Abstract: Based on wavelet analysis method,the linear fitting residual error of GNSS time series from different types of test GNSS monument within 2 km range under the same observation condition is decomposed using sym 8 wavelet decomposition of 8 scales.Different scales (namely different frequencies)noise of residual obtained,then different scales noise of residual was combined into high frequency part with the characteristics of white noise,the medium frequency part with the characteristics of flickering noise and the low frequency part with the characteristics of random walking noise.This article analyzed the noise amplitude dimension and history of the different types of test GNSS monument which is thought to be inherent in the high frequency part (white noise)and the medium frequency (flickering noise).The results show that the sampling by the single day solution under the ITRF 2005 and allow for observation error and the level of data processing,the stability of rock,soil and roof monument has little influence on the amplitude of the white noise and flickering noise.The data of the different types GNSS monument can be used for impending earthquake anomaly analysis or deformation anomaly monitoring after earthquake.