ausblenden:
Schlagwörter:
-
Zusammenfassung:
At present, the most precise method for obtaining reliable observations of Celestial Pole Offsets (CPO) is through the use of Very Long Baseline Interferometry (VLBI). The CPO data includes a variety of components, such as the free core nutation (FCN), trends, and harmonics, which result from shortcomings in the IAU 2006/2000A precession-nutation model, geophysical disturbances and observational noise. In consequence, the possibility of forecasting CPO is limited. One possible step forward in improving these predictions is to utilize more sophisticated models. On the other hand, there is a compelling need to identify next-generation time-series algorithms to be integrated into an operational processing chain to make advanced CPO predictions. Of specific interest is the emergence of machine learning regression algorithms.The goal of this study is to advance our understanding of CPO prediction developing new models/methodologies/tools. With the support of UAVAC (University of Alicante VLBI Analysis Center), we perform a global VLBI analysis to determine empirical corrections to the precession offsets and rates, and to the amplitudes of a wide set of terms included in the IAU 2006/2000A precession-nutation theory. Additionally, we explore new strategies to get more precise FCN models. Finally, based on these empirical corrections and novel FCN models, we use a diversity of advanced machine learning regression algorithms to make CPO prediction ranging from 1 to 365 days. The performance of the trained machine learning models is investigated by comparing their predictions vs. the corrected/updated CPO time series. The validation results suggests that the trained models produce reliable estimates.
