Descriptions of natural language contain abundant geographical spatial information including geographical objects, spatial relations, attribute information and so on. Among them, the understanding of spatial relations tend to play a decisive role in understanding the whole description statement. However, the uncertainty, ambiguity,flexibility and other features of descriptions of natural language make it difficult for computer to understand and process natural language. Taking the natural language as a breakthrough point, we explore description of spatial relationship in natural language and visualization under fuzzy semantics, and realizes the goal of "text-to-figure" transformation. We focus on establishing the mapping relationship between the qualitative description language and quantitative graphic language. According to different levels of "vagueness", descriptions can be divided into three categories, namely completely fuzzy description, interval fuzzy description and quantitative fuzzy description, so as to study the natural language description based on different categories and levels. Furthermore, we put forward two methods of visualization expression: random parameter method and buffer method, which realize the visualization of spatial relations descriptions of natural language. In the example of Xianlin campus of Nanjing Normal University, the system adopts a three-tiered architecture containing user layer, service layer and data layer. We also design and develop the analogous expression system of spatial relations of natural language, which realizes the "text-to-figure" conversion between fuzzy natural language description and point-line features, and verifies the validity of the method proposed in this paper. Most importantly, this system contributes a lot to recognition, reasoning, calculation, formal expression and visualization expression of geographical spatial information.
Geographical regionalization is a traditional and important content of geographical research. Geographers have carried out research on the dominant factors, characteristics of various geographical regions and the relationship of constraints and dependencies among these geographic elements over the years, accumulating a wealth of procedural knowledge as well as producing a large number of geographical regionalization plans. Some of them are multi-angle, multi-time and serialized. Map visualization is an effective tool and information carrier for expressing the zoning knowledge. However, the current research has always focused on the theory, method and technology of regional division and seldom pay attention to the combination of zoning knowledge and visualization method. In the process of compiling the national zoning atlas (the new century edition), we sorted out academic papers, monographs, reports and atlas related to zoning and found that most of the zoning knowledge existed in the form of text. Visualization is only used to represent the partitioned results. The combination of zoning knowledge and visualization theory and method is very weak, and the visualization expressive content is limited as well as the form of visualization, which is not enough to reflect the profound knowledge connotation and association of geographical zoning. It leads to the lack of zoning knowledge utilization. It hinders the knowledge acquisition in the zoning map. The cartography of new period has changed gradually from the presentation of the data to the expression of knowledge and guide the reader to find out rules which can also show the depth of scientific research to promote the re-use of knowledge. This article starts from making a cognition for prior knowledge of geographical zoning, and establishes the expressive content system of the knowledge of geographical zoning. Then, we study the visualization expressive method for the different knowledge of zoning maps, and discuss visual method of the comprehensive multi-dimensional zoning knowledge based on geo-informatic graphic. Finally, we use a visual case study, which shows that this paper provides an efficient strategy for knowledge representation and discovery, and it can be a further support for research decision-making of various regional factors.
People's cognition of the importance of roads is closely related to the facilities around the road. Aiming at the problem of lack of semantic features in the existing study on automatic selection in road generalization, the POI data is introduced into the semantic feature analysis of the road. A method of road automatic selection considering road spatial and semantic feature is proposed. Firstly, three new parameters of the road semantic feature measure are constructed by combining the POI position data. Then, three new parameters are combined with other metric parameters such as connection values and average linear density to calculate the road importance values. Finally, the road importance values, the composition of road stroke and the stroke connectivity are regarded as the constraint condition of automatic selection. The road is selected step by step. The experimental results show that the proposed method takes into account the semantic features of the road while retaining the main road, maintaining the density of the road distribution and the connectivity of the road.
Urban land classification is the foundation of urban planning, whose result is of great significance to the allocation of urban resources and the development of urban construction. Previous researches of urban land classification are mainly focused on the study of macro-scale areas, which is characterized by “sparse road network and large block system”. However, with the development of cities, the planning model featured by macro-scale area has caused problems such as the low efficiency of urban traffic and land development. To solve these problems, the construction of urban blocks with small scales was put forward. To make full use of the potential value of the current big data of traffic in the block planning with small scales, this paper presents a land classification method for blocks with small scales through combining the topic modeling and support vector machine (SVM). The regions near People's Square of Huangpu District in Shanghai was taken as the study area. We firstly divided the study area according to fine road network, and then formed a regional mobility pattern through processing the data on the GPS of taxis in one week. By using the data on points of interest (POI), the model of Latent Dirichlet Allocation (LDA) and SVM model, the land use classes are identified. Accuracy assessment of the proposed method has been made based on classification map visually interpreted, and the obtained result has been approved by the geographic data of Baidu Map. The results indicated that this method enables the possibility of the land classification of small-scaled blocks, and could achieve high classification accuracy by utilizing the big data of traffic.
