Galileo orbit and clock quality of the IGS Multi-GNSS Experiment

Abstract

The Multi-GNSS Experiment (MGEX) of the International GNSS Service (IGS) aims at the data collection and analysis of all available satellite navigation systems. In particular the new global and regional satellite navigation systems are of interest, i.e., the European Galileo, the Chinese BeiDou, the Japanese QZSS as well as satellite based augmentation systems. This article analyzes the orbit and clock quality of the Galileo products of four MGEX analysis centers for a common time period of 20 weeks. Orbit comparisons of the individual analysis centers have a consistency at the 5–30 cm level. Day boundary discontinuities range from 4 to 28 cm whereas 2-day orbit fit RMS values vary between 1 and 7 cm. The accuracy evaluated by satellite laser ranging residuals is on the one decimeter level with a systematic bias of about −5 cm for all analysis centers. In addition, systematic errors on the decimeter level related to solar radiation pressure mismodeling are present in all orbit products. Due to the correlation of radial orbit errors with the clock parameters, these errors are also visible as a bump in the Allan deviation of the Galileo satellite clocks at the orbital frequency.

Introduction

The International GNSS Service (IGS, Dow et al., 2009) aims at the provision of highest quality Global Navigation Satellite System (GNSS) data, products and services (IGS Central Bureau, 2013). Since the end of 1993 the IGS provides precise GPS orbit products on an operational basis. In view of the emerging European Galileo and Chinese BeiDou navigation systems as well as regional navigation and augmentation systems like the Japanese Quasi-Zenith Satellite System (QZSS) and the Indian Regional Navigation Satellite System (IRNSS), the IGS initiated a Multi-GNSS EXperiment (MGEX).

The call for participation was published in August 2011 (Hugentobler, 2011). Major purpose of the experiment is “to collect tracking data from the new satellites and foster engineering experimentation with the new signals”. MGEX is carried out within the IGS Multi-GNSS Working Group and has basically the same structure as the operational IGS:

• Tracking Network: currently consisting of about 90 active tracking stations contributed by about 25 different institutions. An up-to-date station map is available at the MGEX web site http://igs.org/mgex/.

• Data Centers: providing observation data as well as the products generated by the analysis centers. Currently three data centers provide the MGEX products.¹

• Analysis Centers: processing the multi-GNSS observations in order to generate precise orbit and clock products of the new GNSSs.

The activities of the analysis centers (ACs) currently focus on Galileo and QZSS. Altogether six ACs provide orbit and clock products for MGEX, four of them are also involved in the operational GPS/GLONASS processing of the IGS:

• Centre National d’Etudes Spatiales (CNES) and Collecte Localisation Satellites (CLS)

• Center for Orbit Determination in Europe (CODE)

• Deutsches GeoForschungsZentrum (GFZ)

• European Space Agency (ESA)

• Japan Aerospace Exploration Agency (JAXA)

• Technische Universität München (TUM)

The GNSS considered by the ACs as well as the product availability at the end of 2013 are listed in Table 1. The JAXA products contain only QZSS whereas TUM provides Galileo and QZSS products. ESA is the only AC considering the BeiDou MEO satellites but its products are only available for a limited time period of 17 days. Therefore, ESA is not considered in this study. The other AC contributions are limited to Galileo (the inclusion of GPS or GLONASS in the MGEX products is not considered here). JAXA, CNES/CLS, and TUM provide products on an operational basis whereas the other ACs process the GNSS data from time to time in a campaign mode. The common time period of the four Galileo ACs considered in this study are 20 weeks from 28 April till 14 September 2013 (day of year 118–257/2013, GPS week 1738–1757).

In addition to the precise orbit and clock products discussed so far, a merged navigation file is generated on a daily basis by TUM and Deutsches Zentrum für Luft- und Raumfahrt (DLR) from the individual navigation files of 43 globally distributed multi-GNSS receivers. This product is available at the CDDIS² and IGN³ data centers since 1 January 2013 and covers GPS, GLONASS, Galileo, BeiDou, QZSS, and different Satellite Based Augmentation Systems (SBAS). IRNSS is currently not considered due to lack of commercially available receivers. A total of up to 85 navigation satellites is included in this product.

This article focusses on the orbit and clock quality of the four Galileo In-Orbit Validation (IOV) satellites. They are denoted by their RINEX (Gurtner and Estey, 2009) codes, i.e., E11 and E12 for the first pair of satellites launched in October 2011; E19 and E20 for the second pair launched in October 2012. First results and a more detailed description of MGEX are presented in Rizos et al., 2013, Montenbruck et al., 2013, Montenbruck, 2013, Montenbruck et al., 2014b. Montenbruck et al. (2013) analyzed one week of MGEX products and found an orbit consistency at the 14–36 cm level amongst three different ACs as well as a systematic bias in the satellite laser ranging (SLR) residuals of −7 cm for two of the three ACs. Hackel et al. (2014) computed orbit and clock products for Galileo E11 and E12 with a processing scheme similar to that of the TUM AC. They achieved an orbit accuracy on the one decimeter level but also found pronounced errors in the GNSS-only parameter estimation visible as 1 cycle-per-revolution (CPR) signal in the estimated clock parameters. These periodic variations of the apparent clock are highly correlated with the SLR residuals as also shown by Uhlemann et al. (2014). Similar errors were already identified in the orbit and clock parameters of the Galileo In-Orbit Validation Element (GIOVE) satellites by Montenbruck et al. (2012b). Hackel et al. (2014) showed that these errors could be mitigated by including SLR observations or applying clock constraints.

Montenbruck et al. (2014b) compared 8 months of TUM and CODE Galileo MGEX orbits and found a 3D RMS difference of 16 cm. They could confirm the $-5$ cm bias in the SLR residuals for E11 and E12 and also show that E19 and E20 have such a bias. They found a clear dependence of the SLR residuals on the elevation of the Sun above the orbital plane indicating deficiencies in the orbit modeling. Uhlemann et al. (2014) analyzed ten weeks of GFZ’s contribution to MGEX. They report orbit overlap RMS values of 6 cm for the Galileo IOV satellites and also did a first combination of the CODE, GFZ, and TUM orbit products resulting in an orbit agreement on the 3–10 cm level w.r.t. the combined product. Analysis of 13 months of CODE’s contribution to MGEX results in orbit differences at the day boundaries of 6–8 cm and longarc fit 3D-RMS values at the 2 cm level but also revealed systematic errors depending on the elevation of the Sun above the orbital plane in the estimated orbit and clock parameters (Prange et al., 2014).

Hauschild et al. (2013) analyzed the Rubidium Atomic Frequency Standards (RAFSs) of Galileo E11 and E12 and demonstrated an only slightly worse stability compared to the currently most precise RAFS clocks onboard the GPS II-F satellites. Uhlemann et al. (2014) report a clock stability of about 20 ps for the Passive Hydrogen Masers (PHMs) of all four IOV satellites.

Although several studies on Galileo MGEX products are already available as listed above, it is problematic to directly compare the corresponding orbit performance indicators as they were obtained with different approaches or implementations. Therefore, this article evaluates the CNES/CLS, CODE, GFZ and TUM Galileo orbits and clocks with the same methods in order to get an objective impression of their consistency and accuracy. Section 2 gives an overview of the software, models, and parameters of the different ACs. The orbit quality of the precise as well as broadcast orbits is assessed in Section 3 with day boundary discontinuities, orbit fit RMS values, orbit comparisons, and SLR residuals. Section 4 discusses the quality of the precise clock products.