Prediction of Polar Motion Based on Combination of Weighted Least-Squares and Autoregressive Moving Average

High accurate prediction of polar motion has important significant and useful value for high-precision space navigation and positioning. In this paper, the weighted least-squares (WLS) is proposed to use to predict different span of polar motion, as the cycles and trends have the characteristic of time-varying in the observational data of polar motion. Autoregressive Moving Average (ARMA) can be applied to fit the residuals of polar motion as it can be regard as a smooth, zero-mean sequence. The LS + AR model, the LS + ARMA model, the WLS + AR model and the WLS + ARMA model are used to predict the different span of polar motion, the results show that, the application of weighted least-squares can improve the polar motion prediction accuracy effectively. And the WLS + ARMA model is equal to WLS + AR model, and in some days WLS + ARMA model is better than WLS + AR model.