Abstract: The common mode errors (CMEs) existing in the Global Positioning System (GPS) coordinate time series are removed using the stack filter (SF), network inverse filter (NIF) and principal component analysis (PCA) methods, to improve the accuracy of GPS monitoring regional surface displacements. By building GPS coordinate time series model, remove the obvious tectonic movements, and extract the noise residuals. Then, the CMEs in the residuals are extracted with the SF, NIF, PCA methods. Using 2019-2021 GPS coordinate time series in Japan’s Boso peninsular, the SF, NIF, and PCA methods of extracting the CMEs are compared, the effects of different GPS site spatial resolutions on the CMEs are analyzed, and the surface displacements before and after removing the CMEs are analyzed. The results show that the performances of the SF, NIF and PCA methods to extract the CMEs are consistent. The spatial resolution of GPS sites decreases, and the dispersion of the extracted CMEs increases. The CMEs affect the size and direction of the slow slip surface displacements, so they need to be removed.