Archive

  • 2019 Volume 21 Issue 8
    Published: 25 August 2019
      

  • Select all
    |
  • CHENG Bo, LI Weihong, TONG Haoxin
    Download PDF ( ) HTML ( )   Knowledge map   Save

    Chinese word segmentation is a basic step in Chinese text processing and Chinese natural language processing. As a branch of Chinese word segmentation, Chinese address segmentation, which has become one of the hottest issues in Chinese word segmentation and geography research, is an important method to standardize Chinese address and conduct geocoding. Existing studies on Chinese address segmentation are mainly based on Statistical Machine Learning (SML) and Recurrent Neural Network (RNN). However, either of them cannot combine the advantages of the other. Furthermore, the current Chinese address segmentation methods lack address level segmentation and are too dependent on dictionaries and features. Therefore, this paper combined the four-word-position tagging set and Chinese hierarchical address characteristics to construct an address tagging system, and proposed a Chinese hierarchical address segmentation model (BiLSTM-CRF) which combines bidirectional long-term memory networks and conditional random fields algorithm. The proposed model utilizes the BiLSTM model to remember the characteristics of context address, while retaining the ability of the CRF algorithm to control the address tagging output by transferring probability matrix. In so doing, it has more powerful capability than the traditional statistical machine learning algorithms and RNN in the field of sequence labeling and word segmentation. To test the performance of BiLSTM-CRF on address samples marked by the address tagging system, CRF, LSTM, and BiLSTM were used to compare with BiLSTM-CRF and were respectively applied for training under the same condition as BiLSTM-CRF. We found that: (1) The segmentation effect of BiLSTM-CRF which is based on the Chinese address tagging system was better than the models for comparison, and the address tagging was more elaborate, in line with the actual address distribution. (2) The BiLSTM-CRF model had an accuracy of 93.4%, which was higher than the CRF (90.4%), LSTM (89.3%) and BiLSTM (91.2%) models. The overall address word segmentation performance and the effect of BiLSTM-CRF on each level address segmentation were more prominent than the other models. (3) The word segmentation performance of each model was correlated with the address level positively, i.e., the higher the address level, the better the word segmentation effect. The Chinese address tagging system and word segmentation model proposed in this study give a reference for the standardization of Chinese address, and provide the possibility to further improve the accuracy of geocoding technology. Future study can focus on fine-tuning the model to improve the model accuracy.

  • SU Kai, CHENG Changxiu, Nikita Murzintcev, ZHANG Ting
    Download PDF ( ) HTML ( )   Knowledge map   Save

    From 1990 to 2010, the occurrence of natural disasters was increasing in countries along the "One Belt and One Road" where most countries are developing countries with underdeveloped economy and weak disaster resistance. When disasters happen, people in those countries will tweet about the disasters in real time. The tweets contain important information for emergency rescue, disaster assessment, disaster reduction and prevention, etc. Therefore, mining and analyzing relevant tweets can provide powerful support for China's international rescue and relief work. However, twitter data is fragmented and unstructured, and the number of topics that tweets contain are huge and miscellaneous. Therefore, how to rapidly screen out relevant information from tweets becomes a research challenge. Without empirical corpus, topic model can rapidly aggregate information from a large number of disaster-related tweets, which are valuable for disaster relief and assessment. In this paper, the BTM model and LDA model, that are widely used in the study of natural language processing, were adopted to cluster Haiyan typhoon-related tweets at fine granularity topics. Then we verified and compared the accuracy of two models, and tested their ability to distinguish similar disaster topics. In addition, based on the "demand-related" tweets obtained from topic categorization, through place-name matching, we analyzed the spatial distribution of demand degree of materials and medical care in the Philippines during the occurrence of Haiyan typhoon. The result shows that: (1) In classifying Haiyan typhoon-related tweets at fine granularity topics, the overall accuracy of BTM was 0.598, while that of LDA was only 0.321, indicating that BTM can outperform LDA. (2) The F1-measure values of BTM in "disaster location-related" and "blessing-related" tweets were 0.8 and 0.78, indicating that BTM can better identify tweets of those two topics. (3) After preliminary verification, the spatial distribution of material and medical needs generated based on "demand-related" tweets was basically consistent with the actual demand. Our findings can help quickly obtain first-hand disaster information from twitter when China lacks relevant data of disasters occurring in the "One Belt and One Road" region, so to provide data support for China's international rescue work. Besides, our methodology can be used for studying domestic microblog in disasters.

