Nonlinear least-squares method via an isomorphic geometrical setup

The resolution of a nonlinear parametric adjustment model is addressed through an isomorphic geometrical setup with tensor structure and notation, represented by a u-dimensional “model surface” embedded in a flat n-dimensional “observational space”. Then observations correspond to the observational-space coordinates of the pointQ, theu initial parameters correspond to the model-surface coordinates of the “initial” pointP, and theu adjusted parameters correspond to the model-surface coordinates of the “least-squares” point . The least-squares criterion results in a minimum-distance property implying that the vector Q must be orthogonal to the model surface. The geometrical setup leads to the solution of modified normal equations, characterized by a positive-definite matrix. The latter contains second-order and, optionally, thirdorder partial derivatives of the observables with respect to the parameters. This approach significantly shortens the convergence process as compared to the standard (linearized) method.     

The spheroidal fixed–free two-boundary-value problem for geoid determination (the spheroidal Bruns' transform)

The target of the spheroidal Gauss–Listing geoid determination is presented as a solution of the spheroidal fixed–free two-boundary value problem based on a spheroidal Bruns' transformation (“spheroidal Bruns' formula”). The nonlinear spheroidal Bruns' transform (nonlinear spheroidal Bruns' formula), the spheroidal fixed part and the spheroidal free part of the two-boundary value problem are derived. Four different spheroidal gravity models are treated, in particular to determine whether they pass the test to fit to the postulate of a level ellipsoidal gravity field, namely of Somigliana–Pizzetti type. Received: 4 May 1999 / Accepted: 21 May 1999     

On symmetrical three-dimensional datum conversion

A 3-D similarity transformation is frequently used to convert GPS-WGS84-based coordinates to those in a local datum using a set of control points with coordinate values in both systems. In this application, the Gauss-Markov (GM) model is often employed to represent the problem, and a least-squares approach is used to compute the parameters within the mathematical model. However, the Gauss–Markov model considers the source coordinates in the data matrix (A) as fixed or error-free; this is an imprecise assumption since these coordinates are also measured quantities and include random errors. The errors-in-variables (EIV) model assumes that all the variables in the mathematical model are contaminated by random errors. This model may be solved using the relatively new total least-squares (TLS) estimation technique, introduced in 1980 by Golub and Van Loan. In this paper, the similarity transformation problem is analyzed with respect to the EIV model, and a novel algorithm is described to obtain the transformation parameters. It is proved that even with the EIV model, a closed form Procrustes approach can be employed to obtain the rotation matrix and translation parameters. The transformation scale may be calculated by solving the proper quadratic equation. A numerical example and a practical case study are presented to test this new algorithm and compare the EIV and the GM models.     

An iterative solution of weighted total least-squares adjustment

Total least-squares (TLS) adjustment is used to estimate the parameters in the errors-in-variables (EIV) model. However, its exact solution is rather complicated, and the accuracies of estimated parameters are too difficult to analytically compute. Since the EIV model is essentially a non-linear model, it can be solved according to the theory of non-linear least-squares adjustment. In this contribution, we will propose an iterative method of weighted TLS (WTLS) adjustment to solve EIV model based on Newton–Gauss approach of non-linear weighted least-squares (WLS) adjustment. Then the WLS solution to linearly approximated EIV model is derived and its discrepancy is investigated by comparing with WTLS solution. In addition, a numerical method is developed to compute the unbiased variance component estimate and the covariance matrix of the WTLS estimates. Finally, the real and simulation experiments are implemented to demonstrate the performance and efficiency of the presented iterative method and its linearly approximated version as well as the numerical method. The results show that the proposed iterative method can obtain such good solution as WTLS solution of Schaffrin and Wieser (J Geod 82:415–421, 2008) and the presented numerical method can be reasonably applied to evaluate the accuracy of WTLS solution.     

On the multivariate total least-squares approach to empirical coordinate transformations. Three algorithms

The multivariate total least-squares (MTLS) approach aims at estimating a matrix of parameters, Ξ, from a linear model (Y−E _Y = (X−E _X ) · Ξ) that includes an observation matrix, Y, another observation matrix, X, and matrices of randomly distributed errors, E _Y and E _X . Two special cases of the MTLS approach include the standard multivariate least-squares approach where only the observation matrix, Y, is perturbed by random errors and, on the other hand, the data least-squares approach where only the coefficient matrix X is affected by random errors. In a previous contribution, the authors derived an iterative algorithm to solve the MTLS problem by using the nonlinear Euler–Lagrange conditions. In this contribution, new lemmas are developed to analyze the iterative algorithm, modify it, and compare it with a new ‘closed form’ solution that is based on the singular-value decomposition. For an application, the total least-squares approach is used to estimate the affine transformation parameters that convert cadastral data from the old to the new Israeli datum. Technical aspects of this approach, such as scaling the data and fixing the columns in the coefficient matrix are investigated. This case study illuminates the issue of “symmetry” in the treatment of two sets of coordinates for identical point fields, a topic that had already been emphasized by Teunissen (1989, Festschrift to Torben Krarup, Geodetic Institute Bull no. 58, Copenhagen, Denmark, pp 335–342). The differences between the standard least-squares and the TLS approach are analyzed in terms of the estimated variance component and a first-order approximation of the dispersion matrix of the estimated parameters.     

