2018 Vol. 43, No. 7
The snow depth on the Arctic sea ice is not only a significant geophysical variable, but also an important parameter for the study of mass and energy balance, calculating sea ice thickness.To reduce the systematic error from different passive microwave sensor, we calibrated brightness temperature data obtained from the DMSP F17-SSMIS and F13-SSM/I during the overlap period.Forty-eight calibration models at the monthly scale were built up and compared with traditional calibration model at the yearly scale, which formed the basis for the snow depth retrieval and analysis on the Arctic first-year sea ice from 2003 to 2014.The results show that the correlation coefficients of the monthly fitting models of 19H, 19V, 22V, 37V channels are higher than the traditional model from January to May.Based on the calibrated satellite observation data, there is a general decline trend of snow depth on the Arctic first-year sea ice from 2003 to 2014, with the East Siberia Sea, the Laptev Sea and the Barents Sea decreasing obviously during the study period.
Aiming at the low positioning accuracy of image space location in urban environments, this paper proposes a new method of space location of image in urban environments based on C/S structure. First, a series of indoor images are taken in advance. We use the nearest neighbor distance ratio (NNDR) and normalized cross correlation (NCC) to get the SIFT coarse matches, then use RANSAC method to optimize it, and calculate the fundamental matrix and projection matrix. Building 3D point cloud model is established, and the object space feature library including image feature points, image point coordinates and object space point coordinates are got. Second, we regard the building images which are taken by user's mobile phone as the location images. Image feature points of location images are extracted and matched with the feature points of object space feature library, then the object space point coordinates are obtained. Finally, we use collinearity equation model to accurately calculate the exterior orientation elements, and display the user's space position on the phone, thus space positioning is realized. The experimental results show that the positioning method in this paper can reach the centimeter level positioning accuracy, and can meet the requirements of the user positioning accuracy.
Considering the problem that the traditional fuzzy clustering algorithm can not overcome the inherent speckle noise of SAR image, a fuzzy clustering segmentation algorithm using variable shape parameter Gamma distribution and neighborhood correlation is proposed.The variable shape parameter Gamma distribution is used to model the speckle noise of the multi-view SAR intensity image, and its negative logarithm is used as the similarity measure between the pixel and the intensity of the cluster in the feature field.Markov random field (MRF) is used to establish a generic correlation model of neighborhood pixels in label fields.In the framework of fuzzy clustering, the fuzzy objective function is constructed based on the above models, and the optimal result is obtained under the objective function minimization criterion.Experiments show that the variable shape parameter Gamma distribution can more accurately fit the histogram of pixel intensity in the homogeneous region.In order to effectively solve the shape parameter contained in the Gamma function, Newton iteration algorithm is used to estimate its numerical solution.Qualitative and quantitative analysis results show that the algorithm is effective in segmenting synthetic and real multi-view SAR images.
For mobile LiDAR point cloud data, a new hybrid index structure combining global KD-tree and local Octree is proposed to improve the efficiency of data organization and management, which is named as KD-OcTree index. Firstly, global KD-tree reconstructs the spatial neighborhood relations by defining the segmenting dimension and segmenting planes, for the purpose of ensuring the balance of the whole index. Then, local Octree is constructed in the leafs of KD-tree, which can avoid some shortcomings such as the unbalance of point cloud distribution, deeper Octree, large amount of non-point space, and so on. Lastly, we take three real scenes' point clouds as test data to process. The experimental results and comparative analysis show that the KD-OcTree index can not only improve the speed of constructing index and neighborhood searching, but also improve the effect of data-processing and influence the reliability of classification.