  • WANG Jinxin,ZHAO Guangcheng,LU Fengnian,ZHANG Gubin,ZENG Tao,QIAO Tianrong
    Download PDF ( ) HTML ( )   Knowledge map   Save

    Three-dimensional (3D) modeling has always been the subject of research in earth information science. The 3D geological space expression can accurately reveal the spatial structure and distribution pattern of geological phenomena and processes. Under the background of spatiotemporal big data, the contemporary Digital Earth platform is facing new opportunities. The traditional 3D geological modeling methods have the following limitations: local small area, projection data, surface static modeling, difficult 3D spatial query and analysis, and unfavorable organization and management of spatiotemporal big data. The Earth Tessellation Grid provides a new solution to solving the above problems and building a new generation digital earth platform with its global omnidirectional perspective, cyclic recursive splitting mechanism, and organic flexible codec strategy. Taking the Zhengzhou Airport Economic Zone as an example, this paper established a regional true 3D geological model framework based on the Sphere Geodesic Octree Grid bricks (SGOG grids) and conducted spatial analysis using measured geological data. Firstly, the original data was preprocessed. It mainly included data reading, data encryption, and projection and coordinate transformation. Next, the geological data and the SGOG split data was matched. The true 3D geological framework was constructed by matching the brick nodes of SGOG specific splitting level with the geological envelope feature points. Then, the multi-scale and multi-story 3D models were established through vulnerability patching and shade rendering. Among them, the vulnerability filling was achieved by recombining the SGOG brick voxels by their coding logic, and the shaded rendering of the bricks was implemented by the OSG's rendering engine. Finally, based on the above, the spatial analysis of the true 3D geological model was conducted, including the true 3D profile analysis, digital drilling, and geometric feature parameter calculation. The profiles were established by judging the position of the brick relative to the section line, which included three types of warp, weft, and arbitrary lines. Digital drillings were constructed by determining the bricks where the center point of the drillings were located. By calculating the external surface area of the upper and lower bricks of the geological body model and the volume of all the bricks, the upper and lower surface area and volume were determined. The experimental results show that the modeling method proposed in this paper is not only simple in structure and easy to operate, but also suitable for complex and irregular geological bodies. It can flexibly express the precision and scale by using the multi-scale characteristics of SGOG bricks, and convenient for multi-angle true 3D spatial analysis. The earth tessellation grid is an inevitable trend in the development of the digital earth.

  • MA Xiaohui,ZHOU Jieping,GONG Jianhua,HUANG Lin,LI Wenhang,ZOU Yuling
    Download PDF ( ) HTML ( )   Knowledge map   Save

