13:45 〜 14:00
[G06-1-02] A new baseline processing strategy for GNSS meteorology
Though GNSS data processing has been significantly improved over years it is still commonly observed that tropospheric parameters (ZTD and gradients) contain many outliers. Detection and removal of outliers in GNSS time series is an important step before their use in meteorology and climatology. We show that outliers in double difference processing are often caused by defects in the baseline strategy. We elaborate and test a new baseline strategy which solves this problem and significantly reduces the number of outliers.
One year of data is analyzed from a network of 136 GNSS stations. Three baselines strategies are inter-compared: 1) a 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; 2) the widely-used “obs-max" strategy available in Bernese GNSS software (e.g. for processing of regional to global networks); and 3) our new baselines strategy.
It is shown that many outliers are due to data gaps at reference stations which cause disconnections of clusters of stations from the reference network. These outliers are common–mode biases due to the strong correlation between stations in short baselines. They can reach a few centimeters in ZTD and can be detected by a jump in formal errors. The magnitude and sign of the bias is impossible to predict, because it depends on different errors in the observations and on the geometry of the baselines. Several cases are illustrated. This defect is shown to strongly impact the 2 most widely used strategies. Our new strategy ensures that no station is disconnected from the main network and that all clusters include long baselines necessary to estimate absolute tropospheric parameters. This strategy might help to improve significantly the quality of GNSS tropospheric parameters estimated in national to regional networks for meteorology.
One year of data is analyzed from a network of 136 GNSS stations. Three baselines strategies are inter-compared: 1) a 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; 2) the widely-used “obs-max" strategy available in Bernese GNSS software (e.g. for processing of regional to global networks); and 3) our new baselines strategy.
It is shown that many outliers are due to data gaps at reference stations which cause disconnections of clusters of stations from the reference network. These outliers are common–mode biases due to the strong correlation between stations in short baselines. They can reach a few centimeters in ZTD and can be detected by a jump in formal errors. The magnitude and sign of the bias is impossible to predict, because it depends on different errors in the observations and on the geometry of the baselines. Several cases are illustrated. This defect is shown to strongly impact the 2 most widely used strategies. Our new strategy ensures that no station is disconnected from the main network and that all clusters include long baselines necessary to estimate absolute tropospheric parameters. This strategy might help to improve significantly the quality of GNSS tropospheric parameters estimated in national to regional networks for meteorology.