With the rapid increase of data size of remote sensing images, the traditional serial band re-gistration method cannot meet the demand for real-time processing of big-data multispectral images. Therefore, a CPU/GPU cooperative fast band registration method for multispectral imagery is proposed in this paper. Firstly, the computational amount and degree of parallelism are analyzed; point matching and differential rectification are ported to GPU to execute while the affine transformation parameter is still calculated on CPU. Secondly, kernel task assignment and basic settings are made to ensure the two above GPU steps executable. Moreover, three performance optimization methods, including memory access optimization, instruction optimization and transmission/computation overlap, are designed to further improve the efficiency of band registration. The experimental results based on NVIDIA Tesla M2050 GPU and Intel Xeon E5650 CPU show that the running time of YG-26 multispectral image band registration is only 3.25 s with our method, which got a speedup ratio of 32.32 compared with the traditional CPU serial method. The proposed method can provide quasi-real-time processing capability for multispectral imagery with big data size.
In interferogram filtering, traditional threshold filtering algorithm based on wavelet transform does not consider the statistical properties of SAR's interference phase, and the filtering effect obtained in the low coherence regions is not satisfactory.This paper presents a kind of phase noise filtering algorithm combining the shearlet transform and standard deviation of phase.The algorithm uses the phase standard deviation to correct the filter threshold and improve the filtering effect.In addition, in order to evaluate the filtering effect and to select the appropriate filtering method for the measured data, a local mean square error distribution of the simulated interferogram is proposed as the filtering quality evaluation index.Compared with Goldstein filtering, wavelet filtering, optimal direction fusion filtering and shearlet soft threshold filtering, the results show that the proposed method can not only weaken the noise of interferogram, but also keep the details and avoid the weak filtering in low coherence regions.
The running attitude of car-body relative to the track will change because of various vibrations during railway vehicle operation. In this paper, a photogrammetric method based on area array camera and line laser is proposed to measure running attitude of car-body. Firstly, the system is installed at any position of the car-body as long as being able to scan the track and all three positions are non-collinear; the middle point of line laser projection on rail surface is selected as the reference point; and the calibration method is proposed to build the space coordinate transformation of the reference point between image coordinate system and vehicle coordinate system. Secondly, the area array camera will shoot the projection of line laser on rail surface, and the coordinate of the reference point in image coordinate system will be converted into the coordinate in vehicle coordinate system during operation. Then, the mathematical model is established to obtain the running attitude of car-body through coordinates transformation of three reference points in the vehicle coordinate system. Finally, the detection system is built and experiments were carried out both on the car-body vibration simulation platform and the real vehicle. The comparative results verify the reliability of measurement system.
In dam health diagnosis, the information uncertainty and the information fusion of diagnosis indexes are common but difficult problems. An improved cloud fusion algorithm based on the cloud drop concept is proposed. In view of the specific features of the cloud drop in the cloud model, cloud drops from several "atom clouds" are produced with the forward cloud generator firstly. Then, the cloud integration is conducted by weighting operation. The "synthetic cloud" is finally obtained with the backward cloud generator. Thus, the transfer and fusion of the information uncertainty in the cloud model are realized. A practical dam health diagnosis example based on the improved cloud fusion algorithm is presented to verify its feasibility and reasonability. The result shows that the proposed algorithm is particularly suitable for the multi-index, multi-hierarchy and different-weight dam health diagnosis.
In order to meet the future development requirements of independent altimetry satellite, a two-satellites tandem mode is designed which can cover the whole ocean area with 1'×1' resolution after 2.3 years of flying. Firstly, two gravity field inversion methods for independent marine altimetry satellite are proposed, one of which takes the tandem orbital characteristics into account while the other does not. Secondly, the simulation calculation of random error propagation under normal distribution is carried out by using the inverse Vening-Meinesz method, and the corresponding indices of error under the two approaches are obtained.Based on the above error indexes, the corresponding precision indexes of the two kinds of gravity field inversion methods in the tandem flight mode are calculated respectively.Among, the second inversion method makes full use of the characteristics of the relative orbit determination by using the east-west observations in the tandem mode, and takes account of the systematic error characteristics of the propagation error and geophysical correction error under the close-range conditions. Therefore, the vertical deflection accuracy of the second inversion method is higher than that of the first inversion method, and its gravity field inversion also has certain advantages accordingly.The theoretical calculation results show that using the first inversion method, we can get a gravity anomaly accuracy of 6-10 mGal in 2.3 yearsand 4.2-7.1 mGal in 4.6 years; while using the second inversion method, a gravity anomaly accuracy of 3.9 mGal in 2.3 years and 2.8 mGal in 4.6 years can be obtained, respectively.
