Analysis of GNSS Coordinate Time Series in North China by Independent Component Analysis

Common mode error (CME), as a kind of non-tectonic motion, is one of the most important error sources in Global Navigation Satellite System (GNSS) coordinate time series. In order to obtain a precise and reliable velocity at a station, the influence of CME should be removed or decreased. In this paper, independent component analysis (ICA) method is applied to extract CMEs and analyze their influences on the coordinate time series based on 24 GNSS reference stations in North China during 2011–2020 period. The results indicate that, (1) After CMEs filtered out by the ICA, the average correlation coefficients between the residual time series decrease from previously 0.5–0.7 to −0.04 in east, north and up directions at stations, implying that the strong correlation between the residual time series at stations is significantly reduced by the ICA filtering. (2) The RMS of residual time series is effectively suppressed after the ICA filtering, with the RMS decreasing about 43.95%, 52.25% and 38.13% in east, north and up direction, respectively. (3) The averages of uncertainties of velocity estimations are decreasing from 0.30, 0.27 and 0.42 mm/yr to 0.12, 0.09 and 0.22 mm/yr in east, north and up direction before and after the ICA filtering, respectively, showing that the precision of the velocity estimation is improved about 60%, 69% and 49% in the three directions. Therefore, it is necessary to do the filtering among the GNSS coordinate time series to remove the effect of CME and the ICA is demonstrated to be an effective method to apply the filtering, so as to extract the precise tectonic motion in North China.