System-specific systematic errors in earth rotation parameters derived from GPS, GLONASS, and Galileo

Abstract

The earth rotation parameters (ERPs) are time-variable global geodetic parameters with a purely geophysical origin. Theoretically, the estimates of these parameters should be independent of the satellite constellation used in GNSS processing. Nonetheless, clear differences in the time series of ERPs are noticed when using different GNSS constellations. In this study, GPS, GLONASS, and Galileo estimates of ERP have been extensively evaluated in search of system-specific signals. Some of the processing details, such as modeling of the direct solar radiation pressure and length of the orbital arc, also have an impact on the ERP estimates. The GPS-based polar motion estimates are of better quality than those based on GLONASS and Galileo, which are susceptible to deficiencies in the orbit modeling. On the other hand, we observe a systematic bias of GPS-based length-of-day (LoD) with respect to the IERS-C04-14 values with a mean offset of − 22.4 µs/day. The Galileobased solutions are almost entirely free of this issue. The extension of the orbital arc in the GNSS processing from 1 to 3 days is superior for the quality of the ERPs, especially for pole coordinate rates and LoD. The spurious signals inherently influence the Galileo-based and GLONASS-based ERPs at the frequencies which arise from the resonance between the satellite revolution period and earth rotation, e.g., 3.4 days for Galileo and 3.9 days for GLONASS. These and the draconitic signals overshadow the GNSS-based ERP estimates. Although all the system-specific solutions are affected by the artificial signals, the combination of different GNSS mitigates most of the uncertainties and improves the ERP results.

Geocenter coordinates derived from multi-GNSS: a look into the role of solar radiation pressure modeling

“…The higher reduction for the GAL solution is, to a large extent, attributed to the mitigation of the Galileorelated orbital signals close to 3.4 days (Fig. 6) (Zajdel et al 2019b(Zajdel et al , 2020. The 3-day arc has naturally smoothened the signals with periods shorter than 6 days.…”

“…The computations are conducted using a development version of the Bernese GNSS Software (Dach et al 2015). The processing details and the overview of the background models are described in detail by Zajdel et al (2020) following the strategy used by CODE MGEX solution (Prange et al 2017). The GCC are estimated simultaneously with the orbit parameters, station coordinates, troposphere parameters, and ERP.…”

“…The longer orbital arc is tailored to reference frame realization and is designed to overcome some of the inherent weaknesses of the GNSS observations. However, the long arc might as well aggravate the solution quality if some significant orbit mismodeling exists (Zajdel et al 2020). In the case of the 1-day arcs, the stochastic pulses are estimated every midnight.…”

“…However, the box-wing alone is insufficient without additional correction model parameters. Thus, the hybrid strategy of combining both specific terms of empirical and physical models is beneficial as for the orbit quality (Li et al 2019) and the Galileo-based ERP (Zajdel et al 2020).…”

“…With only limited a priori knowledge about the non-conservative forces acting on satellites, we must incorporate additional empirical orbit parameters into the solution, i.e., Empirical CODE Orbit Model (ECOM) or JPL GSPM (Springer et al 1999; Bar-Sever and Kuang 2004). The errors in the orbit model, as well as the correlations between the estimated parameters, may introduce spurious orbit-related signals in the GNSS-based geophysical parameters, i.e., earth rotation parameters (ERP) and GCC (Meindl et al 2013;Zajdel et al 2020). Theoretically, an improvement may come from a decrease in the number of estimated parameters or, equivalently, from putting constraints on some of the estimated parameters.…”