The large amount of calculations with a continuously high-speed growth required in current development of computational technology presents a great challenge to the research of spatial data processing in geography. To accelerate the processing speed of constantly updated real-time data that distributed unevenly, this study developed a dynamic grouping method based on the Number-K Dimension (N-KD) Tree technique. The method employs a data parallel algorithm to preserve spatial proximity and balance the data loads after grouping the real-time data that vary both in quantity and spatial domain. Based on the KD Tree creation, the algorithm measures the data sparsity according to the mathematic variance, addresses the point data near boundaries according to the nearest searching approach, and achieve an unequal data partition with respect to space while having balanced data loads. Experiments reveal that the method is capable of efficiently grouping the point data that have various spatial distributions and balancing the data amount among these groups. The algorithm also supports parallel computing and distributed collaborative working model, which highlights the practical values in its applications.
Integrity authentication is proved to be the prerequisite of remote sensing image applications. In other words, an image with poor integrity would have little value in further utilizations. The special nature of the authentication process is not taken into account in conventional authentication technologies, which makes it difficult for all users to be authorized to discriminate whether the images have been destabilized. Moreover, traditional authentication technologies are based on the binary data, which can not meet the requirements of content authentication for remote sensing image. According to the actual requirements of remote sensing image authentication, a multi-level authorization method for remote sensing image based on perceptual hashing and one-way function is proposed from the perspective of content authentication. Perceptual hash technology is different from the cryptographic hash function, that it can express the content of remote sensing image with little information, at the same time, it can keep the robustness for operations that do not change the content of image. Multi-level authorization management is realized by establishing a key vector corresponding to the permission vector. So the higher privileged users can calculate the keys of users with lower priorities, based on their own keys. First, the key vector was produced for multi-level authorization management based on one-way function. Second, the image is partitioned into regions with different sensitivity levels according to the contained objects. Then, the perceptual hash values of the regions are generated and used for authentication, and are encrypted with hierarchical keys. Users with different priorities can authenticate the corresponding contents with relevant sensitivity levels from the remote sensing image, after decrypting the perceptual hash values. The experiment results show that the proposed approach effectively achieved the multi-level authentication for remote sensing images with high computing efficiency and confidentiality.
Depressions cause errors or breaks when extracting rivers from DEM. Filling the depressions in the initial DEM is important. Low elevated surface areas in space are usually adjacent, and high elevated surface areas in space are always discrete. Therefore, when the grids of low elevated areas are filled, a large number of grids in low elevated areas will have to be recalculated, which makes the process of depression filling more complex. This paper investigates a new ranking algorithm for improving the efficiency of filling the depressions of DEM. First, considering the fact that the depression filling time in different pieces decreases in the form of power function, the algorithm creates a list of variable arrays. Then the data of DEM, which has been ranked using quicksort algorithm, is put into the arrays successively. Finally, grids are filled within the variable arrays, whose lengths are increased with respect to arithmetic sequence. As the lengths of the arrays are increased, the low elevated areas are divided finely. With the increase of elevation, the arrays will contain more numbers of grids, which will reduce the depth of seeking path in iterations for the unfilled grids. The new algorithm keeps the integrity and continuity of the drainage networks extracted by the pits-removed DEM and well reduces the depth of seeking path in iterations to accelerate pits-removing process. To illustrate this approach, the algorithm has been employed in the SRTM DEM at a 90-meter resolution of Chengdu City. The results show a considerable efficiency. The efficiency of the algorithm outperforms the traditional algorithm, as it will improve 50% efficiency compared with previous algorithm when the number of grids of DEM reaches 20 million. And the improvement will be more evident when the amount of DEM data increases.
