Estimating ambiguity fixed satellite orbit, integer clock and daily bias products for GPS L1/L2, L1/L5 and Galileo E1/E5a, E1/E5b signals

Abstract

Ambiguity resolution of a single receiver is becoming more and more popular for precise GNSS (Global Navigation Satellite System) applications. To serve such an approach, dedicated satellite orbit, clock and bias products are needed. However, we need to be sure whether products based on specific frequencies and signals can be used when processing measurements of other frequencies and signals. For instance, for Galileo E5a frequency, some receivers track only the pilot signal (C5Q) while some track only the pilot-data signal (C5X). We cannot compute the differences between C5Q and C5X directly since these two signals are not tracked concurrently by any common receiver. As code measurements contribute equally as phase in the Melbourne-Wuebbena (MelWub) linear combination it is important to investigate whether C5Q and C5X can be mixed in a network to compute a common satellite MelWub bias product. By forming two network clusters tracking Q and X signals, respectively, we confirm that GPS C5Q and C5X signals cannot be mixed together. Because the bias differences between GPS C5Q and C5X can be more than half of one wide-lane cycle. Whereas, mixing of C5Q and C5X signals for Galileo satellites is possible. The RMS of satellite MelWub bias differences between Q and X cluster is about 0.01 wide-lane cycles for both E1/E5a and E1/E5b frequencies. Furthermore, we develop procedures to compute satellite integer clock and narrow-lane bias products using individual dual-frequency types. Same as the finding from previous studies, GPS satellite clock differences between L1/L2 and L1/L5 estimates exist and show a periodical behavior, with a peak-to-peak amplitude of 0.7 ns after removing the daily mean difference of each satellite. For Galileo satellites, the maximum clock difference between E1/E5a and E1/E5b estimates after removing the mean value is 0.04 ns and the mean RMS of differences is 0.015 ns. This is at the same level as the noise of the carrier phase measurement in the ionosphere-free linear combination. Finally, we introduce all the estimated GPS and Galileo satellite products into PPP-AR (precise point positioning, ambiguity resolution) and Sentinel-3A satellite orbit determination. Ambiguity fixed solutions show clear improvement over float solutions. The repeatability of five ground-station coordinates show an improvement of more than 30% in the east direction when using both GPS and Galileo products. The Sentinel-3A satellite tracks only GPS L1/L2 measurements. The standard deviation (STD) of satellite laser ranging (SLR) residuals is reduced by about 10% when fixing ambiguity parameters to integer values.