Abstract： We collect the data of near-field intensive GNSS stations in Nepal, and use GAMIT/GLOBK software to calculate the data of 58 near-field continuous GNSS stations and 41 post-earthquake mobile GNSS stations in Nepal from 2011 to 2017. We analyze and process the periodicity, common mode error and noise of GNSS time series. The results show that: 1) The average WRMS of E, N and U components decreased by 15.52%, 26.41% and 45.06% after GRACE corrected the period term of GNSS time series; 2) In view of the large number of missing data in the time series, the traditional PCA method is improved to calculate the common mode error by using the first principal component, and the efficiency of the three components reached 61.91%, 53.91% and 53.88%, respectively; 3) Noise analysis shows that the optimal noise model of the Nepal earthquake near field GNSS station is WN+FN; compared with the WN model alone, the WN+FN model underestimates interseismic velocity, coseismic displacement and post-seismic displacement by about 6 to 9 times, 4 to 7 times and 2 to 4 times, respectively. The results of this paper are of great significance for the maintenance of regional coordinate frame, and the inversion of geophysical model of co-seismic rupture and post-earthquake residual slip of Nepal earthquake.
