A procedure for testing the assumption of homoscedasticity in least squares residuals: a case study of GPS carrier-phase observations

Due to increased demands on the quality of the results of Global Positioning System (GPS) evaluations, various authors have studied improvements of the stochastic model of GPS carrier-phase observations. These improvements are based on the reasonable assumption that the commonly used stochastic model with independent and homoscedastic (i.e. equal variance) errors is unrealistic. However, this has not been proved rigorously so far. A statistical test procedure based on uncorrelated least–squares residuals, which allows verification of the hypothesis of a heterogeneous variance, is provided. The statistical test procedure is of interest in its own right, and is independent of the practical problem considered. The presented technique is applied to GPS carrier-phase observations. Results show that the variances of the investigated observations are far from homogeneous. It is indicated that the error variances of the presented data increase with decreasing GPS satellite elevation. These results confirm the assumption that the commonly used stochastic model of GPS observations is inadequate and has to be improved.     

Calibration errors on experimental slant total electron content (TEC) determined with GPS

The Global Positioning System (GPS) has become a powerful tool for ionospheric studies. In addition, ionospheric corrections are necessary for the augmentation systems required for Global Navigation Satellite Systems (GNSS) use. Dual-frequency carrier-phase and code-delay GPS observations are combined to obtain ionospheric observables related to the slant total electron content (sTEC) along the satellite-receiver line-of-sight (LoS). This observable is affected by inter-frequency biases [IFB; often called differential code biases (DCB)] due to the transmitting and the receiving hardware. These biases must be estimated and eliminated from the data in order to calibrate the experimental sTEC obtained from GPS observations. Based on the analysis of single differences of the ionospheric observations obtained from pairs of co-located dual-frequency GPS receivers, this research addresses two major issues: (1) assessing the errors translated from the code-delay to the carrier-phase ionospheric observable by the so-called levelling process, applied to reduce carrier-phase ambiguities from the data; and (2) assessing the short-term stability of receiver IFB. The conclusions achieved are: (1) the levelled carrier-phase ionospheric observable is affected by a systematic error, produced by code-delay multi-path through the levelling procedure; and (2) receiver IFB may experience significant changes during 1 day. The magnitude of both effects depends on the receiver/antenna configuration. Levelling errors found in this research vary from 1.4 total electron content units (TECU) to 5.3 TECU. In addition, intra-day vaiations of code-delay receiver IFB ranging from 1.4 to 8.8 TECU were detected.     

GPS观测量先验方差-协方差矩阵实时估计

GPS观测量的先验方差－协方差矩阵的可靠性直接关系到GPS定位结果和可靠性，关系到模糊度初始化时间、模糊度搜索的可靠性及成功率。本文提出了一种GPS观测量的先验方差－协方差矩阵的实时估计方法。其特点是直接利用伪距和载波相位观测值，来实时估计先验方差－协方差矩阵，而且可广泛应用于各种测量型接收机的各种测量模式。该方法应用于模糊度解算中，并与其他方法进行比较，以检验其效果。     

Minimal detectable biases of GPS data

In this contribution closed-form expressions are given for the minimal detectable biases of single- and dual-frequency pseudo-range and carrier-phase data. They are given for three different single-baseline models. These are the geometry-free model and two variants of the geometry-based model, namely the roving and stationary variants. The baselines are considered to be sufficiently short such that orbital uncertainties in the fixed orbits and residual ionospheric and tropospheric delays can be assumed absent. The stochastic model used is one that permits cross-correlation and the use of different variances for individual GPS observables, including the possibility to weigh the observables in dependence on which satellite is tracked. Received: 23 April 1997 / Accepted: 27 October 1997     

Taking correlations into account: a diagonal correlation model

GPS Solutions - The true covariance matrix of the GPS phase observations is unknown and has to be assumed or estimated. The variance of the least-squares residuals was empirically shown to have an...     

Accuracy assessment of the GPS-based slant total electron content

The main scope of this research is to assess the ultimate accuracy that can be achieved for the slant total electron content (sTEC) estimated from dual-frequency global positioning system (GPS) observations which depends, primarily, on the calibration of the inter-frequency biases (IFB). Two different calibration approaches are analyzed: the so-called satellite-by-satellite one, which involves levelling the carrier-phase to the code-delay GPS observations and then the IFB estimation; and the so-called arc-by-arc one, which avoids the use of code-delay observations but requires the estimation of arc-dependent biases. Two strategies are used for the analysis: the first one compares calibrated sTEC from two co-located GPS receivers that serve to assess the levelling errors; and the second one, assesses the model error using synthetic data free of calibration error, produced with a specially developed technique. The results show that the arc-by-arc calibration technique performs better than the satellite-by-satellite one for mid-latitudes, while the opposite happens for low-latitudes.     

