Hadoop-based Multi-version Raster Map Storage And Management System

With the intensive study of geographic information technology and cloud computing technology, Geographic Information System(GIS) develops rapidly. Storage and management of raster map data has become one of the most important researching directions in the field of GIS. Raster map data has the following characteristics, such as simple structure, small size, multi-version, large total volume and spatial feature. According to these characteristics, this paper deigns and implements a storage and management system of multi-version raster map data which is based on Hadoop. The main contents are as follows:1. Analyze and summary the current development situation and existing problems of GIS at home and abroad, and then propose a scheme of storage and management system of multi-version raster map data which is based on Hadoop.2. Analyze the existing raster map data storage technique, and then propose the tile pyramid model of Mercator projection and the tile pyramid model of geography latitude and longitude coordinates projection. Apply the quadtree coding into tile code after analyzing it. To solve the multi-version problem of raster map data, propose the S model and the T model respectively, and finally choose to use the T model.3. Detailedly study and analyze Hadoop. Research the running mechanism and system architecture of HDFS, HBase and Map Reduce, and make a brief analysis on Zoo Keeper and Ambari.4. Propose the storage model of multi-verion tile pyramid. Design a three-layer architecture of storage and management system of multi-verion tile pyramid: the distributed storage layer of multi-version tile pyramid, the management layer of tile pyramid and the external interface layer of tile pyramid.5. Implement the distributed storage layer of the multi-version tile pyramid, including the storage of metadata of tile pyramid, the storage of package file which is formed by picking tile in HDFS and the storage of tile index in HBase after using quadtree coding. To improve the read performance of the distributed storage layer, design and implement the adjusting service of tile storage location proximity based on Map Reduce. Implement the tile re-index service based on Map Reduce to meet the needs of replacing the tile index system and using more efficient tile index coding.6. Implement the management layer of tile pyramid to provide access service for the external interface layer. The external interface layer implements the write interface of tile pyramid, the access interface of tile and the access interface of metadata.The storage and management system of multi-version raster map data based on hadoop has been used in scientific research project. It can provide write interface of tile pyramid for the map tile creation program. Through the adjusting service of tile storage location proximity it can provide slow delay and stable online service of tile and metadata for the map display clients such as OSGEarth, Open Layers and Cesium. It uses the hierarchical design and the modular design which makes it have better expansibility.