Analysis and Normalization of the GPS and GLONASS Satellite Clocks

P. V. Movsesian, S. D. Petrov, D. A. Trofimov, I. V. Chekunov

Transactions of IAA RAS, issue 52, 36–39 (2020)

DOI: 10.32876/ApplAstron.52.36-39

Keywords: GNSS, clock modelling, PPP-method

Abstract

Nowadays, the determination of precise coordinates using GNSS observations can be made with a differential method or with an absolute solution (PPP method – Precise Point Positioning). The precision of navigational data, such as satellite ephemerides and satellite clock biases, is crucial to the PPP method. Usually satellite orbits are approximated with smooth functions. Unlike ephemerides, satellite clock biases are not approximated with smooth function, and clock series are processed not as a continuous time series but as daily fragments, which leads to jumps at 00:00 UTC and during a day. The analysis of clock biases leads us to the conclusion that the aforementioned jumps don’t represent real clock behaviour. They are a result of using incorrect processing methods. This conclusion is also shared by a number of researchers. Different studies propose a variety of satellite clock improvement methods but clock jump removal is performed manually, which is an obstacle for large data array processing. We propose an automatic jump correction algorithm. Determination of quadratic trend and Kalman filter are employed by other methods in some way or another. Besides, we assume the existence of some additional linear trends that were not properly processed originally. We have processed about two years of satellite clock biases and have come to the conclusion that the use of our algorithm allows the removal of clock jumps on time intervals of any length.

Citation