2020 Vol. 45, No. 10
By the end of 2018, China has successfully launched 19 BDS-3 satellites into orbit. These BDS-3 satellites are equipped with high-precision inter-satellite link (ISL) payloads and have successfully realized inter-satellite two-way pseudorange measurements. We introduce the ISL observation model and carry out precise orbit determination (POD) for eight BDS-3 satellites using L-band satellite-ground and Ka-band ISL observations. The L-band satellite-ground data is collected from six international GNSS monitoring and assessment system (iGMAS) stations located in China's territory. The results show that the root mean square (RMS) value of POD residual for the ISL observations is better than 6 cm, and the estimations of ISL equipment time delays vary within ±0.15 ns for these BDS-3 satellites. When the ground L-band tracking station is confined to the territory, the addition of ISL observations can significantly improve the orbit determination accuracy. By the orbit overlap comparison, the errors of satellite orbits obtained by combined POD are about 13 cm in 3D and about 3 cm in radial direction, yielding an improvement of 85% compared to ground-only POD using 6 stations in China.
This paper discusses the calculaion approach of the coordinates of the BeiDou monitoring stations by proposing two schemes, one is precise point positioning (PPP) and the other is net-solution with MGEX. Only BDS observation data is used to verify two schemes. Especially, the improvement for the BeiDou satellites orbit determination accuracy with regional tracking network after coordinates refinement is analyzed. Initial results indicate that the PPP repeatability of station coordinates bias for days is better than 2.3 cm in the horizontal component and 3.4 cm in the vertical component, the net-solution repeatability for days is better than 0.8 cm in the horizontal component and 2.2 cm in the vertical component. BDS orbit determination accuracy with regional tracking network has a great enhancement after coordinates update using net-solution. The accuracy of IGSOs 3D overlap improves by 15.4%, while MEOs improves by 25.8%. The SLR validation reveals that laser residuals of BD-2 satellites reduce from 0.39 m to 0.24 m, the accuracy improves by 38%, the residuals of BD-3 satellites reduce from 0.25 m to 0.18 m, the accuracy improves by 28%.
Based on the comprehensive calculation of CORS network, this paper uses the load field removal-recovery technology to study the crustal vertical deformation and ground gravity spatio-temporal changes caused by environmental load in Wenzhou-Lishui region and compares it with GRACE gravity satellite results. Some conclusions are drown as follows: ① The effects of environmental load on the vertical deformation and gravity change of crust are up to the centimeter and ten micro-gauge levels, respectively, and the seasonal variation characteristics are significant. ② The vertical deformation and gravity change in winter show a decreasing and increasing trend from west to east, respectively. ③ Compared with GRACE results, in addition to local region differences, the overall trend of spatiotemporal changes has a higher consistency. ④The CORS network can monitor the spatiotemporal changes of crustal vertical deformation and gravity field caused by total surface environmental load. The research methods and results of this paper can provide an important reference for environmental dynamics research and geological disaster monitoring and early warning.
Cumulative sum (CUSUM) control chart is a more mature method for outlier identification, which can effectively identify and forewarn the deformation information in the global navigation satellite system (GNSS) coordinate series, but for long-term continuous monitoring, gross errors inevitably exist in the GNSS coordinate series, which has a certain impact on the accuracy of the CUSUM control chart. Therefore, a modified CUSUM control chart based on median was proposed to identify and forewarn the deformation information in the GNSS coordinate series, and its principle and algorithm flow were given. The results show that the accuracy of deformation information identification and forewarning using modified CUSUM control chart is better than traditional method, and the false alarm rate of the disaster warning are reduced, which shows the necessity and superiority of using the modified CUSUM control chart.
The initial orbit determination (IOD) of the very-short-arc (VSA) and the association between them are the key steps for expanding the catalogue with optical observations obtained from ground-based or space-based telescopes. For this reason, the distance search method for IOD and the geometrical method for their association were proposed. With optical observations from ground-based small telescopes array at Changchun Observatory, National Astronomical Observatories, China Academy of Sciences in Changchun and simulated observations from space-based telescope, the IOD and the association between the IODs are performed. The successful rates of getting an IOD solution for an optical arc with the distance search method is about 90%. For two kinds of observations. The IOD association results show that the true positive (assert that they are from same object for IODs from a same object) rate is over 80% in most cases. Ant the effects on the association accuracy of filtered different IODs are analyzed in depth. With actual optical observations obtained at Changchun, the results of IOD and their association show the validity of the IOD method and association method used in this paper and they can be used to expand the space object catalogue.
