (PPP; [1]). Past studies [2, 3] indicated that static positioning precision of several centimeters to a decimeter could be achieved using such predicted values, and that errors in the predicted satellite-clock correction values were the largest contributors to the resulting position error. This study was undertaken therefore in an effort to understand how errors in input satellite clock correction values are converted in PPP processing to errors in receiver position and other estimated quantities. A simulation study was made of the impact of a constant 10 ns error in-a single GPS satellite's clock estimates on the position and clock accuracy obtained in PPP parameter estimation. The 10 ns constant error was applied to the clock estimates of PRN 1 (GPS 63 [4]); the impact was studied at five mid-latitude northern hemisphere locations for test date 21 October 2012. Said input error caused no discernable error in estimated site position. Nor did it create any carrier-phase residuals. The error manifests rather as receiver clock errors, ambiguity errors, and pseudorange residuals. PRN 1 pseudorange residuals accounted for 96-98% of the input error, with the remainder of the input error expressed as receiver clock errors of 220-390 ps, ambiguity errors of the same size but opposite sign, and non-PRN 1 pseudorange residuals. The impact of a PRN 1 10 ns satellite-clock error on position/clock accuracy was further investigated by repeating the above experiment but using only pseudorange measurements to obtain the answers. In this case, cm-to-dm level position errors and ns-level clock errors were induced. Further studies must be conducted in order to understand how these large errors transition to the small errors observed in the tested PPP pseudorange+phase parameter estimation, in which (a) many ambiguity parameters were available to absorb input satellite clock error and (b) the pseudorange measurements were given 10~(-4) weight of the carrier-phase measurements.