基于极限学习机的极移中长期预报

传统的极移预报多是基于最小二乘外推和自回归等线性模型,但极移包含了复杂的非线性成分,线性模型的预报效果往往不甚理想。将一种新型神经网络极限学习机(Extreme Learning Machine, ELM)用于极移中长期预报。首先利用最小二乘外推模型对极移序列进行拟合,获得趋势项外推值,然后采用极限学习机对最小二乘拟合残差进行预报,最终的极移预报值为趋势项外推值与残差预报值之和。将极限学习机的预报结果同反向传播(Back Propagation, BP)神经网络与地球定向参数预报比较活动(Earth Orientation Parameters Prediction Comparison Campaign, EOP PCC)的预报结果进行对比,结果表明,极限学习机用于极移中长期预报是高效可行的。     

利用经验模态分解提高极移预报精度（英文）

经验模态分解(Empirical Mode Decomposition,EMD)是一种数据驱动的自适应非线性、非平稳信号分解方法。为提高极移预报精度,将经验模态分解应用于极移预报中。首先利用经验模态分解方法对极移序列进行分解,获得极移的高频分量和低频分量;然后采用最小二乘(Least Squares,LS)外推模型对极移低频分量进行拟合,获得最小二乘拟合残差;其次采用自回归(Autoregressive,AR)模型对极移高频分量和最小二乘拟合残差之和进行建模预报;最后将最小二乘模型和自回归模型外推值相加获得极移预报值。将经验模态分解和LS+AR组合模型预报结果与LS+AR模型预报以及地球定向参数预报比较竞赛(Earth Orientation Parameters Prediction Comparison Campaign,EOP PCC)的预报结果进行比较,结果表明,将经验模态分解应用于极移预报中,可以明显改善极移预报精度。     

基于有效角动量函数的多参数极移预报

极移的变化与多种激发息息相关,这些激发包括大气表面压力和大气风、海底压力和洋流、陆地水分布以及气候变暖导致的海平面变化,并且可以通过有效角动量函数来估计.在极移预报中,通过刘维尔方程融合有效角动量函数,并利用最小二乘与自回归组合的方法进行拟合及外推,同时,对自回归模型的可调节参数设置更多的选择,在不同的极移预报阶段,对于不同分量的预报匹配更优的参数,有效地提高了极移的预报精度.在441次1–90 d的极移预报实验中,短中期的预报改善更为明显,在1–6 d和7–30 d的极移X预报结果中,分别有56.9%和53.5%优于国际地球自转服务(International Earth Rotation Service, IERS)的预报;在1–6 d和7–30 d的极移Y预报结果中,分别有66.5%和59.7%优于IERS的预报.整体上,极移Y的预报精度比极移X的预报精度有更多的提升,以IERS的地球定向参数(Earth Orientation Parameters, EOP)产品EOP 14C04 (IAU2000A)为参考,极移X预报在第1 d、第5 d的MAE (Mean Absolute...     

一种改正LS+AR模型提高短期极移预报精度的方法

最小二乘(Least Squares,LS)与自回归(Auto Regressive,AR)联合(LS+AR)模型在极移预报(polar motion,PM)中存在以下问题:最小二乘拟合的内部残差值较好而LS外推的残差值较大;LS拟合残差序列是非线性的,故根据预报历元前的残差序列建立的AR模型可能并不适用于待预报的残差序列,存在不匹配预报的情况.针对这两个问题,通过以下方法进行解决:首先对LS拟合数据两端点附加约束条件使其固定到LS拟合曲线上,因此在两端点附近的拟合值与观测值十分接近;然后选取与LS外推残差序列变化趋势接近的内推残差序列作为AR模型的建模对象,进行残差预报.通过实例表明该方法能够有效地提高LS+AR模型在短期极移预报的精度.此外,通过与RLS(Robustified Least Squares)+AR、RLS+ARIMA(Auto Regressive Integrated Moving Average)和LS+ANN(Artificial Neural Network)模型的预报结果对比,证明了该方法在极移预报中的可行性.实例证明了所提出的方法在短期预报中可以取得良好的预报结果,尤其在1–10d超短期的极移预报上可以获得与国际最好预报精度相当的预报结果.     

