利用奇异谱分析探测GPS坐标时间序列时变周期信号

GPS坐标时间序列呈现复杂的非线性特征,准确提取坐标时间序列中的季节性信号对理解区域形变具有重要意义。传统最小二乘拟合(Least Squares Fitting,LSF)无法顾及信号的年际变化特征,难以有效分离噪声和周期信号。介绍利用奇异谱分析(Singular Spectrum Analysis,SSA)方法提取GPS坐标时间序列中时变周期信号。结果表明,SSA可从含有噪声的GPS时间序列中提取各坐标分量的周期信号,相较于LSF,SSA在非线性信号的准确提取上具有更好的效果。     

云南地区GPS垂向时间序列周期项信号的SSA提取

针对传统的GPS时间序列最小二乘(LS)谐波函数模型方法不能精确提取GPS垂向时间序列的周期项信号问题,该文以云南地区2011年1月—2017年6月连续观测的27个GPS基准站数据为例,使用主成分分析(PCA)获得27个测站的共模误差,然后分别使用奇异谱分析(SSA)和LS方法对PCA滤波前后的测站时间序列进行周期项信号提取.实验结果表明,原始时间序列分别减去SSA、PCA-SSA、LS和PCA-LS提取周期项信号的RMS减少百分比平均为25.66％、17.23％、16.76％、16.76％;原始时间序列与4种方法提取周期项信号的相关系数平均为0.66、0.55、0.54、0.54,说明SSA提取的周期项信号要优于其他3种方法,将SSA应用于云南地区GPS垂向时间序列的周期项信号提取是可行和有效的.     

一种基于交叉证认技术的自适应小波变换及其在削减GPS多路径误差中的应用

多路径效应影响是目前限制GPS定位精度进一步提高的瓶颈。提出用交叉证认技术自动识别小波分解的信号层,再通过小波重构实现降噪和信号提取,并将该方法应用于GPS多路径误差的削减中。对模拟数据和实际GPS观测资料的分析表明,该方法能合理分离不同噪声水平下资料序列中的信号和噪声;当噪声水平小于信号振幅的一半时,能成功分离资料序列中的高频信号。同时,运用该方法得到的多路径改正模型和GPS多路径效应的重复性,可有效地削弱多路径效应的影响,提高GPS定位精度。     

山东CORS基准站坐标时间序列噪声分析

针对目前在分析GPS基准站噪声时对其坐标时间序列周期项提取不充分的现状,该文提出用小波谱探测坐标时间序列中的周期项,用小波分析方法提取周期项,得到不含明显周期项噪声序列。以山东CORS中26个基准站两年的坐标时间序列为例,去除了坐标时间序列中阶跃项,确定了长期趋势项,对周期项进行了探测、分离和提取,得到了各测站的噪声序列;用谱指数法分析了噪声的类型,用最大似然估计获取了各种噪声的分量,进而建立了噪声组合模型。实验结果表明,山东CORS基准站噪声模型可以用白噪声加闪烁噪声的形式来描述,其白噪声的平均值为1.25mm,闪烁噪声的平均值为5.46mm。     

顾及MLE的IGS坐标时间序列噪声模型特性分析

以30个GPS基准站坐标序列为对象,提出分别采用赤池信息量准则（AIC）与贝叶斯信息准则（BIC）噪声模型估计准则判定GPS时间序列噪声特性,对比分析GPS时间序列噪声模型特性,探讨不同噪声模型对GPS站速度及其不确定度的影响. 结果表明GPS站坐标序列噪声模型主要表现为FN+WN、PL及FN+RW+WN噪声模型特性;FN+WN噪声模型对GPS站速度估计值的影响相对较小,但在U分量影响最为明显;此外,RW对站速度不确定度的影响不可忽略, 正确获取模型参数估计的实际不确定度及改正噪声分量对于合理应用GPS坐标时间序列数据具有重要的意义.      