Due to the less prior constraint information, the ambiguity resolution and positioning performance are not very good in the GPS single-epoch kinematic positioning parameter estimation. Thus, a new method of GPS single-epoch kinematic positioning based on Doppler velocimetry is proposed in this paper. The coordinate initialization using the velocity information is firstly discussed, the current epoch's coordinate of the moving vehicle is predicted with the priori coordinate and velocity information. Compared with the conventional method of single point positioning (SPP) coordinate initialization, it is expected to obtain higher precision, but poorer robustness as well. Thus, the corresponding strategy for single-epoch ambiguity resolution is further introduced so as to improve the positioning performance of the method in this paper. Contrasting with the conventional algorithm of GPS single-epoch kinematic positioning, the experimental results show that, the method proposed can improve the precision of the float ambiguity, the final ambiguity fixed rate and the average positioning accuracy, especially in the case that, the number of GPS satellites is not sufficient or the geometric structure of the satellites is not very good.
GPS phase delay contains a lot of information that can be used to obtain atmospheric parameters with certain methods. Based on ground-based GPS phase delay, a one-dimensional variational (1DVAR) assimilation algorithm combined with an empirical model is proposed to obtain the atmospheric refractive index. The assimilation experiment is carried out by using simulated GPS phase delay data, and also verified with actual measured data. The influence of the background error settings on the result of assimilation is discussed. The experimental result shows that the 1DVAR assimilation algorithm can get high-precision atmospheric refractive index in 0-60 km height. The background error settings of the lower layers can affect the assimilation result in the whole height range. The atmospheric refractive index obtained from the assimilation experiment is used to correct the radio wave refraction, and the result is very good, which can reach 1 mm scale correction accuracy.
This paper studied the multipath error and noise of space-borne GPS code measurements collected from American BlackJack receiver onboard CHAMP, GRACE-A and Jason-2 satellites and the Chinese-built receiver onboard HY2A, ZY3 and TH1 satellites. The emphasis was paid on the variation characteristics of the multipath error and code noise and their impact on real-time onboard orbit determination. The result demonstrated that the C/A and P1 codes observation precision of the Chinese-built receiver was worse than that of American BlackJack receiver, and the P2 code precision of the Chinese-built receiver was superior than that of American BlackJack receiver. The code multipath errors of HY2A, ZY3 and TH1 satellites increased along with the decrease of elevation angles of GPS satellites, and the maximum errors are up to 3.6 m, 1.8 m and 0.7 m, respectively. The orbit determination experiments showed that the monotone increasing multipath error can lead to the systematic bias on the position of real-time onboard orbit results in the radial and tangential directions.
The spatial position uncertainty influences the availability of the vector spatial data, which is directly controlled by the two-dimensional probability distribution model of the point's position errors. This paper studies the spatial distribution model of GPS RTK positioning random error, which has repeatedly measured the selected points for 20 days. Then the one-and two-dimensional normal distribution tests and two-dimensional normal distribution fitting are employed to analyze the distribution model. The results show that the GPS RTK positioning random error fits directional two-dimensional normal distribution model and the significance is enhanced constantly along with the observation time shorten. The research results can build foundation for analyzing the influencing factors of the distribution model of positioning random error and establishing the mathematic forecasting model. Besides, contribute to promote the theoretical research of position uncertainty for spatial vector data.