利用奇异谱分析提高极移中长期预报精度

由于空间大地观测数据传输耗时及处理过程复杂, 导致极移测量值的获取存在时延, 无法满足对高精度的极移预报值有重大需求的应用领域. 针对极移复杂的时变特性, 提出一种基于奇异谱分析(singular spectrum analysis, SSA)的预报方法. 首先用SSA分离提取极移时序中的高频组分与低频组分; 其次建立最小二乘(least square, LS)外推与自回归(AutoreGressive, AR)模型对极移高频和低频组分进行组合预报. 结果表明, SSA方法能够准确地分离和提取极移低频和高频组分, 对低频和高频组分组合预报可以显著改善极移的中长期(30--365d)预报精度, 与国际地球自转和参考系服务局(International Earth Rotation and Reference Systems Service, IERS)提供的A公报中的极移预报值相比, SSA方法对极移PMX分量(本初子午线方向)和PMY分量(西90$^\circ$子午线方向)的中长期预报精度改进最高分别可达45.97%和62.44%. 研究结果验证了SSA方法对极移中长期预报改进的有效性.     

经验模式分解在极移超短期预报中的应用

经验模式分解(Empirical Mode Decomposition,简称EMD)是一种数据驱动的自适应非线性时变信号分解方法,可以把数据分解成具有物理意义的模式函数分量.采用EMD对极移序列进行分解,去除序列中的高频信号,然后基于最小二乘外推(Least Squares Extrapolation,简称LSE)和广义回归神经网络(General Regression Neural Network,简称GRNN)的组合模型对去除高频信号的极移序列进行1～10d的超短期预报.实验结果表明:将该模型应用到极移超短期预报具有可行性,预报精度有明显改善.     

利用端部效应改善的最小二乘外推模型进行UT1-UTC预报

现有的UT1-UTC预报模式在进行周期项与残差项拟合分离时,通常没有考虑最小二乘拟合序列的端部畸变现象(数据处理中称为端部效应),预报精度难以取得较大改善。针对最小二乘拟合存在的端部畸变现象,首先采用时序分析方法在UT1-UTC序列两端进行数据延拓,形成一个新序列,然后用新序列求解最小二乘外推模型系数,最后再联合最小二乘外推模型及神经网络对UT1-UTC序列进行预测。结果表明,在UT1-UTC序列端部增加延拓数据,可以有效地抑制最小二乘拟合序列的端部畸变,相对于常规的最小二乘外推模型,基于端部效应改善的最小二乘(Edge-effect Corrected Least Squares,ECLS)外推模型的UT1-UTC中长期预报精度改善明显。     

利用端点延拓提高LS + NN模型的UT1-UTC预报精度

现有UT1-UTC预报模式在进行周期项与残差项拟合分离时,通常没有考虑最小二乘拟合序列的端部效应,预报精度难以取得较大提高。针对最小二乘拟合存在的端部效应,首先采用灰色模型在UT1-UTC序列的两端进行数据延拓,形成一个新序列,然后对新序列进行最小二乘拟合,最后再联合最小二乘和神经网络(LS+NN)模型对UT1-UTC原始序列进行外推。结果表明,对UT1-UTC序列进行端点数据延拓再进行最小二乘拟合,能够有效地改善最小二乘拟合序列的端部效应;相对于常规LS+NN模型,端部效应改善的LS+NN模型的UT1-UTC预报精度有一定提高,尤其对中长期预报精度提高更为明显。     

基于高斯过程的日长变化预报

由于日长(length-of-day,LOD)变化具有复杂的时变特性,传统线性模型如最小二乘外推模型、时间序列分析模型等的预报效果往往不甚理想,所以将一种新型的机器学习算法—高斯过程(Gaussian processes,GP)方法用于LOD变化预报,并将预报结果同利用反向传播神经网络(back propagation neural networks,BPNN)和广义回归神经网络(general regression neural networks,GRNN)的预报结果以及地球定向参数预报比较竞赛(Earth Orientation Parameters Prediction Comparison Campaign,EOP PCC)的预报结果进行对比.结果表明,GP用于LOD变化预报是高效可行的.     

考虑LS+AR模型基础数据量的极移预报研究

针对目前极移最小二乘(Least Square, LS)+自回归(AutoRegressive, AR)预报模型的单一数据选取方案, 提出分别考虑LS模型数据量和AR残差数据量的组合数据模式, 并对极移预报时单一数据和组合数据预报结果精度进行分析, 探讨模型输入数据量对极移预报精度的影响. 结果表明, 模型输入数据量的变化对极移预报结果影响较大. 采用组合数据预报的方式相比较于单一数据量预报方式精度更高, 特别是针对30--360 d跨度内的中长期预报, 组合数据量的极移预报精度可比单一数据量预报精度有较大改善. 结论证明组合数据在极移预报时具有一定的优势, 可为以后极移预报数据量选取提供一定的借鉴参考意义.     

