Research On Storage And Retrieval Of Land Cover Data Based On HBase And Multilevel Grid Index

National Geographic State Monitoring is an important work in the field of geo-information for surveying and mapping.As an important object of National Geographic State Monitoring,the land cover provides a reliable data source for the analysis and evaluation model of geographical conditions.Efficient land cover data management methods is a prerequisite for mining the potential value of land cover data.Due to the large amount,frequent update and uneven distribution of the land cover data,the traditional spatial data management methods are difficult to be extended and the retrieval ability is not enough.With the rapid development of cloud computing technology,because of its good scalability,flexibility and high performance,NoSQL database has been widely used in a wide range of areas.As a popular column database of NoSQL databases,HBase can provide seamless data integration for distributed computing platforms such as Hadoop and Spark.In order to meet the demand of high-concurrent access and real-time statistical analysis of online land cover census results,this paper studies the method of HBase-based land cover data storage and retrieval,and mainly completes the following work.Firstly,a strategy of land cover data organization storage is designed.According to administrative region,national geographic state survey content and time edition,the land cover data is organized by regional,classified and temporal division in turn,which realizes the integrated management of spatial,attribute and time information.Secondly,a static multi-level grid structure is introduced,and a land cover data characteristics oriented spatial index based on HBase and multi-level grid is designed.In addition,an index parallel construction algorithm based on MapReduce is proposed to improve the efficiency of index construction.Thirdly,a land cover spatial temporal model is proposed.Rely on the multi-version data storage capability of HBase,this paper realizes the macroscopic retrieval of the historical land cover data at each time node,and also can trace the land cover object from the microscopic point of view during a continuous time period.Fourthly,in order to solve the problem of spatial query in column storage environment,the method of land cover data storage and indexing is used to design the attribute query and spatial query algorithm,which effectively improves the retrieval efficiency of land cover data.Attribute query is realized by constructing multi-threaded query sub-tasks.Spatial query can quickly locate the candidate features with rough matching between query area and land cover data by means of data partitioning performance of spatial index and HBase’s random data access ability,and then excute exact feature matching through spatial relationship operations.Finally,the feasibility and validity of this theoretical research is proved by the construction of the national geographic state publishing prototype system and the efficiency comparison experiments of the indexing and retrieval method.The land cover data storage and retrieval method has been applied in Zhejiang National Geographic State Publishing System Province.The practice shows that this method can meet the requirements of efficient management,real-time retrieval and statistical analysis of large-scale land cover data,and has a good expansion capability.It can provide a reference for other mass spatial vector data management parttens.