To tackle the problem that the targets in the image have few pixels and the targets are adjacent to each other and occluded each other, an aerial multi-target detection algorithm is proposed based on spatial-temporal information and trajectory association. Firstly, the pixel-based background modeling algorithm is used to obtain the target space information, and the neighboring frame difference algorithm is used to obtain the target time information. The time and space information are combined to get a spatial-temporal information map. Secondly, the Kalman predictor is used to predict the target position, and the Hungarian matching algorithm is used to correlate the target to obtain the target trajectory. Based on the target trajectory, the missed detection targets are supplemented to improve the target recall rate. Finally, based on the target trajectory features which are extracted in segments, the false alarm targets are filtered out to improve the target precision rate. Experimental videos with aerial multi-target are adopted for experimental verification, and the experimental results show that the proposed algorithm has good detection performance, with the recall rate higher than 96%, the precision rate higher than 98%, and the F-measure higher than 97%.
Spherical triangle grid is a versatile discrete global grid, which is an effective scheme for spatial information representation in the digital earth domain. The general methods for calculating the area of a spherical triangular grid are not accurate. So a fractal method for area calculation is proposed. Firstly, in the same level, areas of spherical triangular grid recoded to 0 or 1 are computed respectively. Then multi-fractal area calculation functions are fitted, which is used as an area calculation model. Finally, the accuracy of the model is calculated by relative root mean square error. Experimental result shows that, with the increase of the number of multifractals, the accuracy of area calculation of grid cells increases and is controllable.
The deficiencies of reverse distance weighting method and reverse latitude difference weighting method traditionally used in the calculation of diurnal variation of geomagnetic data are analyzed. According to the characteristics that the geomagnetic field has affinity with the changing of latitude, and considering the influence of latitude and longitude in the calculation of diurnal variation of geomagnetic data, the bifactor weight determination method is proposed. The validity of the proposed method is testified using geomagnetic observatory data supplied by Intermagnet website. Compared with the reverse distance weighting method and reverse latitude difference weighting method, the new bifactor weight determination method has more superiority.Experimental result shows that the proposed method is applicable in the calculation of diurnal variation of geomagnetic data, and it may have wide application prospect in diurnal variation correction of satellite, airborne and seaborne geomagnetic measurement data.
The areas of airborne gravity measurement are often irregular and the gravitational measurement data often have vacancy and high frequency noise interference. Conventional airborne gravity data processing generally performs the two steps of interpolation and denoising independently. In this paper, we consider these two problems in a unified way. Learning from seismic data processing methods, we propose an iterative method for interpolation and denoising of airborne gravity data based on the projection onto convex sets method. Using the simulative airborne gravity data based on the EGM2008, the results of numerical calculation examples have clearly demonstrated that our proposed iterative method for interpolation and denoising of gravity data is better than the four classical interpolation methods and some wavelet threshold denoising methods. And the effectiveness of the method is further verified by the measured aerial gravity data of a certain area.
In the process of downward continuation(DWC) of airborne gravity, the model error caused by discretization and systematic error in gravity data is expressed by non⁃parametric components. This paper proposes an inverse Poisson integral DWC method based on regularization method and semi⁃parametric kernel estimation, establishes gravity DWC model based on semi⁃parametric kernel estimation method without external data, reduces the ill⁃conditioned influence of the design matrix after Poisson integral discretization and introduces regularization method. It calculates the simulated airborne gravity anomaly based on the EGM2008 model and performs the simulation experiment using linear term and periodic term systematic error. The simulation experiment and the measured gravity anomaly data in the United States show the effectiveness of the proposed method in improving the ill⁃conditioned and separation system errors. The results show that the proposed method can effectively separate systematic errors and has high precision when there is no external data.