PANGA坐标时间序列噪声模型特性分析

选取PANGA观测网中的200个GPS基准站12年的坐标序列. 采用AIC模型估计准则噪声特性进行分析,并结合不同数据策略对噪声模型建立的影响进行探讨. 结果表明:不同数据策略对噪声模型建立的影响较小,基准站站坐标序列噪声模型主要表现为PL+WN噪声模型特性;不同随机模型对GPS站速度估计值的影响相对较小,但在U分量影响最为明显;对比不同策略处理下的周年振幅变化,E、N、U方向振幅值存在差异,最终确定了振幅位移的尺度.      

联合GRACE与GPS比较山西省垂向地表形变

利用2009-05～2010-12共20mon的GRACE时变重力场反演了山西省境内陆地水储量变化造成的地表垂直位移,并与同期的GPS高程残差进行了比较。结果显示,对大部分CORS站而言,GRACE反演的水文地壳垂向形变与GPS高程残差时间序变化量均在-6～6mm之间;由二者提取的年周期信号平均振幅均为2.5mm左右,但是相位存在较大差异。由于GPS数据处理过程未考虑周日、半周日大气潮汐,高阶电离层,及非潮汐海潮的影响,这三项误差会传播到测站坐标的时间序列,对GPS周年信号产生影响,所以可以认为GRACE与GPS结果的差异更多的是来自GPS数据处理的误差。     

时间跨度对GPS坐标时间序列噪声模型估计影响分析

文中介绍GPS坐标时间序列模型,并给出GPS噪声模型种类、估计方法及最佳噪声模型评价准则。选取来自IGS08的24个IGS核心站的时间序列作为数据基础,对不同时间跨度的时间序列噪声模型进行定量分析,探讨其随着时间变化的演化规律。结果表明:当时间序列长度较短时,噪声模型的不确定性较大,而12.5a以上时间跨度的噪声模型趋于稳定且主要表现为闪烁噪声加白噪声的组合噪声模型;大跨度时间序列中噪声的长周期分量变得显著。     

GPS多路径误差滤波方法比较研究

基于GPS多路径时间序列,分别采用Vondrak滤波、经验模式分解(EMD)和小波滤波3种方法构建了GPS多路径误差修正模型,并将其用于削弱多路径效应.通过对模拟数据及实测数据的分析表明,3种方法都能有效地分离不同噪声水平下时间序列中的信号和噪声.同时,利用3种方法构建的多路径修正模型可有效地削弱多路径效应的影响,提高GPS定位精度,但3种方法的优缺点各有不同.     

地表质量负载对格陵兰岛区域GPS站点坐标时序特征的影响

选取ITRF2008框架下格陵兰岛区域12个GPS站2013年1月-2016年12月期间的日解坐标时间序列作为研究对象，并利用极大似然估计分析地表质量负载改正前后各站点的噪声特性、速度场及周期项振幅。结果表明：站点最优噪声模型主要为白噪声+幂律噪声与白噪声+闪烁噪声，地表质量负载形变修正GPS坐标时序后，明显增加U方向闪烁噪声的成分，平均降低其速度约0.36 mm/a，对水平方向影响较小；同时分别降低高程方向44.1%、14.2%的1 a项、0.5 a项振幅，相反，却增加了水平方向的周期项振幅。     

GPS position time-series analysis based on asymptotic normality of M-estimation

The efficacy of robust M-estimators is a well-known issue when dealing with observational blunders. When the number of observations is considerably large—long time series for instance—one can take advantage of the asymptotic normality of the M-estimation and compute reasonable estimates for the unknown parameters of interest. A few leading M-estimators have been employed to identify the most likely functional model for GPS coordinate time series. This includes the simultaneous detection of periodic patterns and offsets in the GPS time series. Estimates of white noise, flicker noise, and random walk noise components are also achieved using the robust M-estimators of (co)variance components, developed in the framework of the least-squares variance component estimation (LS-VCE) theory. The method allows one to compute confidence interval for the (co)variance components in asymptotic sense. Simulated time series using white noise plus flicker noise show that the estimates of random walk noise fluctuate more than those of flicker noise for different M-estimators. This is because random walk noise is not an appropriate noise structure for the series. The same phenomenon is observed using the results of real GPS time series, which implies that the combination of white plus flicker noise is well described for GPS time series. Some of the estimated noise components of LS-VCE differ significantly from those of other M- estimators. This reveals that there are a large number of outliers in the series. This conclusion is also affirmed by performing the statistical tests, which detect (large) parts of the outliers but can also leave parts to be undetected.     