On outlier-hiding effects in specific Gauss–Markov models: geodetic examples

 A specific subclass of Gauss–Markov models has been defined as containing the models for which the disturbance/response matrix, determined under the assumption of uncorrelated observations, consists of independent diagonal blocks. A proposed modification of reliability assessment procedure for such models is presented By the appropriate reduction of a given full covariance matrix for the observations, the proposal allows the assessment to be made in the resulting model which, in contrast to the initial model, is free from outlier-hiding effects of the type not occurring in ordinary models. The theoretical findings are demonstrated using simple numerical examples. All the proofs supporting the proposal are gathered in Appendixes. The proposal, which is not without its own weak points, is an attempt to associate the reliability assessment in specific Gauss–Markov models with effective outlier detection. Received: 23 June 1998 / Accepted: 5 July 2000     

Temporal correlation for network RTK positioning

Temporal correlation in network real-time kinematic (RTK) data exists due to unmodeled multipath and atmospheric errors, in combination with slowly changing satellite constellation. If this correlation is neglected, the estimated uncertainty of the coordinates might be too optimistic. In this study, we compute temporal correlation lengths for network RTK positioning, i.e., the appropriate time separation between the measurements. This leads to more realistic coordinate uncertainty estimates, and an appropriate surveying strategy to control the measurements can be designed. Two methods to estimate temporal correlation lengths are suggested. Several monitor stations that utilize correction data from two SWEPOS^TM Network RTK services, a standard service and a project-adapted service with the mean distance between the reference stations of approximately 70 and 10–20 km, are evaluated. The correlation lengths for the standard service are estimated as 17 min for the horizontal component and 36–37 min for the vertical component. The corresponding estimates for the project-adapted service are 13–17 and 13–16 min, respectively. According to the F test, the proposed composite first-order Gauss–Markov autocovariance function shows a significantly better least-squared fit to data compared to the commonly used one-component first-order Gauss–Markov model. A second suggested method is proposed that has the potential of providing robust correlation lengths without the need to fit a model to the computed autocovariance function.     

Weighted total least squares: necessary and sufficient conditions, fixed and random parameters

A standard errors-in-variables (EIV) model refers to a Gauss–Markov model with an uncertain model matrix from a geodetic perspective. Least squares within the EIV model is usually called the total least squares (TLS) technique because of its symmetrical adjustment. However, the solutions and computational advantages of the weighted TLS problem with a general weight matrix (WTLS) are mostly unknown. In this study, the WTLS problem was solved using three different approaches: iterative methods based on the normal equation, the iteratively linearized Gauss–Helmert model with algebraic Jacobian matrices, and numerical analysis. Furthermore, sufficient conditions for WTLS optimization were investigated systematically as proposed solutions yield only necessary conditions for optimality. A WTLS solution was considered to treat random parameters within the EIV model. Last, applications to test these novel algorithms are presented.     

A bound for the Euclidean norm of the difference between the best linear unbiased estimator and a linear unbiased estimator

 A bound is established for the Euclidean norm of the difference between the best linear unbiased estimator and any linear unbiased estimator in the general linear model. The bound involves the spectral norm of the difference between the dispersion matrices of the two estimators, and the residual sum of squares, all evaluated at the assumed model, but is independent of the provenance of the observation vector at hand. The bound, a straightforward consequence of first principles in Gauss–Markov theory, generalizes previous results on the difference between the best linear unbiased estimator and the ordinary least-squares estimator. In a numerical example from repeated precise levelling, the bound is used to analyse the sensitivity of estimates of vertical motion to the choice of estimator. Received: 9 September 1999 / Accepted: 15 March 2002     

Minimum mean squared error (MSE) adjustment and the optimal Tykhonov–Phillips regularization parameter via reproducing best invariant quadratic uniformly unbiased estimates (repro-BIQUUE)