Improving carrier-phase ambiguity resolution in global GPS network solutions

Integer carrier-phase ambiguity resolution is one of the critical issues for precise GPS applications in geodesy and geodynamics. To resolve as many integer ambiguities as possible, the ‘most-easy-to-fix’ double-difference ambiguities have to be defined. For this purpose, several strategies are implemented in existing GPS software packages, such as choosing the ambiguities according to the baseline length or the variances of the estimated real-valued ambiguities. Although their efficiencies are demonstrated in practice, it is proven in this paper that they do not reflect all effects of varying data quality, because they are based on theoretical considerations of GPS data processing. Therefore, a new approach is presented, which selects the double-difference ambiguities according to their probability of being fixed to the nearest integer. The probability is computed from estimates and variances of wide-lane and narrow-lane ambiguities. Together with an optimized ambiguity fixing procedure, the new approach is implemented in the routine data processing for the International GPS Service (IGS) at GeoForschungsZentrum (GFZ) Potsdam. Within a sub-network of about 90 IGS stations, it is demonstrated that more than 97% of the independent ambiguities are fixed correctly compared to 75% by a commonly used method, and that the additionally fixed ambiguities improve the repeatability of the station coordinates by 10–26% in regions with sparse site distribution.     

Efficiency of carrier-phase integer ambiguity resolution for precise GPS positioning in noisy environments

Precise GPS positioning relies on tracking the carrier-phase. The fractional part of carrier-phase can be measured directly using a standard phase-locked loop, but the integer part is ambiguous and the ambiguity must be resolved based on sequential carrier-phase measurements to ensure the required positioning precision. In the presence of large phase-measurement noise, as can be expected in a jamming environment for example, the amount of data required to resolve the integer ambiguity can be large, which requires a long time for any generic integer parameter estimation algorithm to converge. A key question of interest in significant applications of GPS where fast and accurate positioning is desired is then how the convergence time depends on the noise amplitude. Here we address this question by investigating integer least-sqaures estimation algorithms. Our theoretical derivation and numerical experiments indicate that the convergence time increases linearly with the noise variance, suggesting a less stringent requirement for the convergence time than intuitively expected, even in a jamming environment where the phase noise amplitude is large. This finding can be useful for practical design of GPS-based systems in a jamming environment, for which the ambiguity resolution time for precise positioning may be critical.     

Helmert验后方差估计在GPS/GLONASS组合单点定位中的应用

在给出组合GPS/GLONASS单点定位原理后,提出了利用Helmert验后方差估计法对GPS/GLONASS观测值定权的方法,并结合应用实例,通过对比分析说明该方法的可行性,实践表明该方法定权比采用固定权或经验定权更具有合理性。     

An integer ambiguity resolution procedure for GPS/pseudolite/INS integration

In order to achieve a precise positioning solution from GPS, the carrier-phase measurements with correctly resolved integer ambiguities must be used. Based on the integration of GPS with pseudolites and Inertial Navigation Systems (INS), this paper proposes an effective procedure for single-frequency carrier-phase integer ambiguity resolution. With the inclusion of pseudolites and INS measurements, the proposed procedure can speed up the ambiguity resolution process and increase the reliability of the resolved ambiguities. In addition, a recently developed ambiguity validation test, and a stochastic modelling scheme (based on-line covariance matrix estimation) are adapted to enhance the quality of ambiguity resolution. The results of simulation studies and field experiments indicate that the proposed procedure indeed improves the performance of single-frequency ambiguity resolution in terms of both reliability and time-to-fix-ambiguity.     

Uncertainty in GPS Networks due to Remaining Systematic Errors: The Interval Approach

A realistic assessment of the total uncertainty budget of Global Positioning System (GPS) observations and its adequate mathematical treatment is a basic requirement for all analysis and interpretation of GPS-derived point positions, in particular GPS heights, and their respective changes. This implies not only the random variability but also the remaining systematic errors. At present in geodesy, the main focus is on stochastic approaches in which errors are modeled by means of random variables. Here, an alternative approach based on interval mathematics is presented. It allows us to model and to quantify the impact of remaining systematic errors in GPS carrier-phase observations on the final results using deterministic error bands. In this paper, emphasis is given to the derivation of the observation intervals based on influence parameters and to the study of the complex linear transfer of this type of uncertainty to estimated point positions yielding zonotopes. From the presented simulation studies of GPS baselines, it turns out that the uncertainty due to remaining systematic effects dominates the total uncertainty budget for baselines longer than 200 km.     

