Abstract: Nonlinear deformation affects Global Navigation Satellite System (GNSS) coordinate timing accuracy. In this paper, improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) method and environmental load correction are combined to study the nonlinear deformation removal of GNSS stations. Firstly, use GMIS software to complete the GNSS coordinate timing and remove the gross error. Then use ICEEMDAN method to decompose the GNSS coordinate timing, and use the permutation entropy algorithm to select the high frequency components containing noise and nonlinear deformation. Finally, the environmental load is used to remove the high-frequency components, and the removal effect is compared with the empirical mode decomposition (EMD) method and the environmental load method. The research results show that the root mean squared (RMS) of GNSS coordinate time series after nonlinear deformation removal changes different, and the up (U) direction is the most obvious, with the maximum value of 6.715 mm, followed by the E direction and the north (N) direction. After combining ICEEMDAN met-hod and environmental load,the nonlinear deformation in N direction was weakened 75% of the nonlinear deformation in east (E) direction was weakened, and 62.5% of the nonlinear deformation in U direction was weakened. The correction effect was better than the combination of EMD method and environmental load.