Gravity navigation algorithm is of great significance to improve matching efficiency and accuracy, and it is a hot issue in matching navigation technology. A new gravity matching navigation algorithm based on constraints correlation theory is put forward by analyzing the algorithm of INS/gravity matching integrated navigation. Based on the existing algorithms, the new algorithm increases valid information according inertial navigation systems by comparing the track direction and sailing distance, it can effectively eliminate a large number of interference matching tracks, improves matching efficiency and matching accuracy. The simulation results show that the matching accuracy of this algorithm is significantly improved compared with the probabilistic neural network method, and the accuracy is improved from kilometer to hectometre. The new algorithm shortens about 50% time for probabilistic neural network, greatly improves the efficiency of gravity matching, and better content the requirements of underwater navigation.
The post-earthquake deformation mechanism of the 2015 Nepal Mw7.8 earthquake is studied by using the relatively perfect theory of spherical displacement and GPS data of one year after Nepal earthquake. Two different models of post-earthquake deformation mechanism are explored: ① Single post-earthquake afterslip model (model 1); ② Combined model of post-earthquake afterslip and viscous relaxation (model 2). The results of model 1 show that the post-earthquake afterslip mainly occurs at the depth of 20-35 km and is located in the downdip area of co-seismic rupture; the main component of afterslip is thrust, accompanied by dextral strike-slip components, of which the largest component is 20 cm and 11 cm, respectively; the moment released by the post-earthquake afterslip is 1.23×1020 Nm, equivalent to Mw7.33 earthquake. The residual slip distribution obtained by model 2 is consistent with model 1, but the cumulative slip is slightly smaller, and the released seismic moment is 1.1×1020 Nm, equivalent to Mw7.32 earthquake. Furthermore, the results of model 2 show that the optimum values of lithospheric elastic layer thickness and mantle viscous coefficient in Nepal seismic source area are 40 km and 2×1019 Pa·s, respectively. To sum up, the post-earthquake afterslip effect plays a dominant role and the viscous relaxation effect plays a secondary role in Nepal within one year after the earthquake.
Focuses on the problem of ranging error diverges with the increase of the starting incidence angle and non-convergence of iteration process at the large incidence angle by using the traditional method to calculate the starting incidence angle, this paper proposes a new method of calculating the starting incidenceangle by Newton method. Firstly, the principle and steps of traditional method are introduced. Secondly, the iteration equation is established according to the relationship between the propagation time and the starting incidence angle. Finally, the two methods are compared by two simulation experiments in deepsea and shallow sea. The results show that the computational efficiency of the traditional method is higher than our proposed method when the starting incidence angle is less than 78°. In addition, it is found out that when the starting incidence angle is greater than 60°, the ranging error calculated by the traditional method begins to diverge. In the shallow water simulation experiment, the traditional method also has the problem of non-convergence in the iteration process when the starting incidence angle is greater than 85°. In contrast, our proposed method can converge to the real value at any starting incidence angle.
An interactive extraction method of high-resolution remote sensing image surface features based on full-connected conditional random field is presented. Through human interaction tag estimation foreground model, combined with color and texture features, and using the simple linear iterative clustering(SLIC) algorithm to over segment the input image, the maximal similarity based on region merging (MSRM) is used to expand the foreground region and establish the global information of the full-connected conditional random field to describe image. The model inference is realized by the high-dimensional Gauss filtering method which is based on the mean-field estimation. The contours of the area features are obtained. The method is proved to be effective by experimental extraction of surface features such as water, woodland and terraced fields in high-resolution remote sensing images.
One of the major problems of multi-sensor information fusion is that sensors frequently produce spurious observations, which have a great impact on the fusion accuracy and are difficult to be modeled and predicted. Bayesian information fusion technology based on entropy theory is an effective method to solve this problem. However, this method needs the integral operation in infinite interval, and the problem of numerical instability is prone to occur. To solve this problem, this paper proposes an improved multi-sensor data adaptive fusion method. In the framework of Bayesian theory, we use the difference between the measured values of sensors to adaptively establish the posterior probability distribution model of the sensor. Combined with the theory of mutual information, the pseudo-measured values can be identified and eliminated in real time, without integral calculation. The simulation and measured data test results show that the proposed method achieves the same results as the simple Bayesian fusion method without any spurious measurement, and the information fusion performance is obviously better than the simple Bayesian fusion method.