一种联合载波相位观测值的GPS/BDS滤波导航算法

提出了一种基于历元间相位差分的GPS/BDS单机实时动态定位算法。该方法采用历元间载波相位差分数据准确计算出载体的位置变化量;并以此描述载体的运动状态变化,建立动态定位滤波模型的状态方程。同时以历元间载波相位差分数据与伪距数据作为主要观测值,采用扩展Kalman滤波实时估计载体的位置和钟差。采用自主编制的软件对静态与车载GPS/BDS实测数据进行处理,结果表明:采用该方法,定位结果精度优于传统的标准单点定位算法与载波相位平滑伪距算法;而且算法具有较好的稳定性,与载体的运动状态无关。     

On-line detection of abrupt changes in the carrier-phase measurements of GPS

The detection of cycle slips in carrier-phase data, outliers in phase rate or in code ranges, or any other type of disorder in the measurements of the GPS system, is one of the major quality-control problems that needs to be addressed, especially where GPS is used for real-time applications or for those applications that require continuous and reliable positioning results. This study is concerned with designing algorithms to detect failures or changes of small magnitude, in order to monitor and control the quality of GPS measurements for critical real-time and/or deformation applications. In this regard, the statistical technique known as the cumulative-sum test is described for the detection of abrupt changes in the GPS carrier-phase measurements. Real GPS data have been used to test and evaluate the algorithm. Received: 27 August 1996 / Accepted: 26 February 1997     

GPS slant total electron content accuracy using the single layer model under different geomagnetic regions and ionospheric conditions

The use of observations from the Global Positioning System (GPS) has significantly impacted the study of the ionosphere. As it is widely known, dual-frequency GPS observations can provide very precise estimation of the slant Total Electron Content (sTEC—the linear integral of the electron density along a ray-path) and that the precision level is bounded by the carrier-phase noise and multi-path effects on both frequencies. Despite its precision, GPS sTEC estimations can be systematically affected by errors in the estimation of the satellites and receivers by Inter-Frequency Biases (IFB) that are simultaneously determined with the sTEC. Thus, the ultimate accuracy of the GPS sTEC estimation is determined by the errors with which the IFBs are estimated. This contribution attempts to assess the accuracy of IFBs estimation techniques based on the single layer model for different ionospheric regions (low, mid and high magnetic latitude); different seasons (summer and winter solstices and spring and autumn equinoxes); different solar activity levels (high and low); and different geomagnetic conditions (quiet and very disturbed). The followed strategy relies upon the generation of a synthetic data set free of IFB, multi-path, measurement noise and of any other error source. Therefore, when a data set with such properties is used as the input of the IFB estimation algorithms, any deviation from zero on the estimated IFBs should be taken as indications of the errors introduced by the estimation technique. The truthfulness of this assessment work is warranted by the fact that the synthetic data sets resemble, as realistically as possible, the different conditions that may happen in the real ionosphere. The results of this work show that during the high solar activity period the accuracy for the estimated sTEC is approximately of ±10 TECu for the low geomagnetic region and of ±2.2 TECu for the mid-latitude. During low solar activity the accuracy can be assumed to be in the order of ±2 TECu. For the geomagnetic high-disturbed period, the results show that the accuracy is degraded for those stations located over the region where the storm has the strongest impact, but for those stations over regions where the storm has a moderate effect, the accuracy is comparable to that obtained in the quiet period.     

Quality aspects of a GPS reference network in Antarctica – a simulation study

GPS has become an essential tool for the precise determination of point positions. Since GPS campaigns for geodynamic purposes – such as the monitoring of recent crustal movements – require major financial efforts, it is essential to ensure already in the planning phase a good network quality and the attainment of the scientific goals in a reasonable time. The paper outlines an operationally oriented procedure for these purposes based on the simulation and processing of GPS carrier-phase observations. Sensitivity analysis techniques are applied to describe both the network strain as it may be induced by gross errors and the detectability of point movements between two sessions or two campaigns. In addition, the eigenvalue decomposition of the variance-covariance matrix of the GPS coordinates will be used to identify weakly determined network components. The design of the SCAR Epoch 95 Campaign GPS network is discussed throughout the paper. It was realized in the Antarctic summer 1994/1995 on the Antarctic peninsula. Received: 30 November 1995 / Accepted: 1 July 1997     