Mine production includes geological survey and excavation, and is the fundamental activity for mine enterprise. Data related to mine production are typically geospatial, showing high space, temporal complexity and multi-scale. Geo-ontology can be applied to clearly describe and formalize geo-concepts, integrate heterogeneous geo-information and share geo-knowledge. However, several problems currently exist in the constructing process of using this new technique: efficiency resulting from extensive human involvement in the building process of geo-ontology, and the low accuracy from less attention for poor data quality and semantic contexts during data integration and knowledge reasoning. Moreover, the multiple relationships of the concepts are unclear as the multidimensional attributes and association patterns of geo-concepts are not sufficiently analyzed and expressed. The geological model and semantic descriptions are problematic without dynamic trajectories for moving objects and geological events. In this study, a new semi-automatic construction method of geo-ontology was proposed, regarding to geo-contexts based on a multi-dimensional and many-valued concept lattice (MDVCL). A formal concept hierarchy was automatically generated for higher efficiency using the multidimensional and many-valued concept lattice after data pre-processing and formal context creation of geodatabase. A semantic model of the geo-context-mediated ontologies (GMO) considering both global and local conditions was then constructed by adding four indexes of geo-contexts and dynamic events with OWL (Web Ontology Language) for more accurate formalizing description. The mapping rules were discussed between the concept lattices and the ontologies, and building mappings, in order to achieve straight forward semantic information from the concept lattices. In the end, a construction process was established through three steps, including knowledge pre-processing, multidimensional and many-valued concept lattices and semantic models. A metal mine containing geographic and geological data was selected to build a vein mining ontology for model verification. The results proved that this method could focus on multiple dimensions and complex backgrounds of the data, in order to reduce the risks of semantic errors, increase accuracy and efficiency of mine production, and provide important reference for other geo-ontology domains.
The Extraterrestrial Solar Radiation (ESR) is the basis of surface solar irradiance, and it is also an important astronomical parameter for calculating solar radiation, assessing solar energy resource and for estimation of agricultural potential productivity. Based on Digital Elevation Model (DEM) data of Fujian Province with a resolution of 30 meters, the spatial distribution of ESR over rugged terrains of Fujian Province was calculated by using parallel computing framework of MATLAB software. Then, the influences of slopes and aspects on the distribution of ESR were quantitatively analyzed. Finally, the spatial scale effects of DEM on ESR were discussed. The results clearly depict that the latitude and complex landforms profoundly disrupt the zonal distribution of ESR in Fujian Province. The terrain factors have more significant effects on the magnitude of ESR compared with the latitude. Annual ESR of Fujian Province was mainly at 10 000~13 000 MJ/m2 and presented a decrease trend from Southeast coastal areas to the Midwestern areas. As affected by the slopes and aspects, the ESR over seasons showed characteristics of differences and dissymmetry. The screening impact of topography was higher on spatial distribution of ESR in winter and considerably low in summer. The total amount of ESR in spring was larger than that in autumn. The influences of different slopes and aspects on the distribution of ESR were consistent with the terrain characteristics of Fujian Province, which is high in the northwest and low in the southeast. The ESR was depleted obviously along with the increase of the slope. The ESR was mainly concentrated in the east, southeast and south. The impact of topographic feature was significant, the ESR in the chine was bigger than the one in the valley and the one in the sunny slope is more than that in the shady slope. Meanwhile, the effect of spatial scale of the DEM was obvious. The ESR of Hills of Midwestern Fujian Province was more sensitive to changes in resolution.
