Though Global Navigation Satellite System (GNSS) data processing has been significantly improved over years it is still commonly observed that Zenith Tropospheric Delay (ZTD) estimates contain many outliers which are detrimental to meteorological and climatological applications. In this paper, we show that ZTD outliers in double difference processing are most of the time caused by sub-daily data gaps at reference stations which cause disconnections of clusters of stations from the reference network and common–mode biases due to the strong correlation between stations in short baselines. They can reach a few centimetres in ZTD and coincide usually with a jump in formal errors. The magnitude and sign of these biases are impossible to predict because they depend on different errors in the observations and on the geometry of the baselines. We elaborate and test a new baseline strategy which solves this problem and significantly reduces the number of outliers compared to the standard strategy commonly used for positioning (e.g. determination of national reference frame) in which the pre-defined network is composed of a skeleton of reference stations to which secondary stations are connected in a star-like structure. The new strategy is also shown to perform better than the widely-used strategy maximising the number of observations which available in many GNSS software. The reason is that observations are maximised before processing whereas the final number of used observations can be dramatically lower because of data rejection (screening) during the processing. The study relies on the analysis of one year of GPS (Global Positioning System) data from a regional network of 136 GNSS stations processed using Bernese GNSS Software v.5.2. A post-processing screening procedure is also proposed to detect and remove a few outliers which may still remain due to short data gaps. It is based on a combination of range checks and outlier checks of ZTD and formal errors. The accuracy of the final screened GPS ZTD estimates is assessed by comparison to ERA-Interim reanalysis.
展开▼