Vehicle routing problem (VRP) has typical characteristics of space-time distribution, thus it is influenced by constrains of space-time conditions. It is necessary to take the space and time factors comprehensively into the consideration of vehicle routing planning. In this paper, we proposed a new perspective to investigate vehicle routing problem and introduced a complete analytic framework from the aspects of time geography. Followed by the introduction of the fundamental theory of time geography, concepts including space-time constraints, space-time path, space-time prism, space-time accessibility, and space-time distance are illustrated in the context of VRP. Meanwhile, the diagrammatic and metric methods of these concepts are also provided. In addition, this paper developed a method to calculate space-time distance. This method considers the characteristics of both locations and time windows for different customers, thus it can be used to easily evaluate the proximity between customers. Finally, we illustrate the value and advantage of time geography by presenting an algorithm to solve a large-scale VRP with soft time windows, embedding the concept of spatio-temporal distance. Further efforts may be devoted to adopt time geography to solve real-time VRP and mobile facility routing problem, etc. The main contribution of this paper is that we find the potential and possibility of time geography in solving VRP problems. Using a series of conceptions and methods of time geography, we can integrate both spatial and temporal features of VRP in a unified frame, which could make solving the VRP problems to be more efficient and effective.
Terrain rendering using gradient color is a common way to express spatial distribution and change of topography or other geological factors. Observation and analysis of geological information in the three-dimensional interactive analysis system require users to change the visual angle and distance of the sight according to the observation location and the extent of the analysis area. When the observation distance is very close, an obvious stratification phenomenon will appear when using a color ribbon of selected colors to render terrain, which is unable to express perfectly the smooth transition effect of terrain color rendering. For operation requirements and to overcome existed problems of terrain gradient color rendering in the three dimensional geological interactive system, this paper analyzes characteristics of two color models: RGB and HSL. Combined with the color smooth transition principle of OpenGL, double-color gradient rendering for terrain is realized through interpolating linearly every color component in the RGB color space respectively. And multi-color gradient rendering is achieved for terrain through interpolating hue linearly and fixing saturation and lightness in the color space HSL. In addition, lighting calculation is proposed to be added to enhance the three-dimensional effect of color gradient shaded terrain. At last, a piece of terrain is selected to test the above methods. The rendered effects of the terrain using the double-color gradient interpolation, the multi-color gradient interpolation and a color ramp with 30 colors are compared with each other. Experimental results show that these methods can provide better compatibility with smooth shading algorithms of the OpenGL library, keep smooth gradual transition between colors in a three-dimensional interactive environment when observing the terrain from any distance, and achieve smoother color gradient rendering for the terrain.
Terrain rendering using gradient color is a common way to express spatial distribution and change of topography or other geological factors. Observation and analysis of geological information in the three-dimensional interactive analysis system require users to change the visual angle and distance of the sight according to the observation location and the extent of the analysis area. When the observation distance is very close, an obvious stratification phenomenon will appear when using a color ribbon of selected colors to render terrain, which is unable to express perfectly the smooth transition effect of terrain color rendering. For operation requirements and to overcome existed problems of terrain gradient color rendering in the three dimensional geological interactive system, this paper analyzes characteristics of two color models: RGB and HSL. Combined with the color smooth transition principle of OpenGL, double-color gradient rendering for terrain is realized through interpolating linearly every color component in the RGB color space respectively. And multi-color gradient rendering is achieved for terrain through interpolating hue linearly and fixing saturation and lightness in the color space HSL. In addition, lighting calculation is proposed to be added to enhance the three-dimensional effect of color gradient shaded terrain. At last, a piece of terrain is selected to test the above methods. The rendered effects of the terrain using the double-color gradient interpolation, the multi-color gradient interpolation and a color ramp with 30 colors are compared with each other. Experimental results show that these methods can provide better compatibility with smooth shading algorithms of the OpenGL library, keep smooth gradual transition between colors in a threedimensional interactive environment when observing the terrain from any distance, and achieve smoother color gradient rendering for the terrain.
The computational fluid dynamics (CFD) method is one of the major ways of wind fields patial pattern simulation at present. This method is more often applied to wind field simulations and analyses on small scales because of the limitations of the hardware and software. Meanwhile, the precision and accuracy of this method depends on the precision of 3D building model sand the accuracy of iterated computation models, and there are significant differences existing between the simulation results and the real wind field situation. With the development of Internet of Things (IoT) technology, we can provide precise parameters to real-time simulation of regional wind fields pace distribution, by using a large number of real-time field sensor nodes. Spatial interpolation can be used to simulate the spatial distribution of all the regional environment factors. In order to determine the optimal method for real-time wind fields imulation, this research took the monthly average wind speed data of November, 2011, which are collected from 32 wind speed sensors in Institute of Urban Environment, CAS, as the example. Then, we make a comprehensive analysis of Inverse Distance Weight method, Global Polynomial method, Local Polynomial method, Radial Basis Function method, Nearest Neighbor method and Ordinary Kriging method, and compare the results of the six different methods by Cross-Validation. The result shows that the Inverse Distance Weight method is better than the other methods in the simulation error range, the simulation accuracy and the ability to reflect extreme value, which provides a reference for wind field simulation on small scales.