In a linear Gauss–Markov model, the parameter estimates from BLUUE (Best Linear Uniformly Unbiased Estimate) are not robust against possible outliers in the observations. Moreover, by giving up the unbiasedness constraint, the mean squared error (MSE) risk may be further reduced, in particular when the problem is ill-posed. In this paper, the α-weighted S-homBLE (Best homogeneously Linear Estimate) is derived via formulas originally used for variance component estimation on the basis of the repro-BIQUUE (reproducing Best Invariant Quadratic Uniformly Unbiased Estimate) principle in a model with stochastic prior information. In the present model, however, such prior information is not included, which allows the comparison of the stochastic approach (α-weighted S-homBLE) with the well-established algebraic approach of Tykhonov–Phillips regularization, also known as R-HAPS (Hybrid APproximation Solution), whenever the inverse of the “substitute matrix” S exists and is chosen as the R matrix that defines the relative impact of the regularizing term on the final result. The delay in publishing this paper is due to a number of unfortunate complications. It was first submitted as a multi-author paper in two parts. Due to some miscommunication among the original authors, it was reassigned to one of the J Geod special issues, but later reassigned at this author’s request to a standard issue of J Geod. This compounded with a difficulty to find willing reviewers to slow the process. We apologize to the author.     

On some problems of the downward continuation of the 5′×5′ mean Helmert gravity disturbance

This research deals with some theoretical and numerical problems of the downward continuation of mean Helmert gravity disturbances. We prove that the downward continuation of the disturbing potential is much smoother, as well as two orders of magnitude smaller than that of the gravity anomaly, and we give the expression in spectral form for calculating the disturbing potential term. Numerical results show that for calculating truncation errors the first 180^∘ of a global potential model suffice. We also discuss the theoretical convergence problem of the iterative scheme. We prove that the 5^′×5^′ mean iterative scheme is convergent and the convergence speed depends on the topographic height; for Canada, to achieve an accuracy of 0.01 mGal, at most 80 iterations are needed. The comparison of the “mean” and “point” schemes shows that the mean scheme should give a more reasonable and reliable solution, while the point scheme brings a large error to the solution. Received: 19 August 1996 / Accepted: 4 February 1998     

The convergence problem of collocation solutions in the framework of the stochastic interpretation

The problem of the convergence of the collocation solution to the true gravity field was defined long ago (Tscherning in Boll Geod Sci Affini 39:221–252, 1978) and some results were derived, in particular by Krarup (Boll Geod Sci Affini 40:225–240, 1981). The problem is taken up again in the context of the stochastic interpretation of collocation theory and some new results are derived, showing that, when the potential T can be really continued down to a Bjerhammar sphere, we have a quite general convergence property in the noiseless case. When noise is present in data, still reasonable convergence results hold true.

“Democrito che ’l mondo a caso pone” “Democritus who made the world stochastic” Dante Alighieri, La Divina Commedia, Inferno, IV – 136     

A new estimator for VLBI baseline length repeatability

The goal of this paper is to introduce a more effective technique to approximate for the “repeatability–baseline length” relationship that is used to evaluate the quality of geodetic VLBI results. Traditionally, this relationship is approximated by a quadratic function of baseline length over all baselines. The new model incorporates the mean number of observed group delays of the reference radio sources (i.e. estimated as global parameters) used in the estimation of each baseline. It is shown that the new method provides a better approximation of the “repeatability–baseline length” relationship than the traditional model. Further development of the new approach comes down to modeling the repeatability as a function of two parameters: baseline length and baseline slewing rate. Within the framework of this new approach the station vertical and horizontal uncertainties can be treated as a function of baseline length. While the previous relationship indicated that the station vertical uncertainties are generally 4–5 times larger than the horizontal uncertainties, the vertical uncertainties as determined by the new method are only larger by a factor of 1.44 over all baseline lengths.     

三维坐标转换的非线性Gauss-Helmert模型及其解法

针对传统的三维坐标转换模型局限于求解小旋转角的三维坐标转换参数的缺点,以及没有同时顾及观测向量和系数矩阵的随机误差,该文提出了一种新的三维坐标转换参数求解模型。基于非线性Gauss-Helmert模型,建立了三维坐标转换的Bursa-Wolf模型,采用Newton-Gauss迭代算法,构建了加权整体最小二乘问题的拉格朗日函数,并给出了该算法的具体推导过程及其精度评定公式。实测数据和仿真数据结果表明:新算法在无须假设条件的前提下,可以获得任意旋转角下的坐标转换参数,且待估参数数目大大降低,易于程序实现。     

An alternative approach to robust collocation

The now classical collocation method in geodesy has been derived byH. Moritz (1970; 1973) within an appropriate Mixed Linear Model. According toB. Schaffrin (1985; 1986) even a generalized form of the collocation solution can be proved to represent a combined estimation/prediction procedure of typeBLUUE (Best Linear Uniformly Unbiased Estimation) for the fixed parameters, and of type inhomBLIP (Best inhomogeneously LInear Prediction) for the random effects with not necessarily zero expectation. Moreover, “robust collocation” has been introduced by means of homBLUP (Best homogeneously Linear weakly Unbiased Prediction) for the random effects together with a suitableLUUE for the fixed parameters. Here we present anequivalence theorem which states that the robust collocation solution in theoriginal Mixed Linear Model can identically be derived as traditionalLESS (LEast Squares Solution) in amodified Mixed Linear Model without using artifacts like “pseudo-observations”. This allows us a nice interpretation of “robust collocation” as an adjustment technique in the presence of “weak prior information”.     

