Long-term prediction of the Earth Orientation Parameters by the artificial neural network technique
► We use feed-forward back-propagation neural network to forecast the Earth Orientation Parameters up to 360 days. ► Predictions include linear and non-linear parts. ► Root mean square errors and mean absolute errors of the discrepancy between predictions and observations are presented. ► UT1 predic...
Gespeichert in:
Veröffentlicht in: | Journal of geodynamics 2012-12, Vol.62, p.87-92 |
---|---|
Hauptverfasser: | , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | ► We use feed-forward back-propagation neural network to forecast the Earth Orientation Parameters up to 360 days. ► Predictions include linear and non-linear parts. ► Root mean square errors and mean absolute errors of the discrepancy between predictions and observations are presented. ► UT1 predictions are particularly obtained directly by using UT1 series and indirectly by using integration of LOD predictions. ► Our results show that indirect method is superior to the direct method for predictions of UT1 in time scales less than 360 days.
There are increasing demands for EOP predictions in science, deep space navigation, etc. Based on previous research on short-term prediction of Earth Orientation Parameters (EOP) by artificial neural networks (ANN), we extend our attempt to long-term predictions of EOPs, i.e. predictions with a lead time up to 360 days. The basic theory and some special considerations for the ANN forecast of EOPs are presented, and finally our preliminary results and their accuracy estimates are shown and compared with those obtained by other authors. |
---|---|
ISSN: | 0264-3707 |
DOI: | 10.1016/j.jog.2011.12.004 |