    In the context of indoor emergency evacuation, it is important to understand how people perceive emergency evacuation signs and how their perceptions affect evacuation behaviour. As a new generation of geospatial cognitive analysis tool, virtual geographic environments (VGE) provides an immersive multi-dimensional information space based on virtual reality (VR) technology. It is a three-dimensional, dynamic, and interactive space that conforms to the laws of human perception and cognition. Moreover, based on virtual geographical environments, virtual geographical perception/cognition experiment has developed rapidly, which plays an important role in many research fields including physical geography, environment and health, crowd evacuation, and human-computer interaction, providing a controlled virtual experimental environment that can be observed quantitatively. By designing a variety of experimental schemes, objective experimental data related to spatial perception and cognitive behaviour can be obtained. By combining immersive virtual environment with eye-tracking technology, a VR eye-tracking perception experiment was designed in this study. Taking the indoor corridor as an example, the data of evacuation time, eye movement fixation point, and individual movement trajectory under three types of different virtual fire escape scenarios were obtained, processed, and analyzed by quantitative observation, data statistics, and visual analysis. Finally, the layout of indoor emergency evacuation signs was evaluated and suggestions were proposed. We found that female participants were more likely to be disturbed by the starting position of escape than male participants, and had a poorer sense of virtual space. The evacuation time differed notably among the three types of scenario (no-sign, sign, and smoke). Emergency evacuation signs, smoke, and the initial position of escape had significant influence on evacuation time. The channel wall sign had the highest perception rate (0.929) and was most easily recognized by pedestrians in fire escape. The safety exit sign had the lowest perception rate (0.333), and most of the eye fixation points were on the safety gate instead of the safety exit sign. Therefore, it is necessary to enhance the visual attraction of safety exit signs to attract the attention of people in advance through appropriate size enlargement, brightness enhancement, color flashing, and other striking effects. Different indoor passage micro-environment like the length of passage has great influence on the change of sign's fixation time. And the layout and design of signs near safety exits need to be improved. Virtual eye-tracking perception experiment provides an easy solution for evaluating the design of indoor emergency evacuation signs.

  • HUANG Mengna,MA Ting
    Download PDF ( ) HTML ( )   Knowledge map   Save

    Road construction often leads to landscape fragmentation and damages ecosystem functions, such ecological impacts and consequences have been widely studied in the field of road ecology and geospatial analysis. This paper aims at exploring the influences of the road network in China. With spatial data analysis and the data set of nationwide roads in 2015, we characterized the landscape fragmentation patterns caused by paved roads, estimated the impacts on protected areas, and then explained the relationship between the degree of impacts and multiple environmental variables. The results show that: (1) The area affected by the paved roads in China have reached 10% of the terrestrial areas. The land surface have been cut into over 30 000 patches. The number of small patches is numerous, and the number of large patches is less. The extent of land surface fragmentation presents obvious east-west differentiation, and the spatial pattern of the roadless patches is similar to the population distribution and economic development level. (2) About 58% of the protected areas suffer from road influences. The degree of influence increases as the level of establishment of the protected areas decreases. The national parks have been interfered stronger than the unprotected areas. (3) The main human activity factors were positively correlated with the degree of disturbance on the protected areas, and the size of protected areas and topographic factors were negatively correlated with the degree of interference. Small areas, low level protected areas, plain areas, climate-friendly protected areas are susceptible to road disturbances and are in a state of being affected seriously by human activities. Therefore, China's road construction should achieve a balance between social development and ecological protection. Road disturbances are affected by natural and human factors, which should be considered comprehensively in the study of relevant impact mechanisms and the formulation of environmental protection policies.

  • QI Wei
    Download PDF ( ) HTML ( )   Knowledge map   Save

    Although the Tibetan Plateau has the lowest level of urbanization in China, it has experienced a significant urbanization development since the "Opening-Up and Reform" in 1978. Based on China's population census data in 1990, 2000, and 2010, this paper constructed a county-level spatial database of population urbanization for the Tibet Autonomous Region and Qinghai Province. Using the methods of urbanization stage classification, LISA types classification, and spatial regression models, this study analyzed the spatiotemporal dynamics and driving factors of the population urbanization in the Tibetan Plateau from 1990 to 2010. The main results are as follows. (1) Overall, the Tibetan Plateau had relatively low urbanization, with the level being 22.03%, 28.11%, and 37.05% in 1990, 2000, and 2010 respectively. However, some counties/cities had high urbanization level, including Xining (provincial capital of Qinghai Province), Lhasa (provincial capital of Tibet Autonomous Region), and some mining cities. The urbanization level of these cities/counties could reach even 100%. Most counties in the Tibetan Plateau remained still at the low-level urbanization stage. In addition, the spatial difference of population urbanization in the Tibetan Plateau changed gradually from 1990 to 2010. (2) The Qaidam Basin in western Qinghai Province was the main cluster of high-level urbanization areas, while the Qiangtang area in western Tibet was the largest cluster of low-level urbanization areas. The former had many mining cities while the latter was famous for inhospitable ecological environment and low population density. When compared to neighboring counties/cities, the prefecture-level capitals usually had higher urbanization, which formed a core-periphery pattern of urbanization. (3) Similar to most places in China, off-farm job opportunity, in the secondary industry and tertiary industry, was one of the key drivers of the urbanization in the Tibetan Plateau. Social public service resources also promoted urbanization development. More and more population migrated into urban areas in the Tibetan Plateau due to urban socioeconomic developments. Natural factors were not the significant factors for the spatial difference of population urbanization in the Tibetan Plateau. In some cities or towns, however, natural factors such as topography, had a constraining role for urban growth. The urbanization level in pastoral areas usually lagged behind. In the future, more attention should be paid on the sustainable development of the mining cities and towns in the Tibetan Plateau. Based on the "Belt and Road Initiative", the border cities and towns should be highlighted as not only frontiers but also trading hubs. Moreover, ecomigrants could be encouraged to move into new homes in cities and towns. This paper is hopefully beneficial for future studies of the human activities and sustainable urbanization policy-making in the Tibetan Plateau.

