The Key Techniques Of Cloud GIS Based On Hadoop

Cloud computing is derived from multiple new computing methods such as parallel computing, grid computing and so on. It provides a good opportunity for the development of spatial information services. Its storage technique which can be infinitely extended can meet the storage space requirment resulted from the rapid growth of spatial data, and its mighty computing capability can provide high-speed services for the spatial information retrieval, processing and analysis etc.Faced with the problems in massive spatial data storage, processing and continuous service, this paper combines the characteristics of GIS and cloud computing to applys the open Hadoop cloud computing platform to the field of spatial information service, which makes use of the distributed storage capability and mighty computing capability provided by cloud computing to build cloud-based GIS applications and research on several key technologies. The main content is as follows:(1) This paper designs the cloud GIS architecture which is based on the Hadoop platform. It includes four layers of physical device layer, platform layer, software layer and application layer. And also includes layers-crossed services of user management, service management, resources management, monitoring system, disaster recovery, and operations management. This architecture establishes the foundation for the subsequent research.(2) This paper lays great emphasis on the research of the vector spatial data modeling method in cloud computing platform and achieves efficient storage, query and analysis. In addition, this paper studies the vector data segmentation and reassmble algorithm based on monotone chain and designs the vector data storage format “GWKT(Grid WKT, GWKT)”; designed the globally unique coding of vector features based on spatial information grid and space-filling curve, and achieved the code generation algorithm; based on HBase, designed the spatial information multi-level grid vector data storage model and query algorithm, extended data types and their filter.(3) The parallel computing model is studied as to the deficiency of massive spatial data processing capacity. The HDFS-based vector data storage format is designed, which achieves the MapReduce-based vector data division warehousing and parallel processing model; the vector data parallel computing model based on MapReduce data filtering is constructed, which is fit for the GWKT format and is verified.(4) For spatial information share and interoperability issues, the combination of cloud computing service and OGC Web Services technology is studied. Based on the OGC standard service, the cloud GIS spatial information service layered architecture is designed and services such as WMS, WMTS, WFS and WPS is achieved based on multi-level spatial information grid. In addition, client-based service interface is designed in order that client and server can be completely decoupled.(5) Hadoop-based cloud GIS prototype system is designed and achieved, the key modules such as massive raster and vector data’s efficient storage, management and retrieval, spatial information parallel computing and OGC-based spatial information share are implemented. Simultanously, the performance of several relevant modules is tested, which validates the feasibility, effectiveness and efficiency of the proposed storage model and computing model.