Long-term polar motion prediction based on diurnal variation parameters of polar motion

This paper proposes a long-term polar motion prediction method using two derived diurnal parameters: daily polar motion amplitude (DR) and phase (DA). In particular, this study proposes a processing method to reorganize DA into a continuous curve. Suitable models and fitting-window lengths are selected for the derived parameters. Using IERS 20C04 observations, we conducted 365 forecast experiments with horizons up to 500 days. In this retrospective evaluation, the proposed scheme improves average prediction accuracy over LS+AR by 39.86% for the X pole and 40.17% for the Y pole. These results indicate the potentially achievable performance under the selected settings.