A Novel Hybrid Approach for UT1-UTC Ultra-Short-Term Prediction Utilizing LOD Series and Sum Series of LOD and First-Order-Difference UT1-UTC

Accurate ultra-short-term prediction of UT1-UTC is crucial for real-time applications in high-precision reference frame conversions. Presently, traditional LS + AR and LS + MAR hybrid methods are commonly employed for UT1-UTC prediction. However, inherent unmodeled errors in fitting residuals of these methods often compromise the prediction performance. Thus, mitigating these common unmodeled errors presents an opportunity to enhance UT1-UTC prediction performance. Consequently, we propose a novel hybrid difference method for UT1-UTC ultra-short-term prediction by integrating LOD prediction and the prediction of the sum of the LOD and the first-order-difference UT1-UTC. The evaluation demonstrated promising results: (1) The mean absolute errors (MAEs) of the proposed method range from 21 to 869 mu s in 1-10-day UT1-UTC predictions. (2) Comparative analysis against zero-/first-/second-order-difference LS + AR and zero-/first-order-difference LS + MAR hybrid method reveals a substantial reduction in MAEs by an average of 54/64/44 mu s, and 47/20 mu s, respectively, with the proposed method. (3) Correspondingly, the proposed method achieves average improvement percentages of 17%/18%/15%, and 13%/3% in 1-10-day UT1-UTC predictions.