Impact of combining GPS and GLONASS for reference frame densification using PPP

Abstract

Over the years, Global Navigation Satellite System (GNSS) has proved to be a space geodetic technique that offers crucial information in defining or realising Terrestrial Reference Frames (TRF). With a rigorous data processing technique and noise analysis, estimating GNSS station positions and velocities with sub-millimetre precision is possible. The positioning can be achieved with observations from the GNSS constellations such as Global Positioning System (GPS), GLObalnaya NAvigatsionnaya Sputnikovaya Sistema (GLONASS/GLO), Galileo (GAL), Beidou (BDS), among others. Positioning can also be achieved by combining two or more different GNSS constellations to increase the number of observations. This research investigated the impact of using a combined GPS+GLO and GLO-only daily PPP solution on time series noise properties and reference frame determination rather than when using a GPS-only solution. To do this, GPS_fixed, GPS_float, GLO_float, and GPS_fixed+GLO_float daily PPP solutions were generated using Centre for Orbit Determination in Europe (CODE/COD) repro3 products in the GipsyX/RTGx (GipsyX) GNSS processing software. The GPS ambiguity fixing was achieved in PPP via estimating GPS Wide-lane and Phase Biases (WLPB) in the GipsyX software using non-JPL (Jet Propulsion Laboratory) multi-constellation products. The stacked mean Power Spectral Density (PSD) show that the CODE GPS_fixed PPP solution has slightly lower noise than JPL GPS_fixed solution generated using the official JPL WLPB for all shorter periods (below 7 days) in all coordinate components. The CODE GPS_fixed+GLO_float solution has a noise level similar to the JPL GPS_fixed solution except at 2.67, 4.00, and 7.80-day periods, coinciding with GLO-specific orbit error periods. The CODE GLO_float solution presents a stacked mean PSD that has dominant noise power peaks around 2.67, 4.00, 5.30, and 7.80-day periods that mainly reduced with increasing latitude of stations. The difference in velocities between the most appropriate noise model combination of CODE GPS_fixed and the rest of the solution sets (CODE GPS_float, GLO_float, and GPS_fixed+GLO_float) were evaluated as residual velocities. GPS_fixed+GLO_float solution sets present residual velocities of -0.1, -0.3 and -0.1 mm/yr in the East, North and Up components. The residual velocities are observed to be even bigger in the GLO_float solution, which has 0.2, -0.2, and -0.4 mm/yr in the East, North, and Up components, respectively. A systematic bias was observed in the Up components for residual velocity between CODE and JPL GPS_fixed, which changes from 0.2 to -0.1 mm/yr with an increasing latitude of stations. Additionally, a systematic bias was observed in the Up component of residual velocities estimated using GLO_float solutions that change from 0.2 to -0.4 mm/yr with an increasing latitude of stations. There is a high impact of up to 0.4 mm/yr in CODE GPS_fixed+GLO_float solution mainly in the Up component of stations situated below 40⁰ absolute latitude region when the velocities were modelled as a white noise process against a combination of a white plus coloured noise model. In contrast, the research found that the impact is even higher in the CODE GPS_fixed solutions, which present up to 0.5 mm/yr velocity residuals mostly in the Up component of stations situated below 40⁰ absolute latitude region. The effect of noise modelling on velocity uncertainties was also examined