拟合推估两步极小解法

在回顾了最小二乘拟合推估的“综合极小”解法（正常拟合推估）后，分析了正常拟合推估存在的问题。考虑到随机场信号不一定完全表现为随机性，其中可能含有趋势性，顾而提出了拟合推估的“两步极小”解法，即将随机场分成趋势性部分和随机性部分，对趋势性部分采用函数拟合，对随机性部分采用协方差函数拟合。给出了“分两步极小”拟合推估的2种解法。计算表明，两步极小解法能部分地改善拟合推估的精度。     

On estimating gravity anomalies— A comparison of least squares collocation with conventional least squares techniques

The least squares collocation algorithm for estimating gravity anomalies from geodetic data is shown to be an application of the well known regression equations which provide the mean and covariance of a random vector (gravity anomalies) given a realization of a correlated random vector (geodetic data). It is also shown that the collocation solution for gravity anomalies is equivalent to the conventional least-squares-Stokes' function solution when the conventional solution utilizes properly weighted zero a priori estimates. The mathematical and physical assumptions underlying the least squares collocation estimator are described.     

Integral formulas for heterogeneous data in physical geodesy

The least squares estimator is derived for a random stochastic process implied by one or two heterogeneous random stochastic processes on a sphere. The solution can be regarded as least squares collocation in the continuous case. When the method is applied in physical geodesy the statistical expectation is usually substituted by the global average and the method will then give the minimum mean squares errors of the estimated quantities. The solutions can also be considered as generalizations of the classical integral formulas in physical geodesy for heterogeneous data information.     

广义平差的概括模型

提出了广义平差的概括模型———附有条件的最小二乘配置模型。该概括模型不仅包括滤波和推估模型 ,扩充了原有的最小二乘配置模型 ,而且经典平差模型都是它的特例。     

二次曲面与最小二乘配置的组合模型在GPS高程异常拟合中的应用

在工程实践应用中,为了有效利用GPS高程数据,减少对传统水准测量的依赖,提高GPS高程异常的拟合精度便显得十分重要。为此,本文在介绍二次曲面拟合和最小二乘配置拟合基本原理分析、算法过程推导的基础上,提出了一种新的高程异常拟合方法。首先在二次曲面拟合的基础上,计算得到原始观测数据与拟合数据之间的残差序列,然后采用最小二乘配置模型对包括二次曲面拟合模型误差的综合误差进行优化减弱,最后得到新的高程异常。通过实例,将二次曲面拟合法,最小二乘配置法与文中提出的新方法进行比较分析。结果表明：新的组合方法的拟合预测精度要明显优于最小二乘配置及二次曲面拟合。     

抗差最小二乘配置在数字化地图几何纠正中的应用讨论

抗差估计具有较好的抗拒异常观测值及粗差的能力,而最小二乘配置又能较好地处理系统误差,本文结合两者的优点,利用抗差最小二乘配置对数字化地图进行几何纠正,其中对协方差函数采用抗差拟合,得到了较好的结果。实验证明在GIS数据处理的扫描数字化地图几何纠正中,抗差最小二乘配置在抗拒异常值和处理系统误差方面优于单纯的最小二乘估计和单纯的最小二乘配置方法。     

STATISTICAL CONCEPTS AND THEIR APPLICATION IN PHOTOGRAMMETRY AND SURVEYING (CONTINUED)

Inner constraints, generalised matrix inverses and S-transformations are described in relation to least squares estimation of spatial positions. A simple example is used to illustrate some of the theory. The various roles and properties of covariance functions and matrices are presented with a discussion of their practical determination and use in least squares interpolation and collocation.     

