Research On Structured Model And Algorithms In Land Use Database Generalization

Land use database is a kind of thematic map databases for land and resources management, which is a digital production of land use status survey and reflects the position, quantity, quality, distribution, type, ownerships and land use status of terra-type parcels. Land use data consist of spatial data and attribute data. Land use spatial data are characteristic of continuously distribution in geographic region. Land use attribute data also contain rich semantic information and thematic information. Therefore, it is important for land use generalization to integrate geometry information, spatial relations, sematic information and thematic information. Current research interests are models, algorithms and knowledge. The research on models addresses to concept frameworks and supporting data models of land use generalization; the research on algorithms pays attention to its feasibility and efficiency; the research on knowledge emphasizes partially on classification and representation of map generalization knowledge. Models, algorithms and knowledge are three issues in land use database generalization. The algorithms require holistic integration of generalization model and data structure. Knowledge will be effective when it is united into generalization rules with generalization operators and algorithms. This thesis discusses aggregation model and automatic and interactive methods of three terra-type features such as sporadic parcels, line features and parcels from scale 1:5,000 to 1:25,000 in land use database generalization, which is based on China Land use Spatial Data Transfer Format (LANDUSE-DAT). Four issues and six questions (5W+1H) are explored in detail.The four issues are as follows:Firstly, aggregation model in land use generalization is presented based on general information transformation model. The principle of structured approach to automated map generalization which is firstly presented by prof. Wu Hehai is broadly committed in cartography community. But precious research work limits to some features with obvious structured information such as rivers and relief. Molenaar also presented a method called structured generalization in his generalization approaches, which is also limited to drainage. Whether structured approach is suited to land use database generalization or not and what is structured information of land use data are prior problems in land use database generalization. Many researchers introduced classification approach to land use database generalization. After analyzing the connotation of structured information, the possibility of applying spatial analysis methods to structured analysis is discussed. On the foundamation of the framework of spatial analysis, structured information of land use data issummarized, the concept of structured object (SO) is defined. Structured object is a DLM object with structured information. The concept framework of structured object aggregation model of land use database generalization is presented based on the general information transformation model.Secondly, the creation of structured object is discussed based on topological relations analysis and shape analysis. Topological relations analysis extracts mutual spatial information between DLM objects, and shape analysis extracts intrinsic spatial information of an object. Topological analysis and shape analysis are prerequisites to create structured objects. Three types of structured objects are classified to land use database generalization transformation. I -class structured object is directly created by SELECT operator; II -class structured object and Ill-class structured object are created by DIVIDE operator which depends on morphology analysis of parcel. The distinguishing of three types of structured object is useful to describe essence of mergence, aggregation and amalgamation of land use data, which is also helpful to clarify the generalization operators of area objects which are often confused in many map generalization literatures.Thirdly, based on the concept of structured object, aggregation transformation prodedures of structured objects are discussed. Land use database generalization is a kind of information transformation in a known context. However, in comparison, terrain map generalization such as rural buildings, lakes and buildings is a kind of information transformation an unknown context. Therefore, the methods in terrain map generalization aren't directly applied to land use database generalization. Base on structure object aggregation model, that is, generalization equal to aggregation, initial DLM objects in land use database are decompounded to structured objects which are direct aggregation objects. By aggregation of structured objects and keeping consistency of topological relations of objects, land use database generalization is implemented. Fourthly, On the foundation of structured generalization model and methods, several designing and implementing issues on land use database generalization software are discussed. The mode of developing land use database generalization software could be classified into three types such as independency mode, dependency mode and mixture mode. Independency mode, just as its name implies, don't depend on any GIS software platform, is originated from the operating system. Dependency mode depends on the components and development tools provided by GIS software platform. The mixture mode integrates the advantages of independency and dependency modes in the facets of difficulty and convenience. The thesis introduced independency mode on the basis of LANUSE-DAT data model, explores many issue about development strategies, functional modules, map database management, layer management, generalization mode, generalization assessment. At last, land use database generalization experiments are executed with landuse generalization software iMap and relevant test results are also listed.In general, this thesis answers six questions about land use database generalization:Firstly, Why to generalization (WHY). Land use database is not only thematic map database, but also a database used for land and resource management. Both are important to explore to the intrinsic objectives of land use database generalization. Some researchers regarded the change of the scale of attribute data as the main cause of land use generalization. In fact, considering the purpose of management, classification system of land use database is unchanged from the basic scale to less scale. Therefore, the change of attribute resolution is not major cause of land use database generalization. Land use database reflects land use status of an administrative region in a big geographic space. The essence of map generalization is transformed from a big space to a small space. However, the change of spatial resolution is one of major causes of land use generalization. In this case, the purpose of generalization is to generate less scale land use maps. On the other hand, land use database is a kind of fundamental databases. Other application database, for example, land use planning database, is deprived from land use database for land use planning, land consolidation, cultivated land preservation, land confiscation and et al.. Therefore, the change of purpose of land use database is another cause of land use database generalization. Secondly, What to generalization (WHAT). Database generalization discussed in this thesis is equal to model generalization discussed widely in the community. The objects of Model generalization are objects stored in Digital Landscape Model (DLM). These objects are referred to point objects, line objects and area objects which are not relevant to map symbolization system. The objects of land use database discussed in this thesis are point objects, line objects and area objects stored in LANDUSE-DAT files. All of DLM objects consist of geometry information and attribute information.Fourthly, What is to generalization (IS WHAT). Based on the concept framework presented in this thesis, land use database generalization is divided into two transformation phases. The first one is implemented by two generalization operators (SELECT and DIVIDE operator) based on structured analysis. The result of transformation is derived structured DLM which consists of structured objects. The second phase is implemented by three generalization operators (AGGREGATE, DELETE and COLLAPSE operator). The result of transformation is generalized DLM objects. In a word, land use database generalization is regarded as the aggregation transformation of structured objects.Fourthly, When to generalization (WHEN). Two aspects are respectively assessed according to the objectives of generalization. The change of spatial resolution will result in spatial conflicts that some objects are not expressed visually in target scale. Spatial conflicts could be calculated bymeasures, such as the least length of line feature, the least area of parcel, and the least width of line feature. The change of purposes depend on specific application requirement, for example, cultivated land distribution in a administrative region, land use status of a ownerships region, cultivated land distribution in a ownerships region and et al..Fifthly, Where to generalization (WHERE). Structured object and its adjacent, included objects are aggregation unit of land use generalization transformation.Lastly, How to generalization (HOW). Five operators are used for the implementation of land use generalization. These are SELECT operator, DIVIDE operator, AGGREGATE operator, DELETE operator, COLLAPSE operator. Four algorithms are presented to implement these operators, the algorithm of creation and maintenance of topoiogical relations between structured objects based on LANDUSE-DAT data model, the algorithm of division of parcels based on identification of the bends of parcel, the algorithm of division of parcels by calculating local skeletons of parcels, the algorithm of aggregation of three terra-type features.