Spatially convergent GPS kinematic positioning

It is a known fact that obtaining accurate GPS carrier-phase measurements involves fixed, unknown whole-cycle ambiguity parameters. As the use of cosine functions to eliminate any double-difference integer ambiguities causes spatial ambiguity problems, both reasonably approximated positions and wavelength-dependent convergence ranges are of the utmost importance. Differential GPS-based position solutions are first smoothed to create a polynomial trajectory, leading to less variable position approximations. Long-wavelength wide-lane phase combinations will then be utilized to facilitate convergent GPS positioning, on a stage-by-stage basis. Although double-difference ionospheric path delays are often interpreted as nuisance parameters, they can be obtained when the respective cosines of the original L1 and L2 carrier phases undergo a simultaneous least-squares estimation. In particular, quadratic forms of the estimated phase residuals will be linked with hypothesis testing to allow for a meaningful statistical inference. Some low-dynamics experiments are then performed to prove the feasibility of the proposed hierarchical positioning concept. Electronic Publication     

Spatiotemporal filtering for regional GPS network in China using independent component analysis

Removal of the common mode error (CME) is a routine procedure in postprocessing regional GPS network observations, which is commonly performed using principal component analysis (PCA). PCA decomposes a network time series into a group of modes, where each mode comprises a common temporal function and corresponding spatial response based on second-order statistics (variance and covariance). However, the probability distribution function of a GPS time series is non-Gaussian; therefore, the largest variances do not correspond to the meaningful axes, and the PCA-derived components may not have an obvious physical meaning. In this study, the CME was assumed statistically independent of other errors, and it was extracted using independent component analysis (ICA), which involves higher-order statistics. First, the ICA performance was tested using a simulated example and compared with PCA and stacking methods. The existence of strong local effects on some stations causes significant large spatial responses and, therefore, a strategy based on median and interquartile range statistics was proposed to identify abnormal sites. After discarding abnormal sites, two indices based on the analysis of the spatial responses of all sites in each independent component (east, north, and vertical) were used to define the CME quantitatively. Continuous GPS coordinate time series spanning \(\sim \)4.5 years obtained from 259 stations of the Tectonic and Environmental Observation Network of Mainland China (CMONOC II) were analyzed using both PCA and ICA methods and their results compared. The results suggest that PCA is susceptible to deriving an artificial spatial structure, whereas ICA separates the CME from other errors reliably. Our results demonstrate that the spatial characteristics of the CME for CMONOC II are not uniform for the east, north, and vertical components, but have an obvious north–south or east–west distribution. After discarding 84 abnormal sites and performing spatiotemporal filtering using ICA, an average reduction in scatter of 6.3% was achieved for all three components.     

Interferometric Baseline Performance Estimations for Multistatic Synthetic Aperture Radar Configurations Derived from GRACE GPS Observations

Recent studies have demonstrated the usefulness of global positioning system (GPS) receivers for relative positioning of formation-flying satellites using dual-frequency carrier-phase observations. The accurate determination of distances or baselines between satellites flying in formation can provide significant benefits to a wide area of geodetic studies. For spaceborne radar interferometry in particular, such measurements will improve the accuracy of interferometric products such as digital elevation models (DEM) or surface deformation maps. The aim of this study is to analyze the impact of relative position errors on the interferometric baseline performance of multistatic synthetic aperture radar (SAR) satellites flying in such a formation. Based on accuracy results obtained from differential GPS (DGPS) observations between the twin gravity recovery and climate experiment (GRACE) satellites, baseline uncertainties are derived for three interferometric scenarios of a dedicated SAR mission. For cross-track interferometry in a bistatic operational mode, a mean 2D baseline error (1σ) of 1.4 mm is derived, whereas baseline estimates necessary for a monostatic acquisition mode with a 50 km along-track separation reveal a 2D uncertainty of approximately 1.7 mm. Absolute orbit solutions based on reduced dynamic orbit determination techniques using GRACE GPS code and carrier-phase data allows a repeat-pass baseline estimation with an accuracy down to 4 cm (2D 1σ). To assess the accuracy with respect to quality requirements of high-resolution DEMs, topographic height errors are derived from the estimated baseline uncertainties. Taking the monostatic pursuit flight configuration as the worst case for baseline performance, the analysis reveals that the induced low-frequency modulation (height bias) fulfills the relative vertical accuracy requirement (σ<1 m linear point-to-point error) according to the digital terrain elevation data level 3 (DTED-3) specifications for most of the baseline constellations. The use of a GPS-based reduced dynamic orbit determination technique improves the baseline performance for repeat-pass interferometry. The problem of fulfilling the DTED-3 horizontal accuracy requirements is still an issue to be investigated. DGPS can be used as an operational navigation tool for high-precision baseline estimation if a geodetic-grade dual-frequency spaceborne GPS receiver is assumed to be the primary instrument onboard the SAR satellites. The possibility of using only single-frequency receivers, however, requires further research effort.Deutsche Forschungsgemeinschaft (DFG) research fellow until Sept. 2004 at the Microwaves and Radar Institute, Deutsche Zentrum für Luft- und Raumfahrt (DLR) e.V., 82234 Weßling, Germany