Research On The Models And Algorithms For High Accuracy Prediction Of Faith Rotation Rotamters

As an important part of Earth observation,Earth Rotation Parameters(ERP)characterizes the whole situation of the Earth rotation,and they are the direct reflection of the interaction between the Earth's various spheres of the Earth's core,mantle,crust and atmosphere.ERP include Polar Motion(PM),UT1-UTC and LOD,which is indispensable and crucial in the transformation between the Celestial Reference Frame and the Terrestrial Reference Frame,and is significant for satellite orbit determination and deep space exploration.Due to the complexity of modern geodetic technology data processing and the inconsistency of the strategies carried out by each technique,the release of high-precision ERP data will be delayed for days to weeks,which brings many challenges to space navigation and deep space exploration.With the continuous development of modern geodesy,spacecraft tracking and other technologies,the demand for long-term high-precision ERP prediction is urgent,as well as the improvement of the prediction algorithm.The Earth's rotation is closely related to the changes of the Earth's environment.Exploring the time-varying and excitation mechanisms of the Earth's rotation can improve the modeling and extrapolation accuracy of ERP,and then achieve higher ERP prediction results.In order to further improve the ERP prediction,a series of improved ERP prediction algorithms are proposed in this paper based on the accurate detection of the time-varying of the ERP inherent period terms.The detailed research work can be described as follows:(1)The differences and connections between EOP 08C04 and EOP 14C04 are analyzed,the causes of the deviation on PMY from March 2011 between two solutions are examined,and the differences of the two solutions are further assessed for their prediction performance.On February 1,2017,the IERS EOP 14C04 solution was released by IERS,and the EOP 08C04 is still being updated.The difference and connection between the two solutions is a subject worthy of scrutiny.In this paper,the differences between the EOP 08C04 and EOP 14C04 solutions are analyzed,firstly.There is a translation rate of-0.1 mm/yr on the Z direction between ITRF2014 and ITRF2008,which leads to a clear trend term on PM between the two solutions.Secondly,it was found that the reason of the significant deviation appeared on PMY between EOP 14C04 and EOP 08C04 from March 2011 to 2015 is caused by the effects of the Fukushima earthquake.Most of the monitoring stations in Japan's region have difference deviation caused by the earthquake,and there is deformation rebound in the following years.Finally,the LS+AR model is used to analyze and compare the ERP prediction performance based on EOP 14C04 and EOP 08C04,and the results show that EOP 14C04 has a better solution in part of the high-frequency signal due to the stability of the ITRF2014 and the increase of the number of global monitoring stations.(2)The characteristics of the Earth's Rotation are analyzed based on the ERP observation data,a Retrograde Semi-annual Wobble(RSAW)is detected and confirmed,and its amplitude and period are determined.The Earth rotation is not uniform and it fluctuates slightly under the influence of the external environment around the Earth.The main period terms of ERP are extracted and reconstructed by the Fourier Transform Band-Pass Filter(FTBPF),and the contributions of each period term for ERP are analyzed and compared.It is particularly noteworthy that there is not only a significant Retrograde Annual Wobble(RAW)but also a Retrograde Semi-annual Wobble(RSAW)in PM,its amplitude and period are determined and analyzed in this paper.Then,the characteristics of Chandler Wobble is analyzed and found that Chandler's amplitude has decayed to its lowest state in 2016,which is also the key reason that the rate of PM has not changed significantly since 2010.In addition,it will appear as dual-frequency or multi-frequency when the Chandler period is determined if the length of basic series is insufficient.Finally,the time-varying characteristics of the UT1-UTC/LOD are analyzed,it is found that there is an 870-days term with wide band in UT1-UTC/LOD except Annual Wobble(AW)and Semi-Annual Wobble(SAW).The amplitudes of AW and SAW in UT1-UTC/LOD are relatively stable.The SAW is maintained at 8 ms and the AW is maintained at 12 ms for a long time for UT1-UTC.For LOD,The SAW is maintained at 0.12 ms and the AW is maintained at 0.2 ms for a long time.The value of the instantaneous period varies slightly with time,180-185 days for SAW and 361-369 days for AW.