Improvement in the Medium- and Long-term Prediction Accuracy of Polar Motion Using Singular Spectrum Analysis

Polar motion cannot be determined in real time owing to the delay caused by data transfer and heavy computation procedures. Polar motion predictions are therefore required for many real-time applications and geodynamics study. Polar motion with respect to the axis of the terrestrial reference system is not constant in time but changes due to external forces and internal processes. This paper proposes a hybrid method to improve polar motion prediction based on singular spectrum analysis (SSA). In this method, the SSA is employed to separate and extract the low- and high-frequency components of polar motion. Next, the least square (LS) extrapolation and autoregressive (AR) models are applied to model and predict the extracted low- and high-frequency components, respectively. The subsequent predictions of the low- and high-frequency components are summed to obtain the predicted values of polar motion. The singular spectrum analysis of the polar motion time-series shows that the low- and high-frequency components can be accurately extracted from the original time-series of polar motion by the SSA technology. Further, the prediction results illustrate that the proposed SSA-based combined model can substantially improve the medium- and long-term predictions of polar motion out 30days in the future. In comparison with the polar motion predictions from the Bulletin A published by the International Earth Rotation and Reference Systems Service (IERS), the highest improvements found for PMX in the prime meridian direction and PMY in the west 90^\circ meridian direction are 45.97% and 62.44% up to 365 days in the future, respectively. It is concluded that the SSA is a potential method to enhance medium- and long-term predictions of polar motion.