A new method for evaluation of the ionospheric modeling error in a VRS based Network RTK service

The overall goal of any Network Real Time Kinematic (NRTK) service is to provide the best position accuracy. However evaluating the NRTK service in the positioning domain does not give a full insight in how well the NRTK service is performing as this also includes errors from the rover. The new method presented in this paper provides a possibility to evaluate a NRTK service, based on the Virtual Reference Station (VRS) technique independent of the rover. This is done by comparing observations from a VRS generated at the exact same location as a true reference station. This paper focus on the ionospheric modeling error in the NRTK service, and therefore the geometry free observation is used. The ionospheric modeling error in the NRTK service is extracted and analyzed. Nine true GNSS stations with observations from a day with low and a day with higher ionospheric activity has been chosen to identify dependencies based on satellite elevation, distance to nearest reference station in the interpolation area, distance to nearest reference station in the extrapolation area, and the dependency on the daily ionospheric activity. The results show that the elevation angle dependency is the largest contribution to the modeling error with a standard deviation just above 0.5 cm for low elevation angles, and decreasing to the mm level for high elevation angles. When the distance to the nearest reference station in the network is increased, the elevation modeling error is also increasing. The results also shown that with increased distance to nearest reference station in the interpolation area, a trend referring to the daily ionospheric cycle can be seen. This trend is magnified for the day with higher ionospheric activity. The ionospheric influence on the model error has shown, that with the current (February 2009) level of ionospheric activity, the station separation can be rather large, before a significant degradation of the accuracy will occur. With the increasing level of ionospheric activity over the next years, a magnified model error can be expected. The station separation can thereby become a limiting factor for the modeling of the ionosphere in a NRTK.