Arcsecond-level pointing calibration for ICESat laser altimetry of ice sheets

 Arcsecond-level accuracy of NASA's ICESat (Ice, Cloud, and land Elevation Satellite) satellite laser altimeter beam pointing angle is required to satisfy the scientific goal of detecting centimeter-level elevation changes, over time, in the Greenland and Antarctic ice sheets. Two different approaches, termed “topographic inferred” and “direct detection”, were examined for calibrating the laser pointing angle (that is, detecting and removing pointing determination bias) at the 1.5-arcsec level, using information independent of the onboard pointing instrumentation. Both approaches entail estimating the beam pointing by differencing the three-dimensional position of the altimeter instrument and the laser-beam spot (or “footprint”) location on the ground. Analytical assessments of the two approaches are discussed, along with recommendations for the ICESat pointing determination calibration strategy. Received: 28 April 2000 / Accepted: 6 November 2000     

The adjustment of intersecting chains of triangulation using a medium sized computer

Summary A method of adjusting intersecting chains of triangulation by direct elimination, using variation of coordinates, is given which is suitable for a computer of modest storage capacity (about 4000–8000 words). The network is considered to be made up of “nodes” (where two or more chains intersect) and “links” (which connect two nodes). The normal equations for the points in each link are computed, the coefficients of the unknown coordinates of the associated nodes being treated as extra right hand sides. These equations are then solved to express the coordinates of any link station in terms of the associated nodes. The process is repeated for all links. The normal equations for the nodes are then set up and substitution made for all “link” terms. The method is particularly powerful for the adjustment of a few, long, interconnected chains since the reduced normal equation coefficients are then banded about the diagonal, the semi band width being a little greater than the largest number of unknowns in the nodes along a single tie chain. For a computer of 7000 words capacity it is possible to solve for a band width of over 100 unknowns i.e. more than 50 stations.     

Regional geoid determination in Antarctica utilizing airborne gravity and topography data

Antarctica is the only continent that suffers major gaps in terrestrial gravity data coverage. To overcome this problem and to close these gaps as well as to densify the global satellite gravity field solutions, the International Association of Geodesy (IAG) Commission Project 2.4 “Antarctic Geoid” was set into action. This paper reviews the current situation concerning the gravity field in Antarctica. It is shown that airborne geophysical surveys are the most promising tools to gain new gravity data in Antarctica. In this context, a number of projects to be carried out during the International Polar Year 2007/2008 will contribute to this goal. To demonstrate the feasibility of the regional geoid improvement in Antarctica, we present a case study using gravity and topography data of the southern Prince Charles Mountains, East Antarctica. During the processing, the remove–compute– restore (RCR) technique and least-squares collocation (LSC) were applied. Adding signal parts of up to 6 m to the global gravity field model that was used as a basis, the calculated regional quasigeoid reveals the dominant features of bedrock topography in that region, namely the graben structure of the Lambert glacier system. The accuracy of the improved regional quasigeoid is estimated to be at the level of 15 cm.     

运用证据权重法进行阳性钉螺-地理因子的空间关系研究(英文)

Schistosomiasis is a serious public health problem in the middle-lower Yangtze River Basin in China. Study of spatial variation of snail distribution that is related to microgeographic factors can help to choose pertinent measures for snail extinguishment and environment rebuilding. This paper studied the theoretical architecture of weights-of-evidence approach. The case study was made for spatial relation between the occurrence of infected snails and geographic factor combinations in Waijiazhou marshland of Poyang Lake region in China. The multievidence data came from the geographical factor combinations by crossing operation of vegetation coverage grade layer, cattle route distance grade layer, and special environment layer (181 combinations in total) in GIS. The calculation of weight contrast index shows that high vegetation coverage, cattle route distance of <45 meters, and special geographic factor “ground depression” had direct spatial relation with the occurrence of infected snails. The verification by crossing operation in GIS indicated 72.45% of the infected snails concentrated on the areas of positive weight contrast index (sequenced in an order of weight contrast index from high to low), demonstrating the high efficiency of the model established in finding infected snails according to the geographic factor combinations that can be explicitly discerned in the study area. Supported by a the National Natural Science Fundation of China (No. 30590370), the Research Project “Spatial Simulation of Schistosomiasis Susceptible Areas in the Poyang Lake Region” Sponsored by Science Research Plan 2007 of Jiangxi Normal University (Natural Science Category).