GNSS coordinate time series analysis and signal separation applied to Earth deformation monitoring

Global Navigation Satellite Systems (GNSS) coordinate time series is utilised in geodesy for different purposes, e.g., Earth deformation monitoring. Different signals, such as annual, semi-annual, and draconitic year signals, as well as noise, are superimposed on the GNSS coordinate time series. In addition, a velocity signal caused by plate tectonic movements is present in the horizontal components, and uplift or subsidence exists in the vertical component of the GNSS coordinate time series. We require performing signal separation algorithms to decompose the time series into meaningful signals. In this study, we utilise Monte Carlo singular spectrum analysis (MC-SSA) for signal separation and hierarchical clustering for grouping the modes derived from SSA. We also use least-squares variance component estimation (LS-VCE) and fast LS-VCE for noise determination throughout the whole data processing. Finally, the velocity is determined using least-squares regression after removing the periodic signals and utilising the covariance matrix determined by LS-VCE and fast LS-VCE as a stochastic model. The results are presented in both spatial and temporal domains, which can be used to detect, for example, the phase shift in both domains. The final velocity field and the uncertainty for the up component are also extracted for the GNSS stations in Europe.