Research on Polar Motion Prediction Based on Radial Basis Function Neural Network Interpolation

Abstract

Accurate prediction of polar motion are crucial for various scientific fields, including astronomy, geoscience, and oceanography. The temporal resolution of the modeling data currently utilized in polar motion prediction research is 1 day. This paper proposes to use multiple interpolation methods to interpolate the polar motion observation data to obtain interpolation data with a resolution of 6 hr, and conducts 480 groups of ultra-short-term experiments based on the combined prediction model of least-squares extrapolation of harmonic and autoregressive modeling. Experimental results demonstrate that: (a) the forecasting approach proposed in this paper, which utilizes 6-hr resolution data, significantly enhances prediction accuracy of polar motion; (b) compared with the forecasting scheme without interpolation, the proposed optimal forecasting scheme in this study achieves average improvement rates of 42.27% and 46.94% in the X and Y directions, respectively; (c) the effectiveness of the proposed scheme in this paper was validated through comparison with IERS Bulletin A and the forecast results from the second Earth Orientation Parameter Prediction Comparison Campaign.