  • LIU Zichuan,FENG Xianfeng,WU Shuang,KONG Lingling,YAO Xuanchu
    Download PDF ( ) HTML ( )   Knowledge map   Save

    :With the development of urbanization, more and more urban and rural construction land has been converted to ecological land on Qinghai-Tibet Plateau. Land use change is a key component for global environment change, it also reflects the impacts of human activities on the environment. To some extent, urban and rural construction land can represent the intensity of human activities. Based on the land use data of 1990, 2000, 2005, 2010, and 2015 on Qinghai-Tibet Plateau, the conversation of ecological land to urban and rural construction land in the past decades were analyzed by land-use conversion matrix. Using kernel density and standard deviational ellipse, the spatial mechanism of the conversion was analyzed. The results show that: (1) In the past 25 years, lots of ecological land have been converted to urban and rural construction land. It is over 50 times the conversation of urban and rural construction land to ecological land. The conversion concentrated in two periods, 2000-2005 and 2010-2015. (2) Through experiments, 10km grid can reflect the spatial distribution regularities more obviously. The conversion between urban and rural construction land and ecological land presented a spatial reverse. The conversion of ecological land to urban and rural construction land occurred in just marginal areas. As time went by, urban and rural construction land gradually invaded the hinterland of Qinghai-Tibet Plateau. On the contrary, the conservation of urban and rural construction land to ecological land appeared in the hinterland of Qinghai-Tibet Plateau, and it gradually spread outward. (3) The towns, the hot spots of land conversion, can be divided into four 4 types. Type one is provincial capitals, located in river valleys; type two is industrial towns; type three towns are always located in natural ecological protection areas; and type four is tourism towns with good natural environment and convenient transportation. (4) Ecological lands that have higher ecological services are more likely to be converted to urban and rural construction land more easily. Nevertheless, it is very difficult for urban and rural construction land to transfer to ecological land. Reverse conversion usually transfers from urban and rural construction land to some low ecological service land. In general, the area of urban and rural construction land has increased a lot during the 25 years. Because of the large area of ecological land, ecological land use dynamic index is very low. On Qinghai-Tibet Plateau urbanization did not have a negative effect on eco-environment. So sustainable urbanization is still a significant development trend on Qinghai-Tibet Plateau in the future.

  • ZUO Qilin,ZHAO Na,DUAN Hongmei
    Download PDF ( ) HTML ( )   Knowledge map   Save

