| The Earth Orientation Parameters (EOP) includes the Polar Motion (PM) parameters,the Length of day (LOD) and the Universal Time (UT1-UTC) or called the Earth’s rotationvariation, the Nutation-precession parameters. The PM, UT1-UTC and LOD are also called asEarth Rotation Parameters (ERP). The variation of EOP is very complicated due to multipleexcitations. Highly accurate EOP are the necessary parameters to achieve mutual conversionof the celestial reference frame and earth reference frame. However, due to the complexity ofdata processing, it is difficult to access to these parameters in real-time, about few days delay.In order to meet the needs of the space navigation and positioning, high-precision predictionfor EOP is urgent research subject. The thesis focuses on the theories and algorithms of ERPPrediction and the physical excitation to predict ERP. In view of this, the main work andcontributions are summarized as follows:1. In order to improve the short-team prediction accuracy of polar motion, an improvedGM (1,1) model based on error compensation is used to predict PM within30days. The newmodel considers the strong correlation of prediction from the adjacent epochs and compensatethe prediction error through the previous prediction residuals of GM(1,1) model. The resultsshow that the new model can greatly improve the prediction accuracy for the polar motion ofshort-term forecast within10days.2. Due to polar motion prediction faces great challenge in case of the variable phase/period of the annual oscillation caused by El Ni o, the relationship between the short-termprediction of polar motion and El Ni o event is studied based on the LS+AR model. There isgreater instability of polar motion in the time of El Ni o occurred which bring big predictionerror. In order to reduce the influence of El Ni o event on polar motion prediction, the lengthselection of polar motion fitting series are studied for polar motion prediction. The resultsshow that the accuracy of polar motion prediction can be improved in the time of El Ni o byextending the length of polar motion fitting series to17-19years.3. The influence of Chandler and annual items in pole observation data on predictionaccuracy is analyzed. The results show that it can improve prediction accuracy applied thecycles determined by the known observational data to polar motion prediction due to thecycles’ characteristic in pole is time-varying.4. The influence of different cycle term on prediction accuracy is analyzed by usingdifferent combination of the main cycle terms based on LS+AR prediction model. The resultsshow that the long-term cycle of18.6-year and9.3-year have great impact on LOD prediction accuracy which should be considered. The influence of one-third year on high-accuracyprediction of LOD can not be ignored. Therefore, it is recommended that the above threecycle terms should be taken into account in high-accuracy LOD prediction.5. An improved weighted least-square is proposed to improve prediction accuracy ofEarth’s Rotation Parameters with combination of AR model.4weighting schemes aredeveloped. The results show that the improved WLS+AR model can improve the ERPprediction accuracy effectively, and for different prediction intervals of ERP different weightscheme should be chosen.6. The weighted least-squares (WLS) combined with multivariate autoregressive (MAR)is proposed to predict UT1-UTC with different span. The new method can efficiently considerthe influence of time-varying for the cycle and trend terms of UT1-UTC, which is closelyrelated to atmospheric angular momentum (AAM). The numerical example shows that theprediction accuracy of WLS+MAR method is better than that of LS+MAR method as well asLS+AR method. The results prove that the WLS+MAR model can effectively improve theprediction accuracy of UT1-UTC.Though comparison and analysis of prediction values, these methods proposed in thispaper have been achieved a lot of good results, these are very good supplement to predictionmethod and theory of ERP. |
PDF Full Text Request