iGPS: IDL tool package for GPS position time series analysis

A new tool package written in Interactive Data Language (IDL) was developed for processing and analyzing daily continuous GPS position time series. This software package can read continuous GPS position time series in various formats, detect outliers, remove abnormal observation spans, locate epochs of offsets or post-seismic relaxation events, and estimate their amplitudes. It also provides functionalities for epoch statistics, site selection, periodic noise analysis, spatial filtering (by regional stacking), etc. This tool is referred to as iGPS.     

GPS坐标时间序列中的异常高频周期性噪声

本文分析了中国地壳运动观测网络GPS基准站及中国周边IGS连续站坐标中的周期性噪声,在北、东和垂向分量坐标时间序列中均发现了周期约351/n(n=1,…,6)天的正弦项,而地表质量负荷造成的位移序列中并没有对应的成分,不能解释异常周期项的来源。在估计台站的运动速率时,如考虑此类"异常"周期项,速率误差略微减小,而周年项振幅的估计误差显著增大,且残差离散度未见明显改善。因此,不推荐在GPS速率矢量反演中加入异常周期项参数。     

联合区域叠加滤波法与小波变换去除GPS站坐标时间序列噪声

GPS站坐标时间序列中存在的周期性与非周期性误差严重影响了对测站运动特征的分析及其非线性变化的物理机制解释。因此,为削弱噪声的影响,本文首先利用区域叠加滤波法去除了南加利福尼亚地区16个测站时间序列的共模误差,以此削弱时间序列中存在的包括周年和半周年误差在内的周期性误差。为去除滤波后残留的噪声,对滤波后的信号进行静态离散小波变换,提取了周期为半周年以上的信号。结果表明,联合区域叠加滤波法与小波变换对GPS站坐标时间序列进行处理,既能够削弱周期性误差对信号的影响,又能较好地提取测站的非线性运动信号。     

Filtering GPS time-series using a Vondrak filter and cross-validation

Multipath disturbance is one of the most important error sources in high-accuracy global positioning system (GPS) positioning and navigation. A new data filtering method, based on the Vondrak filter and the technique of cross-validation, is developed for separating signals from noise in data series, and applied to mitigate GPS multipath effects in applications such as deformation monitoring. Both simulated data series and real GPS observations are used to test the proposed method. It is shown that the method can be used to successfully separate signals from noise at different noise levels, and for varying signal frequencies as long as the noise level is lower than the magnitude of the signals. A multipath model can be derived, based on the current-day GPS observations, with the proposed method and used to remove multipath errors in subsequent days of GPS observations when taking advantage of the sidereal day-to-day repeating characteristics of GPS multipath signals. Tests have shown that the reduction in the root mean square (RMS) values of the GPS errors ranges from 20% to 40% when the method is applied.     

Estimating rate uncertainty with maximum likelihood: differences between power-law and flicker–random-walk models

Recent studies have documented that global positioning system (GPS) time series of position estimates have temporal correlations which have been modeled as a combination of power-law and white noise processes. When estimating quantities such as a constant rate from GPS time series data, the estimated uncertainties on these quantities are more realistic when using a noise model that includes temporal correlations than simply assuming temporally uncorrelated noise. However, the choice of the specific representation of correlated noise can affect the estimate of uncertainty. For many GPS time series, the background noise can be represented by either: (1) a sum of flicker and random-walk noise or, (2) as a power-law noise model that represents an average of the flicker and random-walk noise. For instance, if the underlying noise model is a combination of flicker and random-walk noise, then incorrectly choosing the power-law model could underestimate the rate uncertainty by a factor of two. Distinguishing between the two alternate noise models is difficult since the flicker component can dominate the assessment of the noise properties because it is spread over a significant portion of the measurable frequency band. But, although not necessarily detectable, the random-walk component can be a major constituent of the estimated rate uncertainty. None the less, it is possible to determine the upper bound on the random-walk noise.     

