Insights from the post-operational phase of the Second Earth Orientation Parameters Prediction Comparison Campaign (2nd EOP PCC) and sub-campaign dedicated to machine learning-based prediction approaches (EOP PML)

Earth Orientation Parameters (EOP), including polar motion (PM), universal time variations (UT1-UTC), and celestial pole offsets (CPO), play a critical role in the accurate transformation of coordinates between terrestrial and celestial reference frames. Reliable EOP predictions are indispensable for applications in modern geodesy and astronomy, including precise positioning, navigation on Earth and in space, and determining the orbits of satellites.

The Second Earth Orientation Parameters Prediction Comparison Campaign (2^nd EOP PCC), conducted between 2021 and 2022 by CBK PAN Warsaw in collaboration with GFZ Potsdam, provided a platform to evaluate EOP various prediction capabilities. Following the main campaign, a post-operational phase of the 2^nd EOP PCC was initiated in January 2023, offering participants the opportunity to continue submitting and evaluating their EOP forecasts. This extended phase serves as a platform for additional studies on prediction accuracy, reliability, and robustness, while fostering collaboration among participating institutions.

A key element of the post-operational phase is the EOP PML sub-campaign, which explores the application of machine learning (ML) techniques to EOP prediction. Predictions are evaluated under two scenarios: one excluding and the other including Effective Angular Momentum (EAM) data derived from atmospheric, oceanic, hydrological, and sea-level contributions. These predictions are tested over a 10-day horizon using predefined input data, i.e., IERS 20 C04 solutions and EAM forecasts. The EOP PML aims to assess the potential of ML-based approaches in enhancing prediction accuracy under controlled conditions.

This presentation will provide an overview of the post-operational phase of the 2^nd EOP PCC, with a focus on its goals, methodologies, and preliminary results, including insights gained from the EOP PML sub-campaign. By integrating traditional and ML-based approaches, this effort contributes to advancing EOP prediction techniques and their operational applications.