Since accurate satellite physical models are difficult to conclude, the empirical solar radiation models are key to improve the orbital accuracy. With the measurements from the regional network of BDS ground control, three widely-used solar radiation pressure(SRP) empirical models (T20 model, ECOM5 model, ECOM9 model) are used to determinate the orbits of the three kinds of satellites and evaluate the effectiveness. Result shows that ECOM9 model has a better performance when the sun closes to the GEO satellite during the spring (or autumn) equinox and the standard deviation (STD) of the difference between satellite clock offset from orbit determination and two-way satellite-ground clock offset is better than 2 ns, which is evidently improved over that of T20 model and ECOM5 model. For IGSO/MEO satellite, T20 model has a better performance whether during the period of satellite attitude switch or orbit-normal mode. Different from the previous research, the conclusion of this paper shows that for the different types of BDS satellites and the same satellite at different times, we should use different SRP model to obtain a higher orbit determination and prediction accuracy.
Precise orbit determination is crucial in deep space exploration, and white noise in orbital tracking data can affect orbit determination performance. Based on the analysis of zero phase, we compared three kinds of filters, FRR, RRF and filtfilt in Matlab, and designed a zero-phase low pass filter using Kaiser window. The performance of the filter was verified by simulated and measured tracking data of MEX. After filtering white noise in the MEX measurement, the accuracy of MEX orbit determination could be significantly improved. For the two-way Doppler tracking data, the RMS of the velocity residuals was reduced to about one third of the original, that is, in the level of 0.031 mm/s; the difference of orbital position and velocity with the ESA reconstructed orbit was significantly reduced. The filtering process can be used as data preprocessing to improve orbit determination accuracy, and can also provide some reference for Chinese Mars exploration mission.
In the ground vehicle integrated navigation, GNSS observations are often interfered by complex ground environment and thus, its positioning result is more prone to contain outliers which can seriously affect GNSS/SINS integrated filter solution. This paper, from the perspective of IMU system error feature, studies an outlier detection method of GNSS/SINS integrated navigation based on accelerometer bias stability. According to the outlier of accelerometer bias result, the method detects gross errors in GNSS position, velocity, etc; and then applies the robust strategies of rejection and weight reduction to resist the influence of gross errors. The method is analyzed by a set of vehicle measured data. The results show that the observation outliers can greatly affect the accelerometer bias result and thus taking the accelerometer bias stability as a condition, the outliers can be exactly detected. In every direction of ENU, the RMS of position and the RMS of velocity are improved by 70.8% and 87.9% respectively; the RMS of attitude is improved by 77.7%. The method greatly improves the accuracy and robustness of integrated navigation results and provides a new strategy for robust processing of integrated navigation data.
The research area is located in Shazhenxi town and Xietan town of Three Gorges reservoir area in this paper. In order to obtain better results that discrete the continuous factors of landslide, entropy based on minimal description length principle(Ent-MDLP) method is used. To avoid the influence of correlation between factors, we calculate the Pearson correlation coefficient to remove high correlation factor. In order to obtain more accurate non-landslide sample points, the non-landslide sample points are randomly selected from the very low and low susceptible regions predicted by the entropy method. For the optimized random forests model, the optimal random features and its number are determined by iterative calculation of out-of-bag error estimation. Then the optimized random forest is evaluated for the landslide of the study area, and the landslide susceptibility level is divided. The model is compared with the methods of logistic regression, support vector machine and non-optimized random forest. The accuracy of each model is evaluated by plotting the receiver sensitivity curve of each algorithm. The optimized random forest's area is the highest, which the area under the curve is 91.8%. These show that the random forest model is optimized with more high-predictive power in landslide-prone assessment.
