CMONOC II Common Mode Error Analysis and Structural Region Division Research

Abstract

The modern tectonic deformation of the Chinese mainland is dominated by landmass movements, and active tectonic block regions are geological units with a relatively uniform movement pattern. The removal of CMEs can provide more accurate GPS data for exploring the movement characteristics between active tectonic block regions. In order to improve the effect of CME extraction, we propose that the Crustal Movement Observation Network of China be divided into sub-regions based on the refined definition of active tectonic block regions of the Chinese mainland. In this paper, 247 stations in the CMONOC II network are used to form a large spatial scale GPS network and 6 sub-regions with small spatial scale GPS networks. For the large spatial scale GPS network, we compare and analyze the effects of PCA and ICA filtering, and the study shows that PCA is not suitable for CME extraction in this large spatial scale GPS network, while ICA filtering is better. Subsequently, the large spatial scale GPS network and six small spatial scale GPS networks were used to extract CME using ICA, with the results showing that the RMSE values of the residual time series of the large spatial scale GPS network were reduced by 9.60%, 17.08%, and 16.14% in the directions of E, N, and U, respectively, and that the subregions divided according to the refined and determined first-level active plots of the Chinese continent had their residual time series of the RMSE values were reduced by 26.19%, 26.95%, and 28.32% on average in the three respective directions of E, N, and U. The effect of extracting CMEs by dividing the subregions was 29.16%, 5.44%, and 39.84% higher than the effect of extracting them as a whole in the three directions of E, N, and U, respectively. The experimental results demonstrate that the CMONOC II observation network is an effective and feasible method to extract CMEs according to the finely defined active tectonic block region of the Chinese mainland at the first level.

Keywords

• GPS time series
• common mode error
• active tectonic block region
• improved singular spectrum analysis
• principal component analysis
• independent component analysis