Change in spectral index and improvement in velocity uncertainty due to MULTI-GNSS contribution to geodetic site velocity estimation

Noise characteristics of GPS time series have been documented well so far. Nowadays, researchers study the behavior of the noise from GNSS time series. The examples are yet so few in which the spectral behavior of the noise and differences from the GPS solutions are assessed. For this aim, we are testing our in-house time-series analysis software using GPS solutions from the JPL’s GPS time-series archive. The software is being developed in Yildiz Technical University. GNSS data, which is obtained form the IGS’s MGEX experiment, have been processed using PRIDE GNSS software developed by Wuhan University. About 50 globally well scattered MGEX GNSS stations which are capable of acquiring data from 2-5 different GNSS techniques have been selected. Robust estimators are used to remove outliers and data gaps are filled with AR(1). We do not use large matrix computations while estimating the spectral index, learning lessons from the predecessors of the area, rather an iterative and computationally efficient algorithm is used. Alternative algorithms to automatically fix the offsets which are problematic in trend and velocity uncertainty estimation are also offered during the analysis. We calibrate our velocity estimations with those of the GNSS time-series solutions produced by the JPL using GIPSY-X. Initial results seem to be promising. With the contribution of various GNSS techniques (e.g. GALILEO, GLONASS, GPS, BeiDou, etc.) to the geodetic time series analysis the value of the spectral indice changes and the magnitude of the velocity uncertainty on the average becomes smaller. Namely, the noise shrinks and becomes whiter. The velocity uncertainty is reduced by about 11% for the East and 13% for the Up components while the north component stays neutral.