An iteratively reweighted least-squares approach to adaptive robust adjustment of parameters in linear regression models with autoregressive and <Emphasis Type="Italic">t</Emphasis>-distributed deviations

In this paper, we investigate a linear regression time series model of possibly outlier-afflicted observations and autocorrelated random deviations. This colored noise is represented by a covariance-stationary autoregressive (AR) process, in which the independent error components follow a scaled (Student’s) t-distribution. This error model allows for the stochastic modeling of multiple outliers and for an adaptive robust maximum likelihood (ML) estimation of the unknown regression and AR coefficients, the scale parameter, and the degree of freedom of the t-distribution. This approach is meant to be an extension of known estimators, which tend to focus only on the regression model, or on the AR error model, or on normally distributed errors. For the purpose of ML estimation, we derive an expectation conditional maximization either algorithm, which leads to an easy-to-implement version of iteratively reweighted least squares. The estimation performance of the algorithm is evaluated via Monte Carlo simulations for a Fourier as well as a spline model in connection with AR colored noise models of different orders and with three different sampling distributions generating the white noise components. We apply the algorithm to a vibration dataset recorded by a high-accuracy, single-axis accelerometer, focusing on the evaluation of the estimated AR colored noise model.     

Evaluation of gravitational curvatures of a tesseroid in spherical integral kernels

Proper understanding of how the Earth’s mass distributions and redistributions influence the Earth’s gravity field-related functionals is crucial for numerous applications in geodesy, geophysics and related geosciences. Calculations of the gravitational curvatures (GC) have been proposed in geodesy in recent years. In view of future satellite missions, the sixth-order developments of the gradients are becoming requisite. In this paper, a set of 3D integral GC formulas of a tesseroid mass body have been provided by spherical integral kernels in the spatial domain. Based on the Taylor series expansion approach, the numerical expressions of the 3D GC formulas are provided up to sixth order. Moreover, numerical experiments demonstrate the correctness of the 3D Taylor series approach for the GC formulas with order as high as sixth order. Analogous to other gravitational effects (e.g., gravitational potential, gravity vector, gravity gradient tensor), numerically it is found that there exist the very-near-area problem and polar singularity problem in the GC east–east–radial, north–north–radial and radial–radial–radial components in spatial domain, and compared to the other gravitational effects, the relative approximation errors of the GC components are larger due to not only the influence of the geocentric distance but also the influence of the latitude. This study shows that the magnitude of each term for the nonzero GC functionals by a grid resolution 15\(^{{\prime } }\,\times \) 15\(^{{\prime }}\) at GOCE satellite height can reach of about 10\(^{-16}\) m\(^{-1}\) s\(^{2}\) for zero order, 10\(^{-24 }\) or 10\(^{-23}\) m\(^{-1}\) s\(^{2}\) for second order, 10\(^{-29}\) m\(^{-1}\) s\(^{2}\) for fourth order and 10\(^{-35}\) or 10\(^{-34}\) m\(^{-1}\) s\(^{2}\) for sixth order, respectively.     

Stochastic models in the DORIS position time series: estimates for IDS contribution to ITRF2014

This paper focuses on the investigation of the deterministic and stochastic parts of the Doppler Orbitography and Radiopositioning Integrated by Satellite (DORIS) weekly time series aligned to the newest release of ITRF2014. A set of 90 stations was divided into three groups depending on when the data were collected at an individual station. To reliably describe the DORIS time series, we employed a mathematical model that included the long-term nonlinear signal, linear trend, seasonal oscillations and a stochastic part, all being estimated with maximum likelihood estimation. We proved that the values of the parameters delivered for DORIS data are strictly correlated with the time span of the observations. The quality of the most recent data has significantly improved. Not only did the seasonal amplitudes decrease over the years, but also, and most importantly, the noise level and its type changed significantly. Among several tested models, the power-law process may be chosen as the preferred one for most of the DORIS data. Moreover, the preferred noise model has changed through the years from an autoregressive process to pure power-law noise with few stations characterised by a positive spectral index. For the latest observations, the medians of the velocity errors were equal to 0.3, 0.3 and 0.4 mm/year, respectively, for the North, East and Up components. In the best cases, a velocity uncertainty of DORIS sites of 0.1 mm/year is achievable when the appropriate coloured noise model is taken into consideration.     

A grid-based tropospheric product for China using a GNSS network

Tropospheric delay accounts for one source of error in global navigation satellite systems (GNSS). To better characterize the tropospheric delays in the temporal and spatial domain and facilitate the safety-critical use of GNSS across China, a method is proposed to generate a grid-based tropospheric product (GTP) using the GNSS network with an empirical tropospheric model, known as IGGtrop. The prototype system generates the GTPs in post-processing and real-time modes and is based on the undifferenced and uncombined precise point positioning (UU-PPP) technique. GTPs are constructed for a grid form (\(2.0{^{\circ }}\times 2.5{^{\circ }}\) latitude–longitude) over China with a time resolution of 5 min. The real-time GTP messages are encoded in a self-defined RTCM3 format and broadcast to users using NTRIP (networked transport of RTCM via internet protocol), which enables efficient and safe transmission to real-time users. Our approach for GTP generation consists of three sequential steps. In the first step, GNSS-derived zenith tropospheric delays (ZTDs) for a network of GNSS stations are estimated using UU-PPP. In the second step, vertical adjustments for the GNSS-derived ZTDs are applied to address the height differences between the GNSS stations and grid points. The ZTD height corrections are provided by the IGGtrop model. Finally, an inverse distance weighting method is used to interpolate the GNSS-derived ZTDs from the surrounding GNSS stations to the location of the grid point. A total of 210 global positioning system (GPS) stations from the crustal movement observation network of China are used to generate the GTPs in both post-processing and real-time modes. The accuracies of the GTPs are assessed against with ERA-Interim-derived ZTDs and the GPS-derived ZTDs at 12 test GPS stations, respectively. The results show that the post-processing and real-time GTPs can provide the ZTDs with accuracies of 1.4 and 1.8 cm, respectively. We also apply the GTPs in real-time kinematic GPS PPP, and the results show that the convergence time of the PPP solutions is shortened. These results confirm that the GTPs can act as an efficient information source to augment GNSS positioning over China.     

