ASSESSING THE SEASONAL SIGNALS BETWEEN ENVIRONMENTAL LOADINGS AND GPS COORDINATES WITH SINGULAR SPECTRUM ANALYSIS

Environmental loading effects, such as atmospheric, hydrological, and ocean non-tidal loading, can explain about 40% of the total variance of the annual signal in GPS time series. Ideally one would like to subtract the periodic signals from the GPS observations using a geophysical model. However, the errors in the geophysical models are still significant and in this presentation we assess the accuracy of each type of loading at the annual period. For this purpose, we used atmospheric, hydrological and non-tidal oceanic loading models provided by the EOST Loading Service using: ECMWF (European Centre for Medium Range Weather Forecasts) operational and reanalysis (ERA Interim) surface pressure fields (atmospheric loading), MERRA-Land (Modern-Era Retrospective Analysis for Research and Applications), GLDAS/Noah model, using soil-moisture and snow field and previously mentioned the ECMWF operational and reanalysis (ERA interim) models (hydrological loading) and two versions of ECCO (Estimating the Climate and the Circulation of the Ocean): ECCO1 and ECCO2 (non-tidal oceanic loading). In this research, we analyzed the topocentric time series from 20 selected IGS (International GNSS Service) core stations located worldwide, derived from a PPP (Precise Point Positioning) solution obtained by the JPL (Jet Propulsion Laboratory) with GIPSY-OASIS software. The lengths of GPS time series varied from 10 up to 23 years. We applied the non-parametric Singular Spectrum Analysis (SSA) approach to extract the time-varying seasonal oscillations (from 1 to 4 cycles per year) as a combined seasonal signal from atmospheric, hydrological and non-tidal oceanic loading models from the loading time series. The same method was applied for GPS time series. We noticed that annual signal identified using SSA in the non-tidal oceanic loading model can explain on average 34% of its total variance for stations taken to this research and annual oscillations in the hydrological loading model explain up to 73% of its total variance. The greatest percentage of annual signal variance from atmospheric loadings is noticed for station in East Asia: CHAN about 75% as well as for South Australia: YAR2 about 69%. The cross-correlation analysis showed that SSA-derived curves for loading effects and GPS time series agree at maximum of 80%.