Anomalous harmonics in the spectra of GPS position estimates

Prior studies of the power spectra of GPS position time series have found pervasive seasonal signals against a power-law background of flicker noise plus white noise. Dong et al. (2002) estimated that less than half the observed GPS seasonal power can be explained by redistributions of geophysical fluid mass loads. Much of the residual variation is probably caused by unidentified GPS technique errors and analysis artifacts. Among possible mechanisms, Penna and Stewart (2003) have shown how unmodeled analysis errors at tidal frequencies (near 12- and 24-hour periods) can be aliased to longer periods very efficiently. Signals near fortnightly, semiannual, and annual periods are expected to be most seriously affected. We have examined spectra for the 167 sites of the International GNSS (Global Navigation Satellite Systems) Service (IGS) network having more than 200 weekly measurements during 1996.0–2006.0. The non-linear residuals of the weekly IGS solutions that were included in ITRF2005, the latest version of the International Terrestrial Reference Frame (ITRF), have been used. To improve the detection of common-mode signals, the normalized spectra of all sites have been stacked, then boxcar smoothed for each local north (N), east (E), and height (H) component. The stacked, smoothed spectra are very similar for all three components. Peaks are evident at harmonics of about 1 cycle per year (cpy) up to at least 6 cpy, but the peaks are not all at strictly 1.0 cpy intervals. Based on the 6th harmonic of the N spectrum, which is among the sharpest and largest, and assuming a linear overtone model, then a common fundamental of 1.040 ± 0.008 cpy can explain all peaks well, together with the expected annual and semiannual signals. A flicker noise power-law continuum describes the background spectrum down to periods of a few months, after which the residuals become whiter. Similar sub-seasonal tones are not apparent in the residuals of available satellite laser ranging (SLR) and very long baseline interferometry (VLBI) sites, which are both an order of magnitude less numerous and dominated by white noise. There is weak evidence for a few isolated peaks near 1 cpy harmonics in the spectra of geophysical loadings, but these are much noisier than for GPS positions. Alternative explanations related to the GPS technique are suggested by the close coincidence of the period of the 1.040 cpy frequency, about 351.2 days, to the “GPS year”; i.e., the interval required for the constellation to repeat its inertial orientation with respect to the sun. This could indicate that the harmonics are a type of systematic error related to the satellite orbits. Mechanisms could involve orbit modeling defects or aliasing of site-dependent positioning biases modulated by the varying satellite geometry.     

利用高频GPS数据反演地震要素

介绍了利用高频GPS观测数据反演地震要素的原理和方法. 以2019年7月4日和7月6日发生在美国加州里奇克莱斯特的地震为研究对象，首先利用PRIDE PPP-AR软件进行动态精密单点定位解算以获得测站的位移时间序列，然后使用S变换对位移时间序列去噪，再根据去噪后的时间序列提取地震波的振幅与周期，并提出双重3-sigma法则以分别提取地震波的体波和面波到达时，最后据此反演出地震的震中位置、震级以及发震时刻. 根据本文提出的方法反演得到的震中位置与美国地质勘探局（USGS）发布的位置偏差在10 km左右，震级偏差在0.2级左右，发震时刻偏差在20 s以内. 结果表明，利用高频GPS数据可以较为准确地反演出地震要素.      

Adaptive wavelet transform based on cross-validation method and its application to GPS multipath mitigation

Global positioning system (GPS) multipath disturbance is a bottleneck problem that limits the accuracy of precise GPS positioning applications. A method based on the technique of cross-validation for automatically identifying wavelet signal layers is developed and used for separating noise from signals in data series, and applied to mitigate GPS multipath effects. Experiments with both simulated data series and real GPS observations show that the method is a powerful signal decomposer, which can successfully separate noise from signals as long as the noise level is lower than about half of the magnitude of the signals. A multipath correction model is derived based on the proposed method and the sidereal day-to-day repeating property of GPS multipath signals to remove multipath effects on GPS observations and to improve the quality of the GPS measurements.