Non-tidal ocean loading signals are known to be a significant source of geophysically induced noise in gravimetric and geodetic observations also far-away from the coast and especially during extreme events such as storm surges. Operational products suffer from a low temporal and spatial resolution and reveal only small amplitudes on continental stations. Dedicated high-resolution sea-level modelling of the North and Baltic Sea provides a largely improved prediction of non-tidal ocean loading signals. Superconducting gravimeter and GNSS observations on the small offshore island of Heligoland in the North Sea are used for a thorough evaluation of the model values revealing correlations of up to 0.9 and signal reductions of up to 50 % during a storm surge period of one month in Jan-Feb 2022. Additional continental superconducting gravimeter stations are used to assess the benefits from high-resolution modelling for an improved signal separation further away from the coast.

Rapid provision of Earth Orientation Parameters (EOPs, here polar motion and dUT1) is indispensable in many geodetic applications and also for spacecraft navigation. There are, however, discrepancies between the rapid EOPs and the final EOPs that have a higher latency, but the highest accuracy. To reduce these discrepancies, we focus on a data-driven approach, present a novel method named ResLearner, and use it in the context of deep ensemble learning. Furthermore, we introduce a geophysically-constrained approach for ResLearner. We show that the most important geophysical information to improve the rapid EOPs is the effective angular momentum functions of atmosphere, ocean, land hydrology, and sea level. In addition, semi-diurnal, diurnal, and long-period tides coupled with prograde and retrograde tidal excitations are important features. The influence of some climatic indices on the prediction accuracy of dUT1 is discussed and El Ni\~{n}o Southern Oscillation is found to be influential. We developed an operational framework, providing the improved EOPs on a daily basis with a prediction window of 63 days to fully cover the latency of final EOPs. We show that under the operational conditions and using the rapid EOPs of the International Earth Rotation and Reference Systems Service (IERS) we achieve improvements as high as 60\%, thus significantly reducing the differences between rapid and final EOPs. Furthermore, we discuss how the new final series IERS 20 C04 is preferred over 14 C04. Finally, we compare against EOP hindcast experiments of European Space Agency, on which ResLearner presents comparable improvements.

Different Earth orientation parameter (EOP) time series are publicly available that typically arise from the combination of individual space geodetic technique solutions. The applied processing strategies and choices lead to systematically differing signal and noise characteristics particularly at the shortest periods between 2 and 8 days. We investigate the consequences of typical choices by introducing new experimental EOP solutions obtained from combinations at either normal equation level processed by DGFI-TUM and BKG, or observation level processed by ESA. All those experiments contribute to an effort initiated by ESA to develop an independent capacity for routine EOP processing and prediction in Europe. Results are benchmarked against geophysical model-based effective angular momentum functions processed by ESMGFZ. We find, that a multi-technique combination at normal equation level that explicitly aligns a priori station coordinates to the ITRF2014 frequently outperforms the current IERS standard solution 14C04. A multi-GNSS-only solution already provides very competitive accuracies for the equatorial components. Quite similar results are also obtained from a short combination at observation level experiment using multi-GNSS solutions and SLR from Sentinel-3A and -3B to realize space links. For ΔUT1, however, VLBI information is known to be critically important so that experiments combining only GNSS and possibly SLR at observation level perform worse than combinations of all techniques at normal equation level. The low noise floor and smooth spectra obtained from the multi-GNSS solution nevertheless illustrates the potential of this most rigorous combination approach so that further efforts to include in particular VLBI are strongly recommended.