In view of the existing machine learning methods in the field of high-resolution remote sensing image building extraction and the like, which requires positive and negative training samples to participate at the same time, a one-class building change detection algorithm based on one-class samples without the need for negative samples is proposed. Firstly, it extracts the morphological building index features of the image, and fuse multi-features with the spectral features. Secondly, based on the one-class classification method proposed in this paper, from the object-based perspective, it gets the object-level building change detection results. Finally, it constructs a new shape feature which is refined to obtain the final building change detection result. Through experiments on multi-source high-resolution remote sensing images, it is verified that our proposed algorithm is robust and has better detection accuracy than existing building change detection algorithms.
To address the problem of low accuracy in the urban built-up area extraction method using nighttime light data due to light spillover characteristics, the build-up area extraction method based on neighborhood extremum is proposed. Firstly, the one-dimensional quadratic regression is used to perform relative radiation correction for nighttime light data. Then, the extremum images describing the spatial variation characteristics of gray values are obtained by extremum neighborhood filtering. Finally, the extremum search algorithm was used to obtain the boundary images of built-up areas, and the binary segmentation method is used to extract urban built-up areas. The experimental results show that the means of Kappa coefficients and threshold selection times of our proposed method are 0.85 and 37 s, which are 0.03, 1 503 s and 0.01, 443 s higher than that of the mutation detection method and the statistical analysis method. The spatial morphology of the built-up area extraction results is closer to the reference data, which has better extraction effect and stability.
Rotation, scaling, translation (RST) and boundary micro-deformation are the main factors that affect the matching results of vector graphics, and are also good criteria for evaluating shape description algorithms. In this paper, a vector graphics matching algorithm using multi-dimensional object segmentation ratio is proposed for the influencing factors of shape matching. This algorithm extracts shape characteristic by constructing feature lines for vector graphics. It can accurately describe the shape characteristic of the polygon features to measure the shape similarity between the graphics. The vector graphics of 2 000 real geographical entities are used as the standard library for matching experiment, and half of the graphics are randomly selected as matching graphics. The matching graphics are taken to RST transformation and boundary simplification operations with varying degrees. The transformed graphics are matched with the standard library. The matching results are compared with other vector graphics matching algorithms to test the shape retrieval effect of this algorithm. This experiment demonstrate that the proposed algorithm has higher matching accuracy, RST invariance and deformation robustness. Therefore, it can accurately identify the shape of vector graphics.
A new structure called hybrid index based on improved grid and STR (sort-tile-recursive) R-Tree is proposed to overcome the shortcomings of vector tiles in the retrieval performance of original vector data sources, to improve the efficiency of spatial queries against data sources.The hybrid index improves the spatial query method of the first-level index through vector tile pyramid context information to reduce the space comparison in the query stage. And at the same time, the index structure proposed can effectively decrease the impact of the unbalanced spatial distribution of vector data and optimize the query performance by using STR R-Tree as secondary index. Experimental results show that, the hybrid index proposed in this paper, compared with other spatial indexes of database, adapts well to different types of spatial data and has obviously better performance in data source query stage of vector tile generation process.
Performance of artificial neural network modeling and Kriging interpolation in PM2.5 concentration estimation varies with sample sizes and predictor variables change. This paper analyzes the performance of ordinary Kriging (OK), radical basis function (RBF) networks based on geographic coordinates, CoKriging and RBF with the key factor(s) (CK and CoRBF) selected by correlation analysis and RBF network, using different training sets with various sizes. The spatial distribution of PM2.5 concentration is then estimated by the best performed method. Results show that RBF, CoRBF, OK, and CK can all be used to estimate PM2.5 concentration efficiently, and their accuracies improved unstably as the number of training sites increase. CoRBF with the key factor of population illustrates the largest variation of PM2.5 concentration, while CK has the highest coefficient of determination (R2) and index of agreement (IOA) and the lowest mean square error (MSE), mean absolute error (MAE), and relative error (RE). Correspondingly, the spatial pattern of CK estimated PM2.5 concentration is smoother than CoRBF estimated PM2.5 concentration, while they both are very similar to site measurements and reveal detailed information.