Abstract:We assess the characteristics of noise in the coordinate time series of daily position from August 2008 to April 2013 at 36 north China GPS fiducial stations. Regional stacking filtering is employed to remove the common mode errors from the daily position time series. The noise characteristics of time series and the station velocities are assessed using maximum likelihood estimation. The results indicate that the common mode errors are attributed to the flicker noise and that their variations are 1 mm for the NS component, 1 mm for the EW component and 3 mm for the vertical component. The noise in the unfiltered position time series can be described as a combination of variable white noise plus flicker noise or variable white noise plus power-law noise, while the filtered position time series can be described as a combination of variable noise plus flicker noise plus random walker noise or variable white noise plus power-law noise. The removal of the common mode errors decreases the flicker noise component and highlights the corresponding noise component related to site effects. The velocity uncertainties are about 5-8 times greater than if only variable white noise is considered. The velocity uncertainties decrease by about 40% if the common mode errors are removed.