| Land use data, which generated by the national land survey, is the foundation of land use analysis and evaluation. And it is important data basis for the government to understand the annual land use change. The land use data is generally stored in relational database and use ArcSDE to cope with spatial data. Management system based on such storage approach often works on a single server, resulting in limited storage space. With the deepening of the national land survey, more and more land use data is produced. The traditional data management can’t meet the need of high throughput and high efficiency of land use data organization, storage and process.In order to improve the problem, an in-depth research of cloud computing technology and land use vector data character is conducted and a cloud-storage solution is proposed. It encompasses the following aspects:1.Design the data organization under the cloud environment. The solution stores land use data at county level administrative division, and stores county administrative region boundaries into a file, which is beneficial for the cross-regional operations. Besides, the solution uses the TSV format to store both attribute data and spatial data into a single file, which follows the WKT standard.2. Design the parallel data import mechanism. Using Sqoop and the developed GDAL-based data import tools, the solution support to import land use data from the traditional relational database or data files (shapefile, MDB, etc.) in parallel.3. Design a spatial index creation method. Based on the quad-tree-index theory, the solution can create spatial index in parallel.4. Design an advanced solution to carry out statistics, spatial query and spatial overlay analysis.In the last part, we developed a prototype system based on the solution and gave a performance comparison to the original land use data management system based on Oracle and ArcSDE. Results show that the prototype system is more suitable for the management of massive land use data. |
PDF Full Text Request