Yellow River is the second longest river in China. The soil erosion and sandstorm harms are the severe problems for the management of Yellow River. Rainfall erosivity is a reflection of potential soil erosion caused by rainfall and an important factor of soil erosion models. The research about rainfall erosivity in Yellow River is critical for the preservation and management of Yellow River. This research used daily rainfall data from 166 meteorological stations in Yellow River from 1980 to 2015. The rainfall erosivity was calculated by the daily rainfall erosivity model. It is necessary to use area partition because differences of the climate factors, i.e. rainfall, are significant in a river basin. The annual variation of rainfall erosivity was discussed based on PCA, isodata cluster and maximum likelihood classification method which were used to area partition after the ordinary kriging method, and the gravity center model. The gravity center can be used to reveal migration direction of the rainfall erosivity. The results showed that rainfall erosivity was quite different in different zones of Yellow River. The minimal value of the erosivity was 200 MJ·mm·hm-2·h-1, and the maximum value was 3000 MJ·mm·hm-2·h-1. The rainfall erosivity of Yellow River had an increase from the northwest to the southeast. The zones with large rainfall erosivity also had an intense variation. The gravity center of rainfall erosivity in some zones which included Xining and Huhehaote, had a migration towards the northeast. While some zones which included Taiyuan and Xi'an, had a migration toward the southwest. The migration area of gravity center in each zone dwindled from the northwest to the southeast in temporal and spatial variations. The variety of the same zone's rainfall distribution in different years had a decrease from the northwest to the southeast.
Based on population distribution data of Beijing City at the spatial scale of 1 km×1 km grid in 2005 and 2010 and CA-Markov model, we simulated the spatial distribution of the population in 2015, 2020, 2025 and 2030. Then, we used the population data at the spatial scale of street to verify the simulation accuracy. On the basis of good reliability of the model and from the perspective that population redistribution is driven largely through industries transfer, we combined the data on employment quantity of different industries at the spatial scale of streets of Beijing city with the population redistribution goal and transfer direction of industries. Then, we calculated the decentralization weight of each street, and analyzed the spatial distribution after redistributing 15% of the population in the six central urban districts in 2020. The results indicated that, firstly, during 2005-2010, the region of the low population density at level 1 is accounting for 90%, focusing on Miyun District, Huairou District, Yanqing District and Fangshan District. Population density above level 10 focused on Xicheng District and Dongcheng District. Secondly, from 2015 to 2030, the low population density area shows a downward trend and the middle to high population density area shows an upward trend under natural conditions. Thirdly, under the impact of non-capital function extraction, population distribution in the six central urban districts shows a trend from focusing on middle-high level to low-middle level, and the high population density area shows a downward trend in 2020. Except Dongcheng District, the rest of five districts focus on population density area at level 5-8. In conclusion, this study can be useful for population management, resource allocation, and policy making.
Evapotranspiration (ET) plays an important role in the hydrological processes as it can substantially influence the amount and spatial distribution of water resources at regional scale. Quantitative estimation of spatiotemporal distribution and variation of surface ET is essential for understanding the hydrological cycle and water resources management. In this paper, the spatiotemporal characteristics and variation trend of ET and PET(Potential Evapotranspiration) are studied using MOD16 data during 2000-2014 in the"Wet Island of Tianshan Mountain"—Ili River valley, Xinjiang, China. The results showed that: (1) the accuracy of the MOD16-ET in Ili River valley can meet the requirements, and can be used to study the spatiotemporal distribution of surface ET; (2) the mean annual ET and PET were 392.35 and 1331.19 mm respectively, and ET/PET ratio fluctuates between 0.26 and 0.33. The low ratio of ET/PET indicates that the study area was affected by water deficit. The seasonal variation of ET and PET were a unimodal pattern in the growing season. ET/PET ratio was 0.29 in autumn, and the study area suffered drought in this period. Temperature was the main factor influencing the spatiotemporal distribution of ET and PET; (3) the spatial distribution of ET and PET are opposite. The upstream area of Kashgar river and Künes river in the south had sufficient water supply, while the Yining city, the area of Khorgos river, Qapqal County and the downstream area of Tekkas river suffered from drought and water shortage;(4) During 2000-2014, ET decreased, and PET increased, which showed that the drought in Ili River valley was aggravated in the past 15 years.
A refinement program of high computational complexity is needed to dehaze an image by using dark channel prior. It will avoid haloes at boundaries which is related to the transmission rate. In analyzing halo phenomenon at boundaries, it is founded that highly computational complexity of refinement procedures usually dehaze excessively, and the traditional methods based on dark channel prior for a single image dehazing may cause the color distortion in bright regions. Therefore, a simple and fast neighborhood segmentation method based on the hue is proposed during estimation of original transmittance. Firstly, the source RGB images are converted to HIS color space, In H (Hue) channel, differences in neighborhood of point and center point of the tone of absolute value determine whether those pixels in the neighborhood belong to the same region. Only those pixels belonging to the same areas are used to calculate Dark Channel. Then, transmission value corrects the color of bright region. When the image is restored, hue component remains unchanged in HIS color space. Only the intensity component is defogged using the modified dark channel values. Then, the nonlinear enhancement is performed. Finally, the saturation component is compensated by the color. Experiments show that the proposed algorithm can significantly improve the visual clarity of scenes and get better color fidelity without subsequent image repairing.