    As a key factor in the climate system, precipitation plays an important role in the survival and development of human beings. Accurate precipitation information is essential for climate and environmental research, and the spatial distribution of precipitation data is important for addressing the rational use of water resources. The distribution of national meteorological stations in most areas of China is relatively scattered. In some regions, the terrain is undulating, and the original weather station cannot accurately reflect the actual situation of local precipitation. To accurately reflect the spatial distribution of precipitation in regions of this kind, this paper used the original data of the Heihe river National Meteorological Station, and built virtual sites to establish a precipitation network across the entire basin. The information network entropy and semi-variogram theory were used to optimize the precipitation network, and some existing sites were used to interpolate precipitation across the whole basin. Information entropy can calculate the value of information contained in the precipitation of each station. The larger the amount of information, the larger is the entropy value. To establish an optimal site dataset, we combined the joint entropy and conditional entropy to select sites with a large amount of information, and then combined the nugget value and range of the semi-variogram model to take account of the spatial correlation betweenthese stations. This paper took the annual average precipitation data of 15 national meteorological stations in the Heihe River Basin from 1991 to 2003 as the raw data. According to the characteristics of elevation and vegetation growth, Heihe river is divided into three regions: upstream, midstream, and downstream. The effects of altitude, slope, and aspect on precipitation were considered for each of the three regions, and the relevant factors affecting the precipitation value were determined. We considered multiple factors to establish multiple linear regression equations to invert virtual station precipitation values. Finally, the watershed was interpolated using the drift function KED method and Co-Kriging method to compare the interpolation precision. The results show that the existence of virtual sites has effectively improved the accuracy of precipitation interpolation. Due to the appearance of the drift function, the error between the simulated precipitation value and the observed value in Dingxin, Jinta, and Ma Zongshan is within 5 mm. In this case, the interpolation result using the KED method is closest to the observation value.

  • CHENG Dongya,LI Xudong
    Download PDF ( ) HTML ( )   Knowledge map   Save

    With increasing ecological protection awareness, vegetation restoration has received growing attention. Southwest China is one of the most concentrated areas of the world's karst. In the recent years, rock desertification control and vegetation protection has made good progress. In karst areas,exploring the characteristics of vegetation change is beneficial to the scientific implementation of rocky desertification control and recoverting farmland to forest. The article selected Landsat imagery, digital elevation model data, and population raster data of the Shiqian River Watershed in Guizhou Province, spanning from 1990 to 2016. The spatiotemporal variation characteristics of the watershed vegetation coverage were calculated by the remote sensing imagery; and we expored the topographic gradient characteristics and population impact characteristics of vegetation coverage change. We found that: (1) From 1990 to 2016, the vegetation coverage of the Shiqian River Watershed increased from 62.67% to 75.21%, and increased rapidly, especially after 2002. (2) With increases in the elevation and slope, the vegetation coverage generally rised steadily. With the change of slope direction, the coverage of flat vegetation became the lowest, while the difference of other slopes was not obvious. In 1990, the vegetation coverage in the flat area was 48.48%; while in 2016, the vegetation coverage in the flat area was 53.88%, but the vegetation coverage of other slopes was significantly higher than the flat land. (3) With the increase of population density, the vegetation coverage generally declinined. yet, when the population density was less than 50 people/km 2 in 1990-2016, the vegetation coverage rised by less than 10%. When the population density reached 400 people/km 2 in 1990-2016, the vegetation coverage increased by 20%. But overall, the higher the population density, the faster the vegetation will recover in the study area. The population density is currently greater than 450 people/km 2, and the vegetation coverage is in a downward trend mainly due to urban construction. Our findings can provide guidance for the vegetation protection and reconverting farmland to forest projects in karst areas, and it is hoped that this study can provide reference for related research.

  • LIU Ling,LI Gang,YANG Lan,XUE Shuyan
    Download PDF ( ) HTML ( )   Knowledge map   Save

