Abstract: In view of the shortcomings of low accuracy, poor adaptability, and error-prone P-wave phase picking in high-rate GNSS co-seismic displacements, a method combining complete ensemble empirical mode decomposition (CEEMD) and improved modified energy ratio (IMER) for high-rate GNSS co-seismic displacement P-wave phase identification is proposed to overcome the impact of low signal-to-noise ratio (SNR) and non-stationarity on P-wave phase picking. This method first decomposes the high-rate GNSS co-seismic waveform after wavelet threshold denoising using CEEMD to obtain several intrinsic mode functions (IMFs). Then, multiscale permutation entropy (MPE) is used to group the IMF components into frequency bands, and the signal is reconstructed from the high-frequency components. Finally, the IMER method is applied to accurately pick up the first arrivals of the P-wave in the reconstructed signal. The method is tested using high-rate GNSS data from seismic stations in the Ridgecrest area of California, USA, to pick up the P-wave first arrivals from the acquired co-seismic displacement and velocity waveforms. The results show that, compared with the three traditional methods, CEEMD-IMER has the highest accuracy of P-wave first arrivals pickup and can provide technical support for the early warning of earthquakes and rapid assessment of post-earthquake disasters.