In terms of change detection with high resolution remote sensing images, there are still some unresolved problems such as scattered plots with ragged boundaries in output, being prone to occurrence of “salt-and-pepper” noise, expensive cost of manual annotation in the process of supervised training, redundancy of training samples, underutilization of information in unlabeled samples and so on. In order to address these problems, this paper proposes a new high resolution remote sensing image change detection method by combining the superpixel segmentation technology and Active Learning (AL) approaches. The proposed method consists of the following steps. Firstly the difference image is derived from two temporal remote sensing images. Subsequently the lattice-like homogenous superpixel are obtained by applying the Simple Linear Iterative Clustering (SLIC) algorithm. Simultaneously, we compare the SLIC algorithm with entropy-rate-based and modified-watershed-based superpixel generating algorithms respectively by means of homogeneity of superpixel and their coherence with image object boundaries. Then we compute the means and standard deviations of three bands of superpixel objects as spectral features and extract the entropy, energy and angular second moment by employing Gray-Level Co-occurrence Matrix (GLCM) as texture features. After that, initial training samples are randomly selected and labeled by introducing and following the Margin Sampling (MS) active learning sample selection strategy which is a kind of SVM based AL algorithm taking advantage of SVM geometrical properties and suitable for bipartition problems. A cosine distance based sample similarity measurement called Angle Based Diversity (ABD) is introduced to relief redundancy and ensure diversity of the selected samples. Lastly change detection is carried out according to the extracted information from trained samples. The proposed algorithms (SLIC-MS, SLIC-MS+ABD) are utilized to process WorldViewⅡmultispectral remote sensing data of urban and suburb scenes and the detection result from proposed sampling is compared with that from random sampling to explain detection accuracy of our methods. To illustrate the efficiency of methods proposed in this article, we investigate the iterative times of three techniques for reaching the same detection accuracy. Experimental results confirm that both SLIC-MS and SLIC-MS+ABD can reduce manual labeling cost and achieve better change detection quality than random sampling methods. They also indicate that the two proposed methods can find out samples with high uncertainty, which can be labeled by user themselves, from the unlabeled sample pool by making full use of and mining unlabeled sample information. Compared with the other two methods, SLIC-MS+ABD is more accurate with respect to identical data sets (the same two mentioned remote sensing images) and the same labeled sample number because the diversity of new selected samples has been considered in the learning process. In addition, SLIC-MS+ABD can obviously reduce iterative times to converge for achieving the same detection accuracy than other two approaches. On the basis of the experiment, it can be concluded that our proposed methods greatly relief the amount of user marking and acquire good change detection performance on high resolution remote sensing data sets as well. Experimental results also indicate that the methods implemented in this article saliently exhibit their advantages of manual cost reduction in sample labeling, avoidance of training sample redundancy to reach the same change detection quality for the same data set.
Classification based on remote sensing data has been widely applied in land cover mapping and the dynamic change monitoring research, of which the consequence is always strongly affected by spatial resolution of the used images. However, the response of multi-resolution images to remote sensing classification is still highly uncertain. Satellite observation could supply more and more multi-resolution images covering the same area at the same time and it would provide abundant data and technical support for study of remote sensing classification. In this study, the Hetian basin of Changting County in Fujian Province, was selected as a case to examine the performance of three typical classifiers (Maximum Likelihood Classification, MLC; Support Vector Machine, SVM; Artificial Neural Network, ANN). They were applied to satellite observations of temporal quasi-synchronous and multi-spatial resolution from medium to high spatial resolution (1~50 m) and we investigated the links between spatial resolution and remote sensing classification. Then, we also analyzed the spatial scale difference of spectrum reflectance, recognition accuracy and area extraction of five major land types (including arable land, forest land, water area, bare land and construction land) of the data with seven spatial resolution levels of 1, 2, 4, 8, 16, 30, and 50 m. They were supported with GF-1 PMS (pan and multi-spectra sensor), GF-2 PMS, GF-1 WFV (wide field view), Landsat-8 OLI (operational land imager) and GF-4 PMS data. 1845 recorded points observed in field survey were taken as training samples and validation samples. The results showed that along with the change of image spatial resolution from 1 to 50 m, (1) the mean spectra of bare land and construction land remained stable and no obvious changes occurred to water body, while the mean spectra of arable land and forest land decreased significantly when image resolution coarser than 4 m. The standard deviations of water body, bare land and construction land all increased constantly, while the standard deviations of arable land and forest land almost maintained stable. (2) The overall accuracy gradually decreased from 94.97±2.5% to 79.03±2.25% across the three classifiers, showing a gradually downward trend. Meanwhile, Kappa coefficient also gradually decreased from 0.93±0.03 to 0.72±0.03, which indicated that the accuracy of land cover classification was closely and sensitively related to the resolution of remote sensing images (P<0.05). (3) The calculation errors of the land types area would become larger as the image tend to be coarser, of which the area of arable land, bare land and construction land decreased significantly, the area of forest land increased, and the change of water body was not evident. The results above confirmed that when using multi-resolution images to generate land cover area or making area comparison refer to time serial data results, the errors from spatial database of various multi-scale could not be neglected, which would be more suitable to make the multi-scale transform for spatial effect correction. Our framework demonstrated the regular pattern of multiscale remote sensing classification and provided the prerequisite for scale conversion of classification products with different resolution in the future.