The mare and lunar highland are two major types of lunar morphology. The rapid and reliable identification of these two kinds of lunar morphology is an important basis in lunar research. Currently, major methods for identifying the mare and highland are based on the integrated evaluation index system, which is usually combined with the land surface parameters derived from DEM. Although the mare and highland can be identified by this method, it contains two problems yet. One is the lack of extensibility, because it is difficult for different regions of lunar to share one index system based on the same terrain factors. The other is the significant subjectivity in weight setting for each factor in the index system. To overcome the problems mentioned above, a new method considering the terrain texture features from lunar DEM is proposed by using the 500 m lunar DEM, which is produced from the global moon data obtained by Chinese satellite Chang’E-1(CE-1).Six typical mare sample areas and six typical highland sample areas were selected as the training zones. To construct the different terrain texture eigenvectors between the mare and highland, principal component analysis (PCA) was used to extract the main composition factors after the execution of quantitative analysis based on Gray level co-occurrence matrix (GLCM) model. Then the area located on 40° E~120° W, 0°~30° S was selected as the test area and the same approach in constructing terrain texture eigenvectors was used in this area. At last, supervised classification method was taken to identify those two types of lunar morphology. The recognition rate was about 85.7%. According to the comparative results between the new method and the traditional manual visual interpretation with Chang’E-1(CE-1) remote sensing image (in 120 m resolution),the proposed method is more effective and precise in identifying the mare and highland. Meanwhile, this method is driven by objective data, which spontaneously overcomes the subjectivity deficiency of current methods. Furthermore, this research provides a new thinking strategy of identifying and extracting different geomorphology based on the texture features from DEMs.
Accessibility, as an important index to represent the effectiveness of regional traffic network structure, takes a vital role in transport status research and transport network planning and designing. However, it is hard to find previous researches on the spatial-temporal characteristics of transport network accessibility and its evolution pattern, especially cases concerning the inland mountainous regions. In this paper, those characteristics and the evolution pattern of all nodes (districts) in Chongqing transport network in 1997, 2001, 2004, 2008 and 2012 were analyzed and evaluated by using the weighted average travel time as the indicator. The results show that the accessibility of Chongqing declined gradually from the city center (Yuzhong district) to the city edges. During 1997-2012, with the enhancement of transport network, the accessibility of all nodes was dramatically improved. Comparatively, the growth of accessibility during 1997-2004 is higher than that during 2004-2012. Because the variation of accessibility values has a close relationship with its initial value, the accessibility increase in the marginal areas is greater than that in the central areas. Among the cities, the differences in accessibility gradually narrow, and the accessibility coefficient gradually reaches an equilibrium distribution. Because of the imbalance of the spatial distribution of economy, there is a certain interactive coupling relationship between economic development and accessibility. And the regional effect of the transport network is more complicated. How to properly match the transport network to the urban system is the content of the regional transportation construction planning, and the government should pay more attention into it in the future.
