Modeling Variations in GNSS Precise Frequency Transfer: Focus on Troposphere Effects

Global Navigation Satellite Systems (GNSSs) are indispensable in time and frequency dissemination networks between national metrology institutes worldwide. This is as result of the continuous development in GNSS hardware and satellites’ on-board clocks. Given that receiver clock error is the parameter of interest, significant efforts are therefore focused on enhancing the stability and accuracy of GNSS-based frequency transfer by advancing the estimation algorithms. Also, GNSSs observations are influenced by various error sources and there is urgent need to develop models and algorithms handling those errors. Where two clocks are usually involved in real scenarios for frequency comparison, one clock can also be used as in a common-clock configuration. In which it is used to calibrate and eventually model errors affecting the stability of the links without being affected by stability of frequency sources; such as optical fiber, and GNSS. Troposphere is one error source influencing frequency stability due to its irregular water vapor content, which has been defined as tropospheric wet delay. This study benefits therefore from common-clock experiment over optical fiber and investigates the impact of troposphere delay on frequency transfer stability using GNSS links. Troposphere influence is expected to be more prominent over long baseline, which has been setup to be 52 km in this experiment. Single differences common-view algorithm is proposed in this analysis as the main tool to study the effects of errors. New features have been integrated in the algorithm to have more robust and reliable estimation of the relative tropospheric delay over longer baselines. For this, least-squares adjustment is implemented with constraints to evaluate the stochastic models of the corresponding troposphere delay errors, and to interpret the physical meaning of such fluctuations. Single differences algorithm has achieved excellent preliminary stability in range of 5 · 10?17 at one-day averaging over 52 km baseline. This has been improved by constraining troposphere delay in an extended least-square algorithm. In addition, relative tropospheric delay estimates are evaluated with varying standard deviation, and temporal resolution. Both show impact on the estimated receiver clock errors, hence the stability of GNSS-based frequency transfer.