An adaptive spatiotemporal filtering method for GNSS coordinate time series in CMONOC

Abstract Common mode errors (CMEs) are a persistent challenge in regional GNSS coordinate time series, becoming more difficult to extract as distance increases. This study applied an adaptive spatiotemporal filtering method that divides the area into smaller sub-regions based on predefined thresholds for effective filtering. A total of 213 GNSS stations from the Crustal Movement Observation Network of China (CMONOC) were analyzed, with data from 2013 to 2023 in the North (N), East (E), and Up (U) directions. Three filtering methods were compared: overall spatiotemporal filtering, uniform sub-region filtering, and adaptive sub-region filtering. The results demonstrate that the adaptive spatiotemporal filtering method reduced the RMS values in the N, E, and U directions by an average of 25.46%, 21.52%, and 30.05%, respectively, compared to unfiltered data. Uniform sub-region spatiotemporal filtering achieved reductions of 17.70%, 16.78%, and 19.98% in the RMS values for the same directions. Both methods outperformed the overall PCA spatiotemporal filtering, which resulted in RMS reductions of 15.41%, 14.80%, and 17.19% for the N, E, and U directions, respectively. By evaluating the impact of surface mass loading on the CMEs of GNSS coordinate time series, we found that the amplitude values of CMEs in the N and U directions decreased by 10% and 40% for the adaptive sub-region filtering, and by 8% and 30% for the overall PCA filtering, respectively, after mass loading correction. An amplitude increase was noted in the E direction. Noise analysis indicated significant enhancements, with adaptive filtering further reducing power law noise (PL) by 25.85%, 20.10%, and 25.10% in N, E, and U directions, respectively, compared to PCA filtering. Similarly, uniform sub-region filtering achieved reductions of 14.01%, 9.48%, and 8.54% in these directions.