    With the advancement of information technology, the prevalence of e-commerce is driving the rapid development of the express delivery industry. To solve the distribution problem of the "last kilometer logistics", delivery sites appeared in most cities of China. The delivery sites, including Cainiao Station and China Post, have become the important places that residents visit frequently in everyday life and become an important research topic of urban geography and logistics geography. Based on the point of interest (POI) data of the Cainiao Station and China Post in 10 municipal districts in Shenzhen, this study used text analysis, mathematical statistics, and spatial analysis to examine the organization form, location choice, spatial distribution, agglomeration mode, and influencing factors of the delivery sites in Shenzhen. Results showed that: (1) The delivery sites depend on different types. The Cainiao station is dominated by sole business and joint venture, which relies on professional express companies and convenience stores, etc. China Post is a state-owned enterprise led by the government, generally located in branch service outlets. (2) The delivery sites have a variety of service objects, and both of Cainiao Station and China Post mainly serve communities, supplemented by companies, industrial parks, and hotels. (3) The location of the delivery sites is as close as possible to the entrances and exits of the service objects. Most of the delivery sites are distributed within 200m from the entrances and exits of facilities, and China Post sites are much closer to the service objects. (4) The spatial distribution of the delivery sites is not balanced, more in the central and western regions yet less in the east. It is distributed in the "east-west" direction, with a multi-core agglomeration mode. (5) The spatial pattern of the delivery sites is a result of the combined effects of multiple factors including the level of regional economic development, population distribution, transportation convenience, land use type, and so on. The delivery sites are distributed on the urban residential land but their number is still small in the marginal residential areas. Their spatial relation presents a geographic-adjacent-based coordinated competition development trend. Our findings help explore the comprehensive impact mechanism of the location choice and distribution pattern of the two types of delivery sites, and indicate future prospects of research, including mining their detection function of urban growth and population distribution.

  • DUAN Fei,WANG Jun,CAI Ailing,LI Guicai
    Download PDF ( ) HTML ( )   Knowledge map   Save

    The inland opening-up areas of China are strategic for leading the development of the central and western provinces. Therefore, optimizing spatial structure is significant for the spatial planning of the inland openin-up areas. This study took China's first inland opening-up area ? Liangjiang New District of Chongqing ? as a case study, to analyze changes in the regional spatial structure using both qualitative and quantitative methods and multiple dataset (e.g., population, land use, transportation networks). Moreover, we also simulated the processes of spatial structure changes using multiple spatial models. Results showed that: (1) since the setup of the new district in June, 2010, the regional urban system and spatial structure have changed significantly in terms of population, land use, and transportation networks; (2) natural and regional policies, infrastructure development, the prices of economic elements, and economic aggregation were the four main drivers for industries to move to the study area; (3) the built-up land in Liangjiang New District mainly expanded to places with low accumulative resistance values; (4) the simulation results of cellular automata model could show the spatial structure changes of Liangjiang New District over the past five years and the future spatial structure. Our results can provide a scientific basis for optimizing the development of the inland opening-up area and can also help improve the development of Liangjiang New District. This study took the rapid urbanization of a national inland opening-up area as the object, and researched the spatial structure dynamics of Liangjiang New Area in Chongqing. The explosive growth of industrial land agglomerations in a short period of time and the future construction land demand present a significant feature of large-scale land development with high intensity. The study attempted to provide new ideas and methods for solving the problem of different functional positioning of the interior spatial structure of inland opening-up areas, and to guide their development and internal space structure optimization. Future research can continue to deepen in exploring the formation of industrial functional areas, simulating microeconomic activities, etc.

  • LI Zeyu,MING Dongping,FAN Yinglin,ZHAO Linfeng,LIU Simin
    Download PDF ( ) HTML ( )   Knowledge map   Save

