Research On Spatial-temporal Data Model And Storage Scheme Based On NoSQL

with informatization, digitalization and intellectualization of Human society, spatial-temporal data plays an important role in national economy and daily life, continues to be used in various fields such as event prediction, environmental monitoring, urban evolution, traffic management, earthquake relief, global or regional climate, logistics, real-time GIS, Spatial-temporal data mining for big data. Spatial-temporal data system focus on acquisition, storage, analysis and processing and visualization of data, which not only simulated real world and analyzed, processed spatial data but also as one subject of spatial-temporal data analysis and processing of science and technology. How to store, analyze and process spatial-temporal data for big data is the industry’s research focus.Spatial-temporal database as the core of spatial-temporal information system is the basis of spatial-temporal data systems for storage and computing, the model and storage of traditional spatial-temporal data based on a relational database, it has to convert the problem domain into relational structure model, which results in high price of conversion and stores the spatial and temporal information separately, on the other hand, the relational database is difficult expansion, inflexible and involves a large number of join operations for complex queries, eventually resulting in a sharp decline in efficiency. Therefore we need to find the solution of unified model and storage for spatial-temporal data.No SQL is a no-relational, horizontally scalable, distributed database for big data that can handle semi-structured, structured data, support high concurrent read and write and is modeled in key-value, column, documents, graph, etc. No SQL dispenses with ACID compliance, Schema rigidity, strict data definitions and inefficient join operations, based on CAP theorem, BASE theorem and eventual consistency in favor of high scalability and highly flexibility. No SQL is the foundation of Big Data storage and data analysis, data processing in cloud computing, and can store various, massive, complex spatial-temporal data.First, this paper systematic summarized mainstream spatial-temporal data model and its storage, analyzed their research status. Then, elaborate on the characteristics of temporal data and representative data model, describes No SQL’s theories, data model and the advantage over relational database; The overall architecture of spatial-temporal data model and storage based on No SQL is proposed, analyze spatial-temporal data model and storage based on Mongo DB combined with Hadoop, according to the characteristic of raster and vector data and Mongo DB’s document data model, we design corresponding data transform module and raster, vector data storage, introduce how to combine Mongo DB’s ideal storage with the advantage of Hadoop for big data to process spatial-temporal data and give some use cases; Then, discuss how to use Neo4 j store and process spatial-temporal data with complex relationships, present the temporal expression and spatial expression of spatial-temporal data based on graph database, validate the model in application case, and describe the high performance of Neo4 j Spatial in spatial-temporal data processing and the architecture of Neo4 j cluster; Finally, install Mongo DB and Neo4 j based on Xdata-Hadoop, develop the prototype system of the architecture presented by the paper, and validate the model and architecture.