Velocity estimates derived from time series of space geodetic station positions are influenced by seasonal signals and noise present in these time series. Therefore, the insufficiently modeled trend of these time series will distort station velocity estimates and their uncertainties. In this study, velocities are estimated using two distinct approaches: Singular Spectrum Analysis (SSA) and Wavelet Multiresolution Analysis (WMA), both of which enable the reliable separation of trend from periodic signals and noise. The data used are the combined residual coordinate time series of 45 DORIS (Doppler Orbitography and Radiopositioning Integrated by Satellite) stations, aligned to ITRF2014 and expressed in a topocentric coordinate system (North, East, Up). The SSA results reveal a dominant trend (velocity) in the horizontal components (North and East), with a variance exceeding 99% of the total signal for most stations. The estimates of horizontal velocities by SSA and WMA are almost similar and agree with those calculated from plate tectonic models (GSRM 2.1, ITRF2020, GEODVEL 2010, REVEL 2000, MORVEL 2010, and NUVEL 1A), except six stations: AMSTERDAM, DIONYSOS, EVEREST, MANAGUA, MANGILAO, and MANILLE. These stations, most of which are located at tectonic plate boundaries, have suffered great earthquakes with a magnitude above 6.0. The vertical velocities of DORIS and GPS (Global Positioning System) at 26 co-located sites vary from −3 mm/yr to 2 mm/yr, except for the high-latitude stations THULE (76.54°) and NY-ALESUND (78.92°), which exhibit high velocities of around 7.3 mm/yr and 6.7 mm/yr for DORIS, and 7.2 mm/yr and 8 mm/yr for GPS, respectively.
Khelifa et al. (Sun,) studied this question.