Evaluate the Prediction Level of Polar Motion Interpolation for 6 hr in Ideal and Real Environments

With the increasing demands for global development, deep space exploration missions are taking place more frequently, rendering the traditional 24‐hr Polar Motion (PM) predictions inadequate for supporting the frequent operations of spacecraft. Therefore, this study utilized the mainstream EOP C04 and EOP C04 spliced with the final.daily series to evaluate the performance of interpolating PM to 6‐hr and then forecasting, under both ideal and real‐world environments. Interpolation is performed using the Stable Ultra‐high‐Precision‐Radial‐Basis‐Function (SURBF) and the Seasonal‐Trend‐Residual decomposition with adaptive interpolation (STR). Prediction is carried out using the Least Squares and Autoregressive (LS + AR). Our findings indicate that the two input formats, namely interpolating prior to column assignment in the idealized case and direct column assignment before interpolation in the real‐world environments, greatly affect prediction results. The former takes full advantage of the disclosed subsequent information, achieving a first‐day MAE of less than 0.1mas. The latter adheres to the actual conditions, resulting in a first‐day MAE above 0.2mas and performing worse than traditional forecasting. Additionally, we evaluated the forecasting performance using 6‐hr International GNSS Service Ultra‐rapid (IGU) inputs and found that it not only surpasses traditional methods but also outperforms USNO forecasts over the 1–10 days. Therefore, we do not recommend interpolating 24‐hr to 6‐hr, as interpolation is essentially a guess and does not provide meaningful information. This study only recommends using IGU‐6h derived from real observational data to obtain 6‐hr PM predictions. These findings can assist practical deep space exploration projects in making informed decisions when selecting 6‐hr PM predictions.