最小二乘配置法在GPS高程异常推估中的应用

简述了最小二乘配置法的基本原理,并在水准联测点不多的情况下,直接采用平方根函数作为各随机参数间的协方差函数用于推估GPS高程异常值。由于最小二乘配置法的函数模型中同时考虑了非随机变量和随机变量,使得高程异常值的推估精度更高。实例分析也证明了此模型在精度上优于传统的平面拟合模型和协方差推估模型。残差分析表明此方法更适合于同时存在内插和外推高程异常值的情况。     

Kriging方法结合最小二乘配置在GPS高程拟合中的应用

本文利用Kriging方法结合最小二乘配置将GPS高程转换成正常高.研究了将Kriging方法中的变异函数用于计算最小二乘配置中的协方差的方法,并对一局部GPS水准网的高程作了拟合计算.通过将最小二乘配置法与平面拟合模型和多面函数拟合模型等进行比较,其外符合精度从最大的±0.0277m提高到±0.0162m.     

On the foundation of collocation in physical geodesy

The boundary value problem in physical geodesy is nowadays mostly presented with the use of an advanced stochastic model by Krarup-Moritz. This model includes a primary Gauss-Markov model and an adjoining Wiener-Hopf model. Degenerations of the Wiener-Hopf section are found in thesingular auto-covariance matrix of the residuals. The non-singular inverse of the auto-covariance matrix of the signal is proved to be a generalized inverse of the singular auto-covariance matrix of the residuals. The joint model is given a non-stochastic evaluation for a case with spherical external surface (using a non-singular inverse). These findings will not prevent a successful application of the model, which has important merits, specially when using suitablea priori values for the stochastic parameters in the covariance functions. A method for quadratic unbiased estimation ofa priori variances is presented in an introductory section. It is meant to be of value when using a solution of the boundary value problem with the collocation technique based on the classical Gauss-Markov solution. (Bjerhammar (1963).)     

Least squares adjustment and collocation

Summary For the estimation of parameters in linear models best linear unbiased estimates are derived in case the parameters are random variables. If their expected values are unknown, the well known formulas of least squares adjustment are obtained. If the expected values of the parameters are known, least squares collocation, prediction and filtering are derived. Hence in case of the determination of parameters, a least squares adjustment must precede a collocation because otherwise the collocation gives biased estimates. Since the collocation can be shown to be equivalent to a special case of the least squares adjustment, the variance of unit weight can be estimated for the collocation also. This estimate gives the scale factor for the covariance matrices being used in the collocation. In addition, the methods of testing hypotheses and establishing confidence intervals for the parameters of the least squares adjustment may be applied to the collocation.     

各向异性局部重力场计算的谱方法

本文所采用的基于输入－输出系统论的谱方法在计算结果的精度上与最小二乘配置方法相当,却很容易用于异性场的计算。用该谱方法对卫星测高及海洋重力资料进行组合求解重力场量（大地水准面差距和重力异常）,其误差估计结果表明各向异性场的计算精度优于各向同性场的精度。     

最小二乘配置和普通Kriging法在GPS高程转换中的应用

最小二乘配置法和Kriging法进行高程转换时均兼顾考虑高程异常的系统部分和随机部分,为对比分析两种方法在GPS高程转化时的优劣性,根据最小二乘配置法的协方差函数（或kriging法的变异函数）的不同,采用五种方案对实测数据进行处理。结果分析证明了以距离作为协方差函数的最小二乘配置法在算法和推估结果可靠性要优于其他方案。     

Numerical calculation of covariances between area means

When area means are given on a regular grid, their covariance functions, needed in least squares collocation to estimate quantities statistically related to the data, can be calculated as accurately as desired and very efficiently in the form of a Fourier series truncated at a suitably high frequency.     

On a four-dimensional integrated geodesy

In contrast to continuous global considerations of time dependent boundary value problems an attempt is made to define4D-linear observation equations in the framework of integrated geodesy for discrete, more or less regional and local applications (deformation analysis) where time variations in position and in the gravity field have to be considered. The derivation is a strict analogue and extension of the3D integrated approach. In addition the construction of time dependent covariance functions is discussed, which are necessary to solve for unknown displacements and changes in the gravity potential in the generalized least squares collocation model.     