Comparison of surface radiation data of ECMWF (European Centre for Medium-Range Weather Forecasts) reanalysis data and data from station observation (China Meteorological Administration) is conducted at different time scales to check whether reanalysis data can reflect the characteristics of surface solar radiation over China. Based on the cluster analysis method, China is divided into 8 regions in order to study the regional differences of the surface radiation products of the ECMWF reanalysis data in China. Taking into account the influence of atmospheric factors on the earth's surface radiation and the spatial stratified heterogeneity of the atmospheric distribution, the geographical detector is used to find the causes of errors in different sites of reanalysis data. Overall, ECMWF is higher than the ground observation station data and the monthly deviation is 18.2835W/m2. ECMWF shows seasonal difference, greater deviation in spring and winter, less deviation in summer and autumn. Large relative deviation of the data mainly distributed in December, January, February and March while minor relative deviation of the data mainly concentrated in July, June, August and September. The dominant atmospheric factors in different regions are different in winter and summer. In summer, from zone 1 to 5 the dominant factors are aerosols and the power of determinant is larger. The dominant factors of the zone 6 are albedo and aerosol. The dominant factors of the zone 7 are cloud cover and aerosol but the power of determinant is small, merely 0.0228 and 0.0202, respectively. Failing significance test indicates that the four factors had no significant effect on the relative deviation in the zone 8. In winter, the dominant factors of zone 4, 6, 8 and zone 1, 3, 5, 7 are aerosol and cloud coverage, respectively.
Multi-source remote sensing data fusion models can produce remotely sensed data products with both fine spatial resolution and frequent coverage from multi-source satellite data. It is able to provide strong support for the dynamic monitoring of vegetation with high precision. With the development of different fusion models, evaluating the characteristics and applicability of each model is of great importance for choosing the fittest fusion model in the practice of cropland productivity monitoring. In this paper, we chose Agriculture Comprehensive Exploitation Zone in Lingwu, Ningxia as a focal area. By using linear fitting, time series fusion, and spatial-temporal fusion model to blend the remote sensing data of the Landsat with spatial resolution of 30 m and MODIS with spatial resolution of 500 m, and the time step of 8 days, respectively. Finally, we made a comparison of ability to make a fine description of farmland productivity in spatial pattern, ability of conducting the dynamic monitoring of farmland productivity, and computing speed based on different multi-source remote sensing data fusion models. Results show that: (1) all of the three fusion models can clearly show the differences of NPP between threadiness bare objects such as roads, ridges and cropland. However, time series fusion and spatial-temporal fusion models are clearer than linear fitting model. The spatial-temporal fusion model shows more differences in evenness than time series fusion model for a relatively homogeneous cropland field. (2) Linear fitting model is suitable for estimating the annual variation of farmland productivity only, time series fusion model and spatial-temporal fusion model is suitable for the dynamic monitoring of farmland productivity. What's more, time series fusion model is suitable for monitoring farmland productivity at large or small-scale. (3) There are obvious vatiation in computing speed among the three fusion models. Computing speed of linear fitting model is the fastest, while spatial-temporal fusion model is the slowest. Among them, the computing speed of linear fitting model is 1.5 times faster than time series fusion model and 20 times faster than the spatial-temporal fusion model, respectively.