(3)In view of the existence of RAW and RSAW in PM,the influence of period terms for prediction is analyzed.Meantime,the influence of period terms in UT1-UTC/LOD for prediction is analyzedThe period terms in PM have large amplitude variation and include Retrograde Wobble,which make it more difficult for high-precision modeling and prediction.The influences of different period terms on the prediction of PM is explored in this paper based on LS+AR model.The results show that the inclusion of RAW term can effectively improve the accuracy of PM prediction.Besides,considering the RSAW,the prediction accuracy can be further improved in the medium and long term prediction PMX component.For UT1-UTC/LOD,there are AW,SAW,1/3 AW and 870 days term with wide band after removing the trend.The influence of those terms for UT1-UTC/LOD prediction is analyzed.And the result show that the 1/3 AW and 870-days have a small impact on the UT1-UTC/LOD prediction.Because the 1/3 AW has low amplitude and does not contribute much to UT1-UTC/LOD,while the frequency band of 870-days term is wide.Therefore,1/3 AW and 870-days can be ignored in UT1-UTC/LOD prediction.(4)An Improved Weighted Least-Squares and Autoregressive model is proposed with considering the strong correlation of recent data and the accuracy of observation data,and the influence of the time-varying of period terms on the ERP prediction accuracy is analyzed.ERP prediction based on LS+AR model has the characteristics of low fitting accuracy in boundary and large extrapolation error.An Improved WLS+AR model is proposed for ERP prediction with considering the strong correlation of recent data and the accuracy of observation data.The experimental results show that the weighting method proposed in this paper effectively improves the prediction accuracy of PM.For UT1-UTC/LOD,the medium and long term prediction accuracy is improved.Meantime,in view of the time-varying of ERP period terms,the influence of time variability on prediction accuracy is further analyzed.The balanced period values are determined firstly before ERP prediction in this paper,then better ERP model and extrapolation accuracy are achieved.In the prediction test,the prediction accuracy is greatly improved when the time variability of PM is taken into account.Especially for the PMX component,its accuracy is improved by nearly 40%in the long-term prediction.(5)An additional constrained PCF+WLS+AR prediction model is proposed and is well applied in UT1-UTC prediction.The trend term of UT1-UTC can be simulated by Polynomial Curve Fitting(PCF),which is very important in UT1-UTC modeling,but it has divergence in extrapolation,and it is necessary to choose a suitable PCF order for extrapolation.The annual constraint and interval constraint are designed to optimize PCF order in this paper according to the characteristics of UT1-UTC.The UT1-UTC prediction results show that the additional constrained PCF+WLS+AR model can achieve better prediction accuracy,especially in the medium and long-term prediction,the accuracy is improved by nearly 30%.(6)The IGS PM-rate is considered in PM prediction,and the near real-time prediction accuracy is greatly improved.For UT1-UTC,the LOD and AAM are participated in UT1-UTC model in order to improve the UT1-UTC near-term prediction accuracy.IGS PM-rate is one of ERP products provided by IGS,and it is determined by IGS according to GNSS technology.There is a strong correlation between PM and PM-rate,therefore the PM-rare is added to PM prediction in this paper.Based on this,the WLS+MAR model is proposed with considering IGS PM-rate.The prediction results demonstrate that the PM prediction accuracy in near real-time(1-5 days)is effectively improved.Similarly,the LOD can be thought of as the rate of UT1-UTC,and in view of it has low accuracy in UT1-UTC near-term prediction,a combined UT1-UTC prediction method is proposed with considering the strong correlation between AAM,LOD and UT1-UTC in this paper.The prediction test shows that it has effectively improved the near-term prediction of UT1-UTC when the AAM and LOD are participated in UT1-UTC model.