    The improving spatial resolution of remote sensing imagery provides more abundant information for users, but also increases the difficulty of accurate and efficient extraction of information. Image segmentation is a fundamental step in target extraction from remote sensing imagery. The quality of image segmentation directly affects the accuracy of information extraction from high spatial resolution remote sensing imagery. With various segmentation algorithms, image segmentation evaluation has become one of the research focuses in remote sensing information extraction and target recognition. Aiming at the issue of typical target recognition and from an experimental perspective, this paper compared and analyzed in detail eight representative supervised segmentation indexes: Area Fitness Index (AFI), Similarity of Size (SimSize), Relative Area in Sub-Object (RAsub), Quality Rate (QR), Euclidian Distance 1 (ED1), Euclidian Distance 2 (ED2), area discrepancy index (ADI) and Distance-Based Measure (D). Firstly, we employed a series of experiments to calculate the difference between segmentation image and reference image by using different segmentation methods, then discussed the calculation results and evaluated the advantages and disadvantages of the different supervised segmentation evaluation indexes. The comparison results show that the AFI, ED1, ED2 and D could representatively and synthetically assess the segmentation quality. Further, based on the indexes analysis result, this paper proposed a comprehensive evaluation scheme for remote sensing imagery supervised segmentation evaluation by using weighted calculation of the four representative indexes. Through the experiment of comprehensive evaluation, we conclude that the effect of simple shape objects (such as the baseball field and oil tank) by using various segmentation methods is generally ideal. When the shape of objects is complex and the contour is blurred, the accuracy of image segmentation will be sensitive to the segmentation result to some extent. Meanwhile, the effect of segmentation methods (e.g., Otsu-2D, Regional Growth, and Mean Shift) are in general better than the other segmentation methods (e.g., split and merge, maximal entropy, and fuzzy threshold). In addition, the experiments also suggest that the comprehensive method is helpful for the scientific selection of segmentation methods and it can improve the efficiency of information extraction from high spatial resolution imagery. Finally, this paper systematically analyzed the experimental results from the two aspects of evaluation index and segmentation method, and pointed out the existing problems and development trends of image supervised segmentation evaluation.

  • WANG Yating,KONG Jinling,YANG Liangyan,LI Jianfeng,ZHANG Wenbo
    Download PDF ( ) HTML ( )   Knowledge map   Save

    In the arid and semi-arid regions of northwest China, the precipitation is scarce and evapotranspiration is intense. Soil moisture, as an important ecological factor, affects the energy balance of soil-atmosphere interface. In recent years, Support Vector Regression(SVR) model has been applied in soil moisture inversion for its merits including, high estimation accuracy, good ability to deal with non-linear processing and strong generalization ability. However, existing models rarely consider the influence of surface roughness, which limits of inversion accuracy. Taking the Uxin Banner of Ordos city of Inner Mongolia as a study area, this study aims to construct a suitable soil moisture inversion model through combining Radarsat-2 synthetic aperture radar(SAR) data and GF-1 data. To extract backscattering coefficient of bare soil( σ soil 0 ) from the full polarization Radarsat-2 SAR data, we used Water-Cloud model(WCM) to remove the influence of vegetation-sparse layer on the radar backscattering coefficient. Meanwhile, we constructed backscattering coefficient database of bare soil by using Advanced Integrated Equation Model(AIEM) and Oh Model, and used Look Up Table (LUT) method to simulate effective surface roughness parameters such as root mean square height(S) and correlation length(L). Finally, the soil moisture model was built based on support vector regression, and the soil moisture inversion results of different data sources under the backscattering coefficients of different polarization modes were systematically compared and analyzed. The results showed that the inversion accuracy of the co-polarization data (VV polarization or HH polarization) was higher than that of the cross-polarization data(VH polarization or HV polarization) when the single data source without considering the roughness parameter was used as the model parameter. When the model parameter was the multi-source data with considering the roughness parameter, the inversion accuracy of different polarization data was improved. When the data source was σ vv 0 and roughness parameter, the inversion model had a higher precision, the correlation coefficient between inversion value and measured value was 0.917. The mean absolute error(MAE) and root mean square error(RMSE) were 3.980% and 5.187%, respectively. Our findings can serve as the technical support for remote sensing of surface soil moisture in vegetation-sparse arid areas.

  • ZHANG Junyao,YAO Yonghui,SUONAN Dongzhu,GAO Lijing,WANG Jing,ZHANG Xinghang
    Download PDF ( ) HTML ( )   Knowledge map   Save