A Volterra Adaptive Filtering Method for Polar Motion Prediction Based on Chaotic Time Series

In consideration of the complex time-varying characteristics of polar motion (PM), this paper takes PM as chaotic time series. A Volterra adaptive filter is employed for predicting PM based on the state space reconstruction of delay-coordinate embedding of dynamic system. This method first uses the Least Squares (LS) technology to estimate the harmonic models for the linear trend, Annual and Chandler Wobbles (AW and CW) in PM. The selected LS deterministic models are subsequently used to extrapolate the linear trend, AW, and CW, and obtain the LS residues (the difference between the LS model and PM data themselves). Secondly, the phase space and largest Lyapunov exponent of the LS residues are reconstructed, and calculated by means of the C-C and small data-set algorithm, respectively. Further, a Volterra adaptive filter is designed for generating the extrapolations of the LS residues. The extrapolated LS residues are then added to the LS deterministic models in order to obtain the predicted PM values. The EOP C04 time series released by the International Earth Rotation and Reference Systems Service (IERS) are selected as data base to generate the PM predictions up to 60 days in the future. The results of the predictions are analyzed and compared with those obtained by the Earth Orientation Parameters Prediction Comparison Campaign (EOP PCC) and IERS Bulletin A. The results show that the accuracy of the predictions up to 30 days is comparable with that by the most accurate prediction techniques participating in the EOP PCC for PM, but worse than that by those most accurate techniques beyond 30 days in the future. The results also illustrate that the short-term predictions are better than those published by the IERS Bulletin A. However, the errors of the predictions rapidly increase with the prediction days. It is therefore concluded that the proposed method is a potential technology for short-term PM prediction.     

Temporal Variations in the Second-Degree Stokes Tesseral Geopotential Coefficients from Topex/Poseidon Altimetry

The TOPEX/POSEIDON (T/P) altimetry data set covering the periodof January 1, 1993 to January 3, 2001 was used to derive monthlyseries of the second-degree tesseral geopotential coefficients.To account for the sea water temperature variations, rathersimple models have been devised and discussed, describinglocalized as well as areal variations of sea water temperatureand heights. The second-degree tesseral coefficients have alsobeen shown to be proportional to the pressureportions of the oceanic equatorial effective excitation functions,used in Ocean Angular Momentum (OAM) data. OAM datatogether with Atmospheric Angular Momentum (AAM) data canbe used to study observed polar motion (PM) series.The excess PM rates, derived from the T/P effective excitationfunctions, were compared to the corresponding observed PM rates,derived from the International Earth Rotation Service (IERS)Bulletin A and corrected with AAM also obtainedfrom IERS. The noise of the T/P derived PM rate series was foundto be significantly larger than the corresponding Bulletin A/AAMPM rate residuals as well as the PM rates derived from anindependent OAM series that was also available for the1993–2000 period.     

IGS′92联测期间地极坐标的高频波动

本文分析了ＩＧＳ＇９２联测期间七个ＧＰＳ数据处理中心提供的极坐标序列。通过谱分析、最小二乘拟合和Ｆ检验，表明在这些序列中存在一些共同的高频波动：在Ｘ方向上具有２７．０，１６．５，１３．４和１０．４天的周期，在Ｙ方向上的波动周期约为２０．５，１５．８和１０．０天。并且每个序列与ＥＯＰ（ＩＥＲＳ）９２Ｃ０４之间都存在一个系统差。计算与分析表明，这些系统偏离的主要原因是由于在用ＧＰＳ资料解算Ｘ、Ｙ时，不同分析中心采用了不同系统的台站坐标（或者说只有部分台站采用了固定的台站坐标），从而造成这些序列所在的参考架与ＩＴＲＦ９１之间存在一个平移和旋转。最后，计算了该期间的大气角动量激发函数，可部分地解释该期间的Ｘ、Ｙ高频波动的原因。     

IVS分析中心EOP产品及精度评估

介绍了国际测地/天体测量学甚长基线干涉测量服务(International Very Long Baseline Interferometry (VLBI) Service for Geodesy and Astrometry, IVS)组织机构及下属分析中心概况.系统归纳了目前IVS发布的地球定向参数(Earth Orientation Parameters, EOP)产品类型及不同观测类型的用途.利用2010—2019年公开发布的观测资料,对IVS不同分析中心的EOP日常监测和服务能力进行了评估.通过构造观测台站所构成的几何体积,分析了EOP精度与测站数量、测站网分布的关系,统计了IVS不同观测类型的EOP解算精度.此外,综合公开发布的美国、欧洲等区域网观测数据,分析了不同地区区域网的常规及加强观测结果与IVS结果的差异.结果表明:EOP的解算精度与观测台站的分布密切相关, IVS常规观测确定的极移分量的外符合精度优于0.2 mas,世界时(Universal Time, UT1)与协调世界时(Coordinated Universal Time,UTC)之差(UT1-UTC)的精度在0.015 ms左右,加强观测的UT1-UTC值与国际自转服务组织(International Earth Rotation Service, IERS)的C04之间存在0.02–0.03 ms的差异.区域观测网的精度受观测网形和基线长度制约,总体劣于IVS观测精度,其中,美国甚长基线干涉阵列(Very Long Baseline Array, VLBA)的常规及加强观测结果与IVS全球观测结果最接近.     