Determination of Phobos’ rotational parameters by an inertial frame bundle block adjustment

A functional model for a bundle block adjustment in the inertial reference frame was developed, implemented and tested. This approach enables the determination of rotation parameters of planetary bodies on the basis of photogrammetric observations. Tests with a self-consistent synthetic data set showed that the implementation converges reliably toward the expected values of the introduced unknown parameters of the adjustment, e.g., spin pole orientation, and that it can cope with typical observational errors in the data. We applied the model to a data set of Phobos using images from the Mars Express and the Viking mission. With Phobos being in a locked rotation, we computed a forced libration amplitude of \(1.14^\circ \pm 0.03^\circ \) together with a control point network of 685 points.     

Consistent realization of Celestial and Terrestrial Reference Frames

The Celestial Reference System (CRS) is currently realized only by Very Long Baseline Interferometry (VLBI) because it is the space geodetic technique that enables observations in that frame. In contrast, the Terrestrial Reference System (TRS) is realized by means of the combination of four space geodetic techniques: Global Navigation Satellite System (GNSS), VLBI, Satellite Laser Ranging (SLR), and Doppler Orbitography and Radiopositioning Integrated by Satellite. The Earth orientation parameters (EOP) are the link between the two types of systems, CRS and TRS. The EOP series of the International Earth Rotation and Reference Systems Service were combined of specifically selected series from various analysis centers. Other EOP series were generated by a simultaneous estimation together with the TRF while the CRF was fixed. Those computation approaches entail inherent inconsistencies between TRF, EOP, and CRF, also because the input data sets are different. A combined normal equation (NEQ) system, which consists of all the parameters, i.e., TRF, EOP, and CRF, would overcome such an inconsistency. In this paper, we simultaneously estimate TRF, EOP, and CRF from an inter-technique combined NEQ using the latest GNSS, VLBI, and SLR data (2005–2015). The results show that the selection of local ties is most critical to the TRF. The combination of pole coordinates is beneficial for the CRF, whereas the combination of \(\varDelta \hbox {UT1}\) results in clear rotations of the estimated CRF. However, the standard deviations of the EOP and the CRF improve by the inter-technique combination which indicates the benefits of a common estimation of all parameters. It became evident that the common determination of TRF, EOP, and CRF systematically influences future ICRF computations at the level of several \(\upmu \)as. Moreover, the CRF is influenced by up to \(50~\upmu \)as if the station coordinates and EOP are dominated by the satellite techniques.     

Testing block subdivision algorithms on block designs

Integrated land use–transportation models predict future transportation demand taking into account how households and firms arrange themselves partly as a function of the transportation system. Recent integrated models require parcels as inputs and produce household and employment predictions at the parcel scale. Block subdivision algorithms automatically generate parcel patterns within blocks. Evaluating block subdivision algorithms is done by way of generating parcels and comparing them to those in a parcel database. Three block subdivision algorithms are evaluated on how closely they reproduce parcels of different block types found in a parcel database from Montreal, Canada. While the authors who developed each of the algorithms have evaluated them, they have used their own metrics and block types to evaluate their own algorithms. This makes it difficult to compare their strengths and weaknesses. The contribution of this paper is in resolving this difficulty with the aim of finding a better algorithm suited to subdividing each block type. The proposed hypothesis is that given the different approaches that block subdivision algorithms take, it’s likely that different algorithms are better adapted to subdividing different block types. To test this, a standardized block type classification is used that consists of mutually exclusive and comprehensive categories. A statistical method is used for finding a better algorithm and the probability it will perform well for a given block type. Results suggest the oriented bounding box algorithm performs better for warped non-uniform sites, as well as gridiron and fragmented uniform sites. It also produces more similar parcel areas and widths. The Generalized Parcel Divider 1 algorithm performs better for gridiron non-uniform sites. The Straight Skeleton algorithm performs better for loop and lollipop networks as well as fragmented non-uniform and warped uniform sites. It also produces more similar parcel shapes and patterns.     

What governs the presence of residual vegetation in boreal wildfires?

Wildfires are frequent boreal forest disturbances in Canada, and emulating their patterns with forest harvesting has emerged as a common forest management goal. Wildfires contain many patches of residual vegetation of various size, shape, and composition; understanding their characteristics provides insights for improved emulation criteria. We studied the occurrence of residual vegetation within eleven boreal wildfire events in a natural setting; fires ignited by lightning, no suppression efforts, and no prior anthropogenic disturbances. Relative importance of the measurable geo-environmental factors and their marginal effects on residual presence are studied using Random Forests. These factors included distance from natural firebreaks (wetland, bedrock and non-vegetated areas, and water), land cover, and topographic variables (elevation, slope, and ruggedness index). We present results at spatial resolutions ranging from four to 64 m while emphasizing four and 32 m since they mimic IKONOS- and Landsat-type images. Natural firebreak features, especially the proximity to wetlands, are among the most important variables that explain the likelihood residual occurrence. The majority of residual vegetation areas are concentrated within 100 m of wetlands. Topographic variables, typically important in rugged terrain, are less important in explaining the presence of residuals within our study fires.