    The structural function and ecological characteristics of mountain vegetation can reflect comprehe-nsively the basic characteristics and functional properties of the eco-environment. Mapping of different vegetation types is the basis for the study of vegetation cover dynamics. Therefore, studying vegetation types and their distribution patterns in montane areas is important for understanding the eco-environment and climatic spatial changes. With the development and application of satellite remote sensing technology, remote sensing data have been widely used in the investigation and research of mountain vegetation information extraction. As altitude increases, vegetation presents the characteristic of regular zonal arrangement and combination, which is called the altitudinal belts law of mountain vegetation. Through the altitudinal vegetation belts information, the altitude range of different vegetation groups and the adjacent relationship between the upper and lower layers can be determined. To achieve high-precision extraction of mountain vegetation types, in this paper, we took Taibai Mountain (the main peak of Qinling Mountains) as the experimental area, combined the obvious vertical zonal distribution law of mountain vegetation, and used the data of altitudinal belts of Taibai Mountain vegetation, high-resolution remote sensing imagery (GF1/GF2/ZY3), and 1:10 000 digital surface model (DSM). Followingly, we selected the optimal segmentation scale of different levels by calculating the mean variance, and conducted multi-level and multi-scale image segmentation. Then, we built terrain constraint factors with mountain altitudinal belts information and selected samples. After overlaying terrain constraint factors with altitudinal belts information of vegetation on the high-resolution images, the rough distribution ranges of each vegetation types were clear at a glance, which can make sample selection more efficient and accurate. Lastly, the images were used to extract vegetation information through the object-oriented classification method. The classification result had a total accuracy of 92.9% and a kappa coefficient of 0.9160. To prove the role of terrain constraint factors, some regions in the western part of the north slope were selected for comparing whether terrain constraint factors affected the classification. Compared with the classification without altitudinal vegetation belts information, this method improved the overall accuracy by 10%. The results show that adding terrain constraint factors in the classification process can significantly improve the efficiency and accuracy of sample selection, and provide a guarantee for the accuracy of subsequent vegetation classification. By man-machine interaction, this study applies the knowledge of mountain altitudinal belts to classification, and effectively improves the accuracy of mountain vegetation classification.

  • GENG Renfang,FU Bolin,CAI Jiangtao,CHEN Xiaoyu,LAN Feiwu,YU Hangming,LI Qingxun
    Download PDF ( ) HTML ( )   Knowledge map   Save

    Wetlands are among the most important ecosystems on Earth. They play a key role in alleviating floods and filtering polluted water, and also provide habitats for many plants and animals. A unique wetland type, karst wetland, is widely distributed in southwest China, as influenced by the special soil and water structure of karst landforms. Currently, domestic and foreign scholars pay much less attention to karst wetland than other wetland types, and lack targeted research on high-precision vegetation identification of karst wetland using remote sensing technology. However, like other wetland types, karst wetland has seriously degraded, and many problems need to be solved urgently. Huixian National Wetland Park, located in Guilin, Guangxi province, is a typical karst wetland. In this paper, part of the core area of the Huixian National Wetland Park was selected as the study area, which is greatly affected by human activities and severely degraded. The aerial photography images from an unmanned aerial vehicle (UAV) were used as the data source, and the object-based random forest algorithm was used to realize the high-precision classification of karst wetland vegetation. In so doing, we explored the applicability of UAV RGB remote sensing image and object-based random forest algorithm in karst wetland vegetation recognition, and provided a technical reference for the research and protection of karst wetland by using UAV remote sensing technology. First, the multiscale iterative segmentation algorithm was used to segment the image layers in eCognition Developer 9.0. Then, the texture features calculated based on the grey level co-occurrence matrix (GLCM) and spectral features of the images, the vegetation indexes, geometric features, and the elevation information (DSM) derived from the UAV remote sensing data were fully considered in the feature selection. Finally, the tuning of random forest algorithm parameters, model construction, and classification were implemented in RStudio. Results showed that the object-based random forest algorithm had a high recognition ability for the Huixian wetland vegetation. The overall accuracy was 86.75% and the Kappa coefficient was 0.83 in the 95% confidence interval. In the identification accuracy of vegetation in a single typical karst wetland, the user accuracy of the vegetation cluster of Bermudagrass-Cogongrass-Ludwigia was above 90%, the producer accuracy was over 80%. And the producer accuracy of the Bamboo-Thorny Wingnic-Sweet Olive was higher than 80%, but its user accuracy was only 70.59%.