On the regime of the Earth’s polar motion based on data for the period 1962–2007

The view of the Earth’s polar motion as a completely deterministic process has been called into question in the past decades, because no long-term prediction can be made. At the same time, no fundamental restrictions currently exist in the problem of a long-term prediction of the Earth’s rotation. Determining the boundaries of predictability is related to identifying the regime of the Earth’s polar motion. IERS data for the period 1962–2007 have been used to study the regime of the Earth’s polar motion. Analysis of the plots of polhodes reveals peculiarities in the variations of the pole’s coordinates X and Y in certain intervals along the time axis. The data in the interval from 2003 to 2006 have been analyzed in greatest detail: a model for the Chandler and annual oscillations has been constructed and relations between the parameters of these oscillations have been determined; the shift of the instantaneous pole on the phase plane and the Poincare plane has been investigated. As a result, we have found features inherent in chaotic motion (intermittency) and calculated the period (32 years) of the possible repetitions of such anomalies, as confirmed by our analysis of the plots of polhodes. The intervals where the peculiarities in the motion of the Earth’s instantaneous pole manifest themselves are compared with the intervals of the inflections on the plots of variations in the length of the day (LOD).     

Using the quasar VLBI network for the fundamental time—positioning service of the GLONASS space system

Regular high-precision determinations of the Earth’s orientation parameters (EOPs) on the Quasar VLBI Network were begun in August 2006. The observations are performed within the framework of two national programs: daily sessions at three observatories of the Network to determine all five EOPs (the RU-E program) and 8-h sessions on the Zelenchukskaya-Badary and Svetloe-Badary baselines to determine the Universal Time (the RU-U program). The observations from August 2006 through May 2007 are analyzed. The rms deviations of the EOP values obtained in the RU-E program from the IERS C04 series are 1.1 mas for X _p and Y _p, 37 μs for UT1-UTC, and 0.7 and 0.6 mas for X _c and Y _c, respectively. These results closely match the prospective requirements of GLONASS. The rms deviations of the Universal Times obtained in the RU-U program from the IERS C04 series are 146 μs. We consider the immediate prospects for improving the accuracy of EOP determinations in daily sessions and for implementing the e-VLBI mode for an online determination of the Universal Time. Original Russian Text ? A.M. Finkelstein, E.A. Skurikhina, I.F. Surkis, A.V. Ipatov, I.A. Rakhimov, S.G. Smolentsev, 2008, published in Pis’ma v Astronomicheskiĭ Zhurnal, 2008, Vol. 34, No. 1, pp. 66–76.     

无先验基准方法在SLR资料处理中的应用

SLR（satellite Laser Ranging)资料处理一般来说总是由各SLR测站构成的从标框架里进行。为了克服在SLR资料处理过程中对坐标框架的重复定义，利用全球1999年1月到12月的LAGEOS-1的SLR资料以无先验基准方法解算EOP（Earth Orientation Parameters）和所有SLR站的坐标的试验。在SLR资料处理中用无先验基准方法与GPS（Glaobal Positioning System)的不太一样，由于SLR的资料不能把SLR观测站连结成非常牢固的空间多面体（GPS的资料在每一瞬间可以拟测站联成一完整的空间多面体），因此需要加一些约束，以避免法方程出现秩亏。解得的测站从标用7参数转换到ITRF97坐标系，rms为1.3cm。EOP与IERS的eopc04序列相比，Xp、Yp、的rms分别为0.37mas、0.30mas，LOD（Length Of Day)的rmas为0.019ms。     

利用Lageos 1 SLR资料解算1990-2001年的地球定向参数

利用 12年的Lageos 1卫星激光测距资料 (1990 - 2 0 0 1)解算得到了地球定向参数 (EOP) ,将该序列的结果与同期的EOP(IERS)C0 4进行比较 ,其外符精度为 :极移XP— 0 .4 0mas,YP— 0 .4 2mas ,日长变化Dr— 0 .0 35ms。