Research On Key Technologies In Geographic Information Management Based On Hadoop

In recent years, with the development of Computer Version, Visual Reality and 3D Visualization, geoscience visualization has gradually vitality. Meanwhile, the network technology and cloud computing is booming, which promote the development of distributed geographic, aiming at the entire processing tasks assigned more optimal in the computer cluster. Traditional desktop GIS using centralized storage model, which easily leads to resource bottleneck and can’t be shared, hinders the further development and application of GIS. Spatial data of three-dimensional GIS has the characteristic of massive, the amount of data is generally arriving PB or more. In the network environment, the existing solutions can’t satisfy the requirement of these spatial data, such as storage capacity, management efficiency, tasking scheduling. In this context, the efficient transmission of data is facing serious challenges.Cloud computing broad our scope of mind, we can use distributed storage and parallel computing to solve those problem. On the basis of the comparative analysis of the distributed storage system at home and abroad, the paper chose HDFS as the framework of storage system. The core of the design framework Hadoop is HDFS and MapReduce. HDFS provides storage for the vast amounts of data, MapReduce provides the parallel programming model of the mass data, simplifying the development process. The hardware constructs on the ordinary machine cluster of Linux. Through internal monitor to achieve high fault tolerance, high response, load balancing, it provides external services to meet the high concurrent and high reliability. Therefore, the paper using Hadoop as the framework of space data sharing of Sea-Land. The paper’s main research contents are as follows:First, the paper analyzed the traditional geography concept derived virtual geographic environment under the background of the rapid developing internet. Besides, the means of achieving geographic data are getting increasing rich, data collection gradually increased. In this case, the sharing of traditional data is restricted, then, home and abroad began to develop a virtual information system based on massive data. The paper presented massive spatial data storage solutions based on Hadoop, as the technical support of space sharing of sea-land.Second, this paper analyzes the access requirement of the image data, which is large volume of spatial data, proposes to build the image pyramid to provide users with smooth and efficient data retrieval service. The paper made some researches on the image pyramid building based on the MapReduce, a parallel programming model belongs to Hadoop. In order to manage data efficiently, the paper organized data by using self-defined format. Considering the high concurrency features of Internet, a load balancing method for data access based on Nginx is proposed.Finally, the paper use transit data as an example to make some researches on data mining. Based on HBase, a suitable data model and a query algorithm are brought forward to save thematic data, extract the information of the thematic data, and then make analysis.