Spatial data partitioning plays an important role in the spatial index methods and the data storage strategy for spatial big data. In this paper, to make up the inherent shortcomings of spatial data partitioning and data storage in the Hadoop cloud computing platform, a parallel algorithm based on Hilbert space-filling curve is presented for partitioning the massive spatial vector data. In the spatial vector data partitioning phase, we take more influence factors, including the spatial location relationship between adjacent objects, the size of spatial vector object itself, the number of spatial objects in the same spatial coded block and others, into full consideration. Meanwhile, by following the partitioning principle of merging small coded blocks and sub-splitting large coded blocks, this paper implements the parallel algorithm for partitioning the massive spatial vector data in cloud environment. Experimental results show that the algorithm proposed in this paper can not only improve the efficiency of the spatial R-tree index for massive spatial vector data, but also give a good data balance in Hadoop distributed file system (HDFS).
This paper proposes a relaxation labelling matching approach for multi-scale residential datasets based on neighboring patterns. Firstly, we detect the candidate matching objects and neighboring patterns by buffering analysis and spatial neighboring relations. Secondly, the geometric similarities of candidate matching objects or neighboring patterns are calculated to initialize the matching matrix that contains 1:1, 1:M and M:N relations. After that, the contextual information of neighborhood objects or patterns are explored to heuristically update the matching matrix to achieve a global consistency. The matching pairs with maximum probabilities are finally selected after context consistency detection. The experimental results and contrast analysis show that our method obtains high correct matching rates, efficiently overcomes the problems of shape homogeneity and uneven deviation, and can correctly identify complex 1:M and M:N matching objects in multi-scale residential datasets.
Conventional geoprocessing workflow tools, such as ArcGIS ModelBuilder, focus on the integration of geoprocessing algorithms. The concept of the "Model Web" brings new challenges to these tools. On one hand, existing geoprocessing tools need to adapt to the Web environment to support the plug-in-and-play of distributed geoprocessing services. On the other hand, these services need to couple complex models to support time-step computation. This paper introduces a new method to publish model as service based on WebSocket protocol, and also introduces a workflow-based integrated modelling approach to couple models and services. It integrates OGC services, WebSocket services and OpenMI models, which brings some new features including logical consistency, physical separation, and controllable execution. In this way, traditional geoprocessing workflow tools are extended as tools for integrated modelling. A specific use case demonstrates the applicability of the approach.
With the rapid development of economy, urban internal space structure has been optimized significantly. It is of great significance to identify the spatial distribution and interaction rules of diffe-rent functional regions (DFR) for urban structure analysis and rational planning. We identify the spatial distribution of DFR by analyzing points of interest(POI) data based on kernel density estimation and head/tail breaks. On this basis, we analyze the spatio-temporal discipline of attraction and mutual relationship between typical DFR based on taxi trajectory data. Inside the 5th ring road of urban region of Beijing, the study reveals that:①Typical DFR Xidan, Guomao, Zhongguancun are business-oriented districts, Wangjing is a residential district with a significantly commuting characteristic. ②Guomao has the robust gravity (39.4%) on itself, which indicates that Guomao has more comprehensive urban functions. ③The attraction of DFR within the scope of resident trip distance decreases with the increase of distance, which conforms to the experience cognition and geographic spatial attenuation law. The results show that using kernel density estimation and head/tail breaks to analyze POI data and taxi trajectory data and identify the spatial distribution of DFR is reasonable and effective.
To analyze the interpolation errors spatial distribution characteristics of the typical gully of Yuanmou dry-hot valley, the measured elevation points were interpolated by inverse distance weighting (IDW), local polynomial interpolation (LPI), spline with tension (ST), disjunctive Kriging (DK) and triangulated irregular network (TIN)model to generate DEM. Cross validation, relative difference coefficient and the valley lines discrepancy were used to evaluate the interpolation accuracy. The error points with elevation error greater than 1 m were extracted, and their spatial distribution characteristics were analyzed by coefficient of variation(CV), global Moran's index and Getis-Ord Gi* index. The results show that DK and TIN model had higher interpolation accuracy, the height error points of five interpolation methods were overall aggregating distribution, and the degree was TIN > LPI > DK > ST > IDW. Height errors were significantly positive spatial autocorrelation and TIN model had the highest autocorrelation degree, the hot spots of errors were distributed in the area with large slope.