Strengthening urbanization has now become the main task of China’s modernization. Urban spatial distribution patterns are characterized by complexity and diversity. Through these characteristics, it is easy to notice problems such as the uneven distributions of urban areas within provinces, the excessive implemented levels in administrative divisions, and huge gaps existed between cities and towns in terms of their urbanization levels. Therefore, it is important to find out how to display the spatial distribution of urban area rationally and explore useful information. As one of the spatial data mining tools, Voronoi diagram can be used to implement spatial subdivision and reveal the scope of spatial influence generated by urban area. In this article, a Voronoi diagram of Shanxi Province, which is weighted by the city centricity intensity, is generated using the Weighted Voronoi Diagram Extension for ArcGIS 9.x. It is also used to analyze the spatial influence of cities in Shanxi and to evaluate the rationalities and limitations in developing these cities. Undeveloped areas in these cities are obtained by combining Voronoi diagram with Delaunay triangulation and overlay analysis. Counties to be developed are obtained by implementing overlay analysis on county, street, and river data. Finally, the rationality and feasibility are evaluated by integrating general Voronoi diagram with Cv values. Through the analysis, conclusions could be made as follows: Taiyuan city has a larger spatial influence area than the other cities, and it has a negative impact on the development of other cities; western and eastern areas in Shanxi are developed unevenly; the cities of Fanzhi, Lingshi, and Xinjiang have the advantages and potentials to be developed in priority. Through the integration of general Voronoi Diagram and Cv values, the results correspond well to the actual situations.
With the implementation of new regional development and poverty alleviation strategy, contiguous destitute region shave turned into the main battle field to promote poverty alleviation and development. Selecting the contiguous destitute regions in Wuling Mountain as the study area, taking county as the study unit, this paper selects the main influence factors of poverty from common natural and social factors to build an evaluation index system. Using GIS and BP Neural Network, this paper simulates natural impoverishing index and socio-economic poverty alleviation index, analyses the reason of regional poverty from the perspective of nature and society, and explores the spatial distribution characteristics of poverty in order to provide decision support for establishing policies for poverty alleviation and development, and achieving regional harmonious development. The results show that the natural factors, such as terrain, slope, and disaster, are the main impoverishing index for the study area. The socio-economic factors, such as education, road, and medical care, could alleviate poverty to some extent. The natural poverty degree for most counties in the study area is above average, in which the Tongren and Xiangxi regions have relatively high level of poverty. Most destitute areas have low socio-economic level, and their ability to alleviate poverty is not strong. The degree of poverty in Qianjiang and Zhangjiajie regions is lower, while in Tongren and Xiangxi regions is higher. Large differences exist between these counties’ poverty situations. Guzhang, Longchuan, Wuichuan, Zhengan, Longhui, Xinhua, Daotong, and Chengbu together constitute the poverty distribution pattern of "large dispersion, small aggregation" in Wuling Mountain area. In the process of poverty alleviation and development, considerations should be given to the natural factors, and take advantage of local nature resources, especially the mining resources. According to the poverty type and self-development ability, different regions are compatible with different approaches. Meanwhile, the contiguous destitute regions in Wuling Mountain area should interactively strengthen their exchange and cooperation.
As African terrestrial ecosystem is highly sensitive to climate change, considerable researches during the past several decades have focused on monitoring vegetation activity and its responses to climate change, thus we can thoroughly understand the progress of climate change. It's significant to study the vegetation variation and its control factors in this area.We combined datasets of the satellite-derived Normalized Difference Vegetation Index (NDVI),whose spatial resolution is 8 km, and climate factors (temperature and precipitation) to detect vegetation activity change. Additionally, we analyzed the spatial correlations among NDVI, temperature, precipitation on the spatio-temporal patterns from 1982 to 2006 in Ghana. In order to conduct certain spatial correlation analysis, we interpolated the temperature and precipitation data from Climatic Research Unit (CRU3.0) datasets with respect to the spatial resolution of NDVI. At the regional scale, the results from this study showed the different increase trends occurred in the vegetation area of Ghana in the recently 25 years, which suggested that the vegetation activity has been strengthened. Compare the spatial distribution of NDVI in the early 1980 s (1982-1986) with the early 21st century (2002-2006), we found that the areas whose NDVI value is greater than 0.5 and whose value is between 0.4 and 0.5 all show a trend of increase; area percentage with NDVI value greater than 0.5 had increased from 26% to 38.2%, and the area percentage with NDVI value between 0.4 and 0.5 had increased from 47.5% to 51.9%.Spatial correlation analysis showed that the area with precipitation governed NDVI accounted for 57.2% of the total area, and the area with NDVI controlled by the temperature accounted for 42.8% of the total area. In summary, the vegetation of Ghana is more sensitive to the rainfall changes than the temperature changes.
