The practical example of the GIOVE+GPS satellites estimation with the use of the GIOVE-M network of 13 stations has been reviewed and compared with IGS methodology. A summary of the comparison is provided in Table 6.5 against the state of the art in POD achieved by IGS.
GPS GIOVE
Radial Clock Radial Clock IGS 0.02 0.02
GGSP 0.02 0.03 0.31 0.43
ODTS 0.10 0.25 0.32 0.30
Tab. 6.5: Radial versus clock precision (1σ) for GPS and GIOVE satellites in nanoseconds [ns]
GPS satellites achieve for clock products 0.07 ns (rms) accuracy and 0.02 ns (1σ) precision for orbit and radial components. The precision of GIOVE estimations by GGSP and GIOVE-M is 0.3 ns (1σ) with an estimated accuracy of 0.5 ns (rms), which is in line with SLR residuals.
Better performance for GPS than for GIOVE satellites is expected due to the higher number of sensor stations. This hypothesis has been confirmed by analyzing the GPS estimation depen-dency on the number of sensor stations and by the fact that longer estimation arcs of 5 days increase the accuracy of the orbit estimation with respect to the nominal 1-3 days arcs used in IGS.
The accuracy limit of the geodetic time transfer is expected to be 0.1 ns. It has been reviewed with respect to TWTSFT time transfer. Both have been demonstrated to be consistent at 2 ns level, TWTFST being noisier at short interval times but converging to POD at longer integration times.
Validation of group delays estimations has been demonstrated as a challenging task. GIOVE satellites and new frequency combinations allow for a deeper insight into these values. IFB for GIOVE satellites have been reviewed with uncertainties below 5 nanosecond level with respect to the satellite calibrated values. On the stability side, satellite group delays have been demon-strated to not be as stable as assumed, with absolute variations due to changes at the stations and seasonal and sub-daily variations. Since the station group delays (ISB and IFB) are assumed to be constant during the estimation arc, any variation will be propagated into the reference time scale and satellite clocks.
The average behaviour of the best H-maser estimations can be considered to be the limit of the geodetic time transfer. The analysis of stations with active H-maser confirms that the stability of geodetic time transfer is 1E-12τ−1/2and therefore still noisier than the signal provided by the H-maser (1E-13τ−1/2) and at the limit of new Galileo PHM and Block-IIF frequency standards (1E-12τ−1/2).
Due to the limited number of stations and group delay instabilities at the stations, the achieved geodetic time transfer stability by GIOVE mission (2.2E-12τ−1/2) has been confirmed to be
6.6 Conclusions
noisier than the PHM and best performing RAFS in GIOVE satellites. As explained in Chapter 2, no European atomic clock had previously been launched into space and the verification of its in-orbit performance was the second main objective of the GIOVE mission. Another methodo-logy was required to verify the in-orbit performance of GIOVE clocks.
A new methodology has been proposed within this thesis using carrier phase measurements obtained at a station connected to a H-maser as frequency source. The new proposed method-ology has been described and implemented in a dedicated piece of software and then validated with GPS satellites by comparison against IGS results with an excellent agreement. For the first time, the short term behaviour below 300 seconds not covered by IGS final products has been characterized, allowing for, in combination with POD results, full characterization of GNSS clocks from 1 second on.
Once its suitability to characterize GNSS clocks was confirmed, it has been applied to GIOVE clocks. As a result, it has now been proved how the short term stability of RAFS and PHM are in line with the ground measurements, being even possible to identify the activated RAFS unit from the agreement. Additionally, the same methodology has subsequently been modified and successfully applied to satellite ground tests to validate the stability of the signal clock on ground.
These analyses have allowed the validation of this novel methodology, the first full character-ization of GIOVE and GPS clocks, and the successful achievement of the second main objective of the GIOVE mission.
7 Harmonics in satellite clocks
7.1 Introduction
During the accuracy assessment performed in the previous Chapter 6.6 the harmonics in GNSS apparent clocks have been identified as an ambiguous effect difficult to be attributed to clock or orbit sources due to the coupling between both components. Once the difference between the ground AFS stability and POD stability has been understood it remains to explain the ’bump’
in the Allan Deviation introduced by a harmonic of typically 0.5 amplitude in GIOVE PHM
’apparent’ clock phase.
All GNSS signal clocks show a periodic fluctuation. Harmonics in GPS satellites are a well-known feature since the early estimations of GPS clocks [158]. The impact on the clock predic-tion was also early acknowledged, IGSMAIL-3057 already in 2000 suggested to the different analysis centers to include the harmonic in the prediction for ultrarapid products. Neverthe-less, their characteristics and origin have been only recently characterized [150]. Amplitudes of several nanoseconds are reported for Block-IIA satellites while values lower than 0.2 ns are observed in Block-IIR. Amplitudes for GIOVE satellites are definitively larger with values in the order of 1 ns for GIOVE-A RAFS and 0.5 ns for GIOVE-B. This periodic fluctuation in phase seems to be always present with a period analogous to the orbital one (≈14 hours). To understand if this is a real physical phenomenon in the ’signal clock’ or whether is a residual of the POD in the ’apparent clock’ several analysis steps may be performed.
First the harmonics must be confirmed by different independent estimations. Once the har-monic is confirmed not to be an artifact of a single processing, three possible causes can be identified:
1. Frequency variation originating in the atomic frequency standard (the pure physical clock) due to sensitivity to temperature.
2. Group delay variation (in the signal clock) originating from one or several payload units due to sensitivity to temperature.
3. Orbit residuals (in the apparent clock) due to an estimation error of the orbit. As covered in previous sections radial orbit errors and clock errors are close correlated in the POD process.
The next Chapter will analyze the impact of the prediction to the user in Section 8.6. In this chapter the route source of the harmonic in GNSS clocks is analyzed in order to identify the origin, which shall be understood towards the implementation of possible mitigation strategies at system or user level.