A Quality Evaluation of Precise Point Positioning within the Bernese GPS Software Version 5.0

Soycan M.

ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING, vol.37, no.1, pp.147-162, 2012 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 37 Issue: 1
  • Publication Date: 2012
  • Doi Number: 10.1007/s13369-011-0162-5
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.147-162
  • Keywords: Precise point positioning, Global positioning system, Bernese GPS software, International GNSS service
  • Yıldız Technical University Affiliated: Yes


This study examines the degree of precision of precise point positioning system currently implemented in the Bernese GPS software 5.0 by zero differenced observations. The relationship between the accuracy of PPP and observation duration has been investigated using GPS data from a subset of IGS tracking stations observed over 7 days in different periods of year 2007. In the beginning, short reviews about PPP algorithm principle are given. Then, the results of several processing scenarios are presented and their respective precisions and convergence results are analyzed. Finally, with default parameters, accuracy assessments of PPP solutions are discussed. Investigation results have shown that PPP provides cm-level precision quite easily by using appropriate processing parameters within a few hours of observation. A faster convergence can be available for the north component in any case. The east component requires a convergence period more than two times longer than the north component. The east component provides a better precision than the height component after 2-3 h. For the shorter periods of observation, it is impossible to perform a reliable and accurate positioning with PPP due to the unresolved ambiguity. The repeatability and the precision ratio of the positioning are very low and unreliable for observations of a 5-min period. The mean values of RMS errors are about 0.25-0.30 m for north and east, and 0.5 m for up components. A great improvement can be achieved within a 30-min period, where mean RMS errors of 0.09 cm, 0.04 m and 0.11 m, respectively, in north, east and up were obtained. This precision could be improved up to three to five times by using, respectively, 2-8 h observations. It is possible to obtain at mm to cm-level precision at the end of the 12 h observation for all the coordinate components. Finally, 24 h data processing produced 0.0031, 0.0058 and 0.0088 m RMS errors for the north, east and up components in sequence. For all these cases, north components are more precise than up and east.