ION GNSS 2012
Session E6: Precise Point Positioning 2

Title: Satellite Clock Corrections for Real Time PPP: A Case Study in the Context of the Brazilian Continuous GPS Network
Author(s): H.A. Marques, Federal University of Pernambuco, Brazil; J.F. Galera Monico, M.H. Shimabukuro, R.T. Oyama, Sao Paulo State University, Brazil; M. Aquino, University of Nottingham, UK
Room: 209/210 (NCC)

Among the several possible methods of GNSS positioning, those that provide real time positioning (or near real time) with accuracy near to the decimeter or centimeter level are currently in evidence. In this case, we highlight the real time PPP method, which requires the availability of real time precise orbits and corrections of the satellites clocks. Currently, it is possible to use the predicted IGU ephemerides, which are made available by the IGS centers. However, the satellites clocks corrections available in the IGU do not provide enough accuracy (3 ns 0.9 m) to accomplish real time PPP with the level of the required accuracy. Therefore, for real time PPP application it is necessary to further research and develop appropriate methodologies for estimating the satellite clock corrections in real time with better accuracy. The estimation of satellite clock corrections can be performed based on a GNSS network of reference stations. Some investigations have been proposed for the estimation of the satellite clock corrections using GNSS code and phase observable at the double difference level between satellites and epochs. Another possibility consist of applying a Kalman Filter in the network PPP mode to estimate the satellite clock corrections together with other parameters such as phase ambiguities, zenith tropospheric delays, receiver clock corrections, etc. In the latter, all the systematic effects involved with the GNSS satellite signals must be modeled appropriately for each station of the network. There is another possibility of estimating the satellite clock corrections in real time, which consists of the integration of both methods, using network PPP and observables at double difference level in specific time intervals. The methodology adopted in this work consists in the estimation of the satellite clock corrections based on the adjustment of data in the PPP mode for a network of GNSS stations, the so called ´network PPP´. Once the corrections of the satellite clocks are estimated in real time, they must be made available to the user, for real time PPP to be carried out at the station of interest. Therefore, the system has to be able to communicate in real time PPP mode by using an appropriated protocol, such as NTRIP protocol, which was developed by Federal Agency for Cartography and Geodesy (BKG). To estimate satellite clock correction and to accomplish real time PPP two pieces of software have been developed, respectively, "RT_PPP" and "RT_SAT_CLOCK". The system (RT_PPP) is able to process GNSS code and phase data using precise ephemeris and precise satellites clocks corrections. The PPP processing considers the absolute satellite antenna Phase Center Variation (PCV), Ocean Tide Loading (OTL), Earth Body Tide, Phase Windup, among other effects. The troposphere can be estimated as a random walk process in the Kalman filter or corrected using the Brazilian Weather Forecast Model (BWFM) from CPTEC (Centro de Previsao de Tempo e Estudos Clim ticos) or even the European Weather Forecast Model (ECMWF) data available together with the Vienna Mapping Function (VMF). The ionosphere effect can be eliminated by the ionospheric-free observable or estimated by introducing a pseudo-observable in the Kalman Filter, which is treated as a stochastic process. In the software RT_SAT_CLOCK we apply a Kalman filter to estimate satellite clock correction in the network PPP mode. In this case, all corrections must be applied for each station. The software can process GPS data using two types of observables: code smoothed by carrier phase or undifferenced code together with carrier phase. In the former, we estimate receiver clock error; satellite clock correction and troposphere, considering that the phase ambiguities are eliminated when applying differences between consecutive epochs. However, when using undifferenced code and phase, the ambiguities must be estimated together with receiver clock errors, satellite clock corrections and troposphere. In both strategies it is also possible to correct the troposphere delay from Numerical Weather Forecast Model instead of estimating it. The prediction of the satellite clock correction can be performed using a straight line or a second degree polynomial. When estimating ambiguities together with other parameters, there is a considerable number of parameters to be estimated. To reduce this burden, the adjustment is performed using a technique for matrix optimization. The results were generated in real time and post-processed mode (simulating real time). To estimate the satellite clock corrections we used data from the Brazilian continuous GPS network and also from the IGS network in a global satellite clock solution. The estimation of the satellites clock corrections was based on the measurements of the pseudorange smoothed by carrier phase and also using the undifferenced pseudorange and phase with ambiguities estimation for each satellite available at each station. We have used satellite IGU orbits to estimate the satellite clock corrections performing the updates as soon as new ephemeris files are available. The daily precision of the estimated satellite clock corrections reached the order of 0.15 nanoseconds. The estimated satellite clock was applied in the static and kinematic PPP positioning for Brazilian network stations and the results show that it is possible to accomplish real time PPP in the static and kinematic modes with accuracy of the order of 10 to 20 cm, respectively.

Return to Session E6