In hyperspectral unmixing, PPI algorithm is a relatively mature algorithm, but each projection vector in PPI algorithm is generated randomly, and the endmembers extracted by PPI algorithm are not stable. That is, different endmembers can be obtained from the same image by repeatedly running PPI algorithm. This paper, based on the convex geometry description of linear spectral mixing model, utilized the feature that the endmembers are the endpoints of the single convex body enclosed in the hyperspectral image feature space, and proposed a novel pure pixel index algorithm for endmember extraction based on the maximum distance. The average of the spectral vectors of all the sample points is calculated and used as the center of a hypersphere. Next, we calculate the Euclidean distances of all the sample points to the center of the hypersphere, and design a radius of equal to or greater than the maximum distance for the hypersphere in the feature space to include all of the sample points. We evenly select the reference points on the surface of the hypersphere. The farthest sample point with respect to each reference point can be found by calculating the Euclidean distance. Subsequently, every sample point’s frequency of being the most distant to the reference points is recorded as an index to evaluate whether the sample point is an endmember or not. Finally, we use the AVIRIS data of Nevada Cuprite to testify this algorithm. The experimental results illustrate that the precision of the endmember extraction using the algorithm proposed in this paper is better than N-FINDR algorithm and VCA algorithm in general. Moreover, it has a good robustness and could overcome the instability of PPI algorithm caused by random projection.
Digital elevation model (DEM) is one of the most important data format for geography research in mountainous areas. At present, the widely used digital elevation models include SRTM and ASTER GDEM. However, their relevant errors are not well distributed under the influence of different terrain types. For the accuracy analysis of these two types of data, this paper considers the 1:50 000 DEM and rivers as references to carry out experiments over the eastern Tibetan Plateau. The horizontal accuracy and vertical accuracy are evaluated respectively by using "river-ravine" deviation method and root mean square error method. The conclusions are drawn from this study as follows: (1) the average river-ravine deviation of SRTM3 is 127.8 m, of which the offset direction is southwest; while that of ASTER GDEM is 104.1 m, of which the offset direction is west; (2) absolute error distribution of SRTM3 is relatively concentrated, where the elevation error is 35.3 m, which is less than ASTER GDEM elevation error of 50.2 m. The overall vertical accuracy of SRTER3 is better than ASTER GDEM; (3) at high altitudes over the average elevation of 4500 m, there is a positive correlation between the error and the elevation, in which the SRTM3 error shows slower growth and higher vertical accuracy compared with ASTER GDEM; (4) the vertical accuracy of the two types obviously depends on the slope, with which the error approximately indicates an exponential curve growth, and the error of SRTM3 in the flat area is less than that of ASTER GDEM. This study helps to understand the error distribution of the two data sets in mountain areas as well as to provide a foundation for further studies in data selection and error correction.
Qinghai-Tibet plateau is the world's highest and largest plateau which is surrounded by massive mountain ranges. There are plenty of lakes and glaciers covering this area, and the variation of lake area is closely related to the regional and global climate changes. This paper presents a method using MODIS observations to quantify lake area in each month in Qinghai-Tibet plateau for 14 years, and then analyzes the change of lake area at both monthly and annually scales. MODIS surface reflectance data (MOD09A1) acquired within the period of each month are retrieved and composited to produce the optimal and cloud-free SR data of the month, which are adopted as the input data. In order to achieve better accuracy on water detection, multiple water indices are combined, and then the DEM and summer season water maps are adopted to remove noisy pixels caused by cloud, shadow, snow and ice. With this approach, the monthly and annually lake area is derived for Qinghai-Tibet plateau from 2000 to 2013.Subsequently,the accuracy of these results are evaluated by correlation coefficient calculated with respect to the artificial interpreted Landsat products assigned as reference for Selin Co lake and Zhuonai lake, respectively. The comparisons show high consistency on both spatial and temporal aspects(R2 is 0.99 and 0.97). Finally, the variation tendency of lakes in Qinghai-Tibet plateau is analyzed, the results show: (1) a stable increase is obvious (R2 is 0.96) for the lake area during the past 14 years with an average annual increase rate of 490.98 km2a-1; (2)monthly increment indicates lake expansion is unlikely due to seasonality in Qinghai-Tibet plateau. Stable increases (R2> 0.78, increase rate > 400 km2a-1) are also found in every month of the year except February, March and April.
