A Super-Long-Term Prediction Method of Earth Polar Motion Based on Spectrum Analysis

The super-long-term prediction technology of Earth Orientation Parameter (EOP) is a key foundation in autonomous orbit determination for earth-orbit spacecraft. Currently the classical prediction method of EOP is LS + AR, while the accuracy of periodic components and the corresponding frequency of EOP in the LS model is to affect the prediction performance. This paper focuses on the super-long-term prediction method of the polar motion. Firstly, the spectrum analysis of the 14C04 data series released by IERS is done to confirm the periodic components. Secondly, the periods of Chandler motion and annual motion are estimated by using spectral centroid method, which are about 433.75 days and 365 days respectively. Furthermore, the fence effect of FFT and prediction error are used to testify the correctness of the periods. Finally, the 180-day, 365-day and 730-day prediction of polar motion are conducted and compared with the observed value released by IERS, showing that the corresponding prediction error are 12 mas, 20 mas, and 55 mas. The results of the paper can be used as a good reference for long-term prediction of EOP.