Combination of spatial networks using an estimated covariance matrix

When combining satellite and terrestrial networks, covariance matrices are used which have been estimated from previous data. It can be shown that the least-squares estimator of the unknown parameters using such an estimated covariance matrix is not necessarily the best. There are a number of cases where a more efficient estimator can be obtained in a different way. The problem occurs frequently in geodesy, since in least-squares adjustment of correlated observations estimated covariance matrices are often used. If the general structure of the covariance matrix is known, results can often be improved by a method called covariance adjustment. The statistical model used in least-squares collocation leads to a type of covariance matrix which fits into this framework. It is shown in which way improvements can be made using a modified approach of principal component analysis. As a numerical example the combination of a satellite and a terrestrial network has been computed with varying assumptions on the covariance matrix. It is shown which types of matrices are critical and where the usual least-squares approach can be applied without hesitation. Finally, a simplified representation of covariances for spatial networks by means of a suitable covariance function is suggested. Paper presented at the International Symposium on Computational Methods in Geometrical Geodesy-Oxford, 2–8 September, 1973.     

BRDF模型参数分阶段鲁棒性反演方法

遥感BRDF物理模型均建立于一定的假设或基于某些理想状况，其模拟的数据与观测数据之间多少会存在一些差异（误差）。利用BRDF模型反演地表参数时，如果不加选择地使用所有观测数据，势必会影响模型参数反演的准确度。遥感反演时一般都采用代价函数进行参数拟合。经典的最小二乘（LS）拟合代价函数对正态分布误差具有一定的抗干扰性，但是当观测数据含有异常值时却会导致反演结果的不稳定。最小中值平方（LMS）方法具有鲁棒性特点，反演时若将其作为代价函数，则可以有效地检测出观测数据中含有的异常值，从而可以使模型反演准确度提高。本文以遥感BRDF物理模型——SAIL模型为例，使用模拟数据与真实地面观测数据，构建LMS与LS两种代价函数，分阶段地进行地表参数的反演方法研究。结果显示，针对具有一定误差或模型不能完全表示的观测数据，本文采用的分阶段方法可以对模型参数鲁棒地反演。     

卫星重力梯度向下延拓的频域最小二乘配置法

本文深入研究频域最小二乘配置法的基本原理及其在求解卫星重力梯度向下延拓问题中的应用。与一般空域最小二乘配置法相比，该方法具有高效稳定的特点，特别适合于大规模重力场数据处理，并为利用卫星重力梯度数据精化局部重力场提供了可供参考的方法。     

基于多面核函数配置型模型的参数估计

针对同时出现非随机参数、已测点和未测点信号的配置型模型精确建立协方差函数的困难，在研究和应用最小二乘配置和多面函数拟合法的基础上，将两法综合，提出了多面核函数配置法，导出了这种平差方法的解算公式。并与最小二乘配置、多面函数拟合作了比较分析。     

Spherical cap harmonic model for mapping and predicting regional TEC

An approach to modeling the regional ionospheric total electron content (TEC) based on spherical cap harmonic analysis is presented. This approach not only provides a better regional TEC mapping accuracy, but also the capability for ionospheric model prediction based on spectrum analysis and least squares collocation. Unlike conventional approaches, which predict the immediate TEC with models using current observations, the spherical cap harmonic approach utilizes models using past observations to predict a model which will provide future TEC values. A significant advantage in comparison with conventional approaches is that the spherical cap harmonic approach can be used to predict the long-term TEC with reasonable accuracy. This study processes a set of GPS data with an observation time span of 1 year from two GPS networks in China. The TEC mapping accuracy of the spherical cap harmonic model is compared with the polynomial model and the global ionosphere model from IGS. The results show that the spherical cap harmonic model has a better TEC mapping accuracy with smoother residual distributions in both temporal and spatial domains. The TEC prediction with the spherical cap harmonic model has been investigated for both short- and long-term intervals. For the short-term interval, the prediction accuracies for the latencies of 1-day, 2-days, and 3-days are 2.5 TECU, 3.5 TECU, and 4.5 TECU, respectively. For the long-term interval, the prediction accuracy is 4.5 TECU for a 2-month latency.