Precipitation data with high spatial resolution is deemed necessary for hydrology, meteorology, ecology and others. Currently there are mainly two sources of precipitation estimation: meteorological stations and remote sensing technology. However, a large number of studies demonstrated that the measurements acquired from conventional meteorological stations are single points of data, and they can not reflect the spatial variation of precipitation effectively, especially in studying the more complex areas. While the technology of remote sensing can not only improve the quality of the actual observations, but also be able to produce reasonably high resolution gridded precipitation fields. These products obtained by satellites have been widely used in previous studies. However, when applied to complex topography region, the spatial resolution of these products is too coarse and data accuracy is not high. Therefore, we present a statistical downscaling algorithm based on the relationships between precipitation and other environmental associated factors such as topography and vegetation in the Sichuan-Chongqing region, which was developed for downscaling the spatial precipitation fields with these remote sensing products. This algorithm is demonstrated with the Tropical Rainfall Measuring Mission (TRMM) 3B43 dataset, the Digital Elevation Model (DEM) from ASTER Global Digital Elevation Model (ASTER GDEM) and Moderate resolution Imaging Spectroradiometer (MODIS) 13A3 dataset. The statistical relationship among precipitation, geographical factors and Normalized Difference Vegetation Index (NDVI), which is a representation for vegetation, is variable at different scales; therefore, a multiple non-linear regression model was established under four different scales (0.25°, 0.50°, 0.75° and 1.00°, respectively). By applying a downscaling methodology, TRMM 3B43 0.25°×0.25° precipitation fields were downscaled to 1 km×1 km pixel resolution for each year from 2000 to 2011. By comparing these four regression models, we first select the regression model with the highest accuracy as the final downscaling algorithm, and then apply this downscaling algorithm (0.25°) to 1 km resolution for the estimation of high precision in this study. Second, the calibration of downscaling precipitation was conducted based on Geographical Difference Analysis (GDA) and Geographical Ratio Analysis (GRA). The final downscaling estimation results were validated by applying part of meteorological stations measured precipitation data for a period of 12 years in Sichuan-Chongqing region. As a whole, these results indicated downscaling algorithm is reliable, and can effectively improve spatial resolution of precipitation products. The resultant best value of the 1 km annual precipitation data are achieved through downscaling followed by GDA and GRA calibration for most cases. And the downscaling 1 km annual precipitation has not only been significantly improved in the spatial resolution, but also corresponded well with TRMM 3B43 precipitation data and meteorological stations data achieved for Sichuan-Chongqing region.
The characteristics of the visible and near-infrared reflectance from lunar surface are key information for the inversion and mapping of chemical elements in lunar exploration. Due to the dimply topography, hyperspectral dataobtained from IIM should be introduced with topographic correction before the application of practical inversion to get the actual reflectivity information. There ported longitudes and latitudes of the IIM data generally did not match with the conventional data from the LOLA DEM. Consequently, the accuracy of the topographic correction might be affected and decreased. In this paper, a case study was done related to the lunar crater. More specifically, a certain number of ground control points in a crater with same characteristic were selected for both images to match the IIM data and LOLA DEM data. The results from polynomial correction method were different from the correction results calculated directly from lunar longitude and latitude. There was approximately 3.5 pixels’ offset in the direction of longitude and about 1.95 pixels’ offset in the direction of latitude between IIM data and LOLA DEM data. The conventional topographical correction used on earth (i.e. C correction method) was introduced for a possible topographical correction on the moon, where there is almost no atmosphere scattering. The results showed that the performance of the matched topographical correction data was much better than those from unmatched data. After the application of registration and C correction, the slope of the linear correlation equation between the incident angle cosine value and the reflectance was reduced by 89.4 percent, the terrain effects of the shadow area and the highlight area on the lunar surface was eliminated. We concluded that the actual reflectivity information from lunar surface could be closely assessed after applying registration and topographic correction to IIM data, which could provide a sound basis